wait for successful connection before setting up db

* Specify macro_namespace of global_project dispatch macros

* Dispatch get_custom_alias, too

* Add integration tests

* Add changelog entry

.bumpversion.cfg

@@ -1,23 +1,27 @@
 [bumpversion]
-current_version = 0.16.0b3
+current_version = 0.21.0b1
 parse = (?P<major>\d+)
 	\.(?P<minor>\d+)
 	\.(?P<patch>\d+)
-	((?P<prerelease>[a-z]+)(?P<num>\d+))?
+	((?P<prekind>a|b|rc)
+	(?P<pre>\d+)  # pre-release version num
+	)?
 serialize = 
-	{major}.{minor}.{patch}{prerelease}{num}
+	{major}.{minor}.{patch}{prekind}{pre}
 	{major}.{minor}.{patch}
 commit = False
 tag = False
 
-[bumpversion:part:prerelease]
+[bumpversion:part:prekind]
 first_value = a
+optional_value = final
 values = 
 	a
 	b
 	rc
+	final
 
-[bumpversion:part:num]
+[bumpversion:part:pre]
 first_value = 1
 
 [bumpversion:file:setup.py]
@@ -26,6 +30,8 @@ first_value = 1
 
 [bumpversion:file:core/dbt/version.py]
 
+[bumpversion:file:core/scripts/create_adapter_plugins.py]
+
 [bumpversion:file:plugins/postgres/setup.py]
 
 [bumpversion:file:plugins/redshift/setup.py]
@@ -34,3 +40,11 @@ first_value = 1
 
 [bumpversion:file:plugins/bigquery/setup.py]
 
+[bumpversion:file:plugins/postgres/dbt/adapters/postgres/__version__.py]
+
+[bumpversion:file:plugins/redshift/dbt/adapters/redshift/__version__.py]
+
+[bumpversion:file:plugins/snowflake/dbt/adapters/snowflake/__version__.py]
+
+[bumpversion:file:plugins/bigquery/dbt/adapters/bigquery/__version__.py]
+

.circleci/config.yml

@@ -1,171 +0,0 @@
-version: 2
-jobs:
-  unit:
-    docker: &test_only
-      - image: fishtownjacob/test-container:4
-        environment:
-          DBT_INVOCATION_ENV: circle
-    steps:
-      - checkout
-      - run: tox -e flake8,mypy,unit-py36,unit-py38
-  build-wheels:
-    docker: *test_only
-    steps:
-      - checkout
-      - run:
-          name: Build wheels
-          command: |
-            python3.8 -m venv "${PYTHON_ENV}"
-            export PYTHON_BIN="${PYTHON_ENV}/bin/python"
-            $PYTHON_BIN -m pip install -U pip setuptools
-            $PYTHON_BIN -m pip install -r requirements.txt
-            $PYTHON_BIN -m pip install -r dev_requirements.txt
-            /bin/bash ./scripts/build-wheels.sh
-          environment:
-            PYTHON_ENV: /home/tox/build_venv/
-      - store_artifacts:
-          path: ./dist
-          destination: dist
-  integration-postgres-py36:
-    docker: &test_and_postgres
-      - image: fishtownjacob/test-container:4
-        environment:
-          DBT_INVOCATION_ENV: circle
-      - image: postgres
-        name: database
-        environment: &pgenv
-          POSTGRES_USER: "root"
-          POSTGRES_PASSWORD: "password"
-          POSTGRES_DB: "dbt"
-    steps:
-      - checkout
-      - run: &setupdb
-          name: Setup postgres
-          command: bash test/setup_db.sh
-          environment:
-            PGHOST: database
-            PGUSER: root
-            PGPASSWORD: password
-            PGDATABASE: postgres
-      - run:
-          name: Run tests
-          command: tox -e integration-postgres-py36
-      - store_artifacts:
-          path: ./logs
-  integration-snowflake-py36:
-    docker: *test_only
-    steps:
-      - checkout
-      - run:
-          name: Run tests
-          command: tox -e integration-snowflake-py36
-          no_output_timeout: 1h
-      - store_artifacts:
-          path: ./logs
-  integration-redshift-py36:
-    docker: *test_only
-    steps:
-      - checkout
-      - run:
-          name: Run tests
-          command: tox -e integration-redshift-py36
-      - store_artifacts:
-          path: ./logs
-  integration-bigquery-py36:
-    docker: *test_only
-    steps:
-      - checkout
-      - run:
-          name: Run tests
-          command: tox -e integration-bigquery-py36
-      - store_artifacts:
-          path: ./logs
-  integration-postgres-py38:
-    docker: *test_and_postgres
-    steps:
-      - checkout
-      - run: *setupdb
-      - run:
-          name: Run tests
-          command: tox -e integration-postgres-py38
-      - store_artifacts:
-          path: ./logs
-  integration-snowflake-py38:
-    docker: *test_only
-    steps:
-      - checkout
-      - run:
-          name: Run tests
-          command: tox -e integration-snowflake-py38
-          no_output_timeout: 1h
-      - store_artifacts:
-          path: ./logs
-  integration-redshift-py38:
-    docker: *test_only
-    steps:
-      - checkout
-      - run:
-          name: Run tests
-          command: tox -e integration-redshift-py38
-      - store_artifacts:
-          path: ./logs
-  integration-bigquery-py38:
-    docker: *test_only
-    steps:
-      - checkout
-      - run:
-          name: Run tests
-          command: tox -e integration-bigquery-py38
-      - store_artifacts:
-          path: ./logs
-
-workflows:
-  version: 2
-  test-everything:
-    jobs:
-      - unit
-      - integration-postgres-py36:
-          requires:
-            - unit
-      - integration-redshift-py36:
-          requires:
-            - integration-postgres-py36
-          filters: &no_forks
-            branches:
-              # Forked pull requests have CIRCLE_BRANCH set to pull/XXX
-              ignore: /pull\/[0-9]+/
-      - integration-bigquery-py36:
-          requires:
-            - integration-postgres-py36
-          filters: *no_forks
-      - integration-snowflake-py36:
-          requires:
-            - integration-postgres-py36
-          filters: *no_forks
-      - integration-postgres-py38:
-          requires:
-            - unit
-      - integration-redshift-py38:
-          requires:
-            - integration-postgres-py38
-          filters: *no_forks
-      - integration-bigquery-py38:
-          requires:
-            - integration-postgres-py38
-          filters: *no_forks
-      - integration-snowflake-py38:
-          requires:
-            - integration-postgres-py38
-          filters: *no_forks
-      - build-wheels:
-          requires:
-            - unit
-            - integration-postgres-py36
-            - integration-redshift-py36
-            - integration-bigquery-py36
-            - integration-snowflake-py36
-            - integration-postgres-py38
-            - integration-redshift-py38
-            - integration-bigquery-py38
-            - integration-snowflake-py38
-          filters: *no_forks

.github/ISSUE_TEMPLATE/beta-minor-version-release.md

@@ -0,0 +1,27 @@
+---
+name: Beta minor version release
+about: Creates a tracking checklist of items for a Beta minor version release
+title: "[Tracking] v#.##.#B# release "
+labels: 'release'
+assignees: ''
+
+---
+
+### Release Core
+- [ ] [Engineering] Follow [dbt-release workflow](https://www.notion.so/dbtlabs/Releasing-b97c5ea9a02949e79e81db3566bbc8ef#03ff37da697d4d8ba63d24fae1bfa817) 
+- [ ] [Engineering] Verify new release branch is created in the repo
+- [ ] [Product] Finalize migration guide (next.docs.getdbt.com)
+
+### Release Cloud
+- [ ] [Engineering] Create a platform issue to update dbt Cloud and verify it is completed. [Example issue](https://github.com/dbt-labs/dbt-cloud/issues/3481)
+- [ ] [Engineering] Determine if schemas have changed. If so, generate new schemas and push to schemas.getdbt.com
+
+### Announce
+- [ ] [Product] Announce in dbt Slack
+
+### Post-release
+- [ ] [Engineering] [Bump plugin versions](https://www.notion.so/dbtlabs/Releasing-b97c5ea9a02949e79e81db3566bbc8ef#f01854e8da3641179fbcbe505bdf515c) (dbt-spark + dbt-presto), add compatibility as needed
+   - [ ]  [Spark](https://github.com/dbt-labs/dbt-spark) 
+   - [ ]  [Presto](https://github.com/dbt-labs/dbt-presto)
+- [ ] [Engineering] Create a platform issue to update dbt-spark versions to dbt Cloud. [Example issue](https://github.com/dbt-labs/dbt-cloud/issues/3481)
+- [ ] [Engineering] Create an epic for the RC release

.github/ISSUE_TEMPLATE/feature_request.md

@@ -18,3 +18,6 @@ Is this feature database-specific? Which database(s) is/are relevant? Please inc
 
 ### Who will this benefit?
 What kind of use case will this feature be useful for? Please be specific and provide examples, this will help us prioritize properly.
+
+### Are you interested in contributing this feature?
+Let us know if you want to write some code, and how we can help.

.github/ISSUE_TEMPLATE/final-minor-version-release.md

@@ -0,0 +1,28 @@
+---
+name: Final minor version release
+about: Creates a tracking checklist of items for a final minor version release
+title: "[Tracking] v#.##.# final release "
+labels: 'release'
+assignees: ''
+
+---
+
+### Release Core
+- [ ] [Engineering] Verify all necessary changes exist on the release branch
+- [ ] [Engineering] Follow [dbt-release workflow](https://www.notion.so/dbtlabs/Releasing-b97c5ea9a02949e79e81db3566bbc8ef#03ff37da697d4d8ba63d24fae1bfa817) 
+- [ ] [Product] Merge `next` into `current` for docs.getdbt.com
+
+### Release Cloud
+- [ ] [Engineering] Create a platform issue to update dbt Cloud and verify it is completed. [Example issue](https://github.com/dbt-labs/dbt-cloud/issues/3481)
+- [ ] [Engineering] Determine if schemas have changed. If so, generate new schemas and push to schemas.getdbt.com
+
+### Announce
+- [ ] [Product] Update discourse
+- [ ] [Product] Announce in dbt Slack
+
+### Post-release
+- [ ] [Engineering] [Bump plugin versions](https://www.notion.so/dbtlabs/Releasing-b97c5ea9a02949e79e81db3566bbc8ef#f01854e8da3641179fbcbe505bdf515c) (dbt-spark + dbt-presto), add compatibility as needed
+   - [ ]  [Spark](https://github.com/dbt-labs/dbt-spark) 
+   - [ ]  [Presto](https://github.com/dbt-labs/dbt-presto)
+- [ ] [Engineering] Create a platform issue to update dbt-spark versions to dbt Cloud. [Example issue](https://github.com/dbt-labs/dbt-cloud/issues/3481)
+- [ ] [Product] Release new version of dbt-utils with new dbt version compatibility. If there are breaking changes requiring a minor version, plan upgrades of other packages that depend on dbt-utils.

.github/ISSUE_TEMPLATE/rc-minor-version-release copy.md

@@ -0,0 +1,29 @@
+---
+name: RC minor version release
+about: Creates a tracking checklist of items for a RC minor version release
+title: "[Tracking] v#.##.#RC# release "
+labels: 'release'
+assignees: ''
+
+---
+
+### Release Core
+- [ ] [Engineering] Verify all necessary changes exist on the release branch
+- [ ] [Engineering] Follow [dbt-release workflow](https://www.notion.so/dbtlabs/Releasing-b97c5ea9a02949e79e81db3566bbc8ef#03ff37da697d4d8ba63d24fae1bfa817) 
+- [ ] [Product] Update migration guide (next.docs.getdbt.com)
+
+### Release Cloud
+- [ ] [Engineering] Create a platform issue to update dbt Cloud and verify it is completed. [Example issue](https://github.com/dbt-labs/dbt-cloud/issues/3481)
+- [ ] [Engineering] Determine if schemas have changed. If so, generate new schemas and push to schemas.getdbt.com
+
+### Announce
+- [ ] [Product] Publish discourse
+- [ ] [Product] Announce in dbt Slack
+
+### Post-release
+- [ ] [Engineering] [Bump plugin versions](https://www.notion.so/dbtlabs/Releasing-b97c5ea9a02949e79e81db3566bbc8ef#f01854e8da3641179fbcbe505bdf515c) (dbt-spark + dbt-presto), add compatibility as needed
+   - [ ]  [Spark](https://github.com/dbt-labs/dbt-spark) 
+   - [ ]  [Presto](https://github.com/dbt-labs/dbt-presto)
+- [ ] [Engineering] Create a platform issue to update dbt-spark versions to dbt Cloud. [Example issue](https://github.com/dbt-labs/dbt-cloud/issues/3481)
+- [ ] [Product] Release new version of dbt-utils with new dbt version compatibility. If there are breaking changes requiring a minor version, plan upgrades of other packages that depend on dbt-utils.
+- [ ] [Engineering] Create an epic for the final release

.github/actions/setup-postgres-linux/action.yml

@@ -0,0 +1,10 @@
+name: "Set up postgres (linux)"
+description: "Set up postgres service on linux vm for dbt integration tests"
+runs:
+  using: "composite"
+  steps:
+    - shell: bash
+      run: |
+        sudo systemctl start postgresql.service
+        pg_isready
+        sudo -u postgres bash ${{ github.action_path }}/setup_db.sh

.github/actions/setup-postgres-linux/setup_db.sh

@@ -0,0 +1 @@
+../../../test/setup_db.sh

.github/actions/setup-postgres-macos/action.yml

@@ -0,0 +1,24 @@
+name: "Set up postgres (macos)"
+description: "Set up postgres service on macos vm for dbt integration tests"
+runs:
+  using: "composite"
+  steps:
+    - shell: bash
+      run: |
+        brew services start postgresql
+        echo "Check PostgreSQL service is running"
+        i=10
+        COMMAND='pg_isready'
+        while [ $i -gt -1 ]; do
+            if [ $i == 0 ]; then
+                echo "PostgreSQL service not ready, all attempts exhausted"
+                exit 1
+            fi
+            echo "Check PostgreSQL service status"
+            eval $COMMAND && break
+            echo "PostgreSQL service not ready, wait 10 more sec, attempts left: $i"
+            sleep 10
+            ((i--))
+        done
+        createuser -s postgres
+        bash ${{ github.action_path }}/setup_db.sh

.github/actions/setup-postgres-macos/setup_db.sh

@@ -0,0 +1 @@
+../../../test/setup_db.sh

.github/actions/setup-postgres-windows/action.yml

@@ -0,0 +1,12 @@
+name: "Set up postgres (windows)"
+description: "Set up postgres service on windows vm for dbt integration tests"
+runs:
+  using: "composite"
+  steps:
+    - shell: pwsh
+      run: |
+        $pgService = Get-Service -Name postgresql*
+        Set-Service -InputObject $pgService -Status running -StartupType automatic
+        Start-Process -FilePath "$env:PGBIN\pg_isready" -Wait -PassThru
+        $env:Path += ";$env:PGBIN"
+        bash ${{ github.action_path }}/setup_db.sh

.github/actions/setup-postgres-windows/setup_db.sh

@@ -0,0 +1 @@
+../../../test/setup_db.sh

.github/dependabot.yml

@@ -0,0 +1,45 @@
+version: 2
+updates:
+  # python dependencies
+  - package-ecosystem: "pip"
+    directory: "/"
+    schedule:
+      interval: "daily"
+    rebase-strategy: "disabled"
+  - package-ecosystem: "pip"
+    directory: "/core"
+    schedule:
+      interval: "daily"
+    rebase-strategy: "disabled"
+  - package-ecosystem: "pip"
+    directory: "/plugins/bigquery"
+    schedule:
+      interval: "daily"
+    rebase-strategy: "disabled"
+  - package-ecosystem: "pip"
+    directory: "/plugins/postgres"
+    schedule:
+      interval: "daily"
+    rebase-strategy: "disabled"
+  - package-ecosystem: "pip"
+    directory: "/plugins/redshift"
+    schedule:
+      interval: "daily"
+    rebase-strategy: "disabled"
+  - package-ecosystem: "pip"
+    directory: "/plugins/snowflake"
+    schedule:
+      interval: "daily"
+    rebase-strategy: "disabled"
+
+  # docker dependencies
+  - package-ecosystem: "docker"
+    directory: "/"
+    schedule:
+      interval: "weekly"
+    rebase-strategy: "disabled"
+  - package-ecosystem: "docker"
+    directory: "/docker"
+    schedule:
+      interval: "weekly"
+    rebase-strategy: "disabled"

.github/pull_request_template.md

@@ -9,14 +9,13 @@ resolves #
     resolves #1234
 -->
 
-
 ### Description
 
 <!--- Describe the Pull Request here -->
 
-
 ### Checklist
- - [ ] I have signed the [CLA](https://docs.getdbt.com/docs/contributor-license-agreements)
- - [ ] I have run this code in development and it appears to resolve the stated issue
- - [ ] This PR includes tests, or tests are not required/relevant for this PR
- - [ ] I have updated the `CHANGELOG.md` and added information about my change to the "dbt next" section.
+
+- [ ] I have signed the [CLA](https://docs.getdbt.com/docs/contributor-license-agreements)
+- [ ] I have run this code in development and it appears to resolve the stated issue
+- [ ] This PR includes tests, or tests are not required/relevant for this PR
+- [ ] I have updated the `CHANGELOG.md` and added information about my change to the "dbt next" section.

.github/scripts/integration-test-matrix.js

@@ -0,0 +1,95 @@
+module.exports = ({ context }) => {
+  const defaultPythonVersion = "3.8";
+  const supportedPythonVersions = ["3.6", "3.7", "3.8", "3.9"];
+  const supportedAdapters = ["snowflake", "postgres", "bigquery", "redshift"];
+
+  // if PR, generate matrix based on files changed and PR labels
+  if (context.eventName.includes("pull_request")) {
+    // `changes` is a list of adapter names that have related
+    // file changes in the PR
+    // ex: ['postgres', 'snowflake']
+    const changes = JSON.parse(process.env.CHANGES);
+    const labels = context.payload.pull_request.labels.map(({ name }) => name);
+    console.log("labels", labels);
+    console.log("changes", changes);
+    const testAllLabel = labels.includes("test all");
+    const include = [];
+
+    for (const adapter of supportedAdapters) {
+      if (
+        changes.includes(adapter) ||
+        testAllLabel ||
+        labels.includes(`test ${adapter}`)
+      ) {
+        for (const pythonVersion of supportedPythonVersions) {
+          if (
+            pythonVersion === defaultPythonVersion ||
+            labels.includes(`test python${pythonVersion}`) ||
+            testAllLabel
+          ) {
+            // always run tests on ubuntu by default
+            include.push({
+              os: "ubuntu-latest",
+              adapter,
+              "python-version": pythonVersion,
+            });
+
+            if (labels.includes("test windows") || testAllLabel) {
+              include.push({
+                os: "windows-latest",
+                adapter,
+                "python-version": pythonVersion,
+              });
+            }
+
+            if (labels.includes("test macos") || testAllLabel) {
+              include.push({
+                os: "macos-latest",
+                adapter,
+                "python-version": pythonVersion,
+              });
+            }
+          }
+        }
+      }
+    }
+
+    console.log("matrix", { include });
+
+    return {
+      include,
+    };
+  }
+  // if not PR, generate matrix of python version, adapter, and operating
+  // system to run integration tests on
+
+  const include = [];
+  // run for all adapters and python versions on ubuntu
+  for (const adapter of supportedAdapters) {
+    for (const pythonVersion of supportedPythonVersions) {
+      include.push({
+        os: 'ubuntu-latest',
+        adapter: adapter,
+        "python-version": pythonVersion,
+      });
+    }
+  }
+
+  // additionally include runs for all adapters, on macos and windows,
+  // but only for the default python version
+  for (const adapter of supportedAdapters) {
+    for (const operatingSystem of ["windows-latest", "macos-latest"]) {
+      include.push({
+        os: operatingSystem,
+        adapter: adapter,
+        "python-version": defaultPythonVersion,
+      });
+    }
+  }
+
+  console.log("matrix", { include });
+
+  return {
+    include,
+  };
+};

.github/workflows/integration.yml

@@ -0,0 +1,266 @@
+# **what?**
+# This workflow runs all integration tests for supported OS
+# and python versions and core adapters. If triggered by PR,
+# the workflow will only run tests for adapters related
+# to code changes. Use the `test all` and `test ${adapter}`
+# label to run all or additional tests. Use `ok to test`
+# label to mark PRs from forked repositories that are safe
+# to run integration tests for. Requires secrets to run
+# against different warehouses.
+
+# **why?**
+# This checks the functionality of dbt from a user's perspective
+# and attempts to catch functional regressions.
+
+# **when?**
+# This workflow will run on every push to a protected branch
+# and when manually triggered. It will also run for all PRs, including
+# PRs from forks. The workflow will be skipped until there is a label
+# to mark the PR as safe to run.
+
+name: Adapter Integration Tests
+
+on:
+  # pushes to release branches
+  push:
+    branches:
+      - "main"
+      - "develop"
+      - "*.latest"
+      - "releases/*"
+  # all PRs, important to note that `pull_request_target` workflows
+  # will run in the context of the target branch of a PR
+  pull_request_target:
+  # manual tigger
+  workflow_dispatch:
+
+# explicitly turn off permissions for `GITHUB_TOKEN`
+permissions: read-all
+
+# will cancel previous workflows triggered by the same event and for the same ref for PRs or same SHA otherwise
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event_name }}-${{ contains(github.event_name, 'pull_request') && github.event.pull_request.head.ref || github.sha }}
+  cancel-in-progress: true
+
+# sets default shell to bash, for all operating systems
+defaults:
+  run:
+    shell: bash
+
+jobs:
+  # generate test metadata about what files changed and the testing matrix to use
+  test-metadata:
+    # run if not a PR from a forked repository or has a label to mark as safe to test
+    if: >-
+      github.event_name != 'pull_request_target' ||
+      github.event.pull_request.head.repo.full_name == github.repository ||
+      contains(github.event.pull_request.labels.*.name, 'ok to test')
+
+    runs-on: ubuntu-latest
+
+    outputs:
+      matrix: ${{ steps.generate-matrix.outputs.result }}
+
+    steps:
+      - name: Check out the repository (non-PR)
+        if: github.event_name != 'pull_request_target'
+        uses: actions/checkout@v2
+        with:
+          persist-credentials: false
+
+      - name: Check out the repository (PR)
+        if: github.event_name == 'pull_request_target'
+        uses: actions/checkout@v2
+        with:
+          persist-credentials: false
+          ref: ${{ github.event.pull_request.head.sha }}
+
+      - name: Check if relevant files changed
+        # https://github.com/marketplace/actions/paths-changes-filter
+        # For each filter, it sets output variable named by the filter to the text:
+        #  'true' - if any of changed files matches any of filter rules
+        #  'false' - if none of changed files matches any of filter rules
+        # also, returns:
+        #  `changes` - JSON array with names of all filters matching any of the changed files
+        uses: dorny/paths-filter@v2
+        id: get-changes
+        with:
+          token: ${{ secrets.GITHUB_TOKEN }}
+          filters: |
+            postgres:
+              - 'core/**'
+              - 'plugins/postgres/**'
+              - 'dev-requirements.txt'
+            snowflake:
+              - 'core/**'
+              - 'plugins/snowflake/**'
+            bigquery:
+              - 'core/**'
+              - 'plugins/bigquery/**'
+            redshift:
+              - 'core/**'
+              - 'plugins/redshift/**'
+              - 'plugins/postgres/**'
+
+      - name: Generate integration test matrix
+        id: generate-matrix
+        uses: actions/github-script@v4
+        env:
+          CHANGES: ${{ steps.get-changes.outputs.changes }}
+        with:
+          script: |
+            const script = require('./.github/scripts/integration-test-matrix.js')
+            const matrix = script({ context })
+            console.log(matrix)
+            return matrix
+
+  test:
+    name: ${{ matrix.adapter }} / python ${{ matrix.python-version }} / ${{ matrix.os }}
+
+    # run if not a PR from a forked repository or has a label to mark as safe to test
+    # also checks that the matrix generated is not empty
+    if: >-
+      needs.test-metadata.outputs.matrix &&
+      fromJSON( needs.test-metadata.outputs.matrix ).include[0] &&
+      (
+        github.event_name != 'pull_request_target' ||
+        github.event.pull_request.head.repo.full_name == github.repository ||
+        contains(github.event.pull_request.labels.*.name, 'ok to test')
+      )
+
+    runs-on: ${{ matrix.os }}
+
+    needs: test-metadata
+
+    strategy:
+      fail-fast: false
+      matrix: ${{ fromJSON(needs.test-metadata.outputs.matrix) }}
+
+    env:
+      TOXENV: integration-${{ matrix.adapter }}
+      PYTEST_ADDOPTS: "-v --color=yes -n4 --csv integration_results.csv"
+      DBT_INVOCATION_ENV: github-actions
+
+    steps:
+      - name: Check out the repository
+        if: github.event_name != 'pull_request_target'
+        uses: actions/checkout@v2
+        with:
+          persist-credentials: false
+
+      # explicity checkout the branch for the PR,
+      # this is necessary for the `pull_request_target` event
+      - name: Check out the repository (PR)
+        if: github.event_name == 'pull_request_target'
+        uses: actions/checkout@v2
+        with:
+          persist-credentials: false
+          ref: ${{ github.event.pull_request.head.sha }}
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v2
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Set up postgres (linux)
+        if: |
+          matrix.adapter == 'postgres' &&
+          runner.os == 'Linux'
+        uses: ./.github/actions/setup-postgres-linux
+
+      - name: Set up postgres (macos)
+        if: |
+          matrix.adapter == 'postgres' &&
+          runner.os == 'macOS'
+        uses: ./.github/actions/setup-postgres-macos
+
+      - name: Set up postgres (windows)
+        if: |
+          matrix.adapter == 'postgres' &&
+          runner.os == 'Windows'
+        uses: ./.github/actions/setup-postgres-windows
+
+      - name: Install python dependencies
+        run: |
+          pip install --upgrade pip
+          pip install tox
+          pip --version
+          tox --version
+
+      - name: Run tox (postgres)
+        if: matrix.adapter == 'postgres'
+        run: tox
+
+      - name: Run tox (redshift)
+        if: matrix.adapter == 'redshift'
+        env:
+          REDSHIFT_TEST_DBNAME: ${{ secrets.REDSHIFT_TEST_DBNAME }}
+          REDSHIFT_TEST_PASS: ${{ secrets.REDSHIFT_TEST_PASS }}
+          REDSHIFT_TEST_USER: ${{ secrets.REDSHIFT_TEST_USER }}
+          REDSHIFT_TEST_PORT: ${{ secrets.REDSHIFT_TEST_PORT }}
+          REDSHIFT_TEST_HOST: ${{ secrets.REDSHIFT_TEST_HOST }}
+        run: tox
+
+      - name: Run tox (snowflake)
+        if: matrix.adapter == 'snowflake'
+        env:
+          SNOWFLAKE_TEST_ACCOUNT: ${{ secrets.SNOWFLAKE_TEST_ACCOUNT }}
+          SNOWFLAKE_TEST_PASSWORD: ${{ secrets.SNOWFLAKE_TEST_PASSWORD }}
+          SNOWFLAKE_TEST_USER: ${{ secrets.SNOWFLAKE_TEST_USER }}
+          SNOWFLAKE_TEST_WAREHOUSE: ${{ secrets.SNOWFLAKE_TEST_WAREHOUSE }}
+          SNOWFLAKE_TEST_OAUTH_REFRESH_TOKEN: ${{ secrets.SNOWFLAKE_TEST_OAUTH_REFRESH_TOKEN }}
+          SNOWFLAKE_TEST_OAUTH_CLIENT_ID: ${{ secrets.SNOWFLAKE_TEST_OAUTH_CLIENT_ID }}
+          SNOWFLAKE_TEST_OAUTH_CLIENT_SECRET: ${{ secrets.SNOWFLAKE_TEST_OAUTH_CLIENT_SECRET }}
+          SNOWFLAKE_TEST_ALT_DATABASE: ${{ secrets.SNOWFLAKE_TEST_ALT_DATABASE }}
+          SNOWFLAKE_TEST_ALT_WAREHOUSE: ${{ secrets.SNOWFLAKE_TEST_ALT_WAREHOUSE }}
+          SNOWFLAKE_TEST_DATABASE: ${{ secrets.SNOWFLAKE_TEST_DATABASE }}
+          SNOWFLAKE_TEST_QUOTED_DATABASE: ${{ secrets.SNOWFLAKE_TEST_QUOTED_DATABASE }}
+          SNOWFLAKE_TEST_ROLE: ${{ secrets.SNOWFLAKE_TEST_ROLE }}
+        run: tox
+
+      - name: Run tox (bigquery)
+        if: matrix.adapter == 'bigquery'
+        env:
+          BIGQUERY_TEST_SERVICE_ACCOUNT_JSON: ${{ secrets.BIGQUERY_TEST_SERVICE_ACCOUNT_JSON }}
+          BIGQUERY_TEST_ALT_DATABASE: ${{ secrets.BIGQUERY_TEST_ALT_DATABASE }}
+        run: tox
+
+      - uses: actions/upload-artifact@v2
+        if: always()
+        with:
+          name: logs
+          path: ./logs
+
+      - name: Get current date
+        if: always()
+        id: date
+        run: echo "::set-output name=date::$(date +'%Y-%m-%dT%H_%M_%S')" #no colons allowed for artifacts
+
+      - uses: actions/upload-artifact@v2
+        if: always()
+        with:
+          name: integration_results_${{ matrix.python-version }}_${{ matrix.os }}_${{ matrix.adapter }}-${{ steps.date.outputs.date }}.csv
+          path: integration_results.csv
+
+  require-label-comment:
+    runs-on: ubuntu-latest
+
+    needs: test
+
+    permissions:
+      pull-requests: write
+
+    steps:
+      - name: Needs permission PR comment
+        if: >-
+          needs.test.result == 'skipped' &&
+          github.event_name == 'pull_request_target' &&
+          github.event.pull_request.head.repo.full_name != github.repository
+        uses: unsplash/comment-on-pr@master
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        with:
+          msg: |
+            "You do not have permissions to run integration tests, @dbt-labs/core "\
+            "needs to label this PR with `ok to test` in order to run integration tests!"
+          check_for_duplicate_msg: true

.github/workflows/main.yml

@@ -0,0 +1,206 @@
+# **what?**
+# Runs code quality checks, unit tests, and verifies python build on
+# all code commited to the repository. This workflow should not
+# require any secrets since it runs for PRs from forked repos.
+# By default, secrets are not passed to workflows running from
+# a forked repo.
+
+# **why?**
+# Ensure code for dbt meets a certain quality standard.
+
+# **when?**
+# This will run for all PRs, when code is pushed to a release
+# branch, and when manually triggered.
+
+name: Tests and Code Checks
+
+on:
+  push:
+    branches:
+      - "main"
+      - "develop"
+      - "*.latest"
+      - "releases/*"
+  pull_request:
+  workflow_dispatch:
+
+permissions: read-all
+
+# will cancel previous workflows triggered by the same event and for the same ref for PRs or same SHA otherwise
+concurrency:
+  group: ${{ github.workflow }}-${{ github.event_name }}-${{ contains(github.event_name, 'pull_request') && github.event.pull_request.head.ref || github.sha }}
+  cancel-in-progress: true
+
+defaults:
+  run:
+    shell: bash
+
+jobs:
+  code-quality:
+    name: ${{ matrix.toxenv }}
+
+    runs-on: ubuntu-latest
+
+    strategy:
+      fail-fast: false
+      matrix:
+        toxenv: [flake8, mypy]
+
+    env:
+      TOXENV: ${{ matrix.toxenv }}
+      PYTEST_ADDOPTS: "-v --color=yes"
+
+    steps:
+      - name: Check out the repository
+        uses: actions/checkout@v2
+        with:
+          persist-credentials: false
+
+      - name: Set up Python
+        uses: actions/setup-python@v2
+
+      - name: Install python dependencies
+        run: |
+          pip install --upgrade pip
+          pip install tox
+          pip --version
+          tox --version
+
+      - name: Run tox
+        run: tox
+
+  unit:
+    name: unit test / python ${{ matrix.python-version }}
+
+    runs-on: ubuntu-latest
+
+    strategy:
+      fail-fast: false
+      matrix:
+        python-version: [3.6, 3.7, 3.8] # TODO: support unit testing for python 3.9 (https://github.com/dbt-labs/dbt/issues/3689)
+
+    env:
+      TOXENV: "unit"
+      PYTEST_ADDOPTS: "-v --color=yes --csv unit_results.csv"
+
+    steps:
+      - name: Check out the repository
+        uses: actions/checkout@v2
+        with:
+          persist-credentials: false
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v2
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install python dependencies
+        run: |
+          pip install --upgrade pip
+          pip install tox
+          pip --version
+          tox --version
+
+      - name: Run tox
+        run: tox
+
+      - name: Get current date
+        if: always()
+        id: date
+        run: echo "::set-output name=date::$(date +'%Y-%m-%dT%H_%M_%S')" #no colons allowed for artifacts
+
+      - uses: actions/upload-artifact@v2
+        if: always()
+        with:
+          name: unit_results_${{ matrix.python-version }}-${{ steps.date.outputs.date }}.csv
+          path: unit_results.csv
+
+  build:
+    name: build packages
+
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Check out the repository
+        uses: actions/checkout@v2
+        with:
+          persist-credentials: false
+
+      - name: Set up Python
+        uses: actions/setup-python@v2
+        with:
+          python-version: 3.8
+
+      - name: Install python dependencies
+        run: |
+          pip install --upgrade pip
+          pip install --upgrade setuptools wheel twine check-wheel-contents
+          pip --version
+
+      - name: Build distributions
+        run: ./scripts/build-dist.sh
+
+      - name: Show distributions
+        run: ls -lh dist/
+
+      - name: Check distribution descriptions
+        run: |
+          twine check dist/*
+
+      - name: Check wheel contents
+        run: |
+          check-wheel-contents dist/*.whl --ignore W007,W008
+
+      - uses: actions/upload-artifact@v2
+        with:
+          name: dist
+          path: dist/
+
+  test-build:
+    name: verify packages / python ${{ matrix.python-version }} / ${{ matrix.os }}
+
+    needs: build
+
+    runs-on: ${{ matrix.os }}
+
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-latest, macos-latest, windows-latest]
+        python-version: [3.6, 3.7, 3.8, 3.9]
+
+    steps:
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v2
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install python dependencies
+        run: |
+          pip install --upgrade pip
+          pip install --upgrade wheel
+          pip --version
+
+      - uses: actions/download-artifact@v2
+        with:
+          name: dist
+          path: dist/
+
+      - name: Show distributions
+        run: ls -lh dist/
+
+      - name: Install wheel distributions
+        run: |
+          find ./dist/*.whl -maxdepth 1 -type f | xargs pip install --force-reinstall --find-links=dist/
+
+      - name: Check wheel distributions
+        run: |
+          dbt --version
+
+      - name: Install source distributions
+        run: |
+          find ./dist/*.gz -maxdepth 1 -type f | xargs pip install --force-reinstall --find-links=dist/
+
+      - name: Check source distributions
+        run: |
+          dbt --version

.github/workflows/performance.yml

@@ -0,0 +1,174 @@
+name: Performance Regression Tests
+# Schedule triggers
+on:
+  # runs twice a day at 10:05am and 10:05pm
+  schedule:
+    - cron: "5 10,22 * * *"
+  # Allows you to run this workflow manually from the Actions tab
+  workflow_dispatch:
+
+jobs:
+  # checks fmt of runner code
+  # purposefully not a dependency of any other job
+  # will block merging, but not prevent developing
+  fmt:
+    name: Cargo fmt
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v2
+      - uses: actions-rs/toolchain@v1
+        with:
+          profile: minimal
+          toolchain: stable
+          override: true
+      - run: rustup component add rustfmt
+      - uses: actions-rs/cargo@v1
+        with:
+          command: fmt
+          args: --manifest-path performance/runner/Cargo.toml --all -- --check
+
+  # runs any tests associated with the runner
+  # these tests make sure the runner logic is correct
+  test-runner:
+    name: Test Runner
+    runs-on: ubuntu-latest
+    env:
+      # turns errors into warnings
+      RUSTFLAGS: "-D warnings"
+    steps:
+      - uses: actions/checkout@v2
+      - uses: actions-rs/toolchain@v1
+        with:
+          profile: minimal
+          toolchain: stable
+          override: true
+      - uses: actions-rs/cargo@v1
+        with:
+          command: test
+          args: --manifest-path performance/runner/Cargo.toml
+
+  # build an optimized binary to be used as the runner in later steps
+  build-runner:
+    needs: [test-runner]
+    name: Build Runner
+    runs-on: ubuntu-latest
+    env:
+      RUSTFLAGS: "-D warnings"
+    steps:
+      - uses: actions/checkout@v2
+      - uses: actions-rs/toolchain@v1
+        with:
+          profile: minimal
+          toolchain: stable
+          override: true
+      - uses: actions-rs/cargo@v1
+        with:
+          command: build
+          args: --release --manifest-path performance/runner/Cargo.toml
+      - uses: actions/upload-artifact@v2
+        with:
+          name: runner
+          path: performance/runner/target/release/runner
+
+  # run the performance measurements on the current or default branch
+  measure-dev:
+    needs: [build-runner]
+    name: Measure Dev Branch
+    runs-on: ubuntu-latest
+    steps:
+      - name: checkout dev
+        uses: actions/checkout@v2
+      - name: Setup Python
+        uses: actions/setup-python@v2.2.2
+        with:
+          python-version: "3.8"
+      - name: install dbt
+        run: pip install -r dev-requirements.txt -r editable-requirements.txt
+      - name: install hyperfine
+        run: wget https://github.com/sharkdp/hyperfine/releases/download/v1.11.0/hyperfine_1.11.0_amd64.deb && sudo dpkg -i hyperfine_1.11.0_amd64.deb
+      - uses: actions/download-artifact@v2
+        with:
+          name: runner
+      - name: change permissions
+        run: chmod +x ./runner
+      - name: run
+        run: ./runner measure -b dev -p ${{ github.workspace }}/performance/projects/
+      - uses: actions/upload-artifact@v2
+        with:
+          name: dev-results
+          path: performance/results/
+
+  # run the performance measurements on the release branch which we use
+  # as a performance baseline. This part takes by far the longest, so
+  # we do everything we can first so the job fails fast.
+  # -----
+  # we need to checkout dbt twice in this job: once for the baseline dbt
+  # version, and once to get the latest regression testing projects,
+  # metrics, and runner code from the develop or current branch so that
+  # the calculations match for both versions of dbt we are comparing.
+  measure-baseline:
+    needs: [build-runner]
+    name: Measure Baseline Branch
+    runs-on: ubuntu-latest
+    steps:
+      - name: checkout latest
+        uses: actions/checkout@v2
+        with:
+          ref: "0.20.latest"
+      - name: Setup Python
+        uses: actions/setup-python@v2.2.2
+        with:
+          python-version: "3.8"
+      - name: move repo up a level
+        run: mkdir ${{ github.workspace }}/../baseline/ && cp -r ${{ github.workspace }} ${{ github.workspace }}/../baseline
+      - name: "[debug] ls new dbt location"
+        run: ls ${{ github.workspace }}/../baseline/dbt/
+      # installation creates egg-links so we have to preserve source
+      - name: install dbt from new location
+        run: cd ${{ github.workspace }}/../baseline/dbt/ && pip install -r dev-requirements.txt -r editable-requirements.txt
+      # checkout the current branch to get all the target projects
+      # this deletes the old checked out code which is why we had to copy before
+      - name: checkout dev
+        uses: actions/checkout@v2
+      - name: install hyperfine
+        run: wget https://github.com/sharkdp/hyperfine/releases/download/v1.11.0/hyperfine_1.11.0_amd64.deb && sudo dpkg -i hyperfine_1.11.0_amd64.deb
+      - uses: actions/download-artifact@v2
+        with:
+          name: runner
+      - name: change permissions
+        run: chmod +x ./runner
+      - name: run runner
+        run: ./runner measure -b baseline -p ${{ github.workspace }}/performance/projects/
+      - uses: actions/upload-artifact@v2
+        with:
+          name: baseline-results
+          path: performance/results/
+
+  # detect regressions on the output generated from measuring
+  # the two branches. Exits with non-zero code if a regression is detected.
+  calculate-regressions:
+    needs: [measure-dev, measure-baseline]
+    name: Compare Results
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/download-artifact@v2
+        with:
+          name: dev-results
+      - uses: actions/download-artifact@v2
+        with:
+          name: baseline-results
+      - name: "[debug] ls result files"
+        run: ls
+      - uses: actions/download-artifact@v2
+        with:
+          name: runner
+      - name: change permissions
+        run: chmod +x ./runner
+      - name: run calculation
+        run: ./runner calculate -r ./
+        # always attempt to upload the results even if there were regressions found
+      - uses: actions/upload-artifact@v2
+        if: ${{ always() }}
+        with:
+          name: final-calculations
+          path: ./final_calculations.json

.gitignore

@@ -8,7 +8,8 @@ __pycache__/
 
 # Distribution / packaging
 .Python
-env/
+env*/
+dbt_env/
 build/
 develop-eggs/
 dist/
@@ -42,6 +43,7 @@ htmlcov/
 .coverage
 .coverage.*
 .cache
+.env
 nosetests.xml
 coverage.xml
 *,cover
@@ -83,3 +85,12 @@ target/
 
 # pycharm
 .idea/
+venv/
+
+# AWS credentials
+.aws/
+
+.DS_Store
+
+# vscode
+.vscode/

ARCHITECTURE.md

@@ -0,0 +1,49 @@
+The core function of dbt is SQL compilation and execution. Users create projects of dbt resources (models, tests, seeds, snapshots, ...), defined in SQL and YAML files, and they invoke dbt to create, update, or query associated views and tables. Today, dbt makes heavy use of Jinja2 to enable the templating of SQL, and to construct a DAG (Directed Acyclic Graph) from all of the resources in a project. Users can also extend their projects by installing resources (including Jinja macros) from other projects, called "packages." 
+
+## dbt-core
+
+Most of the python code in the repository is within the `core/dbt` directory. Currently the main subdirectories are:
+- [`adapters`](core/dbt/adapters): Define base classes for behavior that is likely to differ across databases
+- [`clients`](core/dbt/clients): Interface with dependencies (agate, jinja) or across operating systems
+- [`config`](core/dbt/config): Reconcile user-supplied configuration from connection profiles, project files, and Jinja macros
+- [`context`](core/dbt/context): Build and expose dbt-specific Jinja functionality
+- [`contracts`](core/dbt/contracts): Define Python objects (dataclasses) that dbt expects to create and validate
+- [`deps`](core/dbt/deps): Package installation and dependency resolution
+- [`graph`](core/dbt/graph): Produce a `networkx` DAG of project resources, and selecting those resources given user-supplied criteria
+- [`include`](core/dbt/include): The dbt "global project," which defines default implementations of Jinja2 macros
+- [`parser`](core/dbt/parser): Read project files, validate, construct python objects
+- [`rpc`](core/dbt/rpc): Provide remote procedure call server for invoking dbt, following JSON-RPC 2.0 spec
+- [`task`](core/dbt/task): Set forth the actions that dbt can perform when invoked
+
+### Invoking dbt
+
+There are two supported ways of invoking dbt: from the command line and using an RPC server.
+
+The "tasks" map to top-level dbt commands. So `dbt run` => task.run.RunTask, etc. Some are more like abstract base classes (GraphRunnableTask, for example) but all the concrete types outside of task/rpc should map to tasks. Currently one executes at a time. The tasks kick off their “Runners” and those do execute in parallel. The parallelism is managed via a thread pool, in GraphRunnableTask.
+
+core/dbt/include/index.html
+This is the docs website code. It comes from the dbt-docs repository, and is generated when a release is packaged.
+
+## Adapters
+
+dbt uses an adapter-plugin pattern to extend support to different databases, warehouses, query engines, etc. The four core adapters that are in the main repository, contained within the [`plugins`](plugins) subdirectory, are: Postgres Redshift, Snowflake and BigQuery. Other warehouses use adapter plugins defined in separate repositories (e.g. [dbt-spark](https://github.com/dbt-labs/dbt-spark), [dbt-presto](https://github.com/dbt-labs/dbt-presto)).
+
+Each adapter is a mix of python, Jinja2, and SQL. The adapter code also makes heavy use of Jinja2 to wrap modular chunks of SQL functionality, define default implementations, and allow plugins to override it. 
+
+Each adapter plugin is a standalone python package that includes:
+
+- `dbt/include/[name]`: A "sub-global" dbt project, of YAML and SQL files, that reimplements Jinja macros to use the adapter's supported SQL syntax
+- `dbt/adapters/[name]`: Python modules that inherit, and optionally reimplement, the base adapter classes defined in dbt-core
+- `setup.py`
+
+The Postgres adapter code is the most central, and many of its implementations are used as the default defined in the dbt-core global project. The greater the distance of a data technology from Postgres, the more its adapter plugin may need to reimplement.
+
+## Testing dbt
+
+The [`test/`](test/) subdirectory includes unit and integration tests that run as continuous integration checks against open pull requests. Unit tests check mock inputs and outputs of specific python functions. Integration tests perform end-to-end dbt invocations against real adapters (Postgres, Redshift, Snowflake, BigQuery) and assert that the results match expectations. See [the contributing guide](CONTRIBUTING.md) for a step-by-step walkthrough of setting up a local development and testing environment.
+
+## Everything else
+
+- [docker](docker/): All dbt versions are published as Docker images on DockerHub. This subfolder contains the `Dockerfile` (constant) and `requirements.txt` (one for each version).
+- [etc](etc/): Images for README
+- [scripts](scripts/): Helper scripts for testing, releasing, and producing JSON schemas. These are not included in distributions of dbt, not are they rigorously tested—they're just handy tools for the dbt maintainers :)

CONTRIBUTING.md

@@ -1,111 +1,151 @@
-# Getting started with dbt
+# Contributing to `dbt`
+
+1. [About this document](#about-this-document)
+2. [Proposing a change](#proposing-a-change)
+3. [Getting the code](#getting-the-code)
+4. [Setting up an environment](#setting-up-an-environment)
+5. [Running `dbt` in development](#running-dbt-in-development)
+6. [Testing](#testing)
+7. [Submitting a Pull Request](#submitting-a-pull-request)
 
 ## About this document
 
-This document is a guide intended for folks interested in contributing to dbt. It is not intended as a guide for end users of dbt (though if it is helpful, that's great!) and it assumes a certain level of familiarity with Python concepts such as virtualenvs, `pip`, python modules, filesystems, and so on. This guide also documents the process by which community-contributed Pull Requests can be incorporated into this repository. This guide assumes you are using macOS or Linux and are comfortable with the command line. If you get stuck while reading this guide, drop us a line in the #development channel on [slack](slack.getdbt.com).
+This document is a guide intended for folks interested in contributing to `dbt`. Below, we document the process by which members of the community should create issues and submit pull requests (PRs) in this repository. It is not intended as a guide for using `dbt`, and it assumes a certain level of familiarity with Python concepts such as virtualenvs, `pip`, python modules, filesystems, and so on. This guide assumes you are using macOS or Linux and are comfortable with the command line.
 
-## Contributing a change
-
-dbt is Apache 2.0-licensed open source software. dbt is the software that it is today because community members like you have opened issues, provided feedback, and contributed to the knowledge loop for the entire communtiy. Whether you are a seasoned open source contributor or a first time committer, you are welcomed and encouraged to contribute code, documentation, ideas, or problem statements to this project.
-
-### Defining the problem
-
-If you have an idea for a new feature or if you've discovered a bug in dbt, the first step is to open an issue. Please check the list of [open issues](https://github.com/fishtown-analytics/dbt/issues) before creating a new one. If you find a relevant issue, please add a comment to the open issue instead of creating a new one. There are hundreds of open issues in this repository and it can be hard to know where to look for a relevant open issue. **The dbt maintainers are always happy to point contributors in the right direction**, so please err on the side of documenting your idea in a new issue if you are unsure where a problem statement belongs.
-
-**Note:** All community-contributed Pull Requests _must_ be associated with an open issue. If you submit a Pull Request that does not pertain to an open issue, you will be asked to create an issue describing the problem before the Pull Request can be reviewed.
-
-### Discussing the idea
-
-After creating an issue, a dbt maintainer will follow up with you to explore your idea further and advise on how to implement the suggested changes. In many cases, community members will chime in with their own thoughts on the problem statement. If you as the issue creator are interested in submitting a Pull Request to address the issue, you should indicate this in the body of the issue. The dbt maintainers are _always_ happy to help contributors with the implementation of fixes and features, so please also indicate if there's anything you're unsure about or could use guidance around in the issue.
-
-### Submitting a change
-
-If the issue is appropriately well-scoped and describes a beneficial change to the dbt codebase, then anyone may submit a Pull Request to implement the functionality described in the issue (see the sections below on how to do this).
-
-In some cases, the right resolution to an open issue might be tangential to the dbt codebase. The right path forward might be a documentation update or a change that can be made in user-space. In other cases, the issue might describe functionality that the dbt maintainers are unwilling or unable to incorporate into the dbt codebase. When it is determined that an open issue describes functionality that will not translate to a code change in the dbt repository, the issue will be tagged with the `wontfix` label (see below) and closed.
-
-### Using issue labels
-
-The dbt maintainers use labels to categorize open issues. Some labels indicate the databases impacted by the issue, while others describe the domain in the dbt codebase germane to the discussion. While most of these labels are self-explanatory (eg. `snowflake` or `bigquery`), there are others that are worth describing.
-
-| tag | description |
-| --- | ----------- |
-| bug | This issue represents a defect or regression in dbt |
-| enhancement | This issue represents net-new functionality in dbt |
-| good first issue | This issue does not require deep knowledge of the dbt codebase to implement. This issue is appropriate for a first-time contributor to implement. |
-| snoozed | This issue describes a good idea, but one which will probably not be addressed in a six-month time horizon. The dbt maintainers will revist these issues periodically and re-prioritize them accordingly. |
-| triage | This is a new issue which has not yet been reviewed by a dbt maintainer. This label is removed when a maintainer reviews and responds to the issue. |
-| stale | This is an old issue which has not recently been updated. Stale issues will periodically be closed by dbt maintainers, but they can be re-opened if the discussion is restarted. |
-| wontfix | This issue does not require a code change in the dbt repository, or the maintainers are unwilling/unable to merge a Pull Request which implements the behavior described in the issue. |
+If you're new to python development or contributing to open-source software, we encourage you to read this document from start to finish. If you get stuck, drop us a line in the `#dbt-core-development` channel on [slack](https://community.getdbt.com).
 
 ### Signing the CLA
 
-All contributors to dbt must sign the [Contributor License Agreement](https://docs.getdbt.com/docs/contributor-license-agreements) to have their Pull Request merged into the dbt codebase. If you are unable to sign the CLA, then the dbt maintainers will unfortunately be unable to merge your Pull Request.
+Please note that all contributors to `dbt` must sign the [Contributor License Agreement](https://docs.getdbt.com/docs/contributor-license-agreements) to have their Pull Request merged into the `dbt` codebase. If you are unable to sign the CLA, then the `dbt` maintainers will unfortunately be unable to merge your Pull Request. You are, however, welcome to open issues and comment on existing ones.
 
+## Proposing a change
+
+`dbt` is Apache 2.0-licensed open source software. `dbt` is what it is today because community members like you have opened issues, provided feedback, and contributed to the knowledge loop for the entire communtiy. Whether you are a seasoned open source contributor or a first-time committer, we welcome and encourage you to contribute code, documentation, ideas, or problem statements to this project.
+
+### Defining the problem
+
+If you have an idea for a new feature or if you've discovered a bug in `dbt`, the first step is to open an issue. Please check the list of [open issues](https://github.com/dbt-labs/dbt/issues) before creating a new one. If you find a relevant issue, please add a comment to the open issue instead of creating a new one. There are hundreds of open issues in this repository and it can be hard to know where to look for a relevant open issue. **The `dbt` maintainers are always happy to point contributors in the right direction**, so please err on the side of documenting your idea in a new issue if you are unsure where a problem statement belongs.
+
+> **Note:** All community-contributed Pull Requests _must_ be associated with an open issue. If you submit a Pull Request that does not pertain to an open issue, you will be asked to create an issue describing the problem before the Pull Request can be reviewed.
+
+### Discussing the idea
+
+After you open an issue, a `dbt` maintainer will follow up by commenting on your issue (usually within 1-3 days) to explore your idea further and advise on how to implement the suggested changes. In many cases, community members will chime in with their own thoughts on the problem statement. If you as the issue creator are interested in submitting a Pull Request to address the issue, you should indicate this in the body of the issue. The `dbt` maintainers are _always_ happy to help contributors with the implementation of fixes and features, so please also indicate if there's anything you're unsure about or could use guidance around in the issue.
+
+### Submitting a change
+
+If an issue is appropriately well scoped and describes a beneficial change to the `dbt` codebase, then anyone may submit a Pull Request to implement the functionality described in the issue. See the sections below on how to do this.
+
+The `dbt` maintainers will add a `good first issue` label if an issue is suitable for a first-time contributor. This label often means that the required code change is small, limited to one database adapter, or a net-new addition that does not impact existing functionality. You can see the list of currently open issues on the [Contribute](https://github.com/dbt-labs/dbt/contribute) page.
+
+Here's a good workflow:
+- Comment on the open issue, expressing your interest in contributing the required code change
+- Outline your planned implementation. If you want help getting started, ask!
+- Follow the steps outlined below to develop locally. Once you have opened a PR, one of the `dbt` maintainers will work with you to review your code.
+- Add a test! Tests are crucial for both fixes and new features alike. We want to make sure that code works as intended, and that it avoids any bugs  previously encountered. Currently, the best resource for understanding `dbt`'s [unit](test/unit) and [integration](test/integration) tests is the tests themselves. One of the maintainers can help by pointing out relevant examples.
+
+In some cases, the right resolution to an open issue might be tangential to the `dbt` codebase. The right path forward might be a documentation update or a change that can be made in user-space. In other cases, the issue might describe functionality that the `dbt` maintainers are unwilling or unable to incorporate into the `dbt` codebase. When it is determined that an open issue describes functionality that will not translate to a code change in the `dbt` repository, the issue will be tagged with the `wontfix` label (see below) and closed.
+
+### Using issue labels
+
+The `dbt` maintainers use labels to categorize open issues. Some labels indicate the databases impacted by the issue, while others describe the domain in the `dbt` codebase germane to the discussion. While most of these labels are self-explanatory (eg. `snowflake` or `bigquery`), there are others that are worth describing.
+
+| tag | description |
+| --- | ----------- |
+| [triage](https://github.com/dbt-labs/dbt/labels/triage) | This is a new issue which has not yet been reviewed by a `dbt` maintainer. This label is removed when a maintainer reviews and responds to the issue. |
+| [bug](https://github.com/dbt-labs/dbt/labels/bug) | This issue represents a defect or regression in `dbt` |
+| [enhancement](https://github.com/dbt-labs/dbt/labels/enhancement) | This issue represents net-new functionality in `dbt` |
+| [good first issue](https://github.com/dbt-labs/dbt/labels/good%20first%20issue) | This issue does not require deep knowledge of the `dbt` codebase to implement. This issue is appropriate for a first-time contributor. |
+| [help wanted](https://github.com/dbt-labs/dbt/labels/help%20wanted) / [discussion](https://github.com/dbt-labs/dbt/labels/discussion) | Conversation around this issue in ongoing, and there isn't yet a clear path forward. Input from community members is most welcome. |
+| [duplicate](https://github.com/dbt-labs/dbt/issues/duplicate) | This issue is functionally identical to another open issue. The `dbt` maintainers will close this issue and encourage community members to focus conversation on the other one. |
+| [snoozed](https://github.com/dbt-labs/dbt/labels/snoozed) | This issue describes a good idea, but one which will probably not be addressed in a six-month time horizon. The `dbt` maintainers will revist these issues periodically and re-prioritize them accordingly. |
+| [stale](https://github.com/dbt-labs/dbt/labels/stale) | This is an old issue which has not recently been updated. Stale issues will periodically be closed by `dbt` maintainers, but they can be re-opened if the discussion is restarted. |
+| [wontfix](https://github.com/dbt-labs/dbt/labels/wontfix) | This issue does not require a code change in the `dbt` repository, or the maintainers are unwilling/unable to merge a Pull Request which implements the behavior described in the issue. |
+
+#### Branching Strategy
+
+`dbt` has three types of branches:
+
+- **Trunks** are where active development of the next release takes place. There is one trunk named `develop` at the time of writing this, and will be the default branch of the repository.
+- **Release Branches** track a specific, not yet complete release of `dbt`. Each minor version release has a corresponding release branch. For example, the `0.11.x` series of releases has a branch called `0.11.latest`. This allows us to release new patch versions under `0.11` without necessarily needing to pull them into the latest version of `dbt`.
+- **Feature Branches** track individual features and fixes. On completion they should be merged into the trunk brnach or a specific release branch.
 
 ## Getting the code
 
 ### Installing git
 
-You will need `git` in order to download and modify the dbt source code. On macOS, the best way to download git is to just install Xcode.
+You will need `git` in order to download and modify the `dbt` source code. On macOS, the best way to download git is to just install [Xcode](https://developer.apple.com/support/xcode/).
 
 ### External contributors
 
-If you are not a member of the `fishtown-analytics` GitHub organization, you can contribute to dbt by forking the dbt repository. For a detailed overview on forking, check out the [GitHub docs on forking](https://help.github.com/en/articles/fork-a-repo). In short, you will need to:
+If you are not a member of the `dbt-labs` GitHub organization, you can contribute to `dbt` by forking the `dbt` repository. For a detailed overview on forking, check out the [GitHub docs on forking](https://help.github.com/en/articles/fork-a-repo). In short, you will need to:
 
-1. fork the dbt repository
-2. clone your fork
+1. fork the `dbt` repository
+2. clone your fork locally
 3. check out a new branch for your proposed changes
 4. push changes to your fork
-5. open a pull request against `fishtown-analytics/dbt` from your forked repository
+5. open a pull request against `dbt-labs/dbt` from your forked repository
 
 ### Core contributors
 
-If you are a member of the `fishtown-analytics` GitHub organization, you will have push access to the dbt repo. Rather than 
-forking dbt to make your changes, just clone the repository and push directly to a branch.
-
+If you are a member of the `dbt-labs` GitHub organization, you will have push access to the `dbt` repo. Rather than forking `dbt` to make your changes, just clone the repository, check out a new branch, and push directly to that branch.
 
 ## Setting up an environment
 
-To begin developing code in dbt, you should set up the following:
+There are some tools that will be helpful to you in developing locally. While this is the list relevant for `dbt` development, many of these tools are used commonly across open-source python projects.
 
-### virtualenv
+### Tools
 
-We strongly recommend using virtual environments when developing code in dbt. We recommend creating this virtualenv
-in the root of the dbt repository. To create a new virtualenv, run:
-```
+A short list of tools used in `dbt` testing that will be helpful to your understanding:
+
+- [`tox`](https://tox.readthedocs.io/en/latest/) to manage virtualenvs across python versions. We currently target the latest patch releases for Python 3.6, Python 3.7, Python 3.8, and Python 3.9
+- [`pytest`](https://docs.pytest.org/en/latest/) to discover/run tests
+- [`make`](https://users.cs.duke.edu/~ola/courses/programming/Makefiles/Makefiles.html) - but don't worry too much, nobody _really_ understands how make works and our Makefile is super simple
+- [`flake8`](https://flake8.pycqa.org/en/latest/) for code linting
+- [`mypy`](https://mypy.readthedocs.io/en/stable/) for static type checking
+- [CircleCI](https://circleci.com/product/) and [Azure Pipelines](https://azure.microsoft.com/en-us/services/devops/pipelines/)
+
+A deep understanding of these tools in not required to effectively contribute to `dbt`, but we recommend checking out the attached documentation if you're interested in learning more about them.
+
+#### virtual environments
+
+We strongly recommend using virtual environments when developing code in `dbt`. We recommend creating this virtualenv
+in the root of the `dbt` repository. To create a new virtualenv, run:
+```sh
 python3 -m venv env
 source env/bin/activate
 ```
 
 This will create and activate a new Python virtual environment.
 
-### docker and docker-compose
+#### docker and docker-compose
 
-Docker and docker-compose are both used in testing. For macOS, the easiest thing to do is to [download docker for mac](https://store.docker.com/editions/community/docker-ce-desktop-mac). You'll need to make an account. On Linux, you can use one of the packages [here](https://docs.docker.com/install/#server). We recommend installing from docker.com instead of from your package manager. On Linux you also have to install docker-compose separately, follow [these instructions](https://docs.docker.com/compose/install/#install-compose).
+Docker and docker-compose are both used in testing. Specific instructions for you OS can be found [here](https://docs.docker.com/get-docker/).
 
 
-### Installing postgres locally (optional)
+#### postgres (optional)
 
 For testing, and later in the examples in this document, you may want to have `psql` available so you can poke around in the database and see what happened. We recommend that you use [homebrew](https://brew.sh/) for that on macOS, and your package manager on Linux. You can install any version of the postgres client that you'd like. On macOS, with homebrew setup, you can run:
 
-```
+```sh
 brew install postgresql
 ```
 
-## Running dbt in development
+## Running `dbt` in development
 
 ### Installation
 
-First make sure that you set up your `virtualenv` as described in section _Setting up an environment_. Next, install dbt (and it's dependencies) with:
+First make sure that you set up your `virtualenv` as described in [Setting up an environment](#setting-up-an-environment). Next, install `dbt` (and its dependencies) with:
 
-```
-pip install -r editable_requirements.txt
+```sh
+make dev
+# or
+pip install -r dev-requirements.txt -r editable-requirements.txt
 ```
 
-When dbt is installed from source in this way, any changes you make to the dbt source code will be reflected immediately in your next `dbt` run.
+When `dbt` is installed this way, any changes you make to the `dbt` source code will be reflected immediately in your next `dbt` run.
 
-### Running dbt
+### Running `dbt`
 
 With your virtualenv activated, the `dbt` script should point back to the source code you've cloned on your machine. You can verify this by running `which dbt`. This command should show you a path to an executable in your virtualenv.
 
@@ -113,82 +153,79 @@ Configure your [profile](https://docs.getdbt.com/docs/configure-your-profile) as
 
 ## Testing
 
-Getting the dbt integration tests set up in your local environment will be very helpful as you start to make changes to your local version of dbt. The section that follows outlines some helpful tips for setting up the test environment.
+Getting the `dbt` integration tests set up in your local environment will be very helpful as you start to make changes to your local version of `dbt`. The section that follows outlines some helpful tips for setting up the test environment.
 
-### Tools
+Since `dbt` works with a number of different databases, you will need to supply credentials for one or more of these databases in your test environment. Most organizations don't have access to each of a BigQuery, Redshift, Snowflake, and Postgres database, so it's likely that you will be unable to run every integration test locally. Fortunately, dbt Labs provides a CI environment with access to sandboxed Redshift, Snowflake, BigQuery, and Postgres databases. See the section on [_Submitting a Pull Request_](#submitting-a-pull-request) below for more information on this CI setup.
 
-A short list of tools used in dbt testing that will be helpful to your understanding:
+### Initial setup
 
-- [virtualenv](https://virtualenv.pypa.io/en/stable/) to manage dependencies
-- [tox](https://tox.readthedocs.io/en/latest/) to manage virtualenvs across python versions
-- [pytest](https://docs.pytest.org/en/latest/) to discover/run tests
-- [make](https://users.cs.duke.edu/~ola/courses/programming/Makefiles/Makefiles.html) - but don't worry too much, nobody _really_ understands how make works and our Makefile is super simple
-- [flake8](https://gitlab.com/pycqa/flake8) for code linting
-- [CircleCI](https://circleci.com/product/) and [Azure Pipelines](https://azure.microsoft.com/en-us/services/devops/pipelines/)
-
-A deep understanding of these tools in not required to effectively contribute to dbt, but we recommend checking out the attached documentation if you're interested in learning more about them.
-
-
-### Running tests via Docker
-
-dbt's unit and integration tests run in Docker. Because dbt works with a number of different databases, you will need to supply credentials for one or more of these databases in your test environment. Most organizations don't have access to each of a BigQuery, Redshift, Snowflake, and Postgres database, so it's likely that you will be unable to run every integration test locally. Fortunately, Fishtown Analytics provides a CI environment with access to sandboxed Redshift, Snowflake, BigQuery, and Postgres databases. See the section on _Submitting a Pull Request_ below for more information on this CI setup.
-
-
-#### Specifying your test credentials
-
-dbt uses test credentials specified in a `test.env` file in the root of the repository. This `test.env` file is git-ignored, but please be _extra_ careful to never check in credentials or other sensitive information when developing against dbt. To create your `test.env` file, copy the provided sample file, then supply your relevant credentials:
-
-```
-cp test.env.sample test.env
-atom test.env # supply your credentials
-```
-
-We recommend starting with dbt's Postgres tests. These tests cover most of the functionality in dbt, are the fastest to run, and are the easiest to set up. dbt's test suite runs Postgres in a Docker container, so no setup should be required to run these tests. If you additionally want to test Snowflake, Bigquery, or Redshift locally you'll need to get credentials and add them to the `test.env` file. 
-
-#### Running tests
-
-dbt's unit tests and Python linter can be run with:
-
-```
-make test-unit
-```
-
-To run the Postgres + Python 3.6 integration tests, you'll have to do one extra step of setting up the test database:
+We recommend starting with `dbt`'s Postgres tests. These tests cover most of the functionality in `dbt`, are the fastest to run, and are the easiest to set up. To run the Postgres integration tests, you'll have to do one extra step of setting up the test database:
 
+```sh
+make setup-db
 ```
+or, alternatively:
+```sh
 docker-compose up -d database
 PGHOST=localhost PGUSER=root PGPASSWORD=password PGDATABASE=postgres bash test/setup_db.sh
 ```
 
-To run a quick test for Python3 integration tests on Postgres, you can run:
+`dbt` uses test credentials specified in a `test.env` file in the root of the repository for non-Postgres databases. This `test.env` file is git-ignored, but please be _extra_ careful to never check in credentials or other sensitive information when developing against `dbt`. To create your `test.env` file, copy the provided sample file, then supply your relevant credentials. This step is only required to use non-Postgres databases.
 
 ```
-make test-quick
+cp test.env.sample test.env
+$EDITOR test.env
 ```
 
-To run tests for a specific database, invoke `tox` directly with the required flags:
+> In general, it's most important to have successful unit and Postgres tests. Once you open a PR, `dbt` will automatically run integration tests for the other three core database adapters. Of course, if you are a BigQuery user, contributing a BigQuery-only feature, it's important to run BigQuery tests as well.
+
+### Test commands
+
+There are a few methods for running tests locally.
+
+#### Makefile
+
+There are multiple targets in the Makefile to run common test suites and code
+checks, most notably:
+
+```sh
+# Runs unit tests with py38 and code checks in parallel.
+make test
+# Runs postgres integration tests with py38 in "fail fast" mode.
+make integration
 ```
-# Run Postgres py36 tests
-docker-compose run test tox -e integration-postgres-py36 -- -x
+> These make targets assume you have a recent version of [`tox`](https://tox.readthedocs.io/en/latest/) installed locally,
+> unless you use choose a Docker container to run tests. Run `make help` for more info.
 
-# Run Snowflake py36 tests
-docker-compose run test tox -e integration-snowflake-py36 -- -x
+Check out the other targets in the Makefile to see other commonly used test
+suites.
 
-# Run BigQuery py36 tests
-docker-compose run test tox -e integration-bigquery-py36 -- -x
+#### `tox`
 
-# Run Redshift py36 tests
-docker-compose run test tox -e integration-redshift-py36 -- -x
+[`tox`](https://tox.readthedocs.io/en/latest/) takes care of managing virtualenvs and install dependencies in order to run
+tests. You can also run tests in parallel, for example, you can run unit tests
+for Python 3.6, Python 3.7, Python 3.8, `flake8` checks, and `mypy` checks in
+parallel with `tox -p`. Also, you can run unit tests for specific python versions
+with `tox -e py36`. The configuration for these tests in located in `tox.ini`.
+
+#### `pytest`
+
+Finally, you can also run a specific test or group of tests using [`pytest`](https://docs.pytest.org/en/latest/) directly. With a virtualenv
+active and dev dependencies installed you can do things like:
+```sh
+# run specific postgres integration tests
+python -m pytest -m profile_postgres test/integration/001_simple_copy_test
+# run all unit tests in a file
+python -m pytest test/unit/test_graph.py
+# run a specific unit test
+python -m pytest test/unit/test_graph.py::GraphTest::test__dependency_list
 ```
+> [Here](https://docs.pytest.org/en/reorganize-docs/new-docs/user/commandlineuseful.html)
+> is a list of useful command-line options for `pytest` to use while developing.
+## Submitting a Pull Request
 
-See the `Makefile` contents for more some other examples of ways to run `tox`.
+dbt Labs provides a sandboxed Redshift, Snowflake, and BigQuery database for use in a CI environment. When pull requests are submitted to the `dbt-labs/dbt` repo, GitHub will trigger automated tests in CircleCI and Azure Pipelines.
 
-### Submitting a Pull Request
+A `dbt` maintainer will review your PR. They may suggest code revision for style or clarity, or request that you add unit or integration test(s). These are good things! We believe that, with a little bit of help, anyone can contribute high-quality code.
 
-Fishtown Analytics provides a sandboxed Redshift, Snowflake, and BigQuery database for use in a CI environment.
-
-When pull requests are submitted to the `fishtown-analytics/dbt` repo, GitHub will trigger automated tests in CircleCI and Azure Pipelines. If the PR submitter is a member of the `fishtown-analytics` GitHub organization, then the credentials for these databases will be automatically supplied as environment variables in the CI test suite.
-
-**If the PR submitter is not a member of the `fishtown-analytics` organization, then these environment variables will not be automatically supplied in the CI environment**. Once a core maintainer has taken a look at the Pull Request, they will kick off the test suite with the required credentials.
-
-Once your tests are passing and your PR has been reviewed, a dbt maintainer will merge your changes into the active development branch! And that's it! Happy developing :tada:
+Once all tests are passing and your PR has been approved, a `dbt` maintainer will merge your changes into the active development branch. And that's it! Happy developing :tada:

Dockerfile

@@ -1,33 +0,0 @@
-FROM ubuntu:18.04
-
-ENV DEBIAN_FRONTEND noninteractive
-
-RUN apt-get update && \
-    apt-get dist-upgrade -y && \
-    apt-get install -y  --no-install-recommends \
-        netcat postgresql curl git ssh  software-properties-common \
-        make build-essential ca-certificates libpq-dev && \
-    add-apt-repository ppa:deadsnakes/ppa && \
-    apt-get install -y \
-        python python-dev python-pip \
-        python3.6 python3.6-dev python3-pip python3.6-venv \
-        python3.7 python3.7-dev python3.7-venv \
-        python3.8 python3.8-dev python3.8-venv \
-        python3.9 python3.9-dev python3.9-venv && \
-    apt-get clean && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
-
-RUN useradd -mU dbt_test_user
-RUN mkdir /usr/app && chown dbt_test_user /usr/app
-RUN mkdir /home/tox && chown dbt_test_user /home/tox
-
-WORKDIR /usr/app
-VOLUME /usr/app
-
-RUN pip3 install -U "tox==3.14.4" wheel "six>=1.14.0,<1.15.0" "virtualenv==20.0.3" setuptools
-# tox fails if the 'python' interpreter (python2) doesn't have `tox` installed
-RUN pip install -U "tox==3.14.4" "six>=1.14.0,<1.15.0" "virtualenv==20.0.3" setuptools
-
-USER dbt_test_user
-
-ENV PYTHONIOENCODING=utf-8
-ENV LANG C.UTF-8

Dockerfile.test

@@ -0,0 +1,75 @@
+FROM ubuntu:20.04
+
+ENV DEBIAN_FRONTEND noninteractive
+
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+    software-properties-common \
+    && add-apt-repository ppa:git-core/ppa -y \
+    && apt-get dist-upgrade -y \
+    && apt-get install -y --no-install-recommends \
+    netcat \
+    postgresql \
+    curl \
+    git \
+    ssh \
+    software-properties-common \
+    make \
+    build-essential \
+    ca-certificates \
+    libpq-dev \
+    libsasl2-dev \
+    libsasl2-2 \
+    libsasl2-modules-gssapi-mit \
+    libyaml-dev \
+    unixodbc-dev \
+    && add-apt-repository ppa:deadsnakes/ppa \
+    && apt-get install -y \
+    python \
+    python-dev \
+    python-pip \
+    python3.6 \
+    python3.6-dev \
+    python3-pip \
+    python3.6-venv \
+    python3.7 \
+    python3.7-dev \
+    python3.7-venv \
+    python3.8 \
+    python3.8-dev \
+    python3.8-venv \
+    python3.9 \
+    python3.9-dev \
+    python3.9-venv \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
+
+ARG DOCKERIZE_VERSION=v0.6.1
+RUN curl -LO https://github.com/jwilder/dockerize/releases/download/$DOCKERIZE_VERSION/dockerize-linux-amd64-$DOCKERIZE_VERSION.tar.gz \
+    && tar -C /usr/local/bin -xzvf dockerize-linux-amd64-$DOCKERIZE_VERSION.tar.gz \
+    && rm dockerize-linux-amd64-$DOCKERIZE_VERSION.tar.gz
+
+RUN pip3 install -U tox wheel six setuptools
+
+# These args are passed in via docker-compose, which reads then from the .env file.
+# On Linux, run `make .env` to create the .env file for the current user.
+# On MacOS and Windows, these can stay unset.
+ARG USER_ID
+ARG GROUP_ID
+
+RUN if [ ${USER_ID:-0} -ne 0 ] && [ ${GROUP_ID:-0} -ne 0 ]; then \
+    groupadd -g ${GROUP_ID} dbt_test_user && \
+    useradd -m -l -u ${USER_ID} -g ${GROUP_ID} dbt_test_user; \
+    else \
+    useradd -mU -l dbt_test_user; \
+    fi
+RUN mkdir /usr/app && chown dbt_test_user /usr/app
+RUN mkdir /home/tox && chown dbt_test_user /home/tox
+
+WORKDIR /usr/app
+VOLUME /usr/app
+
+USER dbt_test_user
+
+ENV PYTHONIOENCODING=utf-8
+ENV LANG C.UTF-8

License.md

@@ -186,7 +186,7 @@
       same "printed page" as the copyright notice for easier
       identification within third-party archives.
 
-   Copyright {yyyy} {name of copyright owner}
+   Copyright 2021 dbt Labs, Inc.
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.

Makefile

@@ -1,29 +1,94 @@
-.PHONY: install test test-unit test-integration
+.DEFAULT_GOAL:=help
 
-changed_tests := `git status --porcelain | grep '^\(M\| M\|A\| A\)' | awk '{ print $$2 }' | grep '\/test_[a-zA-Z_\-\.]\+.py'`
+# Optional flag to run target in a docker container.
+# (example `make test USE_DOCKER=true`)
+ifeq ($(USE_DOCKER),true)
+	DOCKER_CMD := docker-compose run --rm test
+endif
 
-install:
-	pip install -e .
+.PHONY: dev
+dev: ## Installs dbt-* packages in develop mode along with development dependencies.
+	pip install -r dev-requirements.txt -r editable-requirements.txt
 
-test:
-	@echo "Full test run starting..."
-	@time docker-compose run test tox
+.PHONY: mypy
+mypy: .env ## Runs mypy for static type checking.
+	$(DOCKER_CMD) tox -e mypy
 
-test-unit:
-	@echo "Unit test run starting..."
-	@time docker-compose run test tox -e unit-py36,flake8
+.PHONY: flake8
+flake8: .env ## Runs flake8 to enforce style guide.
+	$(DOCKER_CMD) tox -e flake8
 
-test-integration:
-	@echo "Integration test run starting..."
-	@time docker-compose run test tox -e integration-postgres-py36integration-redshift-py36,integration-snowflake-py36,integration-bigquery-py36
+.PHONY: lint
+lint: .env ## Runs all code checks in parallel.
+	$(DOCKER_CMD) tox -p -e flake8,mypy
 
-test-quick:
-	@echo "Integration test run starting..."
-	@time docker-compose run test tox -e integration-postgres-py36 -- -x
+.PHONY: unit
+unit: .env ## Runs unit tests with py38.
+	$(DOCKER_CMD) tox -e py38
 
-clean:
+.PHONY: test
+test: .env ## Runs unit tests with py38 and code checks in parallel.
+	$(DOCKER_CMD) tox -p -e py38,flake8,mypy
+
+.PHONY: integration
+integration: .env integration-postgres ## Alias for integration-postgres.
+
+.PHONY: integration-fail-fast
+integration-fail-fast: .env integration-postgres-fail-fast ## Alias for integration-postgres-fail-fast.
+
+.PHONY: integration-postgres
+integration-postgres: .env ## Runs postgres integration tests with py38.
+	$(DOCKER_CMD) tox -e py38-postgres -- -nauto
+
+.PHONY: integration-postgres-fail-fast
+integration-postgres-fail-fast: .env ## Runs postgres integration tests with py38 in "fail fast" mode.
+	$(DOCKER_CMD) tox -e py38-postgres -- -x -nauto
+
+.PHONY: integration-redshift
+integration-redshift: .env ## Runs redshift integration tests with py38.
+	$(DOCKER_CMD) tox -e py38-redshift -- -nauto
+
+.PHONY: integration-redshift-fail-fast
+integration-redshift-fail-fast: .env ## Runs redshift integration tests with py38 in "fail fast" mode.
+	$(DOCKER_CMD) tox -e py38-redshift -- -x -nauto
+
+.PHONY: integration-snowflake
+integration-snowflake: .env ## Runs snowflake integration tests with py38.
+	$(DOCKER_CMD) tox -e py38-snowflake -- -nauto
+
+.PHONY: integration-snowflake-fail-fast
+integration-snowflake-fail-fast: .env ## Runs snowflake integration tests with py38 in "fail fast" mode.
+	$(DOCKER_CMD) tox -e py38-snowflake -- -x -nauto
+
+.PHONY: integration-bigquery
+integration-bigquery: .env ## Runs bigquery integration tests with py38.
+	$(DOCKER_CMD) tox -e py38-bigquery -- -nauto
+
+.PHONY: integration-bigquery-fail-fast
+integration-bigquery-fail-fast: .env ## Runs bigquery integration tests with py38 in "fail fast" mode.
+	$(DOCKER_CMD) tox -e py38-bigquery -- -x -nauto
+
+.PHONY: setup-db
+setup-db: ## Setup Postgres database with docker-compose for system testing.
+	docker-compose up -d database
+	PGHOST=localhost PGUSER=root PGPASSWORD=password PGDATABASE=postgres bash test/setup_db.sh
+
+# This rule creates a file named .env that is used by docker-compose for passing
+# the USER_ID and GROUP_ID arguments to the Docker image.
+.env: ## Setup step for using using docker-compose with make target.
+	@touch .env
+ifneq ($(OS),Windows_NT)
+ifneq ($(shell uname -s), Darwin)
+	@echo USER_ID=$(shell id -u) > .env
+	@echo GROUP_ID=$(shell id -g) >> .env
+endif
+endif
+
+.PHONY: clean
+clean: ## Resets development environment.
 	rm -f .coverage
 	rm -rf .eggs/
+	rm -f .env
 	rm -rf .tox/
 	rm -rf build/
 	rm -rf dbt.egg-info/
@@ -34,3 +99,14 @@ clean:
 	rm -rf target/
 	find . -type f -name '*.pyc' -delete
 	find . -type d -name '__pycache__' -depth -delete
+
+.PHONY: help
+help: ## Show this help message.
+	@echo 'usage: make [target] [USE_DOCKER=true]'
+	@echo
+	@echo 'targets:'
+	@grep -E '^[a-zA-Z_-]+:.*?## .*$$' $(MAKEFILE_LIST) | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[36m%-30s\033[0m %s\n", $$1, $$2}'
+	@echo
+	@echo 'options:'
+	@echo 'use USE_DOCKER=true to run target in a docker container'
+

README.md

@@ -1,28 +1,18 @@
 <p align="center">
-  <img src="https://github.com/fishtown-analytics/dbt/blob/master/etc/dbt-horizontal.png?raw=true" alt="dbt logo"/>
+  <img src="https://raw.githubusercontent.com/dbt-labs/dbt/ec7dee39f793aa4f7dd3dae37282cc87664813e4/etc/dbt-logo-full.svg" alt="dbt logo" width="500"/>
 </p>
 <p align="center">
-  <a href="https://codeclimate.com/github/fishtown-analytics/dbt">
-    <img src="https://codeclimate.com/github/fishtown-analytics/dbt/badges/gpa.svg" alt="Code Climate"/>
+  <a href="https://github.com/dbt-labs/dbt/actions/workflows/main.yml">
+    <img src="https://github.com/dbt-labs/dbt/actions/workflows/main.yml/badge.svg?event=push" alt="Unit Tests Badge"/>
   </a>
-  <a href="https://circleci.com/gh/fishtown-analytics/dbt/tree/master">
-    <img src="https://circleci.com/gh/fishtown-analytics/dbt/tree/master.svg?style=svg"  alt="CircleCI" />
-  </a>
-  <a href="https://ci.appveyor.com/project/DrewBanin/dbt/branch/development">
-    <img src="https://ci.appveyor.com/api/projects/status/v01rwd3q91jnwp9m/branch/development?svg=true" alt="AppVeyor" />
-  </a>
-  <a href="https://slack.getdbt.com">
-    <img src="https://slack.getdbt.com/badge.svg" alt="Slack" />
+  <a href="https://github.com/dbt-labs/dbt/actions/workflows/integration.yml">
+    <img src="https://github.com/dbt-labs/dbt/actions/workflows/integration.yml/badge.svg?event=push" alt="Integration Tests Badge"/>
   </a>
 </p>
 
-**[dbt](https://www.getdbt.com/)** (data build tool) enables data analysts and engineers to transform their data using the same practices that software engineers use to build applications.
+**[dbt](https://www.getdbt.com/)** enables data analysts and engineers to transform their data using the same practices that software engineers use to build applications.
 
-dbt is the T in ELT. Organize, cleanse, denormalize, filter, rename, and pre-aggregate the raw data in your warehouse so that it's ready for analysis.
-
-![dbt architecture](https://github.com/fishtown-analytics/dbt/blob/master/etc/dbt-arch.png?raw=true)
-
-dbt can be used to [aggregate pageviews into sessions](https://github.com/fishtown-analytics/snowplow), calculate [ad spend ROI](https://github.com/fishtown-analytics/facebook-ads), or report on [email campaign performance](https://github.com/fishtown-analytics/mailchimp).
+![architecture](https://raw.githubusercontent.com/dbt-labs/dbt/6c6649f9129d5d108aa3b0526f634cd8f3a9d1ed/etc/dbt-arch.png)
 
 ## Understanding dbt
 
@@ -30,29 +20,23 @@ Analysts using dbt can transform their data by simply writing select statements,
 
 These select statements, or "models", form a dbt project. Models frequently build on top of one another – dbt makes it easy to [manage relationships](https://docs.getdbt.com/docs/ref) between models, and [visualize these relationships](https://docs.getdbt.com/docs/documentation), as well as assure the quality of your transformations through [testing](https://docs.getdbt.com/docs/testing).
 
-![dbt dag](https://github.com/fishtown-analytics/dbt/blob/master/etc/dbt-dag.png?raw=true)
+![dbt dag](https://raw.githubusercontent.com/dbt-labs/dbt/6c6649f9129d5d108aa3b0526f634cd8f3a9d1ed/etc/dbt-dag.png)
 
 ## Getting started
 
--   [Install dbt](https://docs.getdbt.com/docs/installation)
--   Read the [documentation](https://docs.getdbt.com/).
--   Productionize your dbt project with [dbt Cloud](https://www.getdbt.com)
+- [Install dbt](https://docs.getdbt.com/docs/installation)
+- Read the [introduction](https://docs.getdbt.com/docs/introduction/) and [viewpoint](https://docs.getdbt.com/docs/about/viewpoint/)
 
-## Find out more
+## Join the dbt Community
 
--   Check out the [Introduction to dbt](https://dbt.readme.io/docs/introduction).
--   Read the [dbt Viewpoint](https://dbt.readme.io/docs/viewpoint).
-
-## Join thousands of analysts in the dbt community
-
--   Join the [chat](http://slack.getdbt.com/) on Slack.
--   Find community posts on [dbt Discourse](https://discourse.getdbt.com).
+- Be part of the conversation in the [dbt Community Slack](http://community.getdbt.com/)
+- Read more on the [dbt Community Discourse](https://discourse.getdbt.com)
 
 ## Reporting bugs and contributing code
 
--   Want to report a bug or request a feature? Let us know on [Slack](http://slack.getdbt.com/), or open [an issue](https://github.com/fishtown-analytics/dbt/issues/new).
--   Want to help us build dbt? Check out the [Contributing Getting Started Guide](/CONTRIBUTING.md)
+- Want to report a bug or request a feature? Let us know on [Slack](http://community.getdbt.com/), or open [an issue](https://github.com/dbt-labs/dbt/issues/new)
+- Want to help us build dbt? Check out the [Contributing Guide](https://github.com/dbt-labs/dbt/blob/HEAD/CONTRIBUTING.md)
 
 ## Code of Conduct
 
-Everyone interacting in the dbt project's codebases, issue trackers, chat rooms, and mailing lists is expected to follow the [PyPA Code of Conduct](https://www.pypa.io/en/latest/code-of-conduct/).
+Everyone interacting in the dbt project's codebases, issue trackers, chat rooms, and mailing lists is expected to follow the [dbt Code of Conduct](https://community.getdbt.com/code-of-conduct).

RELEASE.md

@@ -1,92 +0,0 @@
-### Release Procedure :shipit:
-
-#### Branching Strategy
-
-dbt has three types of branches:
-
-- **Trunks** track the latest release of a minor version of dbt. Historically, we used the `master` branch as the trunk. Each minor version release has a corresponding trunk. For example, the `0.11.x` series of releases has a branch called `0.11.latest`. This allows us to release new patch versions under `0.11` without necessarily needing to pull them into the latest version of dbt.
-- **Release Branches** track a specific, not yet complete release of dbt. These releases are codenamed since we don't always know what their semantic version will be. Example: `dev/lucretia-mott` became `0.11.1`.
-- **Feature Branches** track individual features and fixes. On completion they should be merged into a release branch.
-
-#### Git & PyPI
-
-1. Update CHANGELOG.md with the most recent changes
-2. If this is a release candidate, you want to create it off of your release branch. If it's an actual release, you must first merge to a master branch. Open a Pull Request in Github to merge it into the appropriate trunk (`X.X.latest`)
-3. Bump the version using `bumpversion`:
-  - Dry run first by running `bumpversion --new-version <desired-version> <part>` and checking the diff. If it looks correct, clean up the chanages and move on:
-  - Alpha releases: `bumpversion --commit --no-tag --new-version 0.10.2a1 num`
-  - Patch releases: `bumpversion --commit --no-tag --new-version 0.10.2 patch`
-  - Minor releases: `bumpversion --commit --no-tag --new-version 0.11.0 minor`
-  - Major releases: `bumpversion --commit --no-tag --new-version 1.0.0 major`
-4. (If this is a not a release candidate) Merge to `x.x.latest` and (optionally) `master`.
-5. Update the default branch to the next dev release branch.
-6. Build source distributions for all packages by running `./scripts/build-sdists.sh`. Note that this will clean out your `dist/` folder, so if you have important stuff in there, don't run it!!!
-7. Deploy to pypi
-  - `twine upload dist/*`
-8. Deploy to homebrew (see below)
-9. Deploy to conda-forge (see below)
-10. Git release notes (points to changelog)
-11. Post to slack (point to changelog)
-
-After releasing a new version, it's important to merge the changes back into the other outstanding release branches. This avoids merge conflicts moving forward.
-
-In some cases, where the branches have diverged wildly, it's ok to skip this step. But this means that the changes you just released won't be included in future releases.
-
-#### Homebrew Release Process
-
-1. Clone the `homebrew-dbt` repository:
-
-```
-git clone git@github.com:fishtown-analytics/homebrew-dbt.git
-```
-
-2. For ALL releases (prereleases and version releases), copy the relevant formula. To copy from the latest version release of dbt, do:
-
-```bash
-cp Formula/dbt.rb Formula/dbt@{NEW-VERSION}.rb
-```
-
-To copy from a different version, simply copy the corresponding file.
-
-3. Open the file, and edit the following:
-- the name of the ruby class: this is important, homebrew won't function properly if the class name is wrong. Check historical versions to figure out the right name.
-- under the `bottle` section, remove all of the hashes (lines starting with `sha256`)
-
-4. Create a **Python 3.7** virtualenv, activate it, and then install two packages: `homebrew-pypi-poet`, and the version of dbt you are preparing. I use:
-
-```
-pyenv virtualenv 3.7.0 homebrew-dbt-{VERSION}
-pyenv activate homebrew-dbt-{VERSION}
-pip install dbt=={VERSION} homebrew-pypi-poet
-```
-
-homebrew-pypi-poet is a program that generates a valid homebrew formula for an installed pip package. You want to use it to generate a diff against the existing formula. Then you want to apply the diff for the dependency packages only -- e.g. it will tell you that `google-api-core` has been updated and that you need to use the latest version.
-
-5. reinstall, test, and audit dbt. if the test or audit fails, fix the formula with step 1.
-
-```bash
-brew uninstall --force Formula/{YOUR-FILE}.rb
-brew install Formula/{YOUR-FILE}.rb
-brew test dbt
-brew audit --strict dbt
-```
-
-6. Ask Connor to bottle the change (only his laptop can do it!)
-
-#### Conda Forge Release Process
-
-1. Clone the fork of `conda-forge/dbt-feedstock` [here](https://github.com/fishtown-analytics/dbt-feedstock)
-```bash
-git clone git@github.com:fishtown-analytics/dbt-feedstock.git
-
-```
-2. Update the version and sha256 in `recipe/meta.yml`. To calculate the sha256, run:
-
-```bash
-wget https://github.com/fishtown-analytics/dbt/archive/v{version}.tar.gz
-openssl sha256 v{version}.tar.gz
-```
-
-3. Push the changes and create a PR against `conda-forge/dbt-feedstock`
-
-4. Confirm that all automated conda-forge tests are passing

azure-pipelines.yml

@@ -1,158 +0,0 @@
-# Python package
-# Create and test a Python package on multiple Python versions.
-# Add steps that analyze code, save the dist with the build record, publish to a PyPI-compatible index, and more:
-# https://docs.microsoft.com/azure/devops/pipelines/languages/python
-
-trigger:
-  branches:
-    include:
-    - master
-    - dev/*
-    - pr/*
-
-jobs:
-- job: UnitTest
-  pool:
-    vmImage: 'vs2017-win2016'
-  steps:
-  - task: UsePythonVersion@0
-    inputs:
-      versionSpec: '3.7'
-      architecture: 'x64'
-
-  - script: python -m pip install --upgrade pip && pip install tox
-    displayName: 'Install dependencies'
-
-  - script: python -m tox -e pywin-unit
-    displayName: Run unit tests
-
-- job: PostgresIntegrationTest
-  pool:
-    vmImage: 'vs2017-win2016'
-  dependsOn: UnitTest
-
-  steps:
-  - pwsh: echo "True"
-    displayName: Skip everything
-  - pwsh: |
-      choco install postgresql --params '/Password:password' --params-global --version 10.6
-      Set-Content "c:\program files\postgresql\10\data\pg_hba.conf" "host   all   all   ::1/128        trust"
-      Add-Content "c:\program files\postgresql\10\data\pg_hba.conf" "host   all   all   127.0.0.1/32   trust"
-      # the service name is "postgresql-x64-10", conveniently it's both the display name and the actual name
-      Restart-Service postgresql-x64-10
-
-      & "C:\program files\postgresql\10\bin\createdb.exe" -U postgres dbt
-      & "C:\program files\postgresql\10\bin\psql.exe"  -U postgres -c "CREATE ROLE root WITH PASSWORD 'password';"
-      & "C:\program files\postgresql\10\bin\psql.exe"  -U postgres -c "ALTER ROLE root WITH LOGIN;"
-      & "C:\program files\postgresql\10\bin\psql.exe"  -U postgres -c "GRANT CREATE, CONNECT ON DATABASE dbt TO root WITH GRANT OPTION;"
-      & "C:\program files\postgresql\10\bin\psql.exe"  -U postgres -c "CREATE ROLE noaccess WITH PASSWORD 'password' NOSUPERUSER;"
-      & "C:\program files\postgresql\10\bin\psql.exe"  -U postgres -c "ALTER ROLE noaccess WITH LOGIN;"
-      & "C:\program files\postgresql\10\bin\psql.exe"  -U postgres -c "GRANT CONNECT ON DATABASE dbt TO noaccess;"
-    displayName: Install postgresql and set up database
-
-  - task: UsePythonVersion@0
-    inputs:
-      versionSpec: '3.7'
-      architecture: 'x64'
-
-  - script: python -m pip install --upgrade pip && pip install tox
-    displayName: 'Install dependencies'
-
-  - script: python -m tox -e pywin-postgres
-    displayName: Run integration tests
-
-# These three are all similar except secure environment variables, which MUST be passed along to their tasks,
-# but there's probably a better way to do this!
-- job: SnowflakeIntegrationTest
-  pool:
-    vmImage: 'vs2017-win2016'
-  dependsOn: PostgresIntegrationTest
-  condition: and(succeeded(), ne(variables['SYSTEM.PULLREQUEST.ISFORK'], 'true'))
-  steps:
-  - task: UsePythonVersion@0
-    inputs:
-      versionSpec: '3.7'
-      architecture: 'x64'
-
-  - script: python -m pip install --upgrade pip && pip install tox
-    displayName: 'Install dependencies'
-
-  - script: python -m tox -e pywin-snowflake
-    env:
-      SNOWFLAKE_TEST_ACCOUNT: $(SNOWFLAKE_TEST_ACCOUNT)
-      SNOWFLAKE_TEST_PASSWORD: $(SNOWFLAKE_TEST_PASSWORD)
-      SNOWFLAKE_TEST_USER: $(SNOWFLAKE_TEST_USER)
-      SNOWFLAKE_TEST_WAREHOUSE: $(SNOWFLAKE_TEST_WAREHOUSE)
-      SNOWFLAKE_TEST_OAUTH_REFRESH_TOKEN: $(SNOWFLAKE_TEST_OAUTH_REFRESH_TOKEN)
-      SNOWFLAKE_TEST_OAUTH_CLIENT_ID: $(SNOWFLAKE_TEST_OAUTH_CLIENT_ID)
-      SNOWFLAKE_TEST_OAUTH_CLIENT_SECRET: $(SNOWFLAKE_TEST_OAUTH_CLIENT_SECRET)
-    displayName: Run integration tests
-
-- job: BigQueryIntegrationTest
-  pool:
-    vmImage: 'vs2017-win2016'
-  dependsOn: PostgresIntegrationTest
-  condition: and(succeeded(), ne(variables['SYSTEM.PULLREQUEST.ISFORK'], 'true'))
-  steps:
-  - task: UsePythonVersion@0
-    inputs:
-      versionSpec: '3.7'
-      architecture: 'x64'
-  - script: python -m pip install --upgrade pip && pip install tox
-    displayName: 'Install dependencies'
-  - script: python -m tox -e pywin-bigquery
-    env:
-      BIGQUERY_SERVICE_ACCOUNT_JSON: $(BIGQUERY_SERVICE_ACCOUNT_JSON)
-    displayName: Run integration tests
-
-- job: RedshiftIntegrationTest
-  pool:
-    vmImage: 'vs2017-win2016'
-  dependsOn: PostgresIntegrationTest
-  condition: and(succeeded(), ne(variables['SYSTEM.PULLREQUEST.ISFORK'], 'true'))
-  steps:
-  - task: UsePythonVersion@0
-    inputs:
-      versionSpec: '3.7'
-      architecture: 'x64'
-
-  - script: python -m pip install --upgrade pip && pip install tox
-    displayName: 'Install dependencies'
-
-  - script: python -m tox -e pywin-redshift
-    env:
-      REDSHIFT_TEST_DBNAME: $(REDSHIFT_TEST_DBNAME)
-      REDSHIFT_TEST_PASS: $(REDSHIFT_TEST_PASS)
-      REDSHIFT_TEST_USER: $(REDSHIFT_TEST_USER)
-      REDSHIFT_TEST_PORT: $(REDSHIFT_TEST_PORT)
-      REDSHIFT_TEST_HOST: $(REDSHIFT_TEST_HOST)
-    displayName: Run integration tests
-
-- job: BuildWheel
-  pool:
-    vmImage: 'vs2017-win2016'
-  dependsOn:
-    - UnitTest
-    - PostgresIntegrationTest
-    - RedshiftIntegrationTest
-    - SnowflakeIntegrationTest
-    - BigQueryIntegrationTest
-  condition: and(succeeded(), ne(variables['SYSTEM.PULLREQUEST.ISFORK'], 'true'))
-  steps:
-    - task: UsePythonVersion@0
-      inputs:
-        versionSpec: '3.7'
-        architecture: 'x64'
-    - script: python -m pip install --upgrade pip setuptools && python -m pip install -r requirements.txt && python -m pip install -r dev_requirements.txt
-      displayName: Install dependencies
-    - task: ShellScript@2
-      inputs:
-        scriptPath: scripts/build-wheels.sh
-    - task: CopyFiles@2
-      inputs:
-        contents: 'dist\?(*.whl|*.tar.gz)'
-        TargetFolder: '$(Build.ArtifactStagingDirectory)'
-    - task: PublishBuildArtifacts@1
-      inputs:
-        pathtoPublish: '$(Build.ArtifactStagingDirectory)'
-        artifactName: dists

converter.py

@@ -1,73 +0,0 @@
-#!/usr/bin/env python
-import json
-import yaml
-import sys
-import argparse
-from datetime import datetime, timezone
-import dbt.clients.registry as registry
-
-
-def yaml_type(fname):
-    with open(fname) as f:
-        return yaml.load(f)
-
-
-def parse_args():
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--project", type=yaml_type, default="dbt_project.yml")
-    parser.add_argument("--namespace", required=True)
-    return parser.parse_args()
-
-
-def get_full_name(args):
-    return "{}/{}".format(args.namespace, args.project["name"])
-
-
-def init_project_in_packages(args, packages):
-    full_name = get_full_name(args)
-    if full_name not in packages:
-        packages[full_name] = {
-            "name": args.project["name"],
-            "namespace": args.namespace,
-            "latest": args.project["version"],
-            "assets": {},
-            "versions": {},
-        }
-    return packages[full_name]
-
-
-def add_version_to_package(args, project_json):
-    project_json["versions"][args.project["version"]] = {
-        "id": "{}/{}".format(get_full_name(args), args.project["version"]),
-        "name": args.project["name"],
-        "version": args.project["version"],
-        "description": "",
-        "published_at": datetime.now(timezone.utc).astimezone().isoformat(),
-        "packages": args.project.get("packages") or [],
-        "works_with": [],
-        "_source": {
-            "type": "github",
-            "url": "",
-            "readme": "",
-        },
-        "downloads": {
-            "tarball": "",
-            "format": "tgz",
-            "sha1": "",
-        },
-    }
-
-
-def main():
-    args = parse_args()
-    packages = registry.packages()
-    project_json = init_project_in_packages(args, packages)
-    if args.project["version"] in project_json["versions"]:
-        raise Exception("Version {} already in packages JSON"
-                        .format(args.project["version"]),
-                        file=sys.stderr)
-    add_version_to_package(args, project_json)
-    print(json.dumps(packages, indent=2))
-
-if __name__ == "__main__":
-    main()

core/MANIFEST.in

@@ -1 +1 @@
-recursive-include dbt/include *.py *.sql *.yml *.html *.md
+recursive-include dbt/include *.py *.sql *.yml *.html *.md .gitkeep .gitignore

core/dbt/adapters/base/__init__.py

@@ -4,7 +4,11 @@
 from dbt.contracts.connection import Credentials  # noqa
 from dbt.adapters.base.meta import available  # noqa
 from dbt.adapters.base.connections import BaseConnectionManager  # noqa
-from dbt.adapters.base.relation import BaseRelation, RelationType  # noqa
+from dbt.adapters.base.relation import (  # noqa
+    BaseRelation,
+    RelationType,
+    SchemaSearchMap,
+)
 from dbt.adapters.base.column import Column  # noqa
-from dbt.adapters.base.impl import BaseAdapter  # noqa
+from dbt.adapters.base.impl import AdapterConfig, BaseAdapter  # noqa
 from dbt.adapters.base.plugin import AdapterPlugin  # noqa

core/dbt/adapters/base/column.py

@@ -1,12 +1,12 @@
 from dataclasses import dataclass
-
-from hologram import JsonSchemaMixin
-
+import re
 from typing import Dict, ClassVar, Any, Optional
 
+from dbt.exceptions import RuntimeException
+
 
 @dataclass
-class Column(JsonSchemaMixin):
+class Column:
     TYPE_LABELS: ClassVar[Dict[str, str]] = {
         'STRING': 'TEXT',
         'TIMESTAMP': 'TIMESTAMP',
@@ -74,7 +74,7 @@ class Column(JsonSchemaMixin):
 
     def string_size(self) -> int:
         if not self.is_string():
-            raise RuntimeError("Called string_size() on non-string field!")
+            raise RuntimeException("Called string_size() on non-string field!")
 
         if self.dtype == 'text' or self.char_size is None:
             # char_size should never be None. Handle it reasonably just in case
@@ -108,3 +108,46 @@ class Column(JsonSchemaMixin):
 
     def __repr__(self) -> str:
         return "<Column {} ({})>".format(self.name, self.data_type)
+
+    @classmethod
+    def from_description(cls, name: str, raw_data_type: str) -> 'Column':
+        match = re.match(r'([^(]+)(\([^)]+\))?', raw_data_type)
+        if match is None:
+            raise RuntimeException(
+                f'Could not interpret data type "{raw_data_type}"'
+            )
+        data_type, size_info = match.groups()
+        char_size = None
+        numeric_precision = None
+        numeric_scale = None
+        if size_info is not None:
+            # strip out the parentheses
+            size_info = size_info[1:-1]
+            parts = size_info.split(',')
+            if len(parts) == 1:
+                try:
+                    char_size = int(parts[0])
+                except ValueError:
+                    raise RuntimeException(
+                        f'Could not interpret data_type "{raw_data_type}": '
+                        f'could not convert "{parts[0]}" to an integer'
+                    )
+            elif len(parts) == 2:
+                try:
+                    numeric_precision = int(parts[0])
+                except ValueError:
+                    raise RuntimeException(
+                        f'Could not interpret data_type "{raw_data_type}": '
+                        f'could not convert "{parts[0]}" to an integer'
+                    )
+                try:
+                    numeric_scale = int(parts[1])
+                except ValueError:
+                    raise RuntimeException(
+                        f'Could not interpret data_type "{raw_data_type}": '
+                        f'could not convert "{parts[1]}" to an integer'
+                    )
+
+        return cls(
+            name, data_type, char_size, numeric_precision, numeric_scale
+        )

core/dbt/adapters/base/connections.py

@@ -4,21 +4,22 @@ import os
 from multiprocessing.synchronize import RLock
 from threading import get_ident
 from typing import (
-    Dict, Tuple, Hashable, Optional, ContextManager, List
+    Dict, Tuple, Hashable, Optional, ContextManager, List, Union
 )
 
 import agate
 
 import dbt.exceptions
-import dbt.flags
 from dbt.contracts.connection import (
-    Connection, Identifier, ConnectionState, AdapterRequiredConfig, LazyHandle
+    Connection, Identifier, ConnectionState,
+    AdapterRequiredConfig, LazyHandle, AdapterResponse
 )
 from dbt.contracts.graph.manifest import Manifest
 from dbt.adapters.base.query_headers import (
     MacroQueryStringSetter,
 )
 from dbt.logger import GLOBAL_LOGGER as logger
+from dbt import flags
 
 
 class BaseConnectionManager(metaclass=abc.ABCMeta):
@@ -39,7 +40,7 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
     def __init__(self, profile: AdapterRequiredConfig):
         self.profile = profile
         self.thread_connections: Dict[Hashable, Connection] = {}
-        self.lock: RLock = dbt.flags.MP_CONTEXT.RLock()
+        self.lock: RLock = flags.MP_CONTEXT.RLock()
         self.query_header: Optional[MacroQueryStringSetter] = None
 
     def set_query_header(self, manifest: Manifest) -> None:
@@ -60,7 +61,7 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
                 )
             return self.thread_connections[key]
 
-    def set_thread_connection(self, conn):
+    def set_thread_connection(self, conn: Connection) -> None:
         key = self.get_thread_identifier()
         if key in self.thread_connections:
             raise dbt.exceptions.InternalException(
@@ -88,6 +89,11 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
             self.begin()
             self.commit()
 
+    def rollback_if_open(self) -> None:
+        conn = self.get_if_exists()
+        if conn is not None and conn.handle and conn.transaction_open:
+            self._rollback(conn)
+
     @abc.abstractmethod
     def exception_handler(self, sql: str) -> ContextManager:
         """Create a context manager that handles exceptions caused by database
@@ -139,10 +145,6 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
                 .format(conn.name)
             )
         else:
-            logger.debug(
-                'Opening a new connection, currently in state {}'
-                .format(conn.state)
-            )
             conn.handle = LazyHandle(self.open)
 
         conn.name = conn_name
@@ -176,11 +178,9 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
                 return
 
         try:
-            if conn.state == 'open':
-                if conn.transaction_open is True:
-                    self._rollback(conn)
-            else:
-                self.close(conn)
+            # always close the connection. close() calls _rollback() if there
+            # is an open transaction
+            self.close(conn)
         except Exception:
             # if rollback or close failed, remove our busted connection
             self.clear_thread_connection()
@@ -230,16 +230,15 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
         """Perform the actual close operation."""
         # On windows, sometimes connection handles don't have a close() attr.
         if hasattr(connection.handle, 'close'):
-            logger.debug('On {}: Close'.format(connection.name))
+            logger.debug(f'On {connection.name}: Close')
             connection.handle.close()
         else:
-            logger.debug('On {}: No close available on handle'
-                         .format(connection.name))
+            logger.debug(f'On {connection.name}: No close available on handle')
 
     @classmethod
     def _rollback(cls, connection: Connection) -> None:
         """Roll back the given connection."""
-        if dbt.flags.STRICT_MODE:
+        if flags.STRICT_MODE:
             if not isinstance(connection, Connection):
                 raise dbt.exceptions.CompilerException(
                     f'In _rollback, got {connection} - not a Connection!'
@@ -247,17 +246,18 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
 
         if connection.transaction_open is False:
             raise dbt.exceptions.InternalException(
-                'Tried to rollback transaction on connection "{}", but '
-                'it does not have one open!'.format(connection.name))
+                f'Tried to rollback transaction on connection '
+                f'"{connection.name}", but it does not have one open!'
+            )
 
-        logger.debug('On {}: ROLLBACK'.format(connection.name))
+        logger.debug(f'On {connection.name}: ROLLBACK')
         cls._rollback_handle(connection)
 
         connection.transaction_open = False
 
     @classmethod
     def close(cls, connection: Connection) -> Connection:
-        if dbt.flags.STRICT_MODE:
+        if flags.STRICT_MODE:
             if not isinstance(connection, Connection):
                 raise dbt.exceptions.CompilerException(
                     f'In close, got {connection} - not a Connection!'
@@ -268,6 +268,7 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
             return connection
 
         if connection.transaction_open and connection.handle:
+            logger.debug('On {}: ROLLBACK'.format(connection.name))
             cls._rollback_handle(connection)
         connection.transaction_open = False
 
@@ -290,7 +291,7 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
     @abc.abstractmethod
     def execute(
         self, sql: str, auto_begin: bool = False, fetch: bool = False
-    ) -> Tuple[str, agate.Table]:
+    ) -> Tuple[Union[str, AdapterResponse], agate.Table]:
         """Execute the given SQL.
 
         :param str sql: The sql to execute.
@@ -298,7 +299,7 @@ class BaseConnectionManager(metaclass=abc.ABCMeta):
             transaction, automatically begin one.
         :param bool fetch: If set, fetch results.
         :return: A tuple of the status and the results (empty if fetch=False).
-        :rtype: Tuple[str, agate.Table]
+        :rtype: Tuple[Union[str, AdapterResponse], agate.Table]
         """
         raise dbt.exceptions.NotImplementedException(
             '`execute` is not implemented for this adapter!'

core/dbt/adapters/base/impl.py

@@ -1,11 +1,11 @@
 import abc
-from concurrent.futures import ThreadPoolExecutor, as_completed
-from concurrent.futures import Future  # noqa - we use this for typing only
+from concurrent.futures import as_completed, Future
 from contextlib import contextmanager
 from datetime import datetime
+from itertools import chain
 from typing import (
-    Optional, Tuple, Callable, Container, FrozenSet, Type, Dict, Any, List,
-    Mapping, Iterator, Union, Set
+    Optional, Tuple, Callable, Iterable, Type, Dict, Any, List, Mapping,
+    Iterator, Union, Set
 )
 
 import agate
@@ -16,23 +16,28 @@ from dbt.exceptions import (
     get_relation_returned_multiple_results,
     InternalException, NotImplementedException, RuntimeException,
 )
-import dbt.flags
+from dbt import flags
 
 from dbt import deprecations
-from dbt.clients.agate_helper import empty_table, merge_tables
+from dbt.adapters.protocol import (
+    AdapterConfig,
+    ConnectionManagerProtocol,
+)
+from dbt.clients.agate_helper import empty_table, merge_tables, table_from_rows
 from dbt.clients.jinja import MacroGenerator
-from dbt.contracts.graph.compiled import CompileResultNode, CompiledSeedNode
-from dbt.contracts.graph.manifest import Manifest
+from dbt.contracts.graph.compiled import (
+    CompileResultNode, CompiledSeedNode
+)
+from dbt.contracts.graph.manifest import Manifest, MacroManifest
 from dbt.contracts.graph.parsed import ParsedSeedNode
 from dbt.exceptions import warn_or_error
-from dbt.node_types import NodeType
 from dbt.logger import GLOBAL_LOGGER as logger
-from dbt.utils import filter_null_values
+from dbt.utils import filter_null_values, executor
 
-from dbt.adapters.base.connections import BaseConnectionManager, Connection
+from dbt.adapters.base.connections import Connection, AdapterResponse
 from dbt.adapters.base.meta import AdapterMeta, available
 from dbt.adapters.base.relation import (
-    ComponentName, BaseRelation, InformationSchema
+    ComponentName, BaseRelation, InformationSchema, SchemaSearchMap
 )
 from dbt.adapters.base import Column as BaseColumn
 from dbt.adapters.cache import RelationsCache
@@ -54,21 +59,6 @@ def _expect_row_value(key: str, row: agate.Row):
     return row[key]
 
 
-def _relations_filter_schemas(
-    schemas: Container[str]
-) -> Callable[[agate.Row], bool]:
-    def test(row):
-        referenced_schema = _expect_row_value('referenced_schema', row)
-        dependent_schema = _expect_row_value('dependent_schema', row)
-        # handle the null schema
-        if referenced_schema is not None:
-            referenced_schema = referenced_schema.lower()
-        if dependent_schema is not None:
-            dependent_schema = dependent_schema.lower()
-        return referenced_schema in schemas or dependent_schema in schemas
-    return test
-
-
 def _catalog_filter_schemas(manifest: Manifest) -> Callable[[agate.Row], bool]:
     """Return a function that takes a row and decides if the row should be
     included in the catalog output.
@@ -121,55 +111,6 @@ def _relation_name(rel: Optional[BaseRelation]) -> str:
         return str(rel)
 
 
-class SchemaSearchMap(Dict[InformationSchema, Set[Optional[str]]]):
-    """A utility class to keep track of what information_schema tables to
-    search for what schemas
-    """
-    def add(self, relation: BaseRelation):
-        key = relation.information_schema_only()
-        if key not in self:
-            self[key] = set()
-        lowered: Optional[str] = None
-        if relation.schema is not None:
-            lowered = relation.schema.lower()
-        self[key].add(lowered)
-
-    def search(self):
-        for information_schema_name, schemas in self.items():
-            for schema in schemas:
-                yield information_schema_name, schema
-
-    def schemas_searched(self):
-        result: Set[Tuple[str, str]] = set()
-        for information_schema_name, schemas in self.items():
-            result.update(
-                (information_schema_name.database, schema)
-                for schema in schemas
-            )
-        return result
-
-    def flatten(self):
-        new = self.__class__()
-
-        # make sure we don't have duplicates
-        seen = {r.database.lower() for r in self if r.database}
-        if len(seen) > 1:
-            raise_compiler_error(str(seen))
-
-        for information_schema_name, schema in self.search():
-            path = {
-                'database': information_schema_name.database,
-                'schema': schema
-            }
-            new.add(information_schema_name.incorporate(
-                path=path,
-                quote_policy={'database': False},
-                include_policy={'database': False},
-            ))
-
-        return new
-
-
 class BaseAdapter(metaclass=AdapterMeta):
     """The BaseAdapter provides an abstract base class for adapters.
 
@@ -208,17 +149,17 @@ class BaseAdapter(metaclass=AdapterMeta):
     """
     Relation: Type[BaseRelation] = BaseRelation
     Column: Type[BaseColumn] = BaseColumn
-    ConnectionManager: Type[BaseConnectionManager]
+    ConnectionManager: Type[ConnectionManagerProtocol]
 
     # A set of clobber config fields accepted by this adapter
     # for use in materializations
-    AdapterSpecificConfigs: FrozenSet[str] = frozenset()
+    AdapterSpecificConfigs: Type[AdapterConfig] = AdapterConfig
 
     def __init__(self, config):
         self.config = config
         self.cache = RelationsCache()
         self.connections = self.ConnectionManager(config)
-        self._internal_manifest_lazy: Optional[Manifest] = None
+        self._macro_manifest_lazy: Optional[MacroManifest] = None
 
     ###
     # Methods that pass through to the connection manager
@@ -238,6 +179,9 @@ class BaseAdapter(metaclass=AdapterMeta):
     def commit_if_has_connection(self) -> None:
         self.connections.commit_if_has_connection()
 
+    def debug_query(self) -> None:
+        self.execute('select 1 as id')
+
     def nice_connection_name(self) -> str:
         conn = self.connections.get_if_exists()
         if conn is None or conn.name is None:
@@ -268,7 +212,7 @@ class BaseAdapter(metaclass=AdapterMeta):
     @available.parse(lambda *a, **k: ('', empty_table()))
     def execute(
         self, sql: str, auto_begin: bool = False, fetch: bool = False
-    ) -> Tuple[str, agate.Table]:
+    ) -> Tuple[Union[str, AdapterResponse], agate.Table]:
         """Execute the given SQL. This is a thin wrapper around
         ConnectionManager.execute.
 
@@ -277,7 +221,7 @@ class BaseAdapter(metaclass=AdapterMeta):
             transaction, automatically begin one.
         :param bool fetch: If set, fetch results.
         :return: A tuple of the status and the results (empty if fetch=False).
-        :rtype: Tuple[str, agate.Table]
+        :rtype: Tuple[Union[str, AdapterResponse], agate.Table]
         """
         return self.connections.execute(
             sql=sql,
@@ -285,6 +229,21 @@ class BaseAdapter(metaclass=AdapterMeta):
             fetch=fetch
         )
 
+    @available.parse(lambda *a, **k: ('', empty_table()))
+    def get_partitions_metadata(
+        self, table: str
+    ) -> Tuple[agate.Table]:
+        """Obtain partitions metadata for a BigQuery partitioned table.
+
+        :param str table_id: a partitioned table id, in standard SQL format.
+        :return: a partition metadata tuple, as described in
+            https://cloud.google.com/bigquery/docs/creating-partitioned-tables#getting_partition_metadata_using_meta_tables.
+        :rtype: agate.Table
+        """
+        return self.connections.get_partitions_metadata(
+            table=table
+        )
+
     ###
     # Methods that should never be overridden
     ###
@@ -299,32 +258,38 @@ class BaseAdapter(metaclass=AdapterMeta):
         return cls.ConnectionManager.TYPE
 
     @property
-    def _internal_manifest(self) -> Manifest:
-        if self._internal_manifest_lazy is None:
-            return self.load_internal_manifest()
-        return self._internal_manifest_lazy
+    def _macro_manifest(self) -> MacroManifest:
+        if self._macro_manifest_lazy is None:
+            return self.load_macro_manifest()
+        return self._macro_manifest_lazy
 
-    def check_internal_manifest(self) -> Optional[Manifest]:
+    def check_macro_manifest(self) -> Optional[MacroManifest]:
         """Return the internal manifest (used for executing macros) if it's
         been initialized, otherwise return None.
         """
-        return self._internal_manifest_lazy
+        return self._macro_manifest_lazy
 
-    def load_internal_manifest(self) -> Manifest:
-        if self._internal_manifest_lazy is None:
+    def load_macro_manifest(self) -> MacroManifest:
+        if self._macro_manifest_lazy is None:
             # avoid a circular import
-            from dbt.parser.manifest import load_internal_manifest
-            manifest = load_internal_manifest(self.config)
-            self._internal_manifest_lazy = manifest
-        return self._internal_manifest_lazy
+            from dbt.parser.manifest import ManifestLoader
+            manifest = ManifestLoader.load_macros(
+                self.config, self.connections.set_query_header
+            )
+            self._macro_manifest_lazy = manifest
+        return self._macro_manifest_lazy
+
+    def clear_macro_manifest(self):
+        if self._macro_manifest_lazy is not None:
+            self._macro_manifest_lazy = None
 
     ###
     # Caching methods
     ###
-    def _schema_is_cached(self, database: str, schema: str) -> bool:
+    def _schema_is_cached(self, database: Optional[str], schema: str) -> bool:
         """Check if the schema is cached, and by default logs if it is not."""
 
-        if dbt.flags.USE_CACHE is False:
+        if flags.USE_CACHE is False:
             return False
         elif (database, schema) not in self.cache:
             logger.debug(
@@ -335,9 +300,20 @@ class BaseAdapter(metaclass=AdapterMeta):
         else:
             return True
 
-    def _get_cache_schemas(
-        self, manifest: Manifest, exec_only: bool = False
-    ) -> SchemaSearchMap:
+    def _get_cache_schemas(self, manifest: Manifest) -> Set[BaseRelation]:
+        """Get the set of schema relations that the cache logic needs to
+        populate. This means only executable nodes are included.
+        """
+        # the cache only cares about executable nodes
+        return {
+            self.Relation.create_from(self.config, node).without_identifier()
+            for node in manifest.nodes.values()
+            if (
+                node.is_relational and not node.is_ephemeral_model
+            )
+        }
+
+    def _get_catalog_schemas(self, manifest: Manifest) -> SchemaSearchMap:
         """Get a mapping of each node's "information_schema" relations to a
         set of all schemas expected in that information_schema.
 
@@ -347,9 +323,11 @@ class BaseAdapter(metaclass=AdapterMeta):
         lowercase strings.
         """
         info_schema_name_map = SchemaSearchMap()
-        for node in manifest.nodes.values():
-            if exec_only and node.resource_type not in NodeType.executable():
-                continue
+        nodes: Iterator[CompileResultNode] = chain(
+            manifest.nodes.values(),
+            manifest.sources.values(),
+        )
+        for node in nodes:
             relation = self.Relation.create_from(self.config, node)
             info_schema_name_map.add(relation)
         # result is a map whose keys are information_schema Relations without
@@ -362,19 +340,34 @@ class BaseAdapter(metaclass=AdapterMeta):
         """Populate the relations cache for the given schemas. Returns an
         iterable of the schemas populated, as strings.
         """
-        if not dbt.flags.USE_CACHE:
+        if not flags.USE_CACHE:
             return
 
-        info_schema_name_map = self._get_cache_schemas(manifest,
-                                                       exec_only=True)
-        for db, schema in info_schema_name_map.search():
-            for relation in self.list_relations_without_caching(db, schema):
-                self.cache.add(relation)
+        cache_schemas = self._get_cache_schemas(manifest)
+        with executor(self.config) as tpe:
+            futures: List[Future[List[BaseRelation]]] = []
+            for cache_schema in cache_schemas:
+                fut = tpe.submit_connected(
+                    self,
+                    f'list_{cache_schema.database}_{cache_schema.schema}',
+                    self.list_relations_without_caching,
+                    cache_schema
+                )
+                futures.append(fut)
+
+            for future in as_completed(futures):
+                # if we can't read the relations we need to just raise anyway,
+                # so just call future.result() and let that raise on failure
+                for relation in future.result():
+                    self.cache.add(relation)
 
         # it's possible that there were no relations in some schemas. We want
         # to insert the schemas we query into the cache's `.schemas` attribute
         # so we can check it later
-        self.cache.update_schemas(info_schema_name_map.schemas_searched())
+        cache_update: Set[Tuple[Optional[str], Optional[str]]] = set()
+        for relation in cache_schemas:
+            cache_update.add((relation.database, relation.schema))
+        self.cache.update_schemas(cache_update)
 
     def set_relations_cache(
         self, manifest: Manifest, clear: bool = False
@@ -382,7 +375,7 @@ class BaseAdapter(metaclass=AdapterMeta):
         """Run a query that gets a populated cache of the relations in the
         database and set the cache on this adapter.
         """
-        if not dbt.flags.USE_CACHE:
+        if not flags.USE_CACHE:
             return
 
         with self.cache.lock:
@@ -398,7 +391,7 @@ class BaseAdapter(metaclass=AdapterMeta):
             raise_compiler_error(
                 'Attempted to cache a null relation for {}'.format(name)
             )
-        if dbt.flags.USE_CACHE:
+        if flags.USE_CACHE:
             self.cache.add(relation)
         # so jinja doesn't render things
         return ''
@@ -413,7 +406,7 @@ class BaseAdapter(metaclass=AdapterMeta):
             raise_compiler_error(
                 'Attempted to drop a null relation for {}'.format(name)
             )
-        if dbt.flags.USE_CACHE:
+        if flags.USE_CACHE:
             self.cache.drop(relation)
         return ''
 
@@ -435,7 +428,7 @@ class BaseAdapter(metaclass=AdapterMeta):
                 .format(src_name, dst_name, name)
             )
 
-        if dbt.flags.USE_CACHE:
+        if flags.USE_CACHE:
             self.cache.rename(from_relation, to_relation)
         return ''
 
@@ -518,7 +511,7 @@ class BaseAdapter(metaclass=AdapterMeta):
     def get_columns_in_relation(
         self, relation: BaseRelation
     ) -> List[BaseColumn]:
-        """Get a list of the columns in the given Relation."""
+        """Get a list of the columns in the given Relation. """
         raise NotImplementedException(
             '`get_columns_in_relation` is not implemented for this adapter!'
         )
@@ -553,15 +546,14 @@ class BaseAdapter(metaclass=AdapterMeta):
 
     @abc.abstractmethod
     def list_relations_without_caching(
-        self, information_schema: BaseRelation, schema: str
+        self, schema_relation: BaseRelation
     ) -> List[BaseRelation]:
         """List relations in the given schema, bypassing the cache.
 
         This is used as the underlying behavior to fill the cache.
 
-        :param Relation information_schema: The information schema to list
-            relations from.
-        :param str schema: The name of the schema to list relations from.
+        :param schema_relation: A relation containing the database and schema
+            as appropraite for the underlying data warehouse
         :return: The relations in schema
         :rtype: List[self.Relation]
         """
@@ -673,21 +665,23 @@ class BaseAdapter(metaclass=AdapterMeta):
 
         self.expand_column_types(from_relation, to_relation)
 
-    def list_relations(self, database: str, schema: str) -> List[BaseRelation]:
+    def list_relations(
+        self, database: Optional[str], schema: str
+    ) -> List[BaseRelation]:
         if self._schema_is_cached(database, schema):
             return self.cache.get_relations(database, schema)
 
-        information_schema = self.Relation.create(
+        schema_relation = self.Relation.create(
             database=database,
             schema=schema,
             identifier='',
             quote_policy=self.config.quoting
-        ).information_schema()
+        ).without_identifier()
 
         # we can't build the relations cache because we don't have a
         # manifest so we can't run any operations.
         relations = self.list_relations_without_caching(
-            information_schema, schema
+            schema_relation
         )
 
         logger.debug('with database={}, schema={}, relations={}'
@@ -768,7 +762,7 @@ class BaseAdapter(metaclass=AdapterMeta):
     ###
     @abc.abstractmethod
     @available.parse_none
-    def create_schema(self, database: str, schema: str):
+    def create_schema(self, relation: BaseRelation):
         """Create the given schema if it does not exist."""
         raise NotImplementedException(
             '`create_schema` is not implemented for this adapter!'
@@ -776,7 +770,7 @@ class BaseAdapter(metaclass=AdapterMeta):
 
     @abc.abstractmethod
     @available.parse_none
-    def drop_schema(self, database: str, schema: str):
+    def drop_schema(self, relation: BaseRelation):
         """Drop the given schema (and everything in it) if it exists."""
         raise NotImplementedException(
             '`drop_schema` is not implemented for this adapter!'
@@ -951,6 +945,7 @@ class BaseAdapter(metaclass=AdapterMeta):
         context_override: Optional[Dict[str, Any]] = None,
         kwargs: Dict[str, Any] = None,
         release: bool = False,
+        text_only_columns: Optional[Iterable[str]] = None,
     ) -> agate.Table:
         """Look macro_name up in the manifest and execute its results.
 
@@ -963,15 +958,17 @@ class BaseAdapter(metaclass=AdapterMeta):
             execution context.
         :param kwargs: An optional dict of keyword args used to pass to the
             macro.
-        :param release: If True, release the connection after executing.
+        :param release: Ignored.
         """
+        if release is not False:
+            deprecations.warn('execute-macro-release')
         if kwargs is None:
             kwargs = {}
         if context_override is None:
             context_override = {}
 
         if manifest is None:
-            manifest = self._internal_manifest
+            manifest = self._macro_manifest
 
         macro = manifest.find_macro_by_name(
             macro_name, self.config.project_name, project
@@ -999,11 +996,8 @@ class BaseAdapter(metaclass=AdapterMeta):
 
         macro_function = MacroGenerator(macro, macro_context)
 
-        try:
+        with self.connections.exception_handler(f'macro {macro_name}'):
             result = macro_function(**kwargs)
-        finally:
-            if release:
-                self.release_connection()
         return result
 
     @classmethod
@@ -1013,6 +1007,12 @@ class BaseAdapter(metaclass=AdapterMeta):
         """Filter the table as appropriate for catalog entries. Subclasses can
         override this to change filtering rules on a per-adapter basis.
         """
+        # force database + schema to be strings
+        table = table_from_rows(
+            table.rows,
+            table.column_names,
+            text_only_columns=['table_database', 'table_schema', 'table_name']
+        )
         return table.where(_catalog_filter_schemas(manifest))
 
     def _get_one_catalog(
@@ -1022,24 +1022,17 @@ class BaseAdapter(metaclass=AdapterMeta):
         manifest: Manifest,
     ) -> agate.Table:
 
-        name = '.'.join([
-            str(information_schema.database),
-            'information_schema'
-        ])
-
-        with self.connection_named(name):
-            kwargs = {
-                'information_schema': information_schema,
-                'schemas': schemas
-            }
-            table = self.execute_macro(
-                GET_CATALOG_MACRO_NAME,
-                kwargs=kwargs,
-                release=True,
-                # pass in the full manifest so we get any local project
-                # overrides
-                manifest=manifest,
-            )
+        kwargs = {
+            'information_schema': information_schema,
+            'schemas': schemas
+        }
+        table = self.execute_macro(
+            GET_CATALOG_MACRO_NAME,
+            kwargs=kwargs,
+            # pass in the full manifest so we get any local project
+            # overrides
+            manifest=manifest,
+        )
 
         results = self._catalog_filter_table(table, manifest)
         return results
@@ -1047,15 +1040,24 @@ class BaseAdapter(metaclass=AdapterMeta):
     def get_catalog(
         self, manifest: Manifest
     ) -> Tuple[agate.Table, List[Exception]]:
-        # snowflake is super slow. split it out into the specified threads
-        num_threads = self.config.threads
-        schema_map = self._get_cache_schemas(manifest)
+        schema_map = self._get_catalog_schemas(manifest)
+
+        with executor(self.config) as tpe:
+            futures: List[Future[agate.Table]] = []
+            for info, schemas in schema_map.items():
+                if len(schemas) == 0:
+                    continue
+                name = '.'.join([
+                    str(info.database),
+                    'information_schema'
+                ])
+
+                fut = tpe.submit_connected(
+                    self, name,
+                    self._get_one_catalog, info, schemas, manifest
+                )
+                futures.append(fut)
 
-        with ThreadPoolExecutor(max_workers=num_threads) as executor:
-            futures = [
-                executor.submit(self._get_one_catalog, info, schemas, manifest)
-                for info, schemas in schema_map.items() if len(schemas) > 0
-            ]
             catalogs, exceptions = catch_as_completed(futures)
 
         return catalogs, exceptions
@@ -1082,7 +1084,6 @@ class BaseAdapter(metaclass=AdapterMeta):
         table = self.execute_macro(
             FRESHNESS_MACRO_NAME,
             kwargs=kwargs,
-            release=True,
             manifest=manifest
         )
         # now we have a 1-row table of the maximum `loaded_at_field` value and
@@ -1133,6 +1134,102 @@ class BaseAdapter(metaclass=AdapterMeta):
         """
         pass
 
+    def get_compiler(self):
+        from dbt.compilation import Compiler
+        return Compiler(self.config)
+
+    # Methods used in adapter tests
+    def update_column_sql(
+        self,
+        dst_name: str,
+        dst_column: str,
+        clause: str,
+        where_clause: Optional[str] = None,
+    ) -> str:
+        clause = f'update {dst_name} set {dst_column} = {clause}'
+        if where_clause is not None:
+            clause += f' where {where_clause}'
+        return clause
+
+    def timestamp_add_sql(
+        self, add_to: str, number: int = 1, interval: str = 'hour'
+    ) -> str:
+        # for backwards compatibility, we're compelled to set some sort of
+        # default. A lot of searching has lead me to believe that the
+        # '+ interval' syntax used in postgres/redshift is relatively common
+        # and might even be the SQL standard's intention.
+        return f"{add_to} + interval '{number} {interval}'"
+
+    def string_add_sql(
+        self, add_to: str, value: str, location='append',
+    ) -> str:
+        if location == 'append':
+            return f"{add_to} || '{value}'"
+        elif location == 'prepend':
+            return f"'{value}' || {add_to}"
+        else:
+            raise RuntimeException(
+                f'Got an unexpected location value of "{location}"'
+            )
+
+    def get_rows_different_sql(
+        self,
+        relation_a: BaseRelation,
+        relation_b: BaseRelation,
+        column_names: Optional[List[str]] = None,
+        except_operator: str = 'EXCEPT',
+    ) -> str:
+        """Generate SQL for a query that returns a single row with a two
+        columns: the number of rows that are different between the two
+        relations and the number of mismatched rows.
+        """
+        # This method only really exists for test reasons.
+        names: List[str]
+        if column_names is None:
+            columns = self.get_columns_in_relation(relation_a)
+            names = sorted((self.quote(c.name) for c in columns))
+        else:
+            names = sorted((self.quote(n) for n in column_names))
+        columns_csv = ', '.join(names)
+
+        sql = COLUMNS_EQUAL_SQL.format(
+            columns=columns_csv,
+            relation_a=str(relation_a),
+            relation_b=str(relation_b),
+            except_op=except_operator,
+        )
+
+        return sql
+
+
+COLUMNS_EQUAL_SQL = '''
+with diff_count as (
+    SELECT
+        1 as id,
+        COUNT(*) as num_missing FROM (
+            (SELECT {columns} FROM {relation_a} {except_op}
+             SELECT {columns} FROM {relation_b})
+             UNION ALL
+            (SELECT {columns} FROM {relation_b} {except_op}
+             SELECT {columns} FROM {relation_a})
+        ) as a
+), table_a as (
+    SELECT COUNT(*) as num_rows FROM {relation_a}
+), table_b as (
+    SELECT COUNT(*) as num_rows FROM {relation_b}
+), row_count_diff as (
+    select
+        1 as id,
+        table_a.num_rows - table_b.num_rows as difference
+    from table_a, table_b
+)
+select
+    row_count_diff.difference as row_count_difference,
+    diff_count.num_missing as num_mismatched
+from row_count_diff
+join diff_count using (id)
+'''.strip()
+
 
 def catch_as_completed(
     futures  # typing: List[Future[agate.Table]]

core/dbt/adapters/base/plugin.py

@@ -1,7 +1,19 @@
 from typing import List, Optional, Type
 
-from dbt.adapters.base import BaseAdapter, Credentials
+from dbt.adapters.base import Credentials
 from dbt.exceptions import CompilationException
+from dbt.adapters.protocol import AdapterProtocol
+
+
+def project_name_from_path(include_path: str) -> str:
+    # avoid an import cycle
+    from dbt.config.project import Project
+    partial = Project.partial_load(include_path)
+    if partial.project_name is None:
+        raise CompilationException(
+            f'Invalid project at {include_path}: name not set!'
+        )
+    return partial.project_name
 
 
 class AdapterPlugin:
@@ -13,23 +25,16 @@ class AdapterPlugin:
     """
     def __init__(
         self,
-        adapter: Type[BaseAdapter],
+        adapter: Type[AdapterProtocol],
         credentials: Type[Credentials],
         include_path: str,
         dependencies: Optional[List[str]] = None
     ):
-        # avoid an import cycle
-        from dbt.config.project import Project
 
-        self.adapter: Type[BaseAdapter] = adapter
+        self.adapter: Type[AdapterProtocol] = adapter
         self.credentials: Type[Credentials] = credentials
         self.include_path: str = include_path
-        partial = Project.partial_load(include_path)
-        if partial.project_name is None:
-            raise CompilationException(
-                f'Invalid project at {include_path}: name not set!'
-            )
-        self.project_name: str = partial.project_name
+        self.project_name: str = project_name_from_path(include_path)
         self.dependencies: List[str]
         if dependencies is None:
             self.dependencies = []

core/dbt/adapters/base/query_headers.py

@@ -3,32 +3,11 @@ from typing import Optional, Callable, Dict, Any
 
 from dbt.clients.jinja import QueryStringGenerator
 
-from dbt.context.configured import generate_query_header_context
-from dbt.contracts.connection import AdapterRequiredConfig
+from dbt.context.manifest import generate_query_header_context
+from dbt.contracts.connection import AdapterRequiredConfig, QueryComment
 from dbt.contracts.graph.compiled import CompileResultNode
 from dbt.contracts.graph.manifest import Manifest
 from dbt.exceptions import RuntimeException
-from dbt.helper_types import NoValue
-
-
-DEFAULT_QUERY_COMMENT = '''
-{%- set comment_dict = {} -%}
-{%- do comment_dict.update(
-    app='dbt',
-    dbt_version=dbt_version,
-    profile_name=target.get('profile_name'),
-    target_name=target.get('target_name'),
-) -%}
-{%- if node is not none -%}
-  {%- do comment_dict.update(
-    node_id=node.unique_id,
-  ) -%}
-{% else %}
-  {# in the node context, the connection name is the node_id #}
-  {%- do comment_dict.update(connection_name=connection_name) -%}
-{%- endif -%}
-{{ return(tojson(comment_dict)) }}
-'''
 
 
 class NodeWrapper:
@@ -47,14 +26,24 @@ class _QueryComment(local):
     """
     def __init__(self, initial):
         self.query_comment: Optional[str] = initial
+        self.append = False
 
     def add(self, sql: str) -> str:
         if not self.query_comment:
             return sql
-        else:
-            return '/* {} */\n{}'.format(self.query_comment.strip(), sql)
 
-    def set(self, comment: Optional[str]):
+        if self.append:
+            # replace last ';' with '<comment>;'
+            sql = sql.rstrip()
+            if sql[-1] == ';':
+                sql = sql[:-1]
+                return '{}\n/* {} */;'.format(sql, self.query_comment.strip())
+
+            return '{}\n/* {} */'.format(sql, self.query_comment.strip())
+
+        return '/* {} */\n{}'.format(self.query_comment.strip(), sql)
+
+    def set(self, comment: Optional[str], append: bool):
         if isinstance(comment, str) and '*/' in comment:
             # tell the user "no" so they don't hurt themselves by writing
             # garbage
@@ -62,6 +51,7 @@ class _QueryComment(local):
                 f'query comment contains illegal value "*/": {comment}'
             )
         self.query_comment = comment
+        self.append = append
 
 
 QueryStringFunc = Callable[[str, Optional[NodeWrapper]], str]
@@ -87,18 +77,8 @@ class MacroQueryStringSetter:
         self.comment = _QueryComment(None)
         self.reset()
 
-    def _get_comment_macro(self):
-        if (
-            self.config.query_comment != NoValue() and
-            self.config.query_comment
-        ):
-            return self.config.query_comment
-        # if the query comment is null/empty string, there is no comment at all
-        elif not self.config.query_comment:
-            return None
-        else:
-            # else, the default
-            return DEFAULT_QUERY_COMMENT
+    def _get_comment_macro(self) -> Optional[str]:
+        return self.config.query_comment.comment
 
     def _get_context(self) -> Dict[str, Any]:
         return generate_query_header_context(self.config, self.manifest)
@@ -114,4 +94,8 @@ class MacroQueryStringSetter:
         if node is not None:
             wrapped = NodeWrapper(node)
         comment_str = self.generator(name, wrapped)
-        self.comment.set(comment_str)
+
+        append = False
+        if isinstance(self.config.query_comment, QueryComment):
+            append = self.config.query_comment.append
+        self.comment.set(comment_str, append)

core/dbt/adapters/base/relation.py

@@ -1,131 +1,19 @@
-from dbt.utils import filter_null_values, deep_merge, classproperty
-from dbt.node_types import NodeType
-
-import dbt.exceptions
-
-from collections.abc import Mapping, Hashable
-from dataclasses import dataclass, fields
+from collections.abc import Hashable
+from dataclasses import dataclass
 from typing import (
-    Optional, TypeVar, Generic, Any, Type, Dict, Union, List, Iterator, Tuple
+    Optional, TypeVar, Any, Type, Dict, Union, Iterator, Tuple, Set
 )
-from typing_extensions import Protocol
 
-from hologram import JsonSchemaMixin
-from hologram.helpers import StrEnum
-
-from dbt.contracts.util import Replaceable
 from dbt.contracts.graph.compiled import CompiledNode
 from dbt.contracts.graph.parsed import ParsedSourceDefinition, ParsedNode
+from dbt.contracts.relation import (
+    RelationType, ComponentName, HasQuoting, FakeAPIObject, Policy, Path
+)
 from dbt.exceptions import InternalException
-from dbt import deprecations
+from dbt.node_types import NodeType
+from dbt.utils import filter_null_values, deep_merge, classproperty
 
-
-class RelationType(StrEnum):
-    Table = 'table'
-    View = 'view'
-    CTE = 'cte'
-    MaterializedView = 'materializedview'
-    External = 'external'
-
-
-class ComponentName(StrEnum):
-    Database = 'database'
-    Schema = 'schema'
-    Identifier = 'identifier'
-
-
-class HasQuoting(Protocol):
-    quoting: Dict[str, bool]
-
-
-class FakeAPIObject(JsonSchemaMixin, Replaceable, Mapping):
-    # override the mapping truthiness, len is always >1
-    def __bool__(self):
-        return True
-
-    def __getitem__(self, key):
-        try:
-            return getattr(self, key)
-        except AttributeError:
-            raise KeyError(key) from None
-
-    def __iter__(self):
-        deprecations.warn('not-a-dictionary', obj=self)
-        for _, name in self._get_fields():
-            yield name
-
-    def __len__(self):
-        deprecations.warn('not-a-dictionary', obj=self)
-        return len(fields(self.__class__))
-
-    def incorporate(self, **kwargs):
-        value = self.to_dict()
-        value = deep_merge(value, kwargs)
-        return self.from_dict(value)
-
-
-T = TypeVar('T')
-
-
-@dataclass
-class _ComponentObject(FakeAPIObject, Generic[T]):
-    database: T
-    schema: T
-    identifier: T
-
-    def get_part(self, key: ComponentName) -> T:
-        if key == ComponentName.Database:
-            return self.database
-        elif key == ComponentName.Schema:
-            return self.schema
-        elif key == ComponentName.Identifier:
-            return self.identifier
-        else:
-            raise ValueError(
-                'Got a key of {}, expected one of {}'
-                .format(key, list(ComponentName))
-            )
-
-    def replace_dict(self, dct: Dict[ComponentName, T]):
-        kwargs: Dict[str, T] = {}
-        for k, v in dct.items():
-            kwargs[str(k)] = v
-        return self.replace(**kwargs)
-
-
-@dataclass
-class Policy(_ComponentObject[bool]):
-    database: bool = True
-    schema: bool = True
-    identifier: bool = True
-
-
-@dataclass
-class Path(_ComponentObject[Optional[str]]):
-    database: Optional[str]
-    schema: Optional[str]
-    identifier: Optional[str]
-
-    def __post_init__(self):
-        # handle pesky jinja2.Undefined sneaking in here and messing up render
-        if not isinstance(self.database, (type(None), str)):
-            raise dbt.exceptions.CompilationException(
-                'Got an invalid path database: {}'.format(self.database)
-            )
-        if not isinstance(self.schema, (type(None), str)):
-            raise dbt.exceptions.CompilationException(
-                'Got an invalid path schema: {}'.format(self.schema)
-            )
-        if not isinstance(self.identifier, (type(None), str)):
-            raise dbt.exceptions.CompilationException(
-                'Got an invalid path identifier: {}'.format(self.identifier)
-            )
-
-    def get_lowered_part(self, key: ComponentName) -> Optional[str]:
-        part = self.get_part(key)
-        if part is not None:
-            part = part.lower()
-        return part
+import dbt.exceptions
 
 
 Self = TypeVar('Self', bound='BaseRelation')
@@ -133,8 +21,8 @@ Self = TypeVar('Self', bound='BaseRelation')
 
 @dataclass(frozen=True, eq=False, repr=False)
 class BaseRelation(FakeAPIObject, Hashable):
-    type: Optional[RelationType]
     path: Path
+    type: Optional[RelationType] = None
     quote_character: str = '"'
     include_policy: Policy = Policy()
     quote_policy: Policy = Policy()
@@ -157,12 +45,16 @@ class BaseRelation(FakeAPIObject, Hashable):
     def __eq__(self, other):
         if not isinstance(other, self.__class__):
             return False
-        return self.to_dict() == other.to_dict()
+        return self.to_dict(omit_none=True) == other.to_dict(omit_none=True)
 
     @classmethod
-    def get_default_quote_policy(cls: Type[Self]) -> Policy:
+    def get_default_quote_policy(cls) -> Policy:
         return cls._get_field_named('quote_policy').default
 
+    @classmethod
+    def get_default_include_policy(cls) -> Policy:
+        return cls._get_field_named('include_policy').default
+
     def get(self, key, default=None):
         """Override `.get` to return a metadata object so we don't break
         dbt_utils.
@@ -254,6 +146,16 @@ class BaseRelation(FakeAPIObject, Hashable):
     def information_schema_only(self) -> 'InformationSchema':
         return self.information_schema()
 
+    def without_identifier(self) -> 'BaseRelation':
+        """Return a form of this relation that only has the database and schema
+        set to included. To get the appropriately-quoted form the schema out of
+        the result (for use as part of a query), use `.render()`. To get the
+        raw database or schema name, use `.database` or `.schema`.
+
+        The hash of the returned object is the result of render().
+        """
+        return self.include(identifier=False).replace_path(identifier=None)
+
     def _render_iterator(
         self
     ) -> Iterator[Tuple[Optional[ComponentName], Optional[str]]]:
@@ -267,16 +169,11 @@ class BaseRelation(FakeAPIObject, Hashable):
             yield key, path_part
 
     def render(self) -> str:
-        parts: List[str] = [
-            part for _, part in self._render_iterator() if part is not None
-        ]
-
-        if len(parts) == 0:
-            raise dbt.exceptions.RuntimeException(
-                "No path parts are included! Nothing to render."
-            )
-
-        return '.'.join(parts)
+        # if there is nothing set, this will return the empty string.
+        return '.'.join(
+            part for _, part in self._render_iterator()
+            if part is not None
+        )
 
     def quoted(self, identifier):
         return '{quote_char}{identifier}{quote_char}'.format(
@@ -288,10 +185,10 @@ class BaseRelation(FakeAPIObject, Hashable):
     def create_from_source(
         cls: Type[Self], source: ParsedSourceDefinition, **kwargs: Any
     ) -> Self:
-        source_quoting = source.quoting.to_dict()
+        source_quoting = source.quoting.to_dict(omit_none=True)
         source_quoting.pop('column', None)
         quote_policy = deep_merge(
-            cls.get_default_quote_policy().to_dict(),
+            cls.get_default_quote_policy().to_dict(omit_none=True),
             source_quoting,
             kwargs.get('quote_policy', {}),
         )
@@ -304,6 +201,23 @@ class BaseRelation(FakeAPIObject, Hashable):
             **kwargs
         )
 
+    @staticmethod
+    def add_ephemeral_prefix(name: str):
+        return f'__dbt__cte__{name}'
+
+    @classmethod
+    def create_ephemeral_from_node(
+        cls: Type[Self],
+        config: HasQuoting,
+        node: Union[ParsedNode, CompiledNode],
+    ) -> Self:
+        # Note that ephemeral models are based on the name.
+        identifier = cls.add_ephemeral_prefix(node.name)
+        return cls.create(
+            type=cls.CTE,
+            identifier=identifier,
+        ).quote(identifier=False)
+
     @classmethod
     def create_from_node(
         cls: Type[Self],
@@ -496,3 +410,47 @@ class InformationSchema(BaseRelation):
         for k, v in super()._render_iterator():
             yield k, v
         yield None, self.information_schema_view
+
+
+class SchemaSearchMap(Dict[InformationSchema, Set[Optional[str]]]):
+    """A utility class to keep track of what information_schema tables to
+    search for what schemas. The schema values are all lowercased to avoid
+    duplication.
+    """
+    def add(self, relation: BaseRelation):
+        key = relation.information_schema_only()
+        if key not in self:
+            self[key] = set()
+        schema: Optional[str] = None
+        if relation.schema is not None:
+            schema = relation.schema.lower()
+        self[key].add(schema)
+
+    def search(
+        self
+    ) -> Iterator[Tuple[InformationSchema, Optional[str]]]:
+        for information_schema_name, schemas in self.items():
+            for schema in schemas:
+                yield information_schema_name, schema
+
+    def flatten(self, allow_multiple_databases: bool = False):
+        new = self.__class__()
+
+        # make sure we don't have multiple databases if allow_multiple_databases is set to False
+        if not allow_multiple_databases:
+            seen = {r.database.lower() for r in self if r.database}
+            if len(seen) > 1:
+                dbt.exceptions.raise_compiler_error(str(seen))
+
+        for information_schema_name, schema in self.search():
+            path = {
+                'database': information_schema_name.database,
+                'schema': schema
+            }
+            new.add(information_schema_name.incorporate(
+                path=path,
+                quote_policy={'database': False},
+                include_policy={'database': False},
+            ))
+
+        return new

core/dbt/adapters/cache.py

@@ -4,25 +4,20 @@ from typing import List, Iterable, Optional, Dict, Set, Tuple, Any
 import threading
 
 from dbt.logger import CACHE_LOGGER as logger
+from dbt.utils import lowercase
 import dbt.exceptions
 
 _ReferenceKey = namedtuple('_ReferenceKey', 'database schema identifier')
 
 
-def _lower(value: Optional[str]) -> Optional[str]:
-    """Postgres schemas can be None so we can't just call lower()."""
-    if value is None:
-        return None
-    return value.lower()
-
-
 def _make_key(relation) -> _ReferenceKey:
     """Make _ReferenceKeys with lowercase values for the cache so we don't have
     to keep track of quoting
     """
-    return _ReferenceKey(_lower(relation.database),
-                         _lower(relation.schema),
-                         _lower(relation.identifier))
+    # databases and schemas can both be None
+    return _ReferenceKey(lowercase(relation.database),
+                         lowercase(relation.schema),
+                         lowercase(relation.identifier))
 
 
 def dot_separated(key: _ReferenceKey) -> str:
@@ -53,15 +48,15 @@ class _CachedRelation:
 
     @property
     def database(self) -> Optional[str]:
-        return _lower(self.inner.database)
+        return lowercase(self.inner.database)
 
     @property
     def schema(self) -> Optional[str]:
-        return _lower(self.inner.schema)
+        return lowercase(self.inner.schema)
 
     @property
     def identifier(self) -> Optional[str]:
-        return _lower(self.inner.identifier)
+        return lowercase(self.inner.identifier)
 
     def __copy__(self):
         new = self.__class__(self.inner)
@@ -190,7 +185,7 @@ class RelationsCache:
         :param database: The database name to add.
         :param schema: The schema name to add.
         """
-        self.schemas.add((_lower(database), _lower(schema)))
+        self.schemas.add((lowercase(database), lowercase(schema)))
 
     def drop_schema(
         self, database: Optional[str], schema: Optional[str],
@@ -199,7 +194,7 @@ class RelationsCache:
 
         Then remove all its contents (and their dependents, etc) as well.
         """
-        key = (_lower(database), _lower(schema))
+        key = (lowercase(database), lowercase(schema))
         if key not in self.schemas:
             return
 
@@ -217,7 +212,7 @@ class RelationsCache:
 
         :param schemas: An iterable of the schema names to add.
         """
-        self.schemas.update((_lower(d), s.lower()) for (d, s) in schemas)
+        self.schemas.update((lowercase(d), s.lower()) for (d, s) in schemas)
 
     def __contains__(self, schema_id: Tuple[Optional[str], str]):
         """A schema is 'in' the relations cache if it is in the set of cached
@@ -226,7 +221,7 @@ class RelationsCache:
         :param schema_id: The db name and schema name to look up.
         """
         db, schema = schema_id
-        return (_lower(db), schema.lower()) in self.schemas
+        return (lowercase(db), schema.lower()) in self.schemas
 
     def dump_graph(self):
         """Dump a key-only representation of the schema to a dictionary. Every
@@ -484,13 +479,13 @@ class RelationsCache:
         :return List[BaseRelation]: The list of relations with the given
             schema
         """
-        database = _lower(database)
-        schema = _lower(schema)
+        database = lowercase(database)
+        schema = lowercase(schema)
         with self.lock:
             results = [
                 r.inner for r in self.relations.values()
-                if (_lower(r.schema) == schema and
-                    _lower(r.database) == database)
+                if (lowercase(r.schema) == schema and
+                    lowercase(r.database) == database)
             ]
 
         if None in results:
@@ -509,7 +504,7 @@ class RelationsCache:
         self, database: Optional[str], schema: Optional[str]
     ) -> List[_CachedRelation]:
         """Get the relations in a schema. Callers should hold the lock."""
-        key = (_lower(database), _lower(schema))
+        key = (lowercase(database), lowercase(schema))
 
         to_remove: List[_CachedRelation] = []
         for cachekey, relation in self.relations.items():

core/dbt/adapters/factory.py

@@ -1,47 +1,63 @@
 import threading
+from pathlib import Path
 from importlib import import_module
-from typing import Type, Dict, Any
+from typing import Type, Dict, Any, List, Optional, Set
 
-from dbt.exceptions import RuntimeException
-from dbt.include.global_project import PACKAGES
+from dbt.exceptions import RuntimeException, InternalException
+from dbt.include.global_project import (
+    PACKAGE_PATH as GLOBAL_PROJECT_PATH,
+    PROJECT_NAME as GLOBAL_PROJECT_NAME,
+)
 from dbt.logger import GLOBAL_LOGGER as logger
 from dbt.contracts.connection import Credentials, AdapterRequiredConfig
 
-from dbt.adapters.base.impl import BaseAdapter
+
+from dbt.adapters.protocol import (
+    AdapterProtocol,
+    AdapterConfig,
+    RelationProtocol,
+)
 from dbt.adapters.base.plugin import AdapterPlugin
 
 
-# TODO: we can't import these because they cause an import cycle.
-# Profile has to call into load_plugin to get credentials, so adapter/relation
-# don't work
-BaseRelation = Any
+Adapter = AdapterProtocol
 
 
-Adapter = BaseAdapter
-
-
-class AdpaterContainer:
+class AdapterContainer:
     def __init__(self):
         self.lock = threading.Lock()
         self.adapters: Dict[str, Adapter] = {}
-        self.adapter_types: Dict[str, Type[Adapter]] = {}
+        self.plugins: Dict[str, AdapterPlugin] = {}
+        # map package names to their include paths
+        self.packages: Dict[str, Path] = {
+            GLOBAL_PROJECT_NAME: Path(GLOBAL_PROJECT_PATH),
+        }
 
-    def get_adapter_class_by_name(self, name: str) -> Type[Adapter]:
+    def get_plugin_by_name(self, name: str) -> AdapterPlugin:
         with self.lock:
-            if name in self.adapter_types:
-                return self.adapter_types[name]
-
-            names = ", ".join(self.adapter_types.keys())
+            if name in self.plugins:
+                return self.plugins[name]
+            names = ", ".join(self.plugins.keys())
 
         message = f"Invalid adapter type {name}! Must be one of {names}"
         raise RuntimeException(message)
 
-    def get_relation_class_by_name(self, name: str) -> Type[BaseRelation]:
+    def get_adapter_class_by_name(self, name: str) -> Type[Adapter]:
+        plugin = self.get_plugin_by_name(name)
+        return plugin.adapter
+
+    def get_relation_class_by_name(self, name: str) -> Type[RelationProtocol]:
         adapter = self.get_adapter_class_by_name(name)
         return adapter.Relation
 
+    def get_config_class_by_name(
+        self, name: str
+    ) -> Type[AdapterConfig]:
+        adapter = self.get_adapter_class_by_name(name)
+        return adapter.AdapterSpecificConfigs
+
     def load_plugin(self, name: str) -> Type[Credentials]:
-        # this doesn't need a lock: in the worst case we'll overwrite PACKAGES
+        # this doesn't need a lock: in the worst case we'll overwrite packages
         # and adapter_type entries with the same value, as they're all
         # singletons
         try:
@@ -68,9 +84,9 @@ class AdpaterContainer:
 
         with self.lock:
             # things do hold the lock to iterate over it so we need it to add
-            self.adapter_types[name] = plugin.adapter
+            self.plugins[name] = plugin
 
-        PACKAGES[plugin.project_name] = plugin.include_path
+        self.packages[plugin.project_name] = Path(plugin.include_path)
 
         for dep in plugin.dependencies:
             self.load_plugin(dep)
@@ -108,8 +124,58 @@ class AdpaterContainer:
             for adapter in self.adapters.values():
                 adapter.cleanup_connections()
 
+    def get_adapter_plugins(self, name: Optional[str]) -> List[AdapterPlugin]:
+        """Iterate over the known adapter plugins. If a name is provided,
+        iterate in dependency order over the named plugin and its dependencies.
+        """
+        if name is None:
+            return list(self.plugins.values())
 
-FACTORY: AdpaterContainer = AdpaterContainer()
+        plugins: List[AdapterPlugin] = []
+        seen: Set[str] = set()
+        plugin_names: List[str] = [name]
+        while plugin_names:
+            plugin_name = plugin_names[0]
+            plugin_names = plugin_names[1:]
+            try:
+                plugin = self.plugins[plugin_name]
+            except KeyError:
+                raise InternalException(
+                    f'No plugin found for {plugin_name}'
+                ) from None
+            plugins.append(plugin)
+            seen.add(plugin_name)
+            if plugin.dependencies is None:
+                continue
+            for dep in plugin.dependencies:
+                if dep not in seen:
+                    plugin_names.append(dep)
+        return plugins
+
+    def get_adapter_package_names(self, name: Optional[str]) -> List[str]:
+        package_names: List[str] = [
+            p.project_name for p in self.get_adapter_plugins(name)
+        ]
+        package_names.append(GLOBAL_PROJECT_NAME)
+        return package_names
+
+    def get_include_paths(self, name: Optional[str]) -> List[Path]:
+        paths = []
+        for package_name in self.get_adapter_package_names(name):
+            try:
+                path = self.packages[package_name]
+            except KeyError:
+                raise InternalException(
+                    f'No internal package listing found for {package_name}'
+                )
+            paths.append(path)
+        return paths
+
+    def get_adapter_type_names(self, name: Optional[str]) -> List[str]:
+        return [p.adapter.type() for p in self.get_adapter_plugins(name)]
+
+
+FACTORY: AdapterContainer = AdapterContainer()
 
 
 def register_adapter(config: AdapterRequiredConfig) -> None:
@@ -133,13 +199,29 @@ def cleanup_connections():
     FACTORY.cleanup_connections()
 
 
-def get_adapter_class_by_name(name: str) -> Type[BaseAdapter]:
+def get_adapter_class_by_name(name: str) -> Type[AdapterProtocol]:
     return FACTORY.get_adapter_class_by_name(name)
 
 
-def get_relation_class_by_name(name: str) -> Type[BaseRelation]:
+def get_config_class_by_name(name: str) -> Type[AdapterConfig]:
+    return FACTORY.get_config_class_by_name(name)
+
+
+def get_relation_class_by_name(name: str) -> Type[RelationProtocol]:
     return FACTORY.get_relation_class_by_name(name)
 
 
 def load_plugin(name: str) -> Type[Credentials]:
     return FACTORY.load_plugin(name)
+
+
+def get_include_paths(name: Optional[str]) -> List[Path]:
+    return FACTORY.get_include_paths(name)
+
+
+def get_adapter_package_names(name: Optional[str]) -> List[str]:
+    return FACTORY.get_adapter_package_names(name)
+
+
+def get_adapter_type_names(name: Optional[str]) -> List[str]:
+    return FACTORY.get_adapter_type_names(name)

core/dbt/adapters/protocol.py

@@ -0,0 +1,163 @@
+from dataclasses import dataclass
+from typing import (
+    Type, Hashable, Optional, ContextManager, List, Generic, TypeVar, ClassVar,
+    Tuple, Union, Dict, Any
+)
+from typing_extensions import Protocol
+
+import agate
+
+from dbt.contracts.connection import (
+    Connection, AdapterRequiredConfig, AdapterResponse
+)
+from dbt.contracts.graph.compiled import (
+    CompiledNode, ManifestNode, NonSourceCompiledNode
+)
+from dbt.contracts.graph.parsed import ParsedNode, ParsedSourceDefinition
+from dbt.contracts.graph.model_config import BaseConfig
+from dbt.contracts.graph.manifest import Manifest
+from dbt.contracts.relation import Policy, HasQuoting
+
+from dbt.graph import Graph
+
+
+@dataclass
+class AdapterConfig(BaseConfig):
+    pass
+
+
+class ConnectionManagerProtocol(Protocol):
+    TYPE: str
+
+
+class ColumnProtocol(Protocol):
+    pass
+
+
+Self = TypeVar('Self', bound='RelationProtocol')
+
+
+class RelationProtocol(Protocol):
+    @classmethod
+    def get_default_quote_policy(cls) -> Policy:
+        ...
+
+    @classmethod
+    def create_from(
+        cls: Type[Self],
+        config: HasQuoting,
+        node: Union[CompiledNode, ParsedNode, ParsedSourceDefinition],
+    ) -> Self:
+        ...
+
+
+class CompilerProtocol(Protocol):
+    def compile(self, manifest: Manifest, write=True) -> Graph:
+        ...
+
+    def compile_node(
+        self,
+        node: ManifestNode,
+        manifest: Manifest,
+        extra_context: Optional[Dict[str, Any]] = None,
+    ) -> NonSourceCompiledNode:
+        ...
+
+
+AdapterConfig_T = TypeVar(
+    'AdapterConfig_T', bound=AdapterConfig
+)
+ConnectionManager_T = TypeVar(
+    'ConnectionManager_T', bound=ConnectionManagerProtocol
+)
+Relation_T = TypeVar(
+    'Relation_T', bound=RelationProtocol
+)
+Column_T = TypeVar(
+    'Column_T', bound=ColumnProtocol
+)
+Compiler_T = TypeVar('Compiler_T', bound=CompilerProtocol)
+
+
+class AdapterProtocol(
+    Protocol,
+    Generic[
+        AdapterConfig_T,
+        ConnectionManager_T,
+        Relation_T,
+        Column_T,
+        Compiler_T,
+    ]
+):
+    AdapterSpecificConfigs: ClassVar[Type[AdapterConfig_T]]
+    Column: ClassVar[Type[Column_T]]
+    Relation: ClassVar[Type[Relation_T]]
+    ConnectionManager: ClassVar[Type[ConnectionManager_T]]
+    connections: ConnectionManager_T
+
+    def __init__(self, config: AdapterRequiredConfig):
+        ...
+
+    @classmethod
+    def type(cls) -> str:
+        pass
+
+    def set_query_header(self, manifest: Manifest) -> None:
+        ...
+
+    @staticmethod
+    def get_thread_identifier() -> Hashable:
+        ...
+
+    def get_thread_connection(self) -> Connection:
+        ...
+
+    def set_thread_connection(self, conn: Connection) -> None:
+        ...
+
+    def get_if_exists(self) -> Optional[Connection]:
+        ...
+
+    def clear_thread_connection(self) -> None:
+        ...
+
+    def clear_transaction(self) -> None:
+        ...
+
+    def exception_handler(self, sql: str) -> ContextManager:
+        ...
+
+    def set_connection_name(self, name: Optional[str] = None) -> Connection:
+        ...
+
+    def cancel_open(self) -> Optional[List[str]]:
+        ...
+
+    def open(cls, connection: Connection) -> Connection:
+        ...
+
+    def release(self) -> None:
+        ...
+
+    def cleanup_all(self) -> None:
+        ...
+
+    def begin(self) -> None:
+        ...
+
+    def commit(self) -> None:
+        ...
+
+    def close(cls, connection: Connection) -> Connection:
+        ...
+
+    def commit_if_has_connection(self) -> None:
+        ...
+
+    def execute(
+        self, sql: str, auto_begin: bool = False, fetch: bool = False
+    ) -> Tuple[Union[str, AdapterResponse], agate.Table]:
+        ...
+
+    def get_compiler(self) -> Compiler_T:
+        ...

core/dbt/adapters/sql/connections.py

@@ -1,14 +1,17 @@
 import abc
 import time
-from typing import List, Optional, Tuple, Any, Iterable, Dict
+from typing import List, Optional, Tuple, Any, Iterable, Dict, Union
 
 import agate
 
 import dbt.clients.agate_helper
 import dbt.exceptions
-from dbt.contracts.connection import Connection
 from dbt.adapters.base import BaseConnectionManager
+from dbt.contracts.connection import (
+    Connection, ConnectionState, AdapterResponse
+)
 from dbt.logger import GLOBAL_LOGGER as logger
+from dbt import flags
 
 
 class SQLConnectionManager(BaseConnectionManager):
@@ -17,7 +20,7 @@ class SQLConnectionManager(BaseConnectionManager):
     Methods to implement:
         - exception_handler
         - cancel
-        - get_status
+        - get_response
         - open
     """
     @abc.abstractmethod
@@ -37,7 +40,10 @@ class SQLConnectionManager(BaseConnectionManager):
 
                 # if the connection failed, the handle will be None so we have
                 # nothing to cancel.
-                if connection.handle is not None:
+                if (
+                    connection.handle is not None and
+                    connection.state == ConnectionState.OPEN
+                ):
                     self.cancel(connection)
                 if connection.name is not None:
                     names.append(connection.name)
@@ -72,20 +78,19 @@ class SQLConnectionManager(BaseConnectionManager):
 
             cursor = connection.handle.cursor()
             cursor.execute(sql, bindings)
-
             logger.debug(
                 "SQL status: {status} in {elapsed:0.2f} seconds",
-                status=self.get_status(cursor),
+                status=self.get_response(cursor),
                 elapsed=(time.time() - pre)
             )
 
             return connection, cursor
 
     @abc.abstractclassmethod
-    def get_status(cls, cursor: Any) -> str:
+    def get_response(cls, cursor: Any) -> Union[AdapterResponse, str]:
         """Get the status of the cursor."""
         raise dbt.exceptions.NotImplementedException(
-            '`get_status` is not implemented for this adapter!'
+            '`get_response` is not implemented for this adapter!'
         )
 
     @classmethod
@@ -94,7 +99,14 @@ class SQLConnectionManager(BaseConnectionManager):
         column_names: Iterable[str],
         rows: Iterable[Any]
     ) -> List[Dict[str, Any]]:
-
+        unique_col_names = dict()
+        for idx in range(len(column_names)):
+            col_name = column_names[idx]
+            if col_name in unique_col_names:
+                unique_col_names[col_name] += 1
+                column_names[idx] = f'{col_name}_{unique_col_names[col_name]}'
+            else:
+                unique_col_names[column_names[idx]] = 1
         return [dict(zip(column_names, row)) for row in rows]
 
     @classmethod
@@ -107,19 +119,22 @@ class SQLConnectionManager(BaseConnectionManager):
             rows = cursor.fetchall()
             data = cls.process_results(column_names, rows)
 
-        return dbt.clients.agate_helper.table_from_data(data, column_names)
+        return dbt.clients.agate_helper.table_from_data_flat(
+            data,
+            column_names
+        )
 
     def execute(
         self, sql: str, auto_begin: bool = False, fetch: bool = False
-    ) -> Tuple[str, agate.Table]:
+    ) -> Tuple[Union[AdapterResponse, str], agate.Table]:
         sql = self._add_query_comment(sql)
         _, cursor = self.add_query(sql, auto_begin)
-        status = self.get_status(cursor)
+        response = self.get_response(cursor)
         if fetch:
             table = self.get_result_from_cursor(cursor)
         else:
             table = dbt.clients.agate_helper.empty_table()
-        return status, table
+        return response, table
 
     def add_begin_query(self):
         return self.add_query('BEGIN', auto_begin=False)
@@ -130,7 +145,7 @@ class SQLConnectionManager(BaseConnectionManager):
     def begin(self):
         connection = self.get_thread_connection()
 
-        if dbt.flags.STRICT_MODE:
+        if flags.STRICT_MODE:
             if not isinstance(connection, Connection):
                 raise dbt.exceptions.CompilerException(
                     f'In begin, got {connection} - not a Connection!'
@@ -148,7 +163,7 @@ class SQLConnectionManager(BaseConnectionManager):
 
     def commit(self):
         connection = self.get_thread_connection()
-        if dbt.flags.STRICT_MODE:
+        if flags.STRICT_MODE:
             if not isinstance(connection, Connection):
                 raise dbt.exceptions.CompilerException(
                     f'In commit, got {connection} - not a Connection!'

core/dbt/adapters/sql/impl.py

@@ -4,12 +4,11 @@ from typing import Any, Optional, Tuple, Type, List
 import dbt.clients.agate_helper
 from dbt.contracts.connection import Connection
 import dbt.exceptions
-import dbt.flags
 from dbt.adapters.base import BaseAdapter, available
 from dbt.adapters.sql import SQLConnectionManager
 from dbt.logger import GLOBAL_LOGGER as logger
 
-from dbt.adapters.factory import BaseRelation
+from dbt.adapters.base.relation import BaseRelation
 
 LIST_RELATIONS_MACRO_NAME = 'list_relations_without_caching'
 GET_COLUMNS_IN_RELATION_MACRO_NAME = 'get_columns_in_relation'
@@ -174,31 +173,31 @@ class SQLAdapter(BaseAdapter):
             kwargs={'relation': relation}
         )
 
-    def create_schema(self, database: str, schema: str) -> None:
-        logger.debug('Creating schema "{}"."{}".', database, schema)
+    def create_schema(self, relation: BaseRelation) -> None:
+        relation = relation.without_identifier()
+        logger.debug('Creating schema "{}"', relation)
         kwargs = {
-            'database_name': self.quote_as_configured(database, 'database'),
-            'schema_name': self.quote_as_configured(schema, 'schema'),
+            'relation': relation,
         }
         self.execute_macro(CREATE_SCHEMA_MACRO_NAME, kwargs=kwargs)
         self.commit_if_has_connection()
         # we can't update the cache here, as if the schema already existed we
         # don't want to (incorrectly) say that it's empty
 
-    def drop_schema(self, database: str, schema: str) -> None:
-        logger.debug('Dropping schema "{}"."{}".', database, schema)
+    def drop_schema(self, relation: BaseRelation) -> None:
+        relation = relation.without_identifier()
+        logger.debug('Dropping schema "{}".', relation)
         kwargs = {
-            'database_name': self.quote_as_configured(database, 'database'),
-            'schema_name': self.quote_as_configured(schema, 'schema'),
+            'relation': relation,
         }
         self.execute_macro(DROP_SCHEMA_MACRO_NAME, kwargs=kwargs)
         # we can update the cache here
-        self.cache.drop_schema(database, schema)
+        self.cache.drop_schema(relation.database, relation.schema)
 
     def list_relations_without_caching(
-            self, information_schema, schema
+        self, schema_relation: BaseRelation,
     ) -> List[BaseRelation]:
-        kwargs = {'information_schema': information_schema, 'schema': schema}
+        kwargs = {'schema_relation': schema_relation}
         results = self.execute_macro(
             LIST_RELATIONS_MACRO_NAME,
             kwargs=kwargs

core/dbt/clients/_jinja_blocks.py

@@ -69,7 +69,7 @@ BLOCK_START_PATTERN = regex(''.join((
 
 RAW_BLOCK_PATTERN = regex(''.join((
     r'(?:\s*\{\%\-|\{\%)\s*raw\s*(?:\-\%\}\s*|\%\})',
-    r'(?:.*)',
+    r'(?:.*?)',
     r'(?:\s*\{\%\-|\{\%)\s*endraw\s*(?:\-\%\}\s*|\%\})',
 )))
 
@@ -92,6 +92,18 @@ class TagIterator:
         self._parenthesis_stack = []
         self.pos = 0
 
+    def linepos(self, end=None) -> str:
+        """Given an absolute position in the input data, return a pair of
+        line number + relative position to the start of the line.
+        """
+        end_val: int = self.pos if end is None else end
+        data = self.data[:end_val]
+        # if not found, rfind returns -1, and -1+1=0, which is perfect!
+        last_line_start = data.rfind('\n') + 1
+        # it's easy to forget this, but line numbers are 1-indexed
+        line_number = data.count('\n') + 1
+        return f'{line_number}:{end_val - last_line_start}'
+
     def advance(self, new_position):
         self.pos = new_position
 
@@ -320,20 +332,28 @@ class BlockIterator:
                     dbt.exceptions.raise_compiler_error((
                         'Got an unexpected control flow end tag, got {} but '
                         'never saw a preceeding {} (@ {})'
-                    ).format(tag.block_type_name, expected, tag.start))
+                    ).format(
+                        tag.block_type_name,
+                        expected,
+                        self.tag_parser.linepos(tag.start)
+                    ))
                 expected = _CONTROL_FLOW_TAGS[found]
                 if expected != tag.block_type_name:
                     dbt.exceptions.raise_compiler_error((
                         'Got an unexpected control flow end tag, got {} but '
                         'expected {} next (@ {})'
-                    ).format(tag.block_type_name, expected, tag.start))
+                    ).format(
+                        tag.block_type_name,
+                        expected,
+                        self.tag_parser.linepos(tag.start)
+                    ))
 
             if tag.block_type_name in allowed_blocks:
                 if self.stack:
                     dbt.exceptions.raise_compiler_error((
                         'Got a block definition inside control flow at {}. '
                         'All dbt block definitions must be at the top level'
-                    ).format(tag.start))
+                    ).format(self.tag_parser.linepos(tag.start)))
                 if self.current is not None:
                     dbt.exceptions.raise_compiler_error(
                         duplicate_tags.format(outer=self.current, inner=tag)

core/dbt/clients/agate_helper.py

@@ -4,7 +4,8 @@ import agate
 import datetime
 import isodate
 import json
-from typing import Iterable, List, Dict, Union
+import dbt.utils
+from typing import Iterable, List, Dict, Union, Optional, Any
 
 from dbt.exceptions import RuntimeException
 
@@ -34,7 +35,11 @@ class ISODateTime(agate.data_types.DateTime):
         )
 
 
-def build_type_tester(text_columns: Iterable[str]):
+def build_type_tester(
+    text_columns: Iterable[str],
+    string_null_values: Optional[Iterable[str]] = ('null', '')
+) -> agate.TypeTester:
+
     types = [
         agate.data_types.Number(null_values=('null', '')),
         agate.data_types.Date(null_values=('null', ''),
@@ -45,10 +50,10 @@ def build_type_tester(text_columns: Iterable[str]):
         agate.data_types.Boolean(true_values=('true',),
                                  false_values=('false',),
                                  null_values=('null', '')),
-        agate.data_types.Text(null_values=('null', ''))
+        agate.data_types.Text(null_values=string_null_values)
     ]
     force = {
-        k: agate.data_types.Text(null_values=('null', ''))
+        k: agate.data_types.Text(null_values=string_null_values)
         for k in text_columns
     }
     return agate.TypeTester(force=force, types=types)
@@ -57,7 +62,25 @@ def build_type_tester(text_columns: Iterable[str]):
 DEFAULT_TYPE_TESTER = build_type_tester(())
 
 
-def table_from_data(data, column_names):
+def table_from_rows(
+    rows: List[Any],
+    column_names: Iterable[str],
+    text_only_columns: Optional[Iterable[str]] = None,
+) -> agate.Table:
+    if text_only_columns is None:
+        column_types = DEFAULT_TYPE_TESTER
+    else:
+        # If text_only_columns are present, prevent coercing empty string or
+        # literal 'null' strings to a None representation.
+        column_types = build_type_tester(
+            text_only_columns,
+            string_null_values=()
+        )
+
+    return agate.Table(rows, column_names, column_types=column_types)
+
+
+def table_from_data(data, column_names: Iterable[str]) -> agate.Table:
     "Convert list of dictionaries into an Agate table"
 
     # The agate table is generated from a list of dicts, so the column order
@@ -72,20 +95,35 @@ def table_from_data(data, column_names):
         return table.select(column_names)
 
 
-def table_from_data_flat(data, column_names):
-    "Convert list of dictionaries into an Agate table"
+def table_from_data_flat(data, column_names: Iterable[str]) -> agate.Table:
+    """
+    Convert a list of dictionaries into an Agate table. This method does not
+    coerce string values into more specific types (eg. '005' will not be
+    coerced to '5'). Additionally, this method does not coerce values to
+    None (eg. '' or 'null' will retain their string literal representations).
+    """
 
     rows = []
+    text_only_columns = set()
     for _row in data:
         row = []
-        for value in list(_row.values()):
+        for col_name in column_names:
+            value = _row[col_name]
             if isinstance(value, (dict, list, tuple)):
-                row.append(json.dumps(value))
-            else:
-                row.append(value)
+                # Represent container types as json strings
+                value = json.dumps(value, cls=dbt.utils.JSONEncoder)
+                text_only_columns.add(col_name)
+            elif isinstance(value, str):
+                text_only_columns.add(col_name)
+            row.append(value)
+
         rows.append(row)
 
-    return agate.Table(rows, column_names, column_types=DEFAULT_TYPE_TESTER)
+    return table_from_rows(
+        rows=rows,
+        column_names=column_names,
+        text_only_columns=text_only_columns
+    )
 
 
 def empty_table():

core/dbt/clients/git.py

@@ -4,16 +4,43 @@ import os.path
 from dbt.clients.system import run_cmd, rmdir
 from dbt.logger import GLOBAL_LOGGER as logger
 import dbt.exceptions
+from packaging import version
 
 
-def clone(repo, cwd, dirname=None, remove_git_dir=False):
-    clone_cmd = ['git', 'clone', '--depth', '1', repo]
+def _is_commit(revision: str) -> bool:
+    # match SHA-1 git commit
+    return bool(re.match(r"\b[0-9a-f]{40}\b", revision))
+
+
+def clone(repo, cwd, dirname=None, remove_git_dir=False, revision=None, subdirectory=None):
+    has_revision = revision is not None
+    is_commit = _is_commit(revision or "")
+
+    clone_cmd = ['git', 'clone', '--depth', '1']
+    if subdirectory:
+        logger.debug('  Subdirectory specified: {}, using sparse checkout.'.format(subdirectory))
+        out, _ = run_cmd(cwd, ['git', '--version'], env={'LC_ALL': 'C'})
+        git_version = version.parse(re.search(r"\d+\.\d+\.\d+", out.decode("utf-8")).group(0))
+        if not git_version >= version.parse("2.25.0"):
+            # 2.25.0 introduces --sparse
+            raise RuntimeError(
+                "Please update your git version to pull a dbt package "
+                "from a subdirectory: your version is {}, >= 2.25.0 needed".format(git_version)
+            )
+        clone_cmd.extend(['--filter=blob:none', '--sparse'])
+
+    if has_revision and not is_commit:
+        clone_cmd.extend(['--branch', revision])
+
+    clone_cmd.append(repo)
 
     if dirname is not None:
         clone_cmd.append(dirname)
-
     result = run_cmd(cwd, clone_cmd, env={'LC_ALL': 'C'})
 
+    if subdirectory:
+        run_cmd(os.path.join(cwd, dirname or ''), ['git', 'sparse-checkout', 'set', subdirectory])
+
     if remove_git_dir:
         rmdir(os.path.join(dirname, '.git'))
 
@@ -26,33 +53,38 @@ def list_tags(cwd):
     return tags
 
 
-def _checkout(cwd, repo, branch):
-    logger.debug('  Checking out branch {}.'.format(branch))
+def _checkout(cwd, repo, revision):
+    logger.debug('  Checking out revision {}.'.format(revision))
 
-    run_cmd(cwd, ['git', 'remote', 'set-branches', 'origin', branch])
-    run_cmd(cwd, ['git', 'fetch', '--tags', '--depth', '1', 'origin', branch])
+    fetch_cmd = ["git", "fetch", "origin", "--depth", "1"]
 
-    tags = list_tags(cwd)
-
-    # Prefer tags to branches if one exists
-    if branch in tags:
-        spec = 'tags/{}'.format(branch)
+    if _is_commit(revision):
+        run_cmd(cwd, fetch_cmd + [revision])
     else:
-        spec = 'origin/{}'.format(branch)
+        run_cmd(cwd, ['git', 'remote', 'set-branches', 'origin', revision])
+        run_cmd(cwd, fetch_cmd + ["--tags", revision])
+
+    if _is_commit(revision):
+        spec = revision
+    # Prefer tags to branches if one exists
+    elif revision in list_tags(cwd):
+        spec = 'tags/{}'.format(revision)
+    else:
+        spec = 'origin/{}'.format(revision)
 
     out, err = run_cmd(cwd, ['git', 'reset', '--hard', spec],
                        env={'LC_ALL': 'C'})
     return out, err
 
 
-def checkout(cwd, repo, branch=None):
-    if branch is None:
-        branch = 'master'
+def checkout(cwd, repo, revision=None):
+    if revision is None:
+        revision = 'HEAD'
     try:
-        return _checkout(cwd, repo, branch)
+        return _checkout(cwd, repo, revision)
     except dbt.exceptions.CommandResultError as exc:
         stderr = exc.stderr.decode('utf-8').strip()
-    dbt.exceptions.bad_package_spec(repo, branch, stderr)
+    dbt.exceptions.bad_package_spec(repo, revision, stderr)
 
 
 def get_current_sha(cwd):
@@ -66,11 +98,16 @@ def remove_remote(cwd):
 
 
 def clone_and_checkout(repo, cwd, dirname=None, remove_git_dir=False,
-                       branch=None):
+                       revision=None, subdirectory=None):
     exists = None
     try:
-        _, err = clone(repo, cwd, dirname=dirname,
-                       remove_git_dir=remove_git_dir)
+        _, err = clone(
+            repo,
+            cwd,
+            dirname=dirname,
+            remove_git_dir=remove_git_dir,
+            subdirectory=subdirectory,
+        )
     except dbt.exceptions.CommandResultError as exc:
         err = exc.stderr.decode('utf-8')
         exists = re.match("fatal: destination path '(.+)' already exists", err)
@@ -92,7 +129,7 @@ def clone_and_checkout(repo, cwd, dirname=None, remove_git_dir=False,
         logger.debug('Pulling new dependency {}.', directory)
     full_path = os.path.join(cwd, directory)
     start_sha = get_current_sha(full_path)
-    checkout(full_path, repo, branch)
+    checkout(full_path, repo, revision)
     end_sha = get_current_sha(full_path)
     if exists:
         if start_sha == end_sha:
@@ -102,4 +139,4 @@ def clone_and_checkout(repo, cwd, dirname=None, remove_git_dir=False,
                          start_sha[:7], end_sha[:7])
     else:
         logger.debug('  Checked out at {}.', end_sha[:7])
-    return directory
+    return os.path.join(directory, subdirectory or '')

core/dbt/clients/jinja.py

@@ -1,29 +1,38 @@
 import codecs
 import linecache
 import os
+import re
 import tempfile
 import threading
+from ast import literal_eval
 from contextlib import contextmanager
+from itertools import chain, islice
 from typing import (
-    List, Union, Set, Optional, Dict, Any, Iterator, Type, NoReturn
+    List, Union, Set, Optional, Dict, Any, Iterator, Type, NoReturn, Tuple,
+    Callable
 )
 
 import jinja2
 import jinja2.ext
+import jinja2.nativetypes  # type: ignore
 import jinja2.nodes
 import jinja2.parser
 import jinja2.sandbox
 
 from dbt.utils import (
-    get_dbt_macro_name, get_docs_macro_name, get_materialization_macro_name
+    get_dbt_macro_name, get_docs_macro_name, get_materialization_macro_name,
+    get_test_macro_name, deep_map
 )
 
 from dbt.clients._jinja_blocks import BlockIterator, BlockData, BlockTag
+from dbt.contracts.graph.compiled import CompiledSchemaTestNode
+from dbt.contracts.graph.parsed import ParsedSchemaTestNode
 from dbt.exceptions import (
     InternalException, raise_compiler_error, CompilationException,
-    invalid_materialization_argument, MacroReturn
+    invalid_materialization_argument, MacroReturn, JinjaRenderingException,
+    UndefinedMacroException
 )
-from dbt.flags import MACRO_DEBUGGING
+from dbt import flags
 from dbt.logger import GLOBAL_LOGGER as logger  # noqa
 
 
@@ -86,31 +95,119 @@ class MacroFuzzEnvironment(jinja2.sandbox.SandboxedEnvironment):
         If the value is 'write', also write the files to disk.
         WARNING: This can write a ton of data if you aren't careful.
         """
-        if filename == '<template>' and MACRO_DEBUGGING:
-            write = MACRO_DEBUGGING == 'write'
+        if filename == '<template>' and flags.MACRO_DEBUGGING:
+            write = flags.MACRO_DEBUGGING == 'write'
             filename = _linecache_inject(source, write)
 
         return super()._compile(source, filename)  # type: ignore
 
 
+class NativeSandboxEnvironment(MacroFuzzEnvironment):
+    code_generator_class = jinja2.nativetypes.NativeCodeGenerator
+
+
+class TextMarker(str):
+    """A special native-env marker that indicates that a value is text and is
+    not to be evaluated. Use this to prevent your numbery-strings from becoming
+    numbers!
+    """
+
+
+class NativeMarker(str):
+    """A special native-env marker that indicates the field should be passed to
+    literal_eval.
+    """
+
+
+class BoolMarker(NativeMarker):
+    pass
+
+
+class NumberMarker(NativeMarker):
+    pass
+
+
+def _is_number(value) -> bool:
+    return isinstance(value, (int, float)) and not isinstance(value, bool)
+
+
+def quoted_native_concat(nodes):
+    """This is almost native_concat from the NativeTemplate, except in the
+    special case of a single argument that is a quoted string and returns a
+    string, the quotes are re-inserted.
+    """
+    head = list(islice(nodes, 2))
+
+    if not head:
+        return ''
+
+    if len(head) == 1:
+        raw = head[0]
+        if isinstance(raw, TextMarker):
+            return str(raw)
+        elif not isinstance(raw, NativeMarker):
+            # return non-strings as-is
+            return raw
+    else:
+        # multiple nodes become a string.
+        return "".join([str(v) for v in chain(head, nodes)])
+
+    try:
+        result = literal_eval(raw)
+    except (ValueError, SyntaxError, MemoryError):
+        result = raw
+    if isinstance(raw, BoolMarker) and not isinstance(result, bool):
+        raise JinjaRenderingException(
+            f"Could not convert value '{raw!s}' into type 'bool'"
+        )
+    if isinstance(raw, NumberMarker) and not _is_number(result):
+        raise JinjaRenderingException(
+            f"Could not convert value '{raw!s}' into type 'number'"
+        )
+
+    return result
+
+
+class NativeSandboxTemplate(jinja2.nativetypes.NativeTemplate):  # mypy: ignore
+    environment_class = NativeSandboxEnvironment
+
+    def render(self, *args, **kwargs):
+        """Render the template to produce a native Python type. If the
+        result is a single node, its value is returned. Otherwise, the
+        nodes are concatenated as strings. If the result can be parsed
+        with :func:`ast.literal_eval`, the parsed value is returned.
+        Otherwise, the string is returned.
+        """
+        vars = dict(*args, **kwargs)
+
+        try:
+            return quoted_native_concat(
+                self.root_render_func(self.new_context(vars))
+            )
+        except Exception:
+            return self.environment.handle_exception()
+
+
+NativeSandboxEnvironment.template_class = NativeSandboxTemplate  # type: ignore
+
+
 class TemplateCache:
-
     def __init__(self):
-        self.file_cache = {}
+        self.file_cache: Dict[str, jinja2.Template] = {}
 
-    def get_node_template(self, node):
-        key = (node.package_name, node.original_file_path)
+    def get_node_template(self, node) -> jinja2.Template:
+        key = node.macro_sql
 
         if key in self.file_cache:
             return self.file_cache[key]
 
         template = get_template(
-            string=node.raw_sql,
+            string=node.macro_sql,
             ctx={},
             node=node,
         )
-        self.file_cache[key] = template
 
+        self.file_cache[key] = template
         return template
 
     def clear(self):
@@ -135,6 +232,7 @@ class BaseMacroGenerator:
         template = self.get_template()
         # make the module. previously we set both vars and local, but that's
         # redundant: They both end up in the same place
+        # make_module is in jinja2.environment. It returns a TemplateModule
         module = template.make_module(vars=self.context, shared=False)
         macro = module.__dict__[get_dbt_macro_name(name)]
         module.__dict__.update(self.context)
@@ -148,6 +246,7 @@ class BaseMacroGenerator:
             raise_compiler_error(str(e))
 
     def call_macro(self, *args, **kwargs):
+        # called from __call__ methods
         if self.context is None:
             raise InternalException(
                 'Context is still None in call_macro!'
@@ -210,8 +309,10 @@ class MacroGenerator(BaseMacroGenerator):
             e.stack.append(self.macro)
             raise e
 
+    # This adds the macro's unique id to the node's 'depends_on'
     @contextmanager
     def track_call(self):
+        # This is only called from __call__
         if self.stack is None or self.node is None:
             yield
         else:
@@ -226,6 +327,7 @@ class MacroGenerator(BaseMacroGenerator):
             finally:
                 self.stack.pop(unique_id)
 
+    # this makes MacroGenerator objects callable like functions
     def __call__(self, *args, **kwargs):
         with self.track_call():
             return self.call_macro(*args, **kwargs)
@@ -307,42 +409,36 @@ class DocumentationExtension(jinja2.ext.Extension):
         return node
 
 
+class TestExtension(jinja2.ext.Extension):
+    tags = ['test']
+
+    def parse(self, parser):
+        node = jinja2.nodes.Macro(lineno=next(parser.stream).lineno)
+        test_name = parser.parse_assign_target(name_only=True).name
+
+        parser.parse_signature(node)
+        node.name = get_test_macro_name(test_name)
+        node.body = parser.parse_statements(('name:endtest',),
+                                            drop_needle=True)
+        return node
+
+
 def _is_dunder_name(name):
     return name.startswith('__') and name.endswith('__')
 
 
-def create_macro_capture_env(node):
-
-    class ParserMacroCapture(jinja2.Undefined):
-        """
-        This class sets up the parser to capture macros.
-        """
+def create_undefined(node=None):
+    class Undefined(jinja2.Undefined):
         def __init__(self, hint=None, obj=None, name=None, exc=None):
             super().__init__(hint=hint, name=name)
             self.node = node
             self.name = name
-            self.package_name = node.package_name
+            self.hint = hint
             # jinja uses these for safety, so we have to override them.
             # see https://github.com/pallets/jinja/blob/master/jinja2/sandbox.py#L332-L339 # noqa
             self.unsafe_callable = False
             self.alters_data = False
 
-        def __deepcopy__(self, memo):
-            path = os.path.join(self.node.root_path,
-                                self.node.original_file_path)
-
-            logger.debug(
-                'dbt encountered an undefined variable, "{}" in node {}.{} '
-                '(source path: {})'
-                .format(self.name, self.node.package_name,
-                        self.node.name, path))
-
-            # match jinja's message
-            raise_compiler_error(
-                "{!r} is undefined".format(self.name),
-                node=self.node
-            )
-
         def __getitem__(self, name):
             # Propagate the undefined value if a caller accesses this as if it
             # were a dictionary
@@ -355,29 +451,64 @@ def create_macro_capture_env(node):
                     .format(type(self).__name__, name)
                 )
 
-            self.package_name = self.name
             self.name = name
 
-            return self
+            return self.__class__(hint=self.hint, name=self.name)
 
         def __call__(self, *args, **kwargs):
             return self
 
-    return ParserMacroCapture
+        def __reduce__(self):
+            raise_compiler_error(f'{self.name} is undefined', node=node)
+
+    return Undefined
 
 
-def get_environment(node=None, capture_macros=False):
+NATIVE_FILTERS: Dict[str, Callable[[Any], Any]] = {
+    'as_text': TextMarker,
+    'as_bool': BoolMarker,
+    'as_native': NativeMarker,
+    'as_number': NumberMarker,
+}
+
+
+TEXT_FILTERS: Dict[str, Callable[[Any], Any]] = {
+    'as_text': lambda x: x,
+    'as_bool': lambda x: x,
+    'as_native': lambda x: x,
+    'as_number': lambda x: x,
+}
+
+
+def get_environment(
+    node=None,
+    capture_macros: bool = False,
+    native: bool = False,
+) -> jinja2.Environment:
     args: Dict[str, List[Union[str, Type[jinja2.ext.Extension]]]] = {
         'extensions': ['jinja2.ext.do']
     }
 
     if capture_macros:
-        args['undefined'] = create_macro_capture_env(node)
+        args['undefined'] = create_undefined(node)
 
     args['extensions'].append(MaterializationExtension)
     args['extensions'].append(DocumentationExtension)
+    args['extensions'].append(TestExtension)
 
-    return MacroFuzzEnvironment(**args)
+    env_cls: Type[jinja2.Environment]
+    text_filter: Type
+    if native:
+        env_cls = NativeSandboxEnvironment
+        filters = NATIVE_FILTERS
+    else:
+        env_cls = MacroFuzzEnvironment
+        filters = TEXT_FILTERS
+
+    env = env_cls(**args)
+    env.filters.update(filters)
+
+    return env
 
 
 @contextmanager
@@ -388,7 +519,10 @@ def catch_jinja(node=None) -> Iterator[None]:
         e.translated = False
         raise CompilationException(str(e), node) from e
     except jinja2.exceptions.UndefinedError as e:
-        raise CompilationException(str(e), node) from e
+        raise UndefinedMacroException(str(e), node) from e
+    except CompilationException as exc:
+        exc.add_node(node)
+        raise
 
 
 def parse(string):
@@ -396,22 +530,71 @@ def parse(string):
         return get_environment().parse(str(string))
 
 
-def get_template(string, ctx, node=None, capture_macros=False):
+def get_template(
+    string: str,
+    ctx: Dict[str, Any],
+    node=None,
+    capture_macros: bool = False,
+    native: bool = False,
+):
     with catch_jinja(node):
-        env = get_environment(node, capture_macros)
+        env = get_environment(node, capture_macros, native=native)
 
         template_source = str(string)
         return env.from_string(template_source, globals=ctx)
 
 
-def render_template(template, ctx, node=None):
+def render_template(template, ctx: Dict[str, Any], node=None) -> str:
     with catch_jinja(node):
         return template.render(ctx)
 
 
-def get_rendered(string, ctx, node=None, capture_macros=False):
-    template = get_template(string, ctx, node, capture_macros=capture_macros)
+def _requote_result(raw_value: str, rendered: str) -> str:
+    double_quoted = raw_value.startswith('"') and raw_value.endswith('"')
+    single_quoted = raw_value.startswith("'") and raw_value.endswith("'")
+    if double_quoted:
+        quote_char = '"'
+    elif single_quoted:
+        quote_char = "'"
+    else:
+        quote_char = ''
+    return f'{quote_char}{rendered}{quote_char}'
 
+
+# performance note: Local benmcharking (so take it with a big grain of salt!)
+# on this indicates that it is is on average slightly slower than
+# checking two separate patterns, but the standard deviation is smaller with
+# one pattern. The time difference between the two was ~2 std deviations, which
+# is small enough that I've just chosen the more readable option.
+_HAS_RENDER_CHARS_PAT = re.compile(r'({[{%#]|[#}%]})')
+
+
+def get_rendered(
+    string: str,
+    ctx: Dict[str, Any],
+    node=None,
+    capture_macros: bool = False,
+    native: bool = False,
+) -> str:
+    # performance optimization: if there are no jinja control characters in the
+    # string, we can just return the input. Fall back to jinja if the type is
+    # not a string or if native rendering is enabled (so '1' -> 1, etc...)
+    # If this is desirable in the native env as well, we could handle the
+    # native=True case by passing the input string to ast.literal_eval, like
+    # the native renderer does.
+    if (
+        not native and
+        isinstance(string, str) and
+        _HAS_RENDER_CHARS_PAT.search(string) is None
+    ):
+        return string
+    template = get_template(
+        string,
+        ctx,
+        node,
+        capture_macros=capture_macros,
+        native=native,
+    )
     return render_template(template, ctx, node)
 
 
@@ -442,3 +625,41 @@ def extract_toplevel_blocks(
         allowed_blocks=allowed_blocks,
         collect_raw_data=collect_raw_data
     )
+
+
+SCHEMA_TEST_KWARGS_NAME = '_dbt_schema_test_kwargs'
+
+
+def add_rendered_test_kwargs(
+    context: Dict[str, Any],
+    node: Union[ParsedSchemaTestNode, CompiledSchemaTestNode],
+    capture_macros: bool = False,
+) -> None:
+    """Render each of the test kwargs in the given context using the native
+    renderer, then insert that value into the given context as the special test
+    keyword arguments member.
+    """
+    looks_like_func = r'^\s*(env_var|ref|var|source|doc)\s*\(.+\)\s*$'
+
+    def _convert_function(
+        value: Any, keypath: Tuple[Union[str, int], ...]
+    ) -> Any:
+        if isinstance(value, str):
+            if keypath == ('column_name',):
+                # special case: Don't render column names as native, make them
+                # be strings
+                return value
+
+            if re.match(looks_like_func, value) is not None:
+                # curly braces to make rendering happy
+                value = f'{{{{ {value} }}}}'
+
+            value = get_rendered(
+                value, context, node, capture_macros=capture_macros,
+                native=True
+            )
+
+        return value
+
+    kwargs = deep_map(_convert_function, node.test_metadata.kwargs)
+    context[SCHEMA_TEST_KWARGS_NAME] = kwargs

core/dbt/clients/jinja_static.py

@@ -0,0 +1,225 @@
+import jinja2
+from dbt.clients.jinja import get_environment
+from dbt.exceptions import raise_compiler_error
+
+
+def statically_extract_macro_calls(string, ctx, db_wrapper=None):
+    # set 'capture_macros' to capture undefined
+    env = get_environment(None, capture_macros=True)
+    parsed = env.parse(string)
+
+    standard_calls = ['source', 'ref', 'config']
+    possible_macro_calls = []
+    for func_call in parsed.find_all(jinja2.nodes.Call):
+        func_name = None
+        if hasattr(func_call, 'node') and hasattr(func_call.node, 'name'):
+            func_name = func_call.node.name
+        else:
+            # func_call for dbt_utils.current_timestamp macro
+            # Call(
+            #   node=Getattr(
+            #     node=Name(
+            #       name='dbt_utils',
+            #       ctx='load'
+            #     ),
+            #     attr='current_timestamp',
+            #     ctx='load
+            #   ),
+            #   args=[],
+            #   kwargs=[],
+            #   dyn_args=None,
+            #   dyn_kwargs=None
+            # )
+            if (hasattr(func_call, 'node') and
+                    hasattr(func_call.node, 'node') and
+                    type(func_call.node.node).__name__ == 'Name' and
+                    hasattr(func_call.node, 'attr')):
+                package_name = func_call.node.node.name
+                macro_name = func_call.node.attr
+                if package_name == 'adapter':
+                    if macro_name == 'dispatch':
+                        ad_macro_calls = statically_parse_adapter_dispatch(
+                            func_call, ctx, db_wrapper)
+                        possible_macro_calls.extend(ad_macro_calls)
+                    else:
+                        # This skips calls such as adapter.parse_index
+                        continue
+                else:
+                    func_name = f'{package_name}.{macro_name}'
+            else:
+                continue
+        if not func_name:
+            continue
+        if func_name in standard_calls:
+            continue
+        elif ctx.get(func_name):
+            continue
+        else:
+            if func_name not in possible_macro_calls:
+                possible_macro_calls.append(func_name)
+
+    return possible_macro_calls
+
+
+# Call(
+#   node=Getattr(
+#     node=Name(
+#       name='adapter',
+#       ctx='load'
+#     ),
+#     attr='dispatch',
+#     ctx='load'
+#   ),
+#   args=[
+#     Const(value='test_pkg_and_dispatch')
+#   ],
+#   kwargs=[
+#     Keyword(
+#       key='packages',
+#       value=Call(node=Getattr(node=Name(name='local_utils', ctx='load'),
+#          attr='_get_utils_namespaces', ctx='load'), args=[], kwargs=[],
+#          dyn_args=None, dyn_kwargs=None)
+#     )
+#   ],
+#   dyn_args=None,
+#   dyn_kwargs=None
+# )
+def statically_parse_adapter_dispatch(func_call, ctx, db_wrapper):
+    possible_macro_calls = []
+    # This captures an adapter.dispatch('<macro_name>') call.
+
+    func_name = None
+    # macro_name positional argument
+    if len(func_call.args) > 0:
+        func_name = func_call.args[0].value
+    if func_name:
+        possible_macro_calls.append(func_name)
+
+    # packages positional argument
+    packages = None
+    macro_namespace = None
+    packages_arg = None
+    packages_arg_type = None
+
+    if len(func_call.args) > 1:
+        packages_arg = func_call.args[1]
+        # This can be a List or a Call
+        packages_arg_type = type(func_call.args[1]).__name__
+
+    # keyword arguments
+    if func_call.kwargs:
+        for kwarg in func_call.kwargs:
+            if kwarg.key == 'packages':
+                # The packages keyword will be deprecated and
+                # eventually removed
+                packages_arg = kwarg.value
+                # This can be a List or a Call
+                packages_arg_type = type(kwarg.value).__name__
+            elif kwarg.key == 'macro_name':
+                # This will remain to enable static resolution
+                if type(kwarg.value).__name__ == 'Const':
+                    func_name = kwarg.value.value
+                    possible_macro_calls.append(func_name)
+                else:
+                    raise_compiler_error(f"The macro_name parameter ({kwarg.value.value}) "
+                                         "to adapter.dispatch was not a string")
+            elif kwarg.key == 'macro_namespace':
+                # This will remain to enable static resolution
+                kwarg_type = type(kwarg.value).__name__
+                if kwarg_type == 'Const':
+                    macro_namespace = kwarg.value.value
+                else:
+                    raise_compiler_error("The macro_namespace parameter to adapter.dispatch "
+                                         f"is a {kwarg_type}, not a string")
+
+    # positional arguments
+    if packages_arg:
+        if packages_arg_type == 'List':
+            # This will remain to enable static resolution
+            packages = []
+            for item in packages_arg.items:
+                packages.append(item.value)
+        elif packages_arg_type == 'Const':
+            # This will remain to enable static resolution
+            macro_namespace = packages_arg.value
+        elif packages_arg_type == 'Call':
+            # This is deprecated and should be removed eventually.
+            # It is here to support (hackily) common ways of providing
+            # a packages list to adapter.dispatch
+            if (hasattr(packages_arg, 'node') and
+                    hasattr(packages_arg.node, 'node') and
+                    hasattr(packages_arg.node.node, 'name') and
+                    hasattr(packages_arg.node, 'attr')):
+                package_name = packages_arg.node.node.name
+                macro_name = packages_arg.node.attr
+                if (macro_name.startswith('_get') and 'namespaces' in macro_name):
+                    # noqa: https://github.com/dbt-labs/dbt-utils/blob/9e9407b/macros/cross_db_utils/_get_utils_namespaces.sql
+                    var_name = f'{package_name}_dispatch_list'
+                    # hard code compatibility for fivetran_utils, just a teensy bit different
+                    # noqa: https://github.com/fivetran/dbt_fivetran_utils/blob/0978ba2/macros/_get_utils_namespaces.sql
+                    if package_name == 'fivetran_utils':
+                        default_packages = ['dbt_utils', 'fivetran_utils']
+                    else:
+                        default_packages = [package_name]
+
+                    namespace_names = get_dispatch_list(ctx, var_name, default_packages)
+                    packages = []
+                    if namespace_names:
+                        packages.extend(namespace_names)
+                else:
+                    msg = (
+                        f"As of v0.19.2, custom macros, such as '{macro_name}', are no longer "
+                        "supported in the 'packages' argument of 'adapter.dispatch()'.\n"
+                        f"See https://docs.getdbt.com/reference/dbt-jinja-functions/dispatch "
+                        "for details."
+                    ).strip()
+                    raise_compiler_error(msg)
+        elif packages_arg_type == 'Add':
+            # This logic is for when there is a variable and an addition of a list,
+            # like: packages = (var('local_utils_dispatch_list', []) + ['local_utils2'])
+            # This is deprecated and should be removed eventually.
+            namespace_var = None
+            default_namespaces = []
+            # This might be a single call or it might be the 'left' piece in an addition
+            for var_call in packages_arg.find_all(jinja2.nodes.Call):
+                if (hasattr(var_call, 'node') and
+                        var_call.node.name == 'var' and
+                        hasattr(var_call, 'args')):
+                    namespace_var = var_call.args[0].value
+            if hasattr(packages_arg, 'right'):  # we have a default list of namespaces
+                for item in packages_arg.right.items:
+                    default_namespaces.append(item.value)
+            if namespace_var:
+                namespace_names = get_dispatch_list(ctx, namespace_var, default_namespaces)
+                packages = []
+                if namespace_names:
+                    packages.extend(namespace_names)
+
+    if db_wrapper:
+        macro = db_wrapper.dispatch(
+            func_name,
+            packages=packages,
+            macro_namespace=macro_namespace
+        ).macro
+        func_name = f'{macro.package_name}.{macro.name}'
+        possible_macro_calls.append(func_name)
+    else:  # this is only for test/unit/test_macro_calls.py
+        if macro_namespace:
+            packages = [macro_namespace]
+        if packages is None:
+            packages = []
+        for package_name in packages:
+            possible_macro_calls.append(f'{package_name}.{func_name}')
+
+    return possible_macro_calls
+
+
+def get_dispatch_list(ctx, var_name, default_packages):
+    namespace_list = None
+    try:
+        # match the logic currently used in package _get_namespaces() macro
+        namespace_list = ctx['var'](var_name) + default_packages
+    except Exception:
+        pass
+    namespace_list = namespace_list if namespace_list else default_packages
+    return namespace_list

core/dbt/clients/registry.py

@@ -1,10 +1,9 @@
-from functools import wraps
+import functools
 import requests
-from dbt.exceptions import RegistryException
-from dbt.utils import memoized
+from dbt.utils import memoized, _connection_exception_retry as connection_exception_retry
 from dbt.logger import GLOBAL_LOGGER as logger
+from dbt import deprecations
 import os
-import time
 
 if os.getenv('DBT_PACKAGE_HUB_URL'):
     DEFAULT_REGISTRY_BASE_URL = os.getenv('DBT_PACKAGE_HUB_URL')
@@ -19,31 +18,15 @@ def _get_url(url, registry_base_url=None):
     return '{}{}'.format(registry_base_url, url)
 
 
-def _wrap_exceptions(fn):
-    @wraps(fn)
-    def wrapper(*args, **kwargs):
-        max_attempts = 5
-        attempt = 0
-        while True:
-            attempt += 1
-            try:
-                return fn(*args, **kwargs)
-            except requests.exceptions.ConnectionError as exc:
-                if attempt < max_attempts:
-                    time.sleep(1)
-                    continue
-
-                raise RegistryException(
-                    'Unable to connect to registry hub'
-                ) from exc
-    return wrapper
+def _get_with_retries(path, registry_base_url=None):
+    get_fn = functools.partial(_get, path, registry_base_url)
+    return connection_exception_retry(get_fn, 5)
 
 
-@_wrap_exceptions
 def _get(path, registry_base_url=None):
     url = _get_url(path, registry_base_url)
     logger.debug('Making package registry request: GET {}'.format(url))
-    resp = requests.get(url)
+    resp = requests.get(url, timeout=30)
     logger.debug('Response from registry: GET {} {}'.format(url,
                                                             resp.status_code))
     resp.raise_for_status()
@@ -51,22 +34,44 @@ def _get(path, registry_base_url=None):
 
 
 def index(registry_base_url=None):
-    return _get('api/v1/index.json', registry_base_url)
+    return _get_with_retries('api/v1/index.json', registry_base_url)
 
 
 index_cached = memoized(index)
 
 
 def packages(registry_base_url=None):
-    return _get('api/v1/packages.json', registry_base_url)
+    return _get_with_retries('api/v1/packages.json', registry_base_url)
 
 
 def package(name, registry_base_url=None):
-    return _get('api/v1/{}.json'.format(name), registry_base_url)
+    response = _get_with_retries('api/v1/{}.json'.format(name), registry_base_url)
+
+    # Either redirectnamespace or redirectname in the JSON response indicate a redirect
+    # redirectnamespace redirects based on package ownership
+    # redirectname redirects based on package name
+    # Both can be present at the same time, or neither. Fails gracefully to old name
+
+    if ('redirectnamespace' in response) or ('redirectname' in response):
+
+        if ('redirectnamespace' in response) and response['redirectnamespace'] is not None:
+            use_namespace = response['redirectnamespace']
+        else:
+            use_namespace = response['namespace']
+
+        if ('redirectname' in response) and response['redirectname'] is not None:
+            use_name = response['redirectname']
+        else:
+            use_name = response['name']
+
+        new_nwo = use_namespace + "/" + use_name
+        deprecations.warn('package-redirect', old_name=name, new_name=new_nwo)
+
+    return response
 
 
 def package_version(name, version, registry_base_url=None):
-    return _get('api/v1/{}/{}.json'.format(name, version), registry_base_url)
+    return _get_with_retries('api/v1/{}/{}.json'.format(name, version), registry_base_url)
 
 
 def get_available_versions(name):

core/dbt/clients/system.py

@@ -1,8 +1,10 @@
 import errno
+import functools
 import fnmatch
 import json
 import os
 import os.path
+import re
 import shutil
 import subprocess
 import sys
@@ -10,13 +12,18 @@ import tarfile
 import requests
 import stat
 from typing import (
-    Type, NoReturn, List, Optional, Dict, Any, Tuple, Callable
+    Type, NoReturn, List, Optional, Dict, Any, Tuple, Callable, Union
 )
 
 import dbt.exceptions
-import dbt.utils
-
 from dbt.logger import GLOBAL_LOGGER as logger
+from dbt.utils import _connection_exception_retry as connection_exception_retry
+
+if sys.platform == 'win32':
+    from ctypes import WinDLL, c_bool
+else:
+    WinDLL = None
+    c_bool = None
 
 
 def find_matching(
@@ -40,6 +47,8 @@ def find_matching(
     """
     matching = []
     root_path = os.path.normpath(root_path)
+    regex = fnmatch.translate(file_pattern)
+    reobj = re.compile(regex, re.IGNORECASE)
 
     for relative_path_to_search in relative_paths_to_search:
         absolute_path_to_search = os.path.join(
@@ -52,8 +61,7 @@ def find_matching(
                 relative_path = os.path.relpath(
                     absolute_path, absolute_path_to_search
                 )
-
-                if fnmatch.fnmatch(local_file, file_pattern):
+                if reobj.match(local_file):
                     matching.append({
                         'searched_path': relative_path_to_search,
                         'absolute_path': absolute_path,
@@ -64,6 +72,7 @@ def find_matching(
 
 
 def load_file_contents(path: str, strip: bool = True) -> str:
+    path = convert_path(path)
     with open(path, 'rb') as handle:
         to_return = handle.read().decode('utf-8')
 
@@ -79,6 +88,7 @@ def make_directory(path: str) -> None:
     exist. This function handles the case where two threads try to create
     a directory at once.
     """
+    path = convert_path(path)
     if not os.path.exists(path):
         # concurrent writes that try to create the same dir can fail
         try:
@@ -97,6 +107,7 @@ def make_file(path: str, contents: str = '', overwrite: bool = False) -> bool:
     exists. The file is saved with contents `contents`
     """
     if overwrite or not os.path.exists(path):
+        path = convert_path(path)
         with open(path, 'w') as fh:
             fh.write(contents)
         return True
@@ -119,13 +130,42 @@ def supports_symlinks() -> bool:
 
 
 def write_file(path: str, contents: str = '') -> bool:
-    make_directory(os.path.dirname(path))
-    with open(path, 'w', encoding='utf-8') as f:
-        f.write(str(contents))
-
+    path = convert_path(path)
+    try:
+        make_directory(os.path.dirname(path))
+        with open(path, 'w', encoding='utf-8') as f:
+            f.write(str(contents))
+    except Exception as exc:
+        # note that you can't just catch FileNotFound, because sometimes
+        # windows apparently raises something else.
+        # It's also not sufficient to look at the path length, because
+        # sometimes windows fails to write paths that are less than the length
+        # limit. So on windows, suppress all errors that happen from writing
+        # to disk.
+        if os.name == 'nt':
+            # sometimes we get a winerror of 3 which means the path was
+            # definitely too long, but other times we don't and it means the
+            # path was just probably too long. This is probably based on the
+            # windows/python version.
+            if getattr(exc, 'winerror', 0) == 3:
+                reason = 'Path was too long'
+            else:
+                reason = 'Path was possibly too long'
+            # all our hard work and the path was still too long. Log and
+            # continue.
+            logger.debug(
+                f'Could not write to path {path}({len(path)} characters): '
+                f'{reason}\nexception: {exc}'
+            )
+        else:
+            raise
     return True
 
 
+def read_json(path: str) -> Dict[str, Any]:
+    return json.loads(load_file_contents(path))
+
+
 def write_json(path: str, data: Dict[str, Any]) -> bool:
     return write_file(path, json.dumps(data, cls=dbt.utils.JSONEncoder))
 
@@ -161,7 +201,7 @@ def rmdir(path: str) -> None:
     different permissions on Windows. Otherwise, removing directories (eg.
     cloned via git) can cause rmtree to throw a PermissionError exception
     """
-    logger.debug("DEBUG** Window rmdir sys.platform: {}".format(sys.platform))
+    path = convert_path(path)
     if sys.platform == 'win32':
         onerror = _windows_rmdir_readonly
     else:
@@ -170,15 +210,90 @@ def rmdir(path: str) -> None:
     shutil.rmtree(path, onerror=onerror)
 
 
+def _win_prepare_path(path: str) -> str:
+    """Given a windows path, prepare it for use by making sure it is absolute
+    and normalized.
+    """
+    path = os.path.normpath(path)
+
+    # if a path starts with '\', splitdrive() on it will return '' for the
+    # drive, but the prefix requires a drive letter. So let's add the drive
+    # letter back in.
+    # Unless it starts with '\\'. In that case, the path is a UNC mount point
+    # and splitdrive will be fine.
+    if not path.startswith('\\\\') and path.startswith('\\'):
+        curdrive = os.path.splitdrive(os.getcwd())[0]
+        path = curdrive + path
+
+    # now our path is either an absolute UNC path or relative to the current
+    # directory. If it's relative, we need to make it absolute or the prefix
+    # won't work. `ntpath.abspath` allegedly doesn't always play nice with long
+    # paths, so do this instead.
+    if not os.path.splitdrive(path)[0]:
+        path = os.path.join(os.getcwd(), path)
+
+    return path
+
+
+def _supports_long_paths() -> bool:
+    if sys.platform != 'win32':
+        return True
+    # Eryk Sun says to use `WinDLL('ntdll')` instead of `windll.ntdll` because
+    # of pointer caching in a comment here:
+    # https://stackoverflow.com/a/35097999/11262881
+    # I don't know exaclty what he means, but I am inclined to believe him as
+    # he's pretty active on Python windows bugs!
+    try:
+        dll = WinDLL('ntdll')
+    except OSError:  # I don't think this happens? you need ntdll to run python
+        return False
+    # not all windows versions have it at all
+    if not hasattr(dll, 'RtlAreLongPathsEnabled'):
+        return False
+    # tell windows we want to get back a single unsigned byte (a bool).
+    dll.RtlAreLongPathsEnabled.restype = c_bool
+    return dll.RtlAreLongPathsEnabled()
+
+
+def convert_path(path: str) -> str:
+    """Convert a path that dbt has, which might be >260 characters long, to one
+    that will be writable/readable on Windows.
+
+    On other platforms, this is a no-op.
+    """
+    # some parts of python seem to append '\*.*' to strings, better safe than
+    # sorry.
+    if len(path) < 250:
+        return path
+    if _supports_long_paths():
+        return path
+
+    prefix = '\\\\?\\'
+    # Nothing to do
+    if path.startswith(prefix):
+        return path
+
+    path = _win_prepare_path(path)
+
+    # add the prefix. The check is just in case os.getcwd() does something
+    # unexpected - I believe this if-state should always be True though!
+    if not path.startswith(prefix):
+        path = prefix + path
+    return path
+
+
 def remove_file(path: str) -> None:
+    path = convert_path(path)
     os.remove(path)
 
 
 def path_exists(path: str) -> bool:
+    path = convert_path(path)
     return os.path.lexists(path)
 
 
 def path_is_symlink(path: str) -> bool:
+    path = convert_path(path)
     return os.path.islink(path)
 
 
@@ -301,6 +416,9 @@ def run_cmd(
         full_env.update(env)
 
     try:
+        exe_pth = shutil.which(cmd[0])
+        if exe_pth:
+            cmd = [os.path.abspath(exe_pth)] + list(cmd[1:])
         proc = subprocess.Popen(
             cmd,
             cwd=cwd,
@@ -323,14 +441,27 @@ def run_cmd(
     return out, err
 
 
-def download(url: str, path: str) -> None:
-    response = requests.get(url)
+def download_with_retries(
+    url: str, path: str, timeout: Optional[Union[float, tuple]] = None
+) -> None:
+    download_fn = functools.partial(download, url, path, timeout)
+    connection_exception_retry(download_fn, 5)
+
+
+def download(
+    url: str, path: str, timeout: Optional[Union[float, tuple]] = None
+) -> None:
+    path = convert_path(path)
+    connection_timeout = timeout or float(os.getenv('DBT_HTTP_TIMEOUT', 10))
+    response = requests.get(url, timeout=connection_timeout)
     with open(path, 'wb') as handle:
         for block in response.iter_content(1024 * 64):
             handle.write(block)
 
 
 def rename(from_path: str, to_path: str, force: bool = False) -> None:
+    from_path = convert_path(from_path)
+    to_path = convert_path(to_path)
     is_symlink = path_is_symlink(to_path)
 
     if os.path.exists(to_path) and force:
@@ -345,6 +476,7 @@ def rename(from_path: str, to_path: str, force: bool = False) -> None:
 def untar_package(
     tar_path: str, dest_dir: str, rename_to: Optional[str] = None
 ) -> None:
+    tar_path = convert_path(tar_path)
     tar_dir_name = None
     with tarfile.open(tar_path, 'r') as tarball:
         tarball.extractall(dest_dir)
@@ -381,6 +513,8 @@ def move(src, dst):
     This is almost identical to the real shutil.move, except it uses our rmtree
     and skips handling non-windows OSes since the existing one works ok there.
     """
+    src = convert_path(src)
+    dst = convert_path(dst)
     if os.name != 'nt':
         return shutil.move(src, dst)
 
@@ -415,4 +549,5 @@ def rmtree(path):
     """Recursively remove path. On permissions errors on windows, try to remove
     the read-only flag and try again.
     """
+    path = convert_path(path)
     return shutil.rmtree(path, onerror=chmod_and_retry)

core/dbt/clients/yaml_helper.py

@@ -1,8 +1,20 @@
 import dbt.exceptions
-
+from typing import Any, Dict, Optional
 import yaml
 import yaml.scanner
 
+# the C version is faster, but it doesn't always exist
+try:
+    from yaml import (
+        CLoader as Loader,
+        CSafeLoader as SafeLoader,
+        CDumper as Dumper
+    )
+except ImportError:
+    from yaml import (  # type: ignore  # noqa: F401
+        Loader, SafeLoader, Dumper
+    )
+
 
 YAML_ERROR_MESSAGE = """
 Syntax error near line {line_number}
@@ -44,9 +56,13 @@ def contextualized_yaml_error(raw_contents, error):
                                      raw_error=error)
 
 
+def safe_load(contents) -> Optional[Dict[str, Any]]:
+    return yaml.load(contents, Loader=SafeLoader)
+
+
 def load_yaml_text(contents):
     try:
-        return yaml.safe_load(contents)
+        return safe_load(contents)
     except (yaml.scanner.ScannerError, yaml.YAMLError) as e:
         if hasattr(e, 'problem_mark'):
             error = contextualized_yaml_error(contents, e)

core/dbt/compilation.py

@@ -1,35 +1,45 @@
-import itertools
 import os
 from collections import defaultdict
-from typing import List, Dict
+from typing import List, Dict, Any, Tuple, cast, Optional
 
-import dbt.utils
-import dbt.include
-import dbt.tracking
-
-from dbt.node_types import NodeType
-from dbt.linker import Linker
+import networkx as nx  # type: ignore
+import sqlparse
 
+from dbt import flags
+from dbt.adapters.factory import get_adapter
+from dbt.clients import jinja
+from dbt.clients.system import make_directory
 from dbt.context.providers import generate_runtime_model
-import dbt.contracts.project
-import dbt.exceptions
-import dbt.flags
-import dbt.config
-from dbt.contracts.graph.compiled import InjectedCTE, COMPILED_TYPES
+from dbt.contracts.graph.manifest import Manifest
+from dbt.contracts.graph.compiled import (
+    COMPILED_TYPES,
+    CompiledSchemaTestNode,
+    GraphMemberNode,
+    InjectedCTE,
+    ManifestNode,
+    NonSourceCompiledNode,
+)
 from dbt.contracts.graph.parsed import ParsedNode
-
+from dbt.exceptions import (
+    dependency_not_found,
+    InternalException,
+    RuntimeException,
+)
+from dbt.graph import Graph
 from dbt.logger import GLOBAL_LOGGER as logger
+from dbt.node_types import NodeType
+from dbt.utils import pluralize
+import dbt.tracking
 
 graph_file_name = 'graph.gpickle'
 
 
 def _compiled_type_for(model: ParsedNode):
-    if model.resource_type not in COMPILED_TYPES:
-        raise dbt.exceptions.InternalException(
-            'Asked to compile {} node, but it has no compiled form'
-            .format(model.resource_type)
+    if type(model) not in COMPILED_TYPES:
+        raise InternalException(
+            f'Asked to compile {type(model)} node, but it has no compiled form'
         )
-    return COMPILED_TYPES[model.resource_type]
+    return COMPILED_TYPES[type(model)]
 
 
 def print_compile_stats(stats):
@@ -42,20 +52,26 @@ def print_compile_stats(stats):
         NodeType.Operation: 'operation',
         NodeType.Seed: 'seed file',
         NodeType.Source: 'source',
+        NodeType.Exposure: 'exposure',
     }
 
     results = {k: 0 for k in names.keys()}
     results.update(stats)
 
+    # create tracking event for resource_counts
+    if dbt.tracking.active_user is not None:
+        resource_counts = {k.pluralize(): v for k, v in results.items()}
+        dbt.tracking.track_resource_counts(resource_counts)
+
     stat_line = ", ".join([
-        dbt.utils.pluralize(ct, names.get(t)) for t, ct in results.items()
+        pluralize(ct, names.get(t)) for t, ct in results.items()
         if t in names
     ])
 
     logger.info("Found {}".format(stat_line))
 
 
-def _node_enabled(node):
+def _node_enabled(node: ManifestNode):
     # Disabled models are already excluded from the manifest
     if node.resource_type == NodeType.Test and not node.config.enabled:
         return False
@@ -63,14 +79,18 @@ def _node_enabled(node):
         return True
 
 
-def _generate_stats(manifest):
+def _generate_stats(manifest: Manifest):
     stats: Dict[NodeType, int] = defaultdict(int)
-    for node_name, node in itertools.chain(
-            manifest.nodes.items(),
-            manifest.macros.items()):
+    for node in manifest.nodes.values():
         if _node_enabled(node):
             stats[node.resource_type] += 1
 
+    for source in manifest.sources.values():
+        stats[source.resource_type] += 1
+    for exposure in manifest.exposures.values():
+        stats[exposure.resource_type] += 1
+    for macro in manifest.macros.values():
+        stats[macro.resource_type] += 1
     return stats
 
 
@@ -87,39 +107,45 @@ def _extend_prepended_ctes(prepended_ctes, new_prepended_ctes):
         _add_prepended_cte(prepended_ctes, new_cte)
 
 
-def prepend_ctes(model, manifest):
-    model, _, manifest = recursively_prepend_ctes(model, manifest)
+class Linker:
+    def __init__(self, data=None):
+        if data is None:
+            data = {}
+        self.graph = nx.DiGraph(**data)
 
-    return (model, manifest)
+    def edges(self):
+        return self.graph.edges()
 
+    def nodes(self):
+        return self.graph.nodes()
 
-def recursively_prepend_ctes(model, manifest):
-    if model.extra_ctes_injected:
-        return (model, model.extra_ctes, manifest)
+    def find_cycles(self):
+        try:
+            cycle = nx.find_cycle(self.graph)
+        except nx.NetworkXNoCycle:
+            return None
+        else:
+            # cycles is a List[Tuple[str, ...]]
+            return " --> ".join(c[0] for c in cycle)
 
-    if dbt.flags.STRICT_MODE:
-        if not isinstance(model, tuple(COMPILED_TYPES.values())):
-            raise dbt.exceptions.InternalException(
-                'Bad model type: {}'.format(type(model))
-            )
+    def dependency(self, node1, node2):
+        "indicate that node1 depends on node2"
+        self.graph.add_node(node1)
+        self.graph.add_node(node2)
+        self.graph.add_edge(node2, node1)
 
-    prepended_ctes: List[InjectedCTE] = []
+    def add_node(self, node):
+        self.graph.add_node(node)
 
-    for cte in model.extra_ctes:
-        cte_id = cte.id
-        cte_to_add = manifest.nodes.get(cte_id)
-        cte_to_add, new_prepended_ctes, manifest = recursively_prepend_ctes(
-            cte_to_add, manifest)
-        _extend_prepended_ctes(prepended_ctes, new_prepended_ctes)
-        new_cte_name = '__dbt__CTE__{}'.format(cte_to_add.name)
-        sql = ' {} as (\n{}\n)'.format(new_cte_name, cte_to_add.compiled_sql)
-        _add_prepended_cte(prepended_ctes, InjectedCTE(id=cte_id, sql=sql))
-
-    model.prepend_ctes(prepended_ctes)
-
-    manifest.update_node(model)
-
-    return (model, prepended_ctes, manifest)
+    def write_graph(self, outfile: str, manifest: Manifest):
+        """Write the graph to a gpickle file. Before doing so, serialize and
+        include all nodes in their corresponding graph entries.
+        """
+        out_graph = self.graph.copy()
+        for node_id in self.graph.nodes():
+            data = manifest.expect(node_id).to_dict(omit_none=True)
+            out_graph.add_node(node_id, **data)
+        nx.write_gpickle(out_graph, outfile)
 
 
 class Compiler:
@@ -127,95 +153,274 @@ class Compiler:
         self.config = config
 
     def initialize(self):
-        dbt.clients.system.make_directory(self.config.target_path)
-        dbt.clients.system.make_directory(self.config.modules_path)
+        make_directory(self.config.target_path)
+        make_directory(self.config.modules_path)
 
-    def compile_node(self, node, manifest, extra_context=None):
+    # creates a ModelContext which is converted to
+    # a dict for jinja rendering of SQL
+    def _create_node_context(
+        self,
+        node: NonSourceCompiledNode,
+        manifest: Manifest,
+        extra_context: Dict[str, Any],
+    ) -> Dict[str, Any]:
+
+        context = generate_runtime_model(
+            node, self.config, manifest
+        )
+        context.update(extra_context)
+        if isinstance(node, CompiledSchemaTestNode):
+            # for test nodes, add a special keyword args value to the context
+            jinja.add_rendered_test_kwargs(context, node)
+
+        return context
+
+    def add_ephemeral_prefix(self, name: str):
+        adapter = get_adapter(self.config)
+        relation_cls = adapter.Relation
+        return relation_cls.add_ephemeral_prefix(name)
+
+    def _get_relation_name(self, node: ParsedNode):
+        relation_name = None
+        if node.is_relational and not node.is_ephemeral_model:
+            adapter = get_adapter(self.config)
+            relation_cls = adapter.Relation
+            relation_name = str(relation_cls.create_from(self.config, node))
+        return relation_name
+
+    def _inject_ctes_into_sql(self, sql: str, ctes: List[InjectedCTE]) -> str:
+        """
+        `ctes` is a list of InjectedCTEs like:
+
+            [
+                InjectedCTE(
+                    id="cte_id_1",
+                    sql="__dbt__cte__ephemeral as (select * from table)",
+                ),
+                InjectedCTE(
+                    id="cte_id_2",
+                    sql="__dbt__cte__events as (select id, type from events)",
+                ),
+            ]
+
+        Given `sql` like:
+
+          "with internal_cte as (select * from sessions)
+           select * from internal_cte"
+
+        This will spit out:
+
+          "with __dbt__cte__ephemeral as (select * from table),
+                __dbt__cte__events as (select id, type from events),
+                with internal_cte as (select * from sessions)
+           select * from internal_cte"
+
+        (Whitespace enhanced for readability.)
+        """
+        if len(ctes) == 0:
+            return sql
+
+        parsed_stmts = sqlparse.parse(sql)
+        parsed = parsed_stmts[0]
+
+        with_stmt = None
+        for token in parsed.tokens:
+            if token.is_keyword and token.normalized == 'WITH':
+                with_stmt = token
+                break
+
+        if with_stmt is None:
+            # no with stmt, add one, and inject CTEs right at the beginning
+            first_token = parsed.token_first()
+            with_stmt = sqlparse.sql.Token(sqlparse.tokens.Keyword, 'with')
+            parsed.insert_before(first_token, with_stmt)
+        else:
+            # stmt exists, add a comma (which will come after injected CTEs)
+            trailing_comma = sqlparse.sql.Token(
+                sqlparse.tokens.Punctuation, ','
+            )
+            parsed.insert_after(with_stmt, trailing_comma)
+
+        token = sqlparse.sql.Token(
+            sqlparse.tokens.Keyword,
+            ", ".join(c.sql for c in ctes)
+        )
+        parsed.insert_after(with_stmt, token)
+
+        return str(parsed)
+
+    def _recursively_prepend_ctes(
+        self,
+        model: NonSourceCompiledNode,
+        manifest: Manifest,
+        extra_context: Optional[Dict[str, Any]],
+    ) -> Tuple[NonSourceCompiledNode, List[InjectedCTE]]:
+        """This method is called by the 'compile_node' method. Starting
+        from the node that it is passed in, it will recursively call
+        itself using the 'extra_ctes'.  The 'ephemeral' models do
+        not produce SQL that is executed directly, instead they
+        are rolled up into the models that refer to them by
+        inserting CTEs into the SQL.
+        """
+        if model.compiled_sql is None:
+            raise RuntimeException(
+                'Cannot inject ctes into an unparsed node', model
+            )
+        if model.extra_ctes_injected:
+            return (model, model.extra_ctes)
+
+        # Just to make it plain that nothing is actually injected for this case
+        if not model.extra_ctes:
+            model.extra_ctes_injected = True
+            manifest.update_node(model)
+            return (model, model.extra_ctes)
+
+        # This stores the ctes which will all be recursively
+        # gathered and then "injected" into the model.
+        prepended_ctes: List[InjectedCTE] = []
+
+        # extra_ctes are added to the model by
+        # RuntimeRefResolver.create_relation, which adds an
+        # extra_cte for every model relation which is an
+        # ephemeral model.
+        for cte in model.extra_ctes:
+            if cte.id not in manifest.nodes:
+                raise InternalException(
+                    f'During compilation, found a cte reference that '
+                    f'could not be resolved: {cte.id}'
+                )
+            cte_model = manifest.nodes[cte.id]
+
+            if not cte_model.is_ephemeral_model:
+                raise InternalException(f'{cte.id} is not ephemeral')
+
+            # This model has already been compiled, so it's been
+            # through here before
+            if getattr(cte_model, 'compiled', False):
+                assert isinstance(cte_model, tuple(COMPILED_TYPES.values()))
+                cte_model = cast(NonSourceCompiledNode, cte_model)
+                new_prepended_ctes = cte_model.extra_ctes
+
+            # if the cte_model isn't compiled, i.e. first time here
+            else:
+                # This is an ephemeral parsed model that we can compile.
+                # Compile and update the node
+                cte_model = self._compile_node(
+                    cte_model, manifest, extra_context)
+                # recursively call this method
+                cte_model, new_prepended_ctes = \
+                    self._recursively_prepend_ctes(
+                        cte_model, manifest, extra_context
+                    )
+                # Save compiled SQL file and sync manifest
+                self._write_node(cte_model)
+                manifest.sync_update_node(cte_model)
+
+            _extend_prepended_ctes(prepended_ctes, new_prepended_ctes)
+
+            new_cte_name = self.add_ephemeral_prefix(cte_model.name)
+            rendered_sql = (
+                cte_model._pre_injected_sql or cte_model.compiled_sql
+            )
+            sql = f' {new_cte_name} as (\n{rendered_sql}\n)'
+
+            _add_prepended_cte(prepended_ctes, InjectedCTE(id=cte.id, sql=sql))
+
+        injected_sql = self._inject_ctes_into_sql(
+            model.compiled_sql,
+            prepended_ctes,
+        )
+        model._pre_injected_sql = model.compiled_sql
+        model.compiled_sql = injected_sql
+        model.extra_ctes_injected = True
+        model.extra_ctes = prepended_ctes
+        model.validate(model.to_dict(omit_none=True))
+
+        manifest.update_node(model)
+
+        return model, prepended_ctes
+
+    # creates a compiled_node from the ManifestNode passed in,
+    # creates a "context" dictionary for jinja rendering,
+    # and then renders the "compiled_sql" using the node, the
+    # raw_sql and the context.
+    def _compile_node(
+        self,
+        node: ManifestNode,
+        manifest: Manifest,
+        extra_context: Optional[Dict[str, Any]] = None,
+    ) -> NonSourceCompiledNode:
         if extra_context is None:
             extra_context = {}
 
         logger.debug("Compiling {}".format(node.unique_id))
 
-        data = node.to_dict()
+        data = node.to_dict(omit_none=True)
         data.update({
             'compiled': False,
             'compiled_sql': None,
             'extra_ctes_injected': False,
             'extra_ctes': [],
-            'injected_sql': None,
         })
         compiled_node = _compiled_type_for(node).from_dict(data)
 
-        context = generate_runtime_model(
-            compiled_node, self.config, manifest
+        context = self._create_node_context(
+            compiled_node, manifest, extra_context
         )
-        context.update(extra_context)
 
-        compiled_node.compiled_sql = dbt.clients.jinja.get_rendered(
+        compiled_node.compiled_sql = jinja.get_rendered(
             node.raw_sql,
             context,
-            node)
+            node,
+        )
+
+        compiled_node.relation_name = self._get_relation_name(node)
 
         compiled_node.compiled = True
 
-        injected_node, _ = prepend_ctes(compiled_node, manifest)
+        return compiled_node
 
-        should_wrap = {NodeType.Test, NodeType.Operation}
-        if injected_node.resource_type in should_wrap:
-            # data tests get wrapped in count(*)
-            # TODO : move this somewhere more reasonable
-            if 'data' in injected_node.tags and \
-               injected_node.resource_type == NodeType.Test:
-                injected_node.wrapped_sql = (
-                    "select count(*) as errors "
-                    "from (\n{test_sql}\n) sbq").format(
-                        test_sql=injected_node.injected_sql)
-            else:
-                # don't wrap schema tests or analyses.
-                injected_node.wrapped_sql = injected_node.injected_sql
-
-        elif injected_node.resource_type == NodeType.Snapshot:
-            # unfortunately we do everything automagically for
-            # snapshots. in the future it'd be nice to generate
-            # the SQL at the parser level.
-            pass
-
-        elif(injected_node.resource_type == NodeType.Model and
-             injected_node.get_materialization() == 'ephemeral'):
-            pass
-
-        else:
-            injected_node.wrapped_sql = None
-
-        return injected_node
-
-    def write_graph_file(self, linker, manifest):
+    def write_graph_file(self, linker: Linker, manifest: Manifest):
         filename = graph_file_name
         graph_path = os.path.join(self.config.target_path, filename)
-        if dbt.flags.WRITE_JSON:
+        if flags.WRITE_JSON:
             linker.write_graph(graph_path, manifest)
 
-    def link_node(self, linker, node, manifest):
+    def link_node(
+        self, linker: Linker, node: GraphMemberNode, manifest: Manifest
+    ):
         linker.add_node(node.unique_id)
 
         for dependency in node.depends_on_nodes:
-            if manifest.nodes.get(dependency):
+            if dependency in manifest.nodes:
                 linker.dependency(
                     node.unique_id,
-                    (manifest.nodes.get(dependency).unique_id))
+                    (manifest.nodes[dependency].unique_id)
+                )
+            elif dependency in manifest.sources:
+                linker.dependency(
+                    node.unique_id,
+                    (manifest.sources[dependency].unique_id)
+                )
             else:
-                dbt.exceptions.dependency_not_found(node, dependency)
+                dependency_not_found(node, dependency)
 
-    def link_graph(self, linker, manifest):
+    def link_graph(self, linker: Linker, manifest: Manifest):
+        for source in manifest.sources.values():
+            linker.add_node(source.unique_id)
         for node in manifest.nodes.values():
             self.link_node(linker, node, manifest)
+        for exposure in manifest.exposures.values():
+            self.link_node(linker, exposure, manifest)
+            # linker.add_node(exposure.unique_id)
 
         cycle = linker.find_cycles()
 
         if cycle:
             raise RuntimeError("Found a cycle: {}".format(cycle))
 
-    def compile(self, manifest, write=True):
+    def compile(self, manifest: Manifest, write=True) -> Graph:
+        self.initialize()
         linker = Linker()
 
         self.link_graph(linker, manifest)
@@ -226,37 +431,41 @@ class Compiler:
             self.write_graph_file(linker, manifest)
         print_compile_stats(stats)
 
-        return linker
+        return Graph(linker.graph)
 
+    # writes the "compiled_sql" into the target/compiled directory
+    def _write_node(self, node: NonSourceCompiledNode) -> ManifestNode:
+        if (not node.extra_ctes_injected or
+                node.resource_type == NodeType.Snapshot):
+            return node
+        logger.debug(f'Writing injected SQL for node "{node.unique_id}"')
 
-def compile_manifest(config, manifest, write=True) -> Linker:
-    compiler = Compiler(config)
-    compiler.initialize()
-    return compiler.compile(manifest, write=write)
+        if node.compiled_sql:
+            node.compiled_path = node.write_node(
+                self.config.target_path,
+                'compiled',
+                node.compiled_sql
+            )
+        return node
 
+    def compile_node(
+        self,
+        node: ManifestNode,
+        manifest: Manifest,
+        extra_context: Optional[Dict[str, Any]] = None,
+        write: bool = True,
+    ) -> NonSourceCompiledNode:
+        """This is the main entry point into this code. It's called by
+        CompileRunner.compile, GenericRPCRunner.compile, and
+        RunTask.get_hook_sql. It calls '_compile_node' to convert
+        the node into a compiled node, and then calls the
+        recursive method to "prepend" the ctes.
+        """
+        node = self._compile_node(node, manifest, extra_context)
 
-def _is_writable(node):
-    if not node.injected_sql:
-        return False
-
-    if node.resource_type == NodeType.Snapshot:
-        return False
-
-    return True
-
-
-def compile_node(adapter, config, node, manifest, extra_context, write=True):
-    compiler = Compiler(config)
-    node = compiler.compile_node(node, manifest, extra_context)
-
-    if write and _is_writable(node):
-        logger.debug('Writing injected SQL for node "{}"'.format(
-            node.unique_id))
-
-        node.build_path = node.write_node(
-            config.target_path,
-            'compiled',
-            node.injected_sql
+        node, _ = self._recursively_prepend_ctes(
+            node, manifest, extra_context
         )
-
-    return node
+        if write:
+            self._write_node(node)
+        return node

core/dbt/config/__init__.py

@@ -1,5 +1,4 @@
 # all these are just exports, they need "noqa" so flake8 will not complain.
 from .profile import Profile, PROFILES_DIR, read_user_config  # noqa
-from .project import Project  # noqa
-from .runtime import RuntimeConfig  # noqa
-from .renderer import ConfigRenderer  # noqa
+from .project import Project, IsFQNResource  # noqa
+from .runtime import RuntimeConfig, UnsetProfileConfig  # noqa

core/dbt/config/profile.py

@@ -2,12 +2,13 @@ from dataclasses import dataclass
 from typing import Any, Dict, Optional, Tuple
 import os
 
-from hologram import ValidationError
+from dbt.dataclass_schema import ValidationError
 
 from dbt.clients.system import load_file_contents
 from dbt.clients.yaml_helper import load_yaml_text
 from dbt.contracts.connection import Credentials, HasCredentials
 from dbt.contracts.project import ProfileConfig, UserConfig
+from dbt.exceptions import CompilationException
 from dbt.exceptions import DbtProfileError
 from dbt.exceptions import DbtProjectError
 from dbt.exceptions import ValidationException
@@ -16,7 +17,7 @@ from dbt.exceptions import validator_error_message
 from dbt.logger import GLOBAL_LOGGER as logger
 from dbt.utils import coerce_dict_str
 
-from .renderer import ConfigRenderer
+from .renderer import ProfileRenderer
 
 DEFAULT_THREADS = 1
 DEFAULT_PROFILES_DIR = os.path.join(os.path.expanduser('~'), '.dbt')
@@ -52,7 +53,15 @@ def read_profile(profiles_dir: str) -> Dict[str, Any]:
     if os.path.isfile(path):
         try:
             contents = load_file_contents(path, strip=False)
-            return load_yaml_text(contents)
+            yaml_content = load_yaml_text(contents)
+            if not yaml_content:
+                msg = f'The profiles.yml file at {path} is empty'
+                raise DbtProfileError(
+                    INVALID_PROFILE_MESSAGE.format(
+                        error_string=msg
+                    )
+                )
+            return yaml_content
         except ValidationException as e:
             msg = INVALID_PROFILE_MESSAGE.format(error_string=e)
             raise ValidationException(msg) from e
@@ -66,13 +75,17 @@ def read_user_config(directory: str) -> UserConfig:
         if profile:
             user_cfg = coerce_dict_str(profile.get('config', {}))
             if user_cfg is not None:
+                UserConfig.validate(user_cfg)
                 return UserConfig.from_dict(user_cfg)
     except (RuntimeException, ValidationError):
         pass
     return UserConfig()
 
 
-@dataclass
+# The Profile class is included in RuntimeConfig, so any attribute
+# additions must also be set where the RuntimeConfig class is created
+# `init=False` is a workaround for https://bugs.python.org/issue45081
+@dataclass(init=False)
 class Profile(HasCredentials):
     profile_name: str
     target_name: str
@@ -80,6 +93,23 @@ class Profile(HasCredentials):
     threads: int
     credentials: Credentials
 
+    def __init__(
+        self,
+        profile_name: str,
+        target_name: str,
+        config: UserConfig,
+        threads: int,
+        credentials: Credentials
+    ):
+        """Explicitly defining `__init__` to work around bug in Python 3.9.7
+        https://bugs.python.org/issue45081
+        """
+        self.profile_name = profile_name
+        self.target_name = target_name
+        self.config = config
+        self.threads = threads
+        self.credentials = credentials
+
     def to_profile_info(
         self, serialize_credentials: bool = False
     ) -> Dict[str, Any]:
@@ -99,10 +129,24 @@ class Profile(HasCredentials):
             'credentials': self.credentials,
         }
         if serialize_credentials:
-            result['config'] = self.config.to_dict()
-            result['credentials'] = self.credentials.to_dict()
+            result['config'] = self.config.to_dict(omit_none=True)
+            result['credentials'] = self.credentials.to_dict(omit_none=True)
         return result
 
+    def to_target_dict(self) -> Dict[str, Any]:
+        target = dict(
+            self.credentials.connection_info(with_aliases=True)
+        )
+        target.update({
+            'type': self.credentials.type,
+            'threads': self.threads,
+            'name': self.target_name,
+            'target_name': self.target_name,
+            'profile_name': self.profile_name,
+            'config': self.config.to_dict(omit_none=True),
+        })
+        return target
+
     def __eq__(self, other: object) -> bool:
         if not (isinstance(other, self.__class__) and
                 isinstance(self, other.__class__)):
@@ -112,10 +156,10 @@ class Profile(HasCredentials):
     def validate(self):
         try:
             if self.credentials:
-                self.credentials.to_dict(validate=True)
-            ProfileConfig.from_dict(
-                self.to_profile_info(serialize_credentials=True)
-            )
+                dct = self.credentials.to_dict(omit_none=True)
+                self.credentials.validate(dct)
+            dct = self.to_profile_info(serialize_credentials=True)
+            ProfileConfig.validate(dct)
         except ValidationError as exc:
             raise DbtProfileError(validator_error_message(exc)) from exc
 
@@ -135,7 +179,9 @@ class Profile(HasCredentials):
         typename = profile.pop('type')
         try:
             cls = load_plugin(typename)
-            credentials = cls.from_dict(profile)
+            data = cls.translate_aliases(profile)
+            cls.validate(data)
+            credentials = cls.from_dict(data)
         except (RuntimeException, ValidationError) as e:
             msg = str(e) if isinstance(e, RuntimeException) else e.message
             raise DbtProfileError(
@@ -175,6 +221,14 @@ class Profile(HasCredentials):
                    .format(profile_name, target_name, outputs))
             raise DbtProfileError(msg, result_type='invalid_target')
         profile_data = outputs[target_name]
+
+        if not isinstance(profile_data, dict):
+            msg = (
+                f"output '{target_name}' of profile '{profile_name}' is "
+                f"misconfigured in profiles.yml"
+            )
+            raise DbtProfileError(msg, result_type='invalid_target')
+
         return profile_data
 
     @classmethod
@@ -200,6 +254,7 @@ class Profile(HasCredentials):
         """
         if user_cfg is None:
             user_cfg = {}
+        UserConfig.validate(user_cfg)
         config = UserConfig.from_dict(user_cfg)
 
         profile = cls(
@@ -218,7 +273,7 @@ class Profile(HasCredentials):
         raw_profile: Dict[str, Any],
         profile_name: str,
         target_override: Optional[str],
-        renderer: ConfigRenderer,
+        renderer: ProfileRenderer,
     ) -> Tuple[str, Dict[str, Any]]:
         """This is a containment zone for the hateful way we're rendering
         profiles.
@@ -246,7 +301,10 @@ class Profile(HasCredentials):
             raw_profile, profile_name, target_name
         )
 
-        profile_data = renderer.render_profile_data(raw_profile_data)
+        try:
+            profile_data = renderer.render_data(raw_profile_data)
+        except CompilationException as exc:
+            raise DbtProfileError(str(exc)) from exc
         return target_name, profile_data
 
     @classmethod
@@ -254,7 +312,7 @@ class Profile(HasCredentials):
         cls,
         raw_profile: Dict[str, Any],
         profile_name: str,
-        renderer: ConfigRenderer,
+        renderer: ProfileRenderer,
         user_cfg: Optional[Dict[str, Any]] = None,
         target_override: Optional[str] = None,
         threads_override: Optional[int] = None,
@@ -280,7 +338,6 @@ class Profile(HasCredentials):
         # user_cfg is not rendered.
         if user_cfg is None:
             user_cfg = raw_profile.get('config')
-
         # TODO: should it be, and the values coerced to bool?
         target_name, profile_data = cls.render_profile(
             raw_profile, profile_name, target_override, renderer
@@ -309,7 +366,7 @@ class Profile(HasCredentials):
         cls,
         raw_profiles: Dict[str, Any],
         profile_name: str,
-        renderer: ConfigRenderer,
+        renderer: ProfileRenderer,
         target_override: Optional[str] = None,
         threads_override: Optional[int] = None,
     ) -> 'Profile':
@@ -335,7 +392,15 @@ class Profile(HasCredentials):
         # First, we've already got our final decision on profile name, and we
         # don't render keys, so we can pluck that out
         raw_profile = raw_profiles[profile_name]
-
+        if not raw_profile:
+            msg = (
+                f'Profile {profile_name} in profiles.yml is empty'
+            )
+            raise DbtProfileError(
+                INVALID_PROFILE_MESSAGE.format(
+                    error_string=msg
+                )
+            )
         user_cfg = raw_profiles.get('config')
 
         return cls.from_raw_profile_info(
@@ -351,7 +416,7 @@ class Profile(HasCredentials):
     def render_from_args(
         cls,
         args: Any,
-        renderer: ConfigRenderer,
+        renderer: ProfileRenderer,
         project_profile_name: Optional[str],
     ) -> 'Profile':
         """Given the raw profiles as read from disk and the name of the desired
@@ -373,7 +438,6 @@ class Profile(HasCredentials):
         raw_profiles = read_profile(args.profiles_dir)
         profile_name = cls.pick_profile_name(getattr(args, 'profile', None),
                                              project_profile_name)
-
         return cls.from_raw_profiles(
             raw_profiles=raw_profiles,
             profile_name=profile_name,

core/dbt/config/project.py

@@ -1,7 +1,11 @@
 from copy import deepcopy
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from itertools import chain
-from typing import List, Dict, Any, Optional, TypeVar, Union, Tuple, Callable
+from typing import (
+    List, Dict, Any, Optional, TypeVar, Union, Mapping,
+)
+from typing_extensions import Protocol, runtime_checkable
+
 import hashlib
 import os
 
@@ -9,35 +13,31 @@ from dbt.clients.system import resolve_path_from_base
 from dbt.clients.system import path_exists
 from dbt.clients.system import load_file_contents
 from dbt.clients.yaml_helper import load_yaml_text
+from dbt.contracts.connection import QueryComment
 from dbt.exceptions import DbtProjectError
-from dbt.exceptions import RecursionException
 from dbt.exceptions import SemverException
 from dbt.exceptions import validator_error_message
-from dbt.exceptions import warn_or_error
+from dbt.exceptions import RuntimeException
+from dbt.graph import SelectionSpec
 from dbt.helper_types import NoValue
 from dbt.semver import VersionSpecifier
 from dbt.semver import versions_compatible
 from dbt.version import get_installed_version
-from dbt.ui import printer
-from dbt.utils import deep_map
-from dbt.source_config import SourceConfig
-
+from dbt.utils import MultiDict
+from dbt.node_types import NodeType
+from dbt.config.selectors import SelectorDict
 from dbt.contracts.project import (
     Project as ProjectContract,
     SemverString,
 )
 from dbt.contracts.project import PackageConfig
-
-from hologram import ValidationError
-
-from .renderer import ConfigRenderer
-
-
-UNUSED_RESOURCE_CONFIGURATION_PATH_MESSAGE = """\
-WARNING: Configuration paths exist in your dbt_project.yml file which do not \
-apply to any resources.
-There are {} unused configuration paths:\n{}
-"""
+from dbt.dataclass_schema import ValidationError
+from .renderer import DbtProjectYamlRenderer
+from .selectors import (
+    selector_config_from_data,
+    selector_data_from_root,
+    SelectorConfig,
+)
 
 
 INVALID_VERSION_ERROR = """\
@@ -69,23 +69,11 @@ Validator Error:
 """
 
 
-def _list_if_none(value):
-    if value is None:
-        value = []
-    return value
-
-
-def _dict_if_none(value):
-    if value is None:
-        value = {}
-    return value
-
-
-def _list_if_none_or_string(value):
-    value = _list_if_none(value)
-    if isinstance(value, str):
-        return [value]
-    return value
+@runtime_checkable
+class IsFQNResource(Protocol):
+    fqn: List[str]
+    resource_type: NodeType
+    package_name: str
 
 
 def _load_yaml(path):
@@ -93,32 +81,6 @@ def _load_yaml(path):
     return load_yaml_text(contents)
 
 
-def _get_config_paths(config, path=(), paths=None):
-    if paths is None:
-        paths = set()
-
-    for key, value in config.items():
-        if isinstance(value, dict):
-            if key in SourceConfig.ConfigKeys:
-                if path not in paths:
-                    paths.add(path)
-            else:
-                _get_config_paths(value, path + (key,), paths)
-        else:
-            if path not in paths:
-                paths.add(path)
-
-    return frozenset(paths)
-
-
-def _is_config_used(path, fqns):
-    if fqns:
-        for fqn in fqns:
-            if len(path) <= len(fqn) and fqn[:len(path)] == path:
-                return True
-    return False
-
-
 def package_data_from_root(project_root):
     package_filepath = resolve_path_from_base(
         'packages.yml', project_root
@@ -131,11 +93,12 @@ def package_data_from_root(project_root):
     return packages_dict
 
 
-def package_config_from_data(packages_data):
-    if packages_data is None:
+def package_config_from_data(packages_data: Dict[str, Any]):
+    if not packages_data:
         packages_data = {'packages': []}
 
     try:
+        PackageConfig.validate(packages_data)
         packages = PackageConfig.from_dict(packages_data)
     except ValidationError as e:
         raise DbtProjectError(
@@ -202,111 +165,151 @@ def _raw_project_from(project_root: str) -> Dict[str, Any]:
     return project_dict
 
 
-@dataclass
-class PartialProject:
-    profile_name: Optional[str]
-    project_name: Optional[str]
-    project_root: str
-    project_dict: Dict[str, Any]
+def _query_comment_from_cfg(
+        cfg_query_comment: Union[QueryComment, NoValue, str, None]
+) -> QueryComment:
+    if not cfg_query_comment:
+        return QueryComment(comment='')
 
-    def render(self, renderer):
-        packages_dict = package_data_from_root(self.project_root)
-        return Project.render_from_dict(
-            self.project_root,
-            self.project_dict,
-            packages_dict,
-            renderer,
+    if isinstance(cfg_query_comment, str):
+        return QueryComment(comment=cfg_query_comment)
+
+    if isinstance(cfg_query_comment, NoValue):
+        return QueryComment()
+
+    return cfg_query_comment
+
+
+def validate_version(dbt_version: List[VersionSpecifier], project_name: str):
+    """Ensure this package works with the installed version of dbt."""
+    installed = get_installed_version()
+    if not versions_compatible(*dbt_version):
+        msg = IMPOSSIBLE_VERSION_ERROR.format(
+            package=project_name,
+            version_spec=[
+                x.to_version_string() for x in dbt_version
+            ]
         )
+        raise DbtProjectError(msg)
 
-
-@dataclass
-class Project:
-    project_name: str
-    version: Union[SemverString, float]
-    project_root: str
-    profile_name: str
-    source_paths: List[str]
-    macro_paths: List[str]
-    data_paths: List[str]
-    test_paths: List[str]
-    analysis_paths: List[str]
-    docs_paths: List[str]
-    target_path: str
-    snapshot_paths: List[str]
-    clean_targets: List[str]
-    log_path: str
-    modules_path: str
-    quoting: Dict[str, Any]
-    models: Dict[str, Any]
-    on_run_start: List[str]
-    on_run_end: List[str]
-    seeds: Dict[str, Any]
-    snapshots: Dict[str, Any]
-    dbt_version: List[VersionSpecifier]
-    packages: Dict[str, Any]
-    query_comment: Optional[Union[str, NoValue]]
-
-    @property
-    def all_source_paths(self) -> List[str]:
-        return _all_source_paths(
-            self.source_paths, self.data_paths, self.snapshot_paths,
-            self.analysis_paths, self.macro_paths
+    if not versions_compatible(installed, *dbt_version):
+        msg = INVALID_VERSION_ERROR.format(
+            package=project_name,
+            installed=installed.to_version_string(),
+            version_spec=[
+                x.to_version_string() for x in dbt_version
+            ]
         )
+        raise DbtProjectError(msg)
 
-    @staticmethod
-    def _preprocess(project_dict: Dict[str, Any]) -> Dict[str, Any]:
-        """Pre-process certain special keys to convert them from None values
-        into empty containers, and to turn strings into arrays of strings.
-        """
-        handlers: Dict[Tuple[str, ...], Callable[[Any], Any]] = {
-            ('on-run-start',): _list_if_none_or_string,
-            ('on-run-end',): _list_if_none_or_string,
-        }
 
-        for k in ('models', 'seeds', 'snapshots'):
-            handlers[(k,)] = _dict_if_none
-            handlers[(k, 'vars')] = _dict_if_none
-            handlers[(k, 'pre-hook')] = _list_if_none_or_string
-            handlers[(k, 'post-hook')] = _list_if_none_or_string
-        handlers[('seeds', 'column_types')] = _dict_if_none
+def _get_required_version(
+    project_dict: Dict[str, Any],
+    verify_version: bool,
+) -> List[VersionSpecifier]:
+    dbt_raw_version: Union[List[str], str] = '>=0.0.0'
+    required = project_dict.get('require-dbt-version')
+    if required is not None:
+        dbt_raw_version = required
 
-        def converter(value: Any, keypath: Tuple[str, ...]) -> Any:
-            if keypath in handlers:
-                handler = handlers[keypath]
-                return handler(value)
-            else:
-                return value
+    try:
+        dbt_version = _parse_versions(dbt_raw_version)
+    except SemverException as e:
+        raise DbtProjectError(str(e)) from e
 
-        return deep_map(converter, project_dict)
-
-    @classmethod
-    def from_project_config(
-        cls,
-        project_dict: Dict[str, Any],
-        packages_dict: Optional[Dict[str, Any]] = None,
-    ) -> 'Project':
-        """Create a project from its project and package configuration, as read
-        by yaml.safe_load().
-
-        :param project_dict: The dictionary as read from disk
-        :param packages_dict: If it exists, the packages file as
-            read from disk.
-        :raises DbtProjectError: If the project is missing or invalid, or if
-            the packages file exists and is invalid.
-        :returns: The project, with defaults populated.
-        """
-        try:
-            project_dict = cls._preprocess(project_dict)
-        except RecursionException:
+    if verify_version:
+        # no name is also an error that we want to raise
+        if 'name' not in project_dict:
             raise DbtProjectError(
-                'Cycle detected: Project input has a reference to itself',
-                project=project_dict
+                'Required "name" field not present in project',
             )
+        validate_version(dbt_version, project_dict['name'])
+
+    return dbt_version
+
+
+@dataclass
+class RenderComponents:
+    project_dict: Dict[str, Any] = field(
+        metadata=dict(description='The project dictionary')
+    )
+    packages_dict: Dict[str, Any] = field(
+        metadata=dict(description='The packages dictionary')
+    )
+    selectors_dict: Dict[str, Any] = field(
+        metadata=dict(description='The selectors dictionary')
+    )
+
+
+@dataclass
+class PartialProject(RenderComponents):
+    profile_name: Optional[str] = field(metadata=dict(
+        description='The unrendered profile name in the project, if set'
+    ))
+    project_name: Optional[str] = field(metadata=dict(
+        description=(
+            'The name of the project. This should always be set and will not '
+            'be rendered'
+        )
+    ))
+    project_root: str = field(
+        metadata=dict(description='The root directory of the project'),
+    )
+    verify_version: bool = field(
+        metadata=dict(description=(
+            'If True, verify the dbt version matches the required version'
+        ))
+    )
+
+    def render_profile_name(self, renderer) -> Optional[str]:
+        if self.profile_name is None:
+            return None
+        return renderer.render_value(self.profile_name)
+
+    def get_rendered(
+        self,
+        renderer: DbtProjectYamlRenderer,
+    ) -> RenderComponents:
+
+        rendered_project = renderer.render_project(
+            self.project_dict, self.project_root
+        )
+        rendered_packages = renderer.render_packages(self.packages_dict)
+        rendered_selectors = renderer.render_selectors(self.selectors_dict)
+
+        return RenderComponents(
+            project_dict=rendered_project,
+            packages_dict=rendered_packages,
+            selectors_dict=rendered_selectors,
+        )
+
+    def render(self, renderer: DbtProjectYamlRenderer) -> 'Project':
         try:
-            cfg = ProjectContract.from_dict(project_dict)
+            rendered = self.get_rendered(renderer)
+            return self.create_project(rendered)
+        except DbtProjectError as exc:
+            if exc.path is None:
+                exc.path = os.path.join(self.project_root, 'dbt_project.yml')
+            raise
+
+    def create_project(self, rendered: RenderComponents) -> 'Project':
+        unrendered = RenderComponents(
+            project_dict=self.project_dict,
+            packages_dict=self.packages_dict,
+            selectors_dict=self.selectors_dict,
+        )
+        dbt_version = _get_required_version(
+            rendered.project_dict,
+            verify_version=self.verify_version,
+        )
+
+        try:
+            ProjectContract.validate(rendered.project_dict)
+            cfg = ProjectContract.from_dict(
+                rendered.project_dict
+            )
         except ValidationError as e:
             raise DbtProjectError(validator_error_message(e)) from e
-
         # name/version are required in the Project definition, so we can assume
         # they are present
         name = cfg.name
@@ -334,6 +337,7 @@ class Project:
         )
 
         docs_paths: List[str] = value_or(cfg.docs_paths, all_source_paths)
+        asset_paths: List[str] = value_or(cfg.asset_paths, [])
         target_path: str = value_or(cfg.target_path, 'target')
         clean_targets: List[str] = value_or(cfg.clean_targets, [target_path])
         log_path: str = value_or(cfg.log_path, 'logs')
@@ -343,32 +347,43 @@ class Project:
         # break many things
         quoting: Dict[str, Any] = {}
         if cfg.quoting is not None:
-            quoting = cfg.quoting.to_dict()
+            quoting = cfg.quoting.to_dict(omit_none=True)
 
-        models: Dict[str, Any] = cfg.models
-        seeds: Dict[str, Any] = cfg.seeds
-        snapshots: Dict[str, Any] = cfg.snapshots
+        dispatch: List[Dict[str, Any]]
+        models: Dict[str, Any]
+        seeds: Dict[str, Any]
+        snapshots: Dict[str, Any]
+        sources: Dict[str, Any]
+        tests: Dict[str, Any]
+        vars_value: VarProvider
 
+        dispatch = cfg.dispatch
+        models = cfg.models
+        seeds = cfg.seeds
+        snapshots = cfg.snapshots
+        sources = cfg.sources
+        tests = cfg.tests
+        if cfg.vars is None:
+            vars_dict: Dict[str, Any] = {}
+        else:
+            vars_dict = cfg.vars
+
+        vars_value = VarProvider(vars_dict)
         on_run_start: List[str] = value_or(cfg.on_run_start, [])
         on_run_end: List[str] = value_or(cfg.on_run_end, [])
 
-        # weird type handling: no value_or use
-        dbt_raw_version: Union[List[str], str] = '>=0.0.0'
-        if cfg.require_dbt_version is not None:
-            dbt_raw_version = cfg.require_dbt_version
-        query_comment = cfg.query_comment
+        query_comment = _query_comment_from_cfg(cfg.query_comment)
 
-        try:
-            dbt_version = _parse_versions(dbt_raw_version)
-        except SemverException as e:
-            raise DbtProjectError(str(e)) from e
+        packages = package_config_from_data(rendered.packages_dict)
+        selectors = selector_config_from_data(rendered.selectors_dict)
+        manifest_selectors: Dict[str, Any] = {}
+        if rendered.selectors_dict and rendered.selectors_dict['selectors']:
+            # this is a dict with a single key 'selectors' pointing to a list
+            # of dicts.
+            manifest_selectors = SelectorDict.parse_from_selectors_list(
+                rendered.selectors_dict['selectors'])
 
-        try:
-            packages = package_config_from_data(packages_dict)
-        except ValidationError as e:
-            raise DbtProjectError(validator_error_message(e)) from e
-
-        project = cls(
+        project = Project(
             project_name=name,
             version=version,
             project_root=project_root,
@@ -379,6 +394,7 @@ class Project:
             test_paths=test_paths,
             analysis_paths=analysis_paths,
             docs_paths=docs_paths,
+            asset_paths=asset_paths,
             target_path=target_path,
             snapshot_paths=snapshot_paths,
             clean_targets=clean_targets,
@@ -388,16 +404,139 @@ class Project:
             models=models,
             on_run_start=on_run_start,
             on_run_end=on_run_end,
+            dispatch=dispatch,
             seeds=seeds,
             snapshots=snapshots,
             dbt_version=dbt_version,
             packages=packages,
+            manifest_selectors=manifest_selectors,
+            selectors=selectors,
             query_comment=query_comment,
+            sources=sources,
+            tests=tests,
+            vars=vars_value,
+            config_version=cfg.config_version,
+            unrendered=unrendered,
         )
         # sanity check - this means an internal issue
         project.validate()
         return project
 
+    @classmethod
+    def from_dicts(
+        cls,
+        project_root: str,
+        project_dict: Dict[str, Any],
+        packages_dict: Dict[str, Any],
+        selectors_dict: Dict[str, Any],
+        *,
+        verify_version: bool = False,
+    ):
+        """Construct a partial project from its constituent dicts.
+        """
+        project_name = project_dict.get('name')
+        profile_name = project_dict.get('profile')
+
+        return cls(
+            profile_name=profile_name,
+            project_name=project_name,
+            project_root=project_root,
+            project_dict=project_dict,
+            packages_dict=packages_dict,
+            selectors_dict=selectors_dict,
+            verify_version=verify_version,
+        )
+
+    @classmethod
+    def from_project_root(
+        cls, project_root: str, *, verify_version: bool = False
+    ) -> 'PartialProject':
+        project_root = os.path.normpath(project_root)
+        project_dict = _raw_project_from(project_root)
+        config_version = project_dict.get('config-version', 1)
+        if config_version != 2:
+            raise DbtProjectError(
+                f'Invalid config version: {config_version}, expected 2',
+                path=os.path.join(project_root, 'dbt_project.yml')
+            )
+
+        packages_dict = package_data_from_root(project_root)
+        selectors_dict = selector_data_from_root(project_root)
+        return cls.from_dicts(
+            project_root=project_root,
+            project_dict=project_dict,
+            selectors_dict=selectors_dict,
+            packages_dict=packages_dict,
+            verify_version=verify_version,
+        )
+
+
+class VarProvider:
+    """Var providers are tied to a particular Project."""
+
+    def __init__(
+        self,
+        vars: Dict[str, Dict[str, Any]]
+    ) -> None:
+        self.vars = vars
+
+    def vars_for(
+        self, node: IsFQNResource, adapter_type: str
+    ) -> Mapping[str, Any]:
+        # in v2, vars are only either project or globally scoped
+        merged = MultiDict([self.vars])
+        merged.add(self.vars.get(node.package_name, {}))
+        return merged
+
+    def to_dict(self):
+        return self.vars
+
+
+# The Project class is included in RuntimeConfig, so any attribute
+# additions must also be set where the RuntimeConfig class is created
+@dataclass
+class Project:
+    project_name: str
+    version: Union[SemverString, float]
+    project_root: str
+    profile_name: Optional[str]
+    source_paths: List[str]
+    macro_paths: List[str]
+    data_paths: List[str]
+    test_paths: List[str]
+    analysis_paths: List[str]
+    docs_paths: List[str]
+    asset_paths: List[str]
+    target_path: str
+    snapshot_paths: List[str]
+    clean_targets: List[str]
+    log_path: str
+    modules_path: str
+    quoting: Dict[str, Any]
+    models: Dict[str, Any]
+    on_run_start: List[str]
+    on_run_end: List[str]
+    dispatch: List[Dict[str, Any]]
+    seeds: Dict[str, Any]
+    snapshots: Dict[str, Any]
+    sources: Dict[str, Any]
+    tests: Dict[str, Any]
+    vars: VarProvider
+    dbt_version: List[VersionSpecifier]
+    packages: Dict[str, Any]
+    manifest_selectors: Dict[str, Any]
+    selectors: SelectorConfig
+    query_comment: QueryComment
+    config_version: int
+    unrendered: RenderComponents
+
+    @property
+    def all_source_paths(self) -> List[str]:
+        return _all_source_paths(
+            self.source_paths, self.data_paths, self.snapshot_paths,
+            self.analysis_paths, self.macro_paths
+        )
+
     def __str__(self):
         cfg = self.to_project_config(with_packages=True)
         return str(cfg)
@@ -428,6 +567,7 @@ class Project:
             'test-paths': self.test_paths,
             'analysis-paths': self.analysis_paths,
             'docs-paths': self.docs_paths,
+            'asset-paths': self.asset_paths,
             'target-path': self.target_path,
             'snapshot-paths': self.snapshot_paths,
             'clean-targets': self.clean_targets,
@@ -436,124 +576,85 @@ class Project:
             'models': self.models,
             'on-run-start': self.on_run_start,
             'on-run-end': self.on_run_end,
+            'dispatch': self.dispatch,
             'seeds': self.seeds,
             'snapshots': self.snapshots,
+            'sources': self.sources,
+            'tests': self.tests,
+            'vars': self.vars.to_dict(),
             'require-dbt-version': [
                 v.to_version_string() for v in self.dbt_version
             ],
+            'config-version': self.config_version,
         })
+        if self.query_comment:
+            result['query-comment'] = \
+                self.query_comment.to_dict(omit_none=True)
+
         if with_packages:
-            result.update(self.packages.to_dict())
-        if self.query_comment != NoValue():
-            result['query-comment'] = self.query_comment
+            result.update(self.packages.to_dict(omit_none=True))
 
         return result
 
     def validate(self):
         try:
-            ProjectContract.from_dict(self.to_project_config())
+            ProjectContract.validate(self.to_project_config())
         except ValidationError as e:
             raise DbtProjectError(validator_error_message(e)) from e
 
+    @classmethod
+    def partial_load(
+        cls, project_root: str, *, verify_version: bool = False
+    ) -> PartialProject:
+        return PartialProject.from_project_root(
+            project_root,
+            verify_version=verify_version,
+        )
+
     @classmethod
     def render_from_dict(
         cls,
         project_root: str,
         project_dict: Dict[str, Any],
         packages_dict: Dict[str, Any],
-        renderer: ConfigRenderer,
+        selectors_dict: Dict[str, Any],
+        renderer: DbtProjectYamlRenderer,
+        *,
+        verify_version: bool = False
     ) -> 'Project':
-        rendered_project = renderer.render_project(project_dict)
-        rendered_project['project-root'] = project_root
-        rendered_packages = renderer.render_packages_data(packages_dict)
-        return cls.from_project_config(rendered_project, rendered_packages)
-
-    @classmethod
-    def partial_load(
-        cls, project_root: str
-    ) -> PartialProject:
-        project_root = os.path.normpath(project_root)
-        project_dict = _raw_project_from(project_root)
-
-        project_name = project_dict.get('name')
-        profile_name = project_dict.get('profile')
-
-        return PartialProject(
-            profile_name=profile_name,
-            project_name=project_name,
+        partial = PartialProject.from_dicts(
             project_root=project_root,
             project_dict=project_dict,
+            packages_dict=packages_dict,
+            selectors_dict=selectors_dict,
+            verify_version=verify_version,
         )
+        return partial.render(renderer)
 
     @classmethod
     def from_project_root(
-        cls, project_root: str, renderer: ConfigRenderer
+        cls,
+        project_root: str,
+        renderer: DbtProjectYamlRenderer,
+        *,
+        verify_version: bool = False,
     ) -> 'Project':
-        partial = cls.partial_load(project_root)
+        partial = cls.partial_load(project_root, verify_version=verify_version)
         return partial.render(renderer)
 
     def hashed_name(self):
         return hashlib.md5(self.project_name.encode('utf-8')).hexdigest()
 
-    def get_resource_config_paths(self):
-        """Return a dictionary with 'seeds' and 'models' keys whose values are
-        lists of lists of strings, where each inner list of strings represents
-        a configured path in the resource.
-        """
-        return {
-            'models': _get_config_paths(self.models),
-            'seeds': _get_config_paths(self.seeds),
-            'snapshots': _get_config_paths(self.snapshots),
-        }
-
-    def get_unused_resource_config_paths(self, resource_fqns, disabled):
-        """Return a list of lists of strings, where each inner list of strings
-        represents a type + FQN path of a resource configuration that is not
-        used.
-        """
-        disabled_fqns = frozenset(tuple(fqn) for fqn in disabled)
-        resource_config_paths = self.get_resource_config_paths()
-        unused_resource_config_paths = []
-        for resource_type, config_paths in resource_config_paths.items():
-            used_fqns = resource_fqns.get(resource_type, frozenset())
-            fqns = used_fqns | disabled_fqns
-
-            for config_path in config_paths:
-                if not _is_config_used(config_path, fqns):
-                    unused_resource_config_paths.append(
-                        (resource_type,) + config_path
-                    )
-        return unused_resource_config_paths
-
-    def warn_for_unused_resource_config_paths(self, resource_fqns, disabled):
-        unused = self.get_unused_resource_config_paths(resource_fqns, disabled)
-        if len(unused) == 0:
-            return
-
-        msg = UNUSED_RESOURCE_CONFIGURATION_PATH_MESSAGE.format(
-            len(unused),
-            '\n'.join('- {}'.format('.'.join(u)) for u in unused)
-        )
-        warn_or_error(msg, log_fmt=printer.yellow('{}'))
-
-    def validate_version(self):
-        """Ensure this package works with the installed version of dbt."""
-        installed = get_installed_version()
-        if not versions_compatible(*self.dbt_version):
-            msg = IMPOSSIBLE_VERSION_ERROR.format(
-                package=self.project_name,
-                version_spec=[
-                    x.to_version_string() for x in self.dbt_version
-                ]
+    def get_selector(self, name: str) -> SelectionSpec:
+        if name not in self.selectors:
+            raise RuntimeException(
+                f'Could not find selector named {name}, expected one of '
+                f'{list(self.selectors)}'
             )
-            raise DbtProjectError(msg)
+        return self.selectors[name]
 
-        if not versions_compatible(installed, *self.dbt_version):
-            msg = INVALID_VERSION_ERROR.format(
-                package=self.project_name,
-                installed=installed.to_version_string(),
-                version_spec=[
-                    x.to_version_string() for x in self.dbt_version
-                ]
-            )
-            raise DbtProjectError(msg)
+    def get_macro_search_order(self, macro_namespace: str):
+        for dispatch_entry in self.dispatch:
+            if dispatch_entry['macro_namespace'] == macro_namespace:
+                return dispatch_entry['search_order']
+        return None

core/dbt/config/renderer.py

@@ -1,134 +1,238 @@
-from typing import Dict, Any
+from typing import Dict, Any, Tuple, Optional, Union, Callable
 
-from dbt.clients.jinja import get_rendered
+from dbt.clients.jinja import get_rendered, catch_jinja
 
-from dbt.exceptions import DbtProfileError
-from dbt.exceptions import DbtProjectError
-from dbt.exceptions import RecursionException
+from dbt.exceptions import (
+    DbtProjectError, CompilationException, RecursionException
+)
+from dbt.node_types import NodeType
 from dbt.utils import deep_map
 
 
-class ConfigRenderer:
-    """A renderer provides configuration rendering for a given set of cli
-    variables and a render type.
-    """
-    def __init__(self, context: Dict[str, Any]):
+Keypath = Tuple[Union[str, int], ...]
+
+
+class BaseRenderer:
+    def __init__(self, context: Dict[str, Any]) -> None:
         self.context = context
 
-    @staticmethod
-    def _is_deferred_render(keypath):
-        if not keypath:
-            return False
+    @property
+    def name(self):
+        return 'Rendering'
 
-        first = keypath[0]
-        # run hooks
-        if first in {'on-run-start', 'on-run-end', 'query-comment'}:
-            return True
-        # models have two things to avoid
-        if first in {'seeds', 'models', 'snapshots'}:
-            # model-level hooks
-            if 'pre-hook' in keypath or 'post-hook' in keypath:
-                return True
-            # model-level 'vars' declarations
-            if 'vars' in keypath:
-                return True
+    def should_render_keypath(self, keypath: Keypath) -> bool:
+        return True
 
-        return False
-
-    def _render_project_entry(self, value, keypath):
-        """Render an entry, in case it's jinja. This is meant to be passed to
-        deep_map.
-
-        If the parsed entry is a string and has the name 'port', this will
-        attempt to cast it to an int, and on failure will return the parsed
-        string.
-
-        :param value Any: The value to potentially render
-        :param key str: The key to convert on.
-        :return Any: The rendered entry.
-        """
-        # query comments and hooks should be treated as raw sql, they'll get
-        # rendered later.
-        # Same goes for 'vars' declarations inside 'models'/'seeds'
-        if self._is_deferred_render(keypath):
+    def render_entry(self, value: Any, keypath: Keypath) -> Any:
+        if not self.should_render_keypath(keypath):
             return value
 
-        return self.render_value(value)
+        return self.render_value(value, keypath)
 
-    def render_value(self, value, keypath=None):
+    def render_value(
+        self, value: Any, keypath: Optional[Keypath] = None
+    ) -> Any:
         # keypath is ignored.
         # if it wasn't read as a string, ignore it
         if not isinstance(value, str):
             return value
-        return str(get_rendered(value, self.context))
-
-    def _render_profile_data(self, value, keypath):
-        result = self.render_value(value)
-        if len(keypath) == 1 and keypath[-1] == 'port':
-            try:
-                result = int(result)
-            except ValueError:
-                # let the validator or connection handle this
-                pass
-        return result
-
-    @staticmethod
-    def _is_schema_test(keypath) -> bool:
-        # we got passed an UnparsedSourceDefinition
-        if len(keypath) > 2 and keypath[0] == 'tables':
-            if keypath[2] == 'tests':
-                return True
-            elif keypath[2] == 'columns':
-                if len(keypath) > 4 and keypath[4] == 'tests':
-                    return True
-        return False
-
-    def _render_schema_source_data(self, value, keypath):
-        # things to not render:
-        # - descriptions
-        # - test arguments
-        if len(keypath) > 0 and keypath[-1] == 'description':
-            return value
-        elif self._is_schema_test(keypath):
-            return value
-
-        return self.render_value(value)
-
-    def render_project(self, as_parsed):
-        """Render the parsed data, returning a new dict (or whatever was read).
-        """
         try:
-            return deep_map(self._render_project_entry, as_parsed)
+            with catch_jinja():
+                return get_rendered(value, self.context, native=True)
+        except CompilationException as exc:
+            msg = f'Could not render {value}: {exc.msg}'
+            raise CompilationException(msg) from exc
+
+    def render_data(
+        self, data: Dict[str, Any]
+    ) -> Dict[str, Any]:
+        try:
+            return deep_map(self.render_entry, data)
         except RecursionException:
             raise DbtProjectError(
-                'Cycle detected: Project input has a reference to itself',
-                project=as_parsed
+                f'Cycle detected: {self.name} input has a reference to itself',
+                project=data
             )
 
-    def render_profile_data(self, as_parsed):
-        """Render the chosen profile entry, as it was parsed."""
-        try:
-            return deep_map(self._render_profile_data, as_parsed)
-        except RecursionException:
-            raise DbtProfileError(
-                'Cycle detected: Profile input has a reference to itself',
-                project=as_parsed
-            )
 
-    def render_schema_source(self, as_parsed):
-        try:
-            return deep_map(self._render_schema_source_data, as_parsed)
-        except RecursionException:
-            raise DbtProfileError(
-                'Cycle detected: schema.yml input has a reference to itself',
-                project=as_parsed
-            )
+def _list_if_none(value):
+    if value is None:
+        value = []
+    return value
 
-    def render_packages_data(self, as_parsed):
-        try:
-            return deep_map(self.render_value, as_parsed)
-        except RecursionException:
-            raise DbtProfileError(
-                'Cycle detected: schema.yml input has a reference to itself',
-                project=as_parsed
-            )
+
+def _dict_if_none(value):
+    if value is None:
+        value = {}
+    return value
+
+
+def _list_if_none_or_string(value):
+    value = _list_if_none(value)
+    if isinstance(value, str):
+        return [value]
+    return value
+
+
+class ProjectPostprocessor(Dict[Keypath, Callable[[Any], Any]]):
+    def __init__(self):
+        super().__init__()
+
+        self[('on-run-start',)] = _list_if_none_or_string
+        self[('on-run-end',)] = _list_if_none_or_string
+
+        for k in ('models', 'seeds', 'snapshots'):
+            self[(k,)] = _dict_if_none
+            self[(k, 'vars')] = _dict_if_none
+            self[(k, 'pre-hook')] = _list_if_none_or_string
+            self[(k, 'post-hook')] = _list_if_none_or_string
+        self[('seeds', 'column_types')] = _dict_if_none
+
+    def postprocess(self, value: Any, key: Keypath) -> Any:
+        if key in self:
+            handler = self[key]
+            return handler(value)
+
+        return value
+
+
+class DbtProjectYamlRenderer(BaseRenderer):
+    _KEYPATH_HANDLERS = ProjectPostprocessor()
+
+    @property
+    def name(self):
+        'Project config'
+
+    def get_package_renderer(self) -> BaseRenderer:
+        return PackageRenderer(self.context)
+
+    def get_selector_renderer(self) -> BaseRenderer:
+        return SelectorRenderer(self.context)
+
+    def render_project(
+        self,
+        project: Dict[str, Any],
+        project_root: str,
+    ) -> Dict[str, Any]:
+        """Render the project and insert the project root after rendering."""
+        rendered_project = self.render_data(project)
+        rendered_project['project-root'] = project_root
+        return rendered_project
+
+    def render_packages(self, packages: Dict[str, Any]):
+        """Render the given packages dict"""
+        package_renderer = self.get_package_renderer()
+        return package_renderer.render_data(packages)
+
+    def render_selectors(self, selectors: Dict[str, Any]):
+        selector_renderer = self.get_selector_renderer()
+        return selector_renderer.render_data(selectors)
+
+    def render_entry(self, value: Any, keypath: Keypath) -> Any:
+        result = super().render_entry(value, keypath)
+        return self._KEYPATH_HANDLERS.postprocess(result, keypath)
+
+    def should_render_keypath(self, keypath: Keypath) -> bool:
+        if not keypath:
+            return True
+
+        first = keypath[0]
+        # run hooks are not rendered
+        if first in {'on-run-start', 'on-run-end', 'query-comment'}:
+            return False
+
+        # don't render vars blocks until runtime
+        if first == 'vars':
+            return False
+
+        if first in {'seeds', 'models', 'snapshots', 'tests'}:
+            keypath_parts = {
+                (k.lstrip('+ ') if isinstance(k, str) else k)
+                for k in keypath
+            }
+            # model-level hooks
+            if 'pre-hook' in keypath_parts or 'post-hook' in keypath_parts:
+                return False
+
+        return True
+
+
+class ProfileRenderer(BaseRenderer):
+    @property
+    def name(self):
+        'Profile'
+
+
+class SchemaYamlRenderer(BaseRenderer):
+    DOCUMENTABLE_NODES = frozenset(
+        n.pluralize() for n in NodeType.documentable()
+    )
+
+    @property
+    def name(self):
+        return 'Rendering yaml'
+
+    def _is_norender_key(self, keypath: Keypath) -> bool:
+        """
+        models:
+            - name: blah
+            - description: blah
+              tests: ...
+            - columns:
+                - name:
+                - description: blah
+                  tests: ...
+
+        Return True if it's tests or description - those aren't rendered
+        """
+        if len(keypath) >= 2 and keypath[1] in ('tests', 'description'):
+            return True
+
+        if (
+            len(keypath) >= 4 and
+            keypath[1] == 'columns' and
+            keypath[3] in ('tests', 'description')
+        ):
+            return True
+
+        return False
+
+    # don't render descriptions or test keyword arguments
+    def should_render_keypath(self, keypath: Keypath) -> bool:
+        if len(keypath) < 2:
+            return True
+
+        if keypath[0] not in self.DOCUMENTABLE_NODES:
+            return True
+
+        if len(keypath) < 3:
+            return True
+
+        if keypath[0] == NodeType.Source.pluralize():
+            if keypath[2] == 'description':
+                return False
+            if keypath[2] == 'tables':
+                if self._is_norender_key(keypath[3:]):
+                    return False
+        elif keypath[0] == NodeType.Macro.pluralize():
+            if keypath[2] == 'arguments':
+                if self._is_norender_key(keypath[3:]):
+                    return False
+            elif self._is_norender_key(keypath[1:]):
+                return False
+        else:  # keypath[0] in self.DOCUMENTABLE_NODES:
+            if self._is_norender_key(keypath[1:]):
+                return False
+        return True
+
+
+class PackageRenderer(BaseRenderer):
+    @property
+    def name(self):
+        return 'Packages config'
+
+
+class SelectorRenderer(BaseRenderer):
+    @property
+    def name(self):
+        return 'Selector config'

core/dbt/config/runtime.py

@@ -1,35 +1,71 @@
-from copy import deepcopy
-from dataclasses import dataclass
+import itertools
 import os
-from typing import Dict, Any
+from copy import deepcopy
+from dataclasses import dataclass, fields
+from pathlib import Path
+from typing import (
+    Dict, Any, Optional, Mapping, Iterator, Iterable, Tuple, List, MutableSet,
+    Type
+)
 
 from .profile import Profile
 from .project import Project
-from .renderer import ConfigRenderer
-from dbt.utils import parse_cli_vars
+from .renderer import DbtProjectYamlRenderer, ProfileRenderer
+from .utils import parse_cli_vars
+from dbt import tracking
+from dbt.adapters.factory import get_relation_class_by_name, get_include_paths
+from dbt.helper_types import FQNPath, PathSet
 from dbt.context.base import generate_base_context
 from dbt.context.target import generate_target_context
-from dbt.contracts.connection import AdapterRequiredConfig
+from dbt.contracts.connection import AdapterRequiredConfig, Credentials
 from dbt.contracts.graph.manifest import ManifestMetadata
-from dbt.contracts.project import Configuration
-from dbt.exceptions import DbtProjectError
-from dbt.exceptions import validator_error_message
-from dbt.adapters.factory import get_relation_class_by_name
+from dbt.contracts.relation import ComponentName
+from dbt.logger import GLOBAL_LOGGER as logger
+from dbt.ui import warning_tag
 
-from hologram import ValidationError
+from dbt.contracts.project import Configuration, UserConfig
+from dbt.exceptions import (
+    RuntimeException,
+    DbtProfileError,
+    DbtProjectError,
+    validator_error_message,
+    warn_or_error,
+    raise_compiler_error
+)
+
+from dbt.dataclass_schema import ValidationError
+
+
+def _project_quoting_dict(
+    proj: Project, profile: Profile
+) -> Dict[ComponentName, bool]:
+    src: Dict[str, Any] = profile.credentials.translate_aliases(proj.quoting)
+    result: Dict[ComponentName, bool] = {}
+    for key in ComponentName:
+        if key in src:
+            value = src[key]
+            if isinstance(value, bool):
+                result[key] = value
+    return result
 
 
 @dataclass
 class RuntimeConfig(Project, Profile, AdapterRequiredConfig):
     args: Any
+    profile_name: str
     cli_vars: Dict[str, Any]
+    dependencies: Optional[Mapping[str, 'RuntimeConfig']] = None
 
     def __post_init__(self):
         self.validate()
 
     @classmethod
     def from_parts(
-        cls, project: Project, profile: Profile, args: Any,
+        cls,
+        project: Project,
+        profile: Profile,
+        args: Any,
+        dependencies: Optional[Mapping[str, 'RuntimeConfig']] = None,
     ) -> 'RuntimeConfig':
         """Instantiate a RuntimeConfig from its components.
 
@@ -41,8 +77,8 @@ class RuntimeConfig(Project, Profile, AdapterRequiredConfig):
         quoting: Dict[str, Any] = (
             get_relation_class_by_name(profile.credentials.type)
             .get_default_quote_policy()
-            .replace_dict(project.quoting)
-        ).to_dict()
+            .replace_dict(_project_quoting_dict(project, profile))
+        ).to_dict(omit_none=True)
 
         cli_vars: Dict[str, Any] = parse_cli_vars(getattr(args, 'vars', '{}'))
 
@@ -56,6 +92,7 @@ class RuntimeConfig(Project, Profile, AdapterRequiredConfig):
             test_paths=project.test_paths,
             analysis_paths=project.analysis_paths,
             docs_paths=project.docs_paths,
+            asset_paths=project.asset_paths,
             target_path=project.target_path,
             snapshot_paths=project.snapshot_paths,
             clean_targets=project.clean_targets,
@@ -65,11 +102,19 @@ class RuntimeConfig(Project, Profile, AdapterRequiredConfig):
             models=project.models,
             on_run_start=project.on_run_start,
             on_run_end=project.on_run_end,
+            dispatch=project.dispatch,
             seeds=project.seeds,
             snapshots=project.snapshots,
             dbt_version=project.dbt_version,
             packages=project.packages,
+            manifest_selectors=project.manifest_selectors,
+            selectors=project.selectors,
             query_comment=project.query_comment,
+            sources=project.sources,
+            tests=project.tests,
+            vars=project.vars,
+            config_version=project.config_version,
+            unrendered=project.unrendered,
             profile_name=profile.profile_name,
             target_name=profile.target_name,
             config=profile.config,
@@ -77,6 +122,7 @@ class RuntimeConfig(Project, Profile, AdapterRequiredConfig):
             credentials=profile.credentials,
             args=args,
             cli_vars=cli_vars,
+            dependencies=dependencies,
         )
 
     def new_project(self, project_root: str) -> 'RuntimeConfig':
@@ -93,9 +139,13 @@ class RuntimeConfig(Project, Profile, AdapterRequiredConfig):
         profile.validate()
 
         # load the new project and its packages. Don't pass cli variables.
-        renderer = ConfigRenderer(generate_target_context(profile, {}))
+        renderer = DbtProjectYamlRenderer(generate_target_context(profile, {}))
 
-        project = Project.from_project_root(project_root, renderer)
+        project = Project.from_project_root(
+            project_root,
+            renderer,
+            verify_version=getattr(self.args, 'version_check', False),
+        )
 
         cfg = self.from_parts(
             project=project,
@@ -126,12 +176,48 @@ class RuntimeConfig(Project, Profile, AdapterRequiredConfig):
         :raises DbtProjectError: If the configuration fails validation.
         """
         try:
-            Configuration.from_dict(self.serialize())
+            Configuration.validate(self.serialize())
         except ValidationError as e:
             raise DbtProjectError(validator_error_message(e)) from e
 
-        if getattr(self.args, 'version_check', False):
-            self.validate_version()
+    @classmethod
+    def _get_rendered_profile(
+        cls,
+        args: Any,
+        profile_renderer: ProfileRenderer,
+        profile_name: Optional[str],
+    ) -> Profile:
+        return Profile.render_from_args(
+            args, profile_renderer, profile_name
+        )
+
+    @classmethod
+    def collect_parts(
+        cls: Type['RuntimeConfig'], args: Any
+    ) -> Tuple[Project, Profile]:
+        # profile_name from the project
+        project_root = args.project_dir if args.project_dir else os.getcwd()
+        version_check = getattr(args, 'version_check', False)
+        partial = Project.partial_load(
+            project_root,
+            verify_version=version_check
+        )
+
+        # build the profile using the base renderer and the one fact we know
+        cli_vars: Dict[str, Any] = parse_cli_vars(getattr(args, 'vars', '{}'))
+        profile_renderer = ProfileRenderer(generate_base_context(cli_vars))
+        profile_name = partial.render_profile_name(profile_renderer)
+
+        profile = cls._get_rendered_profile(
+            args, profile_renderer, profile_name
+        )
+
+        # get a new renderer using our target information and render the
+        # project
+        ctx = generate_target_context(profile, cli_vars)
+        project_renderer = DbtProjectYamlRenderer(ctx)
+        project = partial.render(project_renderer)
+        return (project, profile)
 
     @classmethod
     def from_args(cls, args: Any) -> 'RuntimeConfig':
@@ -144,20 +230,7 @@ class RuntimeConfig(Project, Profile, AdapterRequiredConfig):
         :raises DbtProfileError: If the profile is invalid or missing.
         :raises ValidationException: If the cli variables are invalid.
         """
-        # profile_name from the project
-        partial = Project.partial_load(os.getcwd())
-
-        # build the profile using the base renderer and the one fact we know
-        cli_vars: Dict[str, Any] = parse_cli_vars(getattr(args, 'vars', '{}'))
-        renderer = ConfigRenderer(generate_base_context(cli_vars=cli_vars))
-        profile = Profile.render_from_args(
-            args, renderer, partial.profile_name
-        )
-
-        # get a new renderer using our target information and render the
-        # project
-        renderer = ConfigRenderer(generate_target_context(profile, cli_vars))
-        project = partial.render(renderer)
+        project, profile = cls.collect_parts(args)
 
         return cls.from_parts(
             project=project,
@@ -170,3 +243,343 @@ class RuntimeConfig(Project, Profile, AdapterRequiredConfig):
             project_id=self.hashed_name(),
             adapter_type=self.credentials.type
         )
+
+    def _get_v2_config_paths(
+        self,
+        config,
+        path: FQNPath,
+        paths: MutableSet[FQNPath],
+    ) -> PathSet:
+        for key, value in config.items():
+            if isinstance(value, dict) and not key.startswith('+'):
+                self._get_v2_config_paths(value, path + (key,), paths)
+            else:
+                paths.add(path)
+        return frozenset(paths)
+
+    def _get_config_paths(
+        self,
+        config: Dict[str, Any],
+        path: FQNPath = (),
+        paths: Optional[MutableSet[FQNPath]] = None,
+    ) -> PathSet:
+        if paths is None:
+            paths = set()
+
+        for key, value in config.items():
+            if isinstance(value, dict) and not key.startswith('+'):
+                self._get_v2_config_paths(value, path + (key,), paths)
+            else:
+                paths.add(path)
+        return frozenset(paths)
+
+    def get_resource_config_paths(self) -> Dict[str, PathSet]:
+        """Return a dictionary with resource type keys whose values are
+        lists of lists of strings, where each inner list of strings represents
+        a configured path in the resource.
+        """
+        return {
+            'models': self._get_config_paths(self.models),
+            'seeds': self._get_config_paths(self.seeds),
+            'snapshots': self._get_config_paths(self.snapshots),
+            'sources': self._get_config_paths(self.sources),
+            'tests': self._get_config_paths(self.tests),
+        }
+
+    def get_unused_resource_config_paths(
+        self,
+        resource_fqns: Mapping[str, PathSet],
+        disabled: PathSet,
+    ) -> List[FQNPath]:
+        """Return a list of lists of strings, where each inner list of strings
+        represents a type + FQN path of a resource configuration that is not
+        used.
+        """
+        disabled_fqns = frozenset(tuple(fqn) for fqn in disabled)
+        resource_config_paths = self.get_resource_config_paths()
+        unused_resource_config_paths = []
+        for resource_type, config_paths in resource_config_paths.items():
+            used_fqns = resource_fqns.get(resource_type, frozenset())
+            fqns = used_fqns | disabled_fqns
+
+            for config_path in config_paths:
+                if not _is_config_used(config_path, fqns):
+                    unused_resource_config_paths.append(
+                        (resource_type,) + config_path
+                    )
+        return unused_resource_config_paths
+
+    def warn_for_unused_resource_config_paths(
+        self,
+        resource_fqns: Mapping[str, PathSet],
+        disabled: PathSet,
+    ) -> None:
+        unused = self.get_unused_resource_config_paths(resource_fqns, disabled)
+        if len(unused) == 0:
+            return
+
+        msg = UNUSED_RESOURCE_CONFIGURATION_PATH_MESSAGE.format(
+            len(unused),
+            '\n'.join('- {}'.format('.'.join(u)) for u in unused)
+        )
+
+        warn_or_error(msg, log_fmt=warning_tag('{}'))
+
+    def load_dependencies(self) -> Mapping[str, 'RuntimeConfig']:
+        if self.dependencies is None:
+            all_projects = {self.project_name: self}
+            internal_packages = get_include_paths(self.credentials.type)
+            # raise exception if fewer installed packages than in packages.yml
+            count_packages_specified = len(self.packages.packages)  # type: ignore
+            count_packages_installed = len(tuple(self._get_project_directories()))
+            if count_packages_specified > count_packages_installed:
+                raise_compiler_error(
+                    f'dbt found {count_packages_specified} package(s) '
+                    f'specified in packages.yml, but only '
+                    f'{count_packages_installed} package(s) installed '
+                    f'in {self.modules_path}. Run "dbt deps" to '
+                    f'install package dependencies.'
+                )
+            project_paths = itertools.chain(
+                internal_packages,
+                self._get_project_directories()
+            )
+            for project_name, project in self.load_projects(project_paths):
+                if project_name in all_projects:
+                    raise_compiler_error(
+                        f'dbt found more than one package with the name '
+                        f'"{project_name}" included in this project. Package '
+                        f'names must be unique in a project. Please rename '
+                        f'one of these packages.'
+                    )
+                all_projects[project_name] = project
+            self.dependencies = all_projects
+        return self.dependencies
+
+    def clear_dependencies(self):
+        self.dependencies = None
+
+    def load_projects(
+        self, paths: Iterable[Path]
+    ) -> Iterator[Tuple[str, 'RuntimeConfig']]:
+        for path in paths:
+            try:
+                project = self.new_project(str(path))
+            except DbtProjectError as e:
+                raise DbtProjectError(
+                    f'Failed to read package: {e}',
+                    result_type='invalid_project',
+                    path=path,
+                ) from e
+            else:
+                yield project.project_name, project
+
+    def _get_project_directories(self) -> Iterator[Path]:
+        root = Path(self.project_root) / self.modules_path
+
+        if root.exists():
+            for path in root.iterdir():
+                if path.is_dir() and not path.name.startswith('__'):
+                    yield path
+
+
+class UnsetCredentials(Credentials):
+    def __init__(self):
+        super().__init__('', '')
+
+    @property
+    def type(self):
+        return None
+
+    @property
+    def unique_field(self):
+        return None
+
+    def connection_info(self, *args, **kwargs):
+        return {}
+
+    def _connection_keys(self):
+        return ()
+
+
+class UnsetConfig(UserConfig):
+    def __getattribute__(self, name):
+        if name in {f.name for f in fields(UserConfig)}:
+            raise AttributeError(
+                f"'UnsetConfig' object has no attribute {name}"
+            )
+
+    def __post_serialize__(self, dct):
+        return {}
+
+
+class UnsetProfile(Profile):
+    def __init__(self):
+        self.credentials = UnsetCredentials()
+        self.config = UnsetConfig()
+        self.profile_name = ''
+        self.target_name = ''
+        self.threads = -1
+
+    def to_target_dict(self):
+        return {}
+
+    def __getattribute__(self, name):
+        if name in {'profile_name', 'target_name', 'threads'}:
+            raise RuntimeException(
+                f'Error: disallowed attribute "{name}" - no profile!'
+            )
+
+        return Profile.__getattribute__(self, name)
+
+
+@dataclass
+class UnsetProfileConfig(RuntimeConfig):
+    """This class acts a lot _like_ a RuntimeConfig, except if your profile is
+    missing, any access to profile members results in an exception.
+    """
+
+    def __post_init__(self):
+        # instead of futzing with InitVar overrides or rewriting __init__, just
+        # `del` the attrs we don't want  users touching.
+        del self.profile_name
+        del self.target_name
+        # don't call super().__post_init__(), as that calls validate(), and
+        # this object isn't very valid
+
+    def __getattribute__(self, name):
+        # Override __getattribute__ to check that the attribute isn't 'banned'.
+        if name in {'profile_name', 'target_name'}:
+            raise RuntimeException(
+                f'Error: disallowed attribute "{name}" - no profile!'
+            )
+
+        # avoid every attribute access triggering infinite recursion
+        return RuntimeConfig.__getattribute__(self, name)
+
+    def to_target_dict(self):
+        # re-override the poisoned profile behavior
+        return {}
+
+    @classmethod
+    def from_parts(
+        cls,
+        project: Project,
+        profile: Profile,
+        args: Any,
+        dependencies: Optional[Mapping[str, 'RuntimeConfig']] = None,
+    ) -> 'RuntimeConfig':
+        """Instantiate a RuntimeConfig from its components.
+
+        :param profile: Ignored.
+        :param project: A parsed dbt Project.
+        :param args: The parsed command-line arguments.
+        :returns RuntimeConfig: The new configuration.
+        """
+        cli_vars: Dict[str, Any] = parse_cli_vars(getattr(args, 'vars', '{}'))
+
+        return cls(
+            project_name=project.project_name,
+            version=project.version,
+            project_root=project.project_root,
+            source_paths=project.source_paths,
+            macro_paths=project.macro_paths,
+            data_paths=project.data_paths,
+            test_paths=project.test_paths,
+            analysis_paths=project.analysis_paths,
+            docs_paths=project.docs_paths,
+            asset_paths=project.asset_paths,
+            target_path=project.target_path,
+            snapshot_paths=project.snapshot_paths,
+            clean_targets=project.clean_targets,
+            log_path=project.log_path,
+            modules_path=project.modules_path,
+            quoting=project.quoting,  # we never use this anyway.
+            models=project.models,
+            on_run_start=project.on_run_start,
+            on_run_end=project.on_run_end,
+            dispatch=project.dispatch,
+            seeds=project.seeds,
+            snapshots=project.snapshots,
+            dbt_version=project.dbt_version,
+            packages=project.packages,
+            manifest_selectors=project.manifest_selectors,
+            selectors=project.selectors,
+            query_comment=project.query_comment,
+            sources=project.sources,
+            tests=project.tests,
+            vars=project.vars,
+            config_version=project.config_version,
+            unrendered=project.unrendered,
+            profile_name='',
+            target_name='',
+            config=UnsetConfig(),
+            threads=getattr(args, 'threads', 1),
+            credentials=UnsetCredentials(),
+            args=args,
+            cli_vars=cli_vars,
+            dependencies=dependencies,
+        )
+
+    @classmethod
+    def _get_rendered_profile(
+        cls,
+        args: Any,
+        profile_renderer: ProfileRenderer,
+        profile_name: Optional[str],
+    ) -> Profile:
+        try:
+            profile = Profile.render_from_args(
+                args, profile_renderer, profile_name
+            )
+        except (DbtProjectError, DbtProfileError) as exc:
+            logger.debug(
+                'Profile not loaded due to error: {}', exc, exc_info=True
+            )
+            logger.info(
+                'No profile "{}" found, continuing with no target',
+                profile_name
+            )
+            # return the poisoned form
+            profile = UnsetProfile()
+            # disable anonymous usage statistics
+            tracking.disable_tracking()
+        return profile
+
+    @classmethod
+    def from_args(cls: Type[RuntimeConfig], args: Any) -> 'RuntimeConfig':
+        """Given arguments, read in dbt_project.yml from the current directory,
+        read in packages.yml if it exists, and use them to find the profile to
+        load.
+
+        :param args: The arguments as parsed from the cli.
+        :raises DbtProjectError: If the project is invalid or missing.
+        :raises DbtProfileError: If the profile is invalid or missing.
+        :raises ValidationException: If the cli variables are invalid.
+        """
+        project, profile = cls.collect_parts(args)
+        if not isinstance(profile, UnsetProfile):
+            # if it's a real profile, return a real config
+            cls = RuntimeConfig
+
+        return cls.from_parts(
+            project=project,
+            profile=profile,
+            args=args
+        )
+
+
+UNUSED_RESOURCE_CONFIGURATION_PATH_MESSAGE = """\
+Configuration paths exist in your dbt_project.yml file which do not \
+apply to any resources.
+There are {} unused configuration paths:
+{}
+"""
+
+
+def _is_config_used(path, fqns):
+    if fqns:
+        for fqn in fqns:
+            if len(path) <= len(fqn) and fqn[:len(path)] == path:
+                return True
+    return False

core/dbt/config/selectors.py

@@ -0,0 +1,182 @@
+from pathlib import Path
+from typing import Dict, Any
+from dbt.clients.yaml_helper import (  # noqa: F401
+    yaml, Loader, Dumper, load_yaml_text
+)
+from dbt.dataclass_schema import ValidationError
+
+from .renderer import SelectorRenderer
+
+from dbt.clients.system import (
+    load_file_contents,
+    path_exists,
+    resolve_path_from_base,
+)
+from dbt.contracts.selection import SelectorFile
+from dbt.exceptions import DbtSelectorsError, RuntimeException
+from dbt.graph import parse_from_selectors_definition, SelectionSpec
+from dbt.graph.selector_spec import SelectionCriteria
+
+MALFORMED_SELECTOR_ERROR = """\
+The selectors.yml file in this project is malformed. Please double check
+the contents of this file and fix any errors before retrying.
+
+You can find more information on the syntax for this file here:
+https://docs.getdbt.com/docs/package-management
+
+Validator Error:
+{error}
+"""
+
+
+class SelectorConfig(Dict[str, SelectionSpec]):
+
+    @classmethod
+    def selectors_from_dict(cls, data: Dict[str, Any]) -> 'SelectorConfig':
+        try:
+            SelectorFile.validate(data)
+            selector_file = SelectorFile.from_dict(data)
+            selectors = parse_from_selectors_definition(selector_file)
+        except ValidationError as exc:
+            yaml_sel_cfg = yaml.dump(exc.instance)
+            raise DbtSelectorsError(
+                f"Could not parse selector file data: \n{yaml_sel_cfg}\n"
+                f"Valid root-level selector definitions: "
+                f"union, intersection, string, dictionary. No lists. "
+                f"\nhttps://docs.getdbt.com/reference/node-selection/"
+                f"yaml-selectors",
+                result_type='invalid_selector'
+            ) from exc
+        except RuntimeException as exc:
+            raise DbtSelectorsError(
+                f'Could not read selector file data: {exc}',
+                result_type='invalid_selector',
+            ) from exc
+
+        return cls(selectors)
+
+    @classmethod
+    def render_from_dict(
+        cls,
+        data: Dict[str, Any],
+        renderer: SelectorRenderer,
+    ) -> 'SelectorConfig':
+        try:
+            rendered = renderer.render_data(data)
+        except (ValidationError, RuntimeException) as exc:
+            raise DbtSelectorsError(
+                f'Could not render selector data: {exc}',
+                result_type='invalid_selector',
+            ) from exc
+        return cls.selectors_from_dict(rendered)
+
+    @classmethod
+    def from_path(
+        cls, path: Path, renderer: SelectorRenderer,
+    ) -> 'SelectorConfig':
+        try:
+            data = load_yaml_text(load_file_contents(str(path)))
+        except (ValidationError, RuntimeException) as exc:
+            raise DbtSelectorsError(
+                f'Could not read selector file: {exc}',
+                result_type='invalid_selector',
+                path=path,
+            ) from exc
+
+        try:
+            return cls.render_from_dict(data, renderer)
+        except DbtSelectorsError as exc:
+            exc.path = path
+            raise
+
+
+def selector_data_from_root(project_root: str) -> Dict[str, Any]:
+    selector_filepath = resolve_path_from_base(
+        'selectors.yml', project_root
+    )
+
+    if path_exists(selector_filepath):
+        selectors_dict = load_yaml_text(load_file_contents(selector_filepath))
+    else:
+        selectors_dict = None
+    return selectors_dict
+
+
+def selector_config_from_data(
+    selectors_data: Dict[str, Any]
+) -> SelectorConfig:
+    if not selectors_data:
+        selectors_data = {'selectors': []}
+
+    try:
+        selectors = SelectorConfig.selectors_from_dict(selectors_data)
+    except ValidationError as e:
+        raise DbtSelectorsError(
+            MALFORMED_SELECTOR_ERROR.format(error=str(e.message)),
+            result_type='invalid_selector',
+        ) from e
+    return selectors
+
+
+# These are utilities to clean up the dictionary created from
+# selectors.yml by turning the cli-string format entries into
+# normalized dictionary entries. It parallels the flow in
+# dbt/graph/cli.py. If changes are made there, it might
+# be necessary to make changes here. Ideally it would be
+# good to combine the two flows into one at some point.
+class SelectorDict:
+
+    @classmethod
+    def parse_dict_definition(cls, definition):
+        key = list(definition)[0]
+        value = definition[key]
+        if isinstance(value, list):
+            new_values = []
+            for sel_def in value:
+                new_value = cls.parse_from_definition(sel_def)
+                new_values.append(new_value)
+            value = new_values
+        if key == 'exclude':
+            definition = {key: value}
+        elif len(definition) == 1:
+            definition = {'method': key, 'value': value}
+        return definition
+
+    @classmethod
+    def parse_a_definition(cls, def_type, definition):
+        # this definition must be a list
+        new_dict = {def_type: []}
+        for sel_def in definition[def_type]:
+            if isinstance(sel_def, dict):
+                sel_def = cls.parse_from_definition(sel_def)
+                new_dict[def_type].append(sel_def)
+            elif isinstance(sel_def, str):
+                sel_def = SelectionCriteria.dict_from_single_spec(sel_def)
+                new_dict[def_type].append(sel_def)
+            else:
+                new_dict[def_type].append(sel_def)
+        return new_dict
+
+    @classmethod
+    def parse_from_definition(cls, definition):
+        if isinstance(definition, str):
+            definition = SelectionCriteria.dict_from_single_spec(definition)
+        elif 'union' in definition:
+            definition = cls.parse_a_definition('union', definition)
+        elif 'intersection' in definition:
+            definition = cls.parse_a_definition('intersection', definition)
+        elif isinstance(definition, dict):
+            definition = cls.parse_dict_definition(definition)
+        return definition
+
+    # This is the normal entrypoint of this code. Give it the
+    # list of selectors generated from the selectors.yml file.
+    @classmethod
+    def parse_from_selectors_list(cls, selectors):
+        selector_dict = {}
+        for selector in selectors:
+            sel_name = selector['name']
+            selector_dict[sel_name] = selector
+            definition = cls.parse_from_definition(selector['definition'])
+            selector_dict[sel_name]['definition'] = definition
+        return selector_dict

core/dbt/config/utils.py

@@ -0,0 +1,23 @@
+from typing import Dict, Any
+
+from dbt.clients import yaml_helper
+from dbt.exceptions import raise_compiler_error, ValidationException
+from dbt.logger import GLOBAL_LOGGER as logger
+
+
+def parse_cli_vars(var_string: str) -> Dict[str, Any]:
+    try:
+        cli_vars = yaml_helper.load_yaml_text(var_string)
+        var_type = type(cli_vars)
+        if var_type is dict:
+            return cli_vars
+        else:
+            type_name = var_type.__name__
+            raise_compiler_error(
+                "The --vars argument must be a YAML dictionary, but was "
+                "of type '{}'".format(type_name))
+    except ValidationException:
+        logger.error(
+            "The YAML provided in the --vars argument is not valid.\n"
+        )
+        raise

core/dbt/context/base.py

@@ -1,22 +1,25 @@
 import json
 import os
 from typing import (
-    Any, Dict, NoReturn, Optional
+    Any, Dict, NoReturn, Optional, Mapping
 )
 
 from dbt import flags
 from dbt import tracking
 from dbt.clients.jinja import undefined_error, get_rendered
+from dbt.clients.yaml_helper import (  # noqa: F401
+    yaml, safe_load, SafeLoader, Loader, Dumper
+)
+from dbt.contracts.graph.compiled import CompiledResource
 from dbt.exceptions import raise_compiler_error, MacroReturn
 from dbt.logger import GLOBAL_LOGGER as logger
-from dbt.utils import merge
 from dbt.version import __version__ as dbt_version
 
-import yaml
 # These modules are added to the context. Consider alternative
 # approaches which will extend well to potentially many modules
 import pytz
 import datetime
+import re
 
 
 def get_pytz_module_context() -> Dict[str, Any]:
@@ -41,10 +44,19 @@ def get_datetime_module_context() -> Dict[str, Any]:
     }
 
 
+def get_re_module_context() -> Dict[str, Any]:
+    context_exports = re.__all__
+
+    return {
+        name: getattr(re, name) for name in context_exports
+    }
+
+
 def get_context_modules() -> Dict[str, Dict[str, Any]]:
     return {
         'pytz': get_pytz_module_context(),
         'datetime': get_datetime_module_context(),
+        're': get_re_module_context(),
     }
 
 
@@ -98,48 +110,48 @@ class Var:
                         "supplied to {} = {}"
     _VAR_NOTSET = object()
 
-    def __init__(self, model, context, overrides):
-        self.model = model
-        self.context = context
+    def __init__(
+        self,
+        context: Mapping[str, Any],
+        cli_vars: Mapping[str, Any],
+        node: Optional[CompiledResource] = None
+    ) -> None:
+        self._context: Mapping[str, Any] = context
+        self._cli_vars: Mapping[str, Any] = cli_vars
+        self._node: Optional[CompiledResource] = node
+        self._merged: Mapping[str, Any] = self._generate_merged()
 
-        # These are hard-overrides (eg. CLI vars) that should take
-        # precedence over context-based var definitions
-        self.overrides = overrides
+    def _generate_merged(self) -> Mapping[str, Any]:
+        return self._cli_vars
 
-        if model is None:
-            # during config parsing we have no model and no local vars
-            self.model_name = '<Configuration>'
-            local_vars = {}
+    @property
+    def node_name(self):
+        if self._node is not None:
+            return self._node.name
         else:
-            self.model_name = model.name
-            local_vars = model.local_vars()
-
-        self.local_vars = merge(local_vars, overrides)
-
-    def pretty_dict(self, data):
-        return json.dumps(data, sort_keys=True, indent=4)
+            return '<Configuration>'
 
     def get_missing_var(self, var_name):
-        pretty_vars = self.pretty_dict(self.local_vars)
+        dct = {k: self._merged[k] for k in self._merged}
+        pretty_vars = json.dumps(dct, sort_keys=True, indent=4)
         msg = self.UndefinedVarError.format(
-            var_name, self.model_name, pretty_vars
+            var_name, self.node_name, pretty_vars
         )
-        raise_compiler_error(msg, self.model)
+        raise_compiler_error(msg, self._node)
 
-    def assert_var_defined(self, var_name, default):
-        if var_name not in self.local_vars and default is self._VAR_NOTSET:
-            return self.get_missing_var(var_name)
+    def has_var(self, var_name: str):
+        return var_name in self._merged
 
     def get_rendered_var(self, var_name):
-        raw = self.local_vars[var_name]
+        raw = self._merged[var_name]
         # if bool/int/float/etc are passed in, don't compile anything
         if not isinstance(raw, str):
             return raw
 
-        return get_rendered(raw, self.context)
+        return get_rendered(raw, self._context)
 
     def __call__(self, var_name, default=_VAR_NOTSET):
-        if var_name in self.local_vars:
+        if self.has_var(var_name):
             return self.get_rendered_var(var_name)
         elif default is not self._VAR_NOTSET:
             return default
@@ -161,6 +173,7 @@ class BaseContext(metaclass=ContextMeta):
             builtins[key] = value
         return builtins
 
+    # no dbtClassMixin so this is not an actual override
     def to_dict(self):
         self._ctx['context'] = self._ctx
         builtins = self.generate_builtins()
@@ -170,15 +183,101 @@ class BaseContext(metaclass=ContextMeta):
 
     @contextproperty
     def dbt_version(self) -> str:
+        """The `dbt_version` variable returns the installed version of dbt that
+        is currently running. It can be used for debugging or auditing
+        purposes.
+
+        > macros/get_version.sql
+
+            {% macro get_version() %}
+              {% set msg = "The installed version of dbt is: " ~ dbt_version %}
+              {% do log(msg, info=true) %}
+            {% endmacro %}
+
+        Example output:
+
+            $ dbt run-operation get_version
+            The installed version of dbt is 0.16.0
+        """
         return dbt_version
 
     @contextproperty
     def var(self) -> Var:
-        return Var(None, self._ctx, self.cli_vars)
+        """Variables can be passed from your `dbt_project.yml` file into models
+        during compilation. These variables are useful for configuring packages
+        for deployment in multiple environments, or defining values that should
+        be used across multiple models within a package.
+
+        To add a variable to a model, use the `var()` function:
+
+        > my_model.sql:
+
+            select * from events where event_type = '{{ var("event_type") }}'
+
+        If you try to run this model without supplying an `event_type`
+        variable, you'll receive a compilation error that looks like this:
+
+            Encountered an error:
+            ! Compilation error while compiling model package_name.my_model:
+            ! Required var 'event_type' not found in config:
+            Vars supplied to package_name.my_model = {
+            }
+
+        To supply a variable to a given model, add one or more `vars`
+        dictionaries to the `models` config in your `dbt_project.yml` file.
+        These `vars` are in-scope for all models at or below where they are
+        defined, so place them where they make the most sense. Below are three
+        different placements of the `vars` dict, all of which will make the
+        `my_model` model compile.
+
+        > dbt_project.yml:
+
+            # 1) scoped at the model level
+            models:
+              package_name:
+                my_model:
+                  materialized: view
+                  vars:
+                    event_type: activation
+            # 2) scoped at the package level
+            models:
+              package_name:
+                vars:
+                  event_type: activation
+                my_model:
+                  materialized: view
+            # 3) scoped globally
+            models:
+              vars:
+                event_type: activation
+              package_name:
+                my_model:
+                  materialized: view
+
+        ## Variable default values
+
+        The `var()` function takes an optional second argument, `default`. If
+        this argument is provided, then it will be the default value for the
+        variable if one is not explicitly defined.
+
+        > my_model.sql:
+
+            -- Use 'activation' as the event_type if the variable is not
+            -- defined.
+            select *
+            from events
+            where event_type = '{{ var("event_type", "activation") }}'
+        """
+        return Var(self._ctx, self.cli_vars)
 
     @contextmember
     @staticmethod
     def env_var(var: str, default: Optional[str] = None) -> str:
+        """The env_var() function. Return the environment variable named 'var'.
+        If there is no such environment variable set, return the default.
+
+        If the default is None, raise an exception for an undefined variable.
+        """
         if var in os.environ:
             return os.environ[var]
         elif default is not None:
@@ -191,6 +290,7 @@ class BaseContext(metaclass=ContextMeta):
         @contextmember
         @staticmethod
         def debug():
+            """Enter a debugger at this line in the compiled jinja code."""
             import sys
             import ipdb  # type: ignore
             frame = sys._getframe(3)
@@ -199,12 +299,48 @@ class BaseContext(metaclass=ContextMeta):
 
     @contextmember('return')
     @staticmethod
-    def _return(value: Any) -> NoReturn:
-        raise MacroReturn(value)
+    def _return(data: Any) -> NoReturn:
+        """The `return` function can be used in macros to return data to the
+        caller. The type of the data (`dict`, `list`, `int`, etc) will be
+        preserved through the return call.
+
+        :param data: The data to return to the caller
+
+
+        > macros/example.sql:
+
+            {% macro get_data() %}
+              {{ return([1,2,3]) }}
+            {% endmacro %}
+
+        > models/my_model.sql:
+
+            select
+              -- getdata() returns a list!
+              {% for i in getdata() %}
+                {{ i }}
+                {% if not loop.last %},{% endif %}
+              {% endfor %}
+
+        """
+        raise MacroReturn(data)
 
     @contextmember
     @staticmethod
     def fromjson(string: str, default: Any = None) -> Any:
+        """The `fromjson` context method can be used to deserialize a json
+        string into a Python object primitive, eg. a `dict` or `list`.
+
+        :param value: The json string to deserialize
+        :param default: A default value to return if the `string` argument
+            cannot be deserialized (optional)
+
+        Usage:
+
+            {% set my_json_str = '{"abc": 123}' %}
+            {% set my_dict = fromjson(my_json_str) %}
+            {% do log(my_dict['abc']) %}
+        """
         try:
             return json.loads(string)
         except ValueError:
@@ -215,6 +351,21 @@ class BaseContext(metaclass=ContextMeta):
     def tojson(
         value: Any, default: Any = None, sort_keys: bool = False
     ) -> Any:
+        """The `tojson` context method can be used to serialize a Python
+        object primitive, eg. a `dict` or `list` to a json string.
+
+        :param value: The value serialize to json
+        :param default: A default value to return if the `value` argument
+            cannot be serialized
+        :param sort_keys: If True, sort the keys.
+
+
+        Usage:
+
+            {% set my_dict = {"abc": 123} %}
+            {% set my_json_string = tojson(my_dict) %}
+            {% do log(my_json_string) %}
+        """
         try:
             return json.dumps(value, sort_keys=sort_keys)
         except ValueError:
@@ -223,8 +374,29 @@ class BaseContext(metaclass=ContextMeta):
     @contextmember
     @staticmethod
     def fromyaml(value: str, default: Any = None) -> Any:
+        """The fromyaml context method can be used to deserialize a yaml string
+        into a Python object primitive, eg. a `dict` or `list`.
+
+        :param value: The yaml string to deserialize
+        :param default: A default value to return if the `string` argument
+            cannot be deserialized (optional)
+
+        Usage:
+
+            {% set my_yml_str -%}
+            dogs:
+             - good
+             - bad
+            {%- endset %}
+            {% set my_dict = fromyaml(my_yml_str) %}
+            {% do log(my_dict['dogs'], info=true) %}
+            -- ["good", "bad"]
+            {% do my_dict['dogs'].pop() }
+            {% do log(my_dict['dogs'], info=true) %}
+            -- ["good"]
+        """
         try:
-            return yaml.safe_load(value)
+            return safe_load(value)
         except (AttributeError, ValueError, yaml.YAMLError):
             return default
 
@@ -235,6 +407,21 @@ class BaseContext(metaclass=ContextMeta):
     def toyaml(
         value: Any, default: Optional[str] = None, sort_keys: bool = False
     ) -> Optional[str]:
+        """The `tojson` context method can be used to serialize a Python
+        object primitive, eg. a `dict` or `list` to a yaml string.
+
+        :param value: The value serialize to yaml
+        :param default: A default value to return if the `value` argument
+            cannot be serialized
+        :param sort_keys: If True, sort the keys.
+
+
+        Usage:
+
+            {% set my_dict = {"abc": 123} %}
+            {% set my_yaml_string = toyaml(my_dict) %}
+            {% do log(my_yaml_string) %}
+        """
         try:
             return yaml.safe_dump(data=value, sort_keys=sort_keys)
         except (ValueError, yaml.YAMLError):
@@ -243,6 +430,18 @@ class BaseContext(metaclass=ContextMeta):
     @contextmember
     @staticmethod
     def log(msg: str, info: bool = False) -> str:
+        """Logs a line to either the log file or stdout.
+
+        :param msg: The message to log
+        :param info: If `False`, write to the log file. If `True`, write to
+            both the log file and stdout.
+
+        > macros/my_log_macro.sql
+
+            {% macro some_macro(arg1, arg2) %}
+              {{ log("Running some_macro: " ~ arg1 ~ ", " ~ arg2) }}
+            {% endmacro %}"
+        """
         if info:
             logger.info(msg)
         else:
@@ -251,6 +450,27 @@ class BaseContext(metaclass=ContextMeta):
 
     @contextproperty
     def run_started_at(self) -> Optional[datetime.datetime]:
+        """`run_started_at` outputs the timestamp that this run started, e.g.
+        `2017-04-21 01:23:45.678`. The `run_started_at` variable is a Python
+        `datetime` object. As of 0.9.1, the timezone of this variable defaults
+        to UTC.
+
+        > run_started_at_example.sql
+
+            select
+                '{{ run_started_at.strftime("%Y-%m-%d") }}' as date_day
+            from ...
+
+
+        To modify the timezone of this variable, use the the `pytz` module:
+
+        > run_started_at_utc.sql
+
+            {% set est = modules.pytz.timezone("America/New_York") %}
+            select
+                '{{ run_started_at.astimezone(est) }}' as run_started_est
+            from ...
+        """
         if tracking.active_user is not None:
             return tracking.active_user.run_started_at
         else:
@@ -258,6 +478,9 @@ class BaseContext(metaclass=ContextMeta):
 
     @contextproperty
     def invocation_id(self) -> Optional[str]:
+        """invocation_id outputs a UUID generated for this dbt run (useful for
+        auditing)
+        """
         if tracking.active_user is not None:
             return tracking.active_user.invocation_id
         else:
@@ -265,13 +488,55 @@ class BaseContext(metaclass=ContextMeta):
 
     @contextproperty
     def modules(self) -> Dict[str, Any]:
+        """The `modules` variable in the Jinja context contains useful Python
+        modules for operating on data.
+
+        # datetime
+
+        This variable is a pointer to the Python datetime module.
+
+        Usage:
+
+            {% set dt = modules.datetime.datetime.now() %}
+
+        # pytz
+
+        This variable is a pointer to the Python pytz module.
+
+        Usage:
+
+            {% set dt = modules.datetime.datetime(2002, 10, 27, 6, 0, 0) %}
+            {% set dt_local = modules.pytz.timezone('US/Eastern').localize(dt) %}
+            {{ dt_local }}
+        """  # noqa
         return get_context_modules()
 
     @contextproperty
     def flags(self) -> Any:
+        """The `flags` variable contains true/false values for flags provided
+        on the command line.
+
+        > flags.sql:
+
+            {% if flags.FULL_REFRESH %}
+            drop table ...
+            {% else %}
+            -- no-op
+            {% endif %}
+
+        The list of valid flags are:
+
+        - `flags.STRICT_MODE`: True if `--strict` (or `-S`) was provided on the
+            command line
+        - `flags.FULL_REFRESH`: True if `--full-refresh` was provided on the
+            command line
+        - `flags.NON_DESTRUCTIVE`: True if `--non-destructive` was provided on
+            the command line
+        """
         return flags
 
 
 def generate_base_context(cli_vars: Dict[str, Any]) -> Dict[str, Any]:
     ctx = BaseContext(cli_vars)
+    # This is not a Mashumaro to_dict call
     return ctx.to_dict()

core/dbt/context/configured.py

@@ -1,15 +1,11 @@
-from typing import Any, Dict, Iterable, Union, Optional
+from typing import Any, Dict
 
-from dbt.clients.jinja import MacroGenerator, MacroStack
 from dbt.contracts.connection import AdapterRequiredConfig
-from dbt.contracts.graph.manifest import Manifest
-from dbt.contracts.graph.parsed import ParsedMacro
-from dbt.include.global_project import PACKAGES
-from dbt.include.global_project import PROJECT_NAME as GLOBAL_PROJECT_NAME
+from dbt.node_types import NodeType
+from dbt.utils import MultiDict
 
-from dbt.context.base import contextproperty
+from dbt.context.base import contextproperty, Var
 from dbt.context.target import TargetContext
-from dbt.exceptions import raise_duplicate_macro_name
 
 
 class ConfiguredContext(TargetContext):
@@ -25,112 +21,80 @@ class ConfiguredContext(TargetContext):
         return self.config.project_name
 
 
-FlatNamespace = Dict[str, MacroGenerator]
-NamespaceMember = Union[FlatNamespace, MacroGenerator]
-FullNamespace = Dict[str, NamespaceMember]
+class FQNLookup:
+    def __init__(self, package_name: str):
+        self.package_name = package_name
+        self.fqn = [package_name]
+        self.resource_type = NodeType.Model
 
 
-class MacroNamespace:
-    def __init__(
-        self,
-        root_package: str,
-        search_package: str,
-        thread_ctx: MacroStack,
-        node: Optional[Any] = None,
-    ) -> None:
-        self.root_package = root_package
-        self.search_package = search_package
-        self.globals: FlatNamespace = {}
-        self.locals: FlatNamespace = {}
-        self.packages: Dict[str, FlatNamespace] = {}
-        self.thread_ctx = thread_ctx
-        self.node = node
-
-    def add_macro(self, macro: ParsedMacro, ctx: Dict[str, Any]):
-        macro_name: str = macro.name
-        macro_func: MacroGenerator = MacroGenerator(
-            macro, ctx, self.node, self.thread_ctx
-        )
-
-        # put plugin macros into the root namespace
-        if macro.package_name in PACKAGES:
-            namespace: str = GLOBAL_PROJECT_NAME
-        else:
-            namespace = macro.package_name
-
-        if namespace not in self.packages:
-            value: Dict[str, MacroGenerator] = {}
-            self.packages[namespace] = value
-
-        if macro_name in self.packages[namespace]:
-            raise_duplicate_macro_name(macro_func.macro, macro, namespace)
-        self.packages[namespace][macro_name] = macro_func
-
-        if namespace == self.search_package:
-            self.locals[macro_name] = macro_func
-        elif namespace in {self.root_package, GLOBAL_PROJECT_NAME}:
-            self.globals[macro_name] = macro_func
-
-    def add_macros(self, macros: Iterable[ParsedMacro], ctx: Dict[str, Any]):
-        for macro in macros:
-            self.add_macro(macro, ctx)
-
-    def get_macro_dict(self) -> FullNamespace:
-        root_namespace: FullNamespace = {}
-
-        root_namespace.update(self.packages)
-        root_namespace.update(self.globals)
-        root_namespace.update(self.locals)
-
-        return root_namespace
-
-
-class ManifestContext(ConfiguredContext):
-    """The Macro context has everything in the target context, plus the macros
-    in the manifest.
-
-    The given macros can override any previous context values, which will be
-    available as if they were accessed relative to the package name.
-    """
+class ConfiguredVar(Var):
     def __init__(
         self,
+        context: Dict[str, Any],
         config: AdapterRequiredConfig,
-        manifest: Manifest,
-        search_package: str,
-    ) -> None:
-        super().__init__(config)
-        self.manifest = manifest
-        self.search_package = search_package
-        self.macro_stack = MacroStack()
+        project_name: str,
+    ):
+        super().__init__(context, config.cli_vars)
+        self._config = config
+        self._project_name = project_name
 
-    def _get_namespace(self):
-        return MacroNamespace(
-            self.config.project_name,
-            self.search_package,
-            self.macro_stack,
-            None,
+    def __call__(self, var_name, default=Var._VAR_NOTSET):
+        my_config = self._config.load_dependencies()[self._project_name]
+
+        # cli vars > active project > local project
+        if var_name in self._config.cli_vars:
+            return self._config.cli_vars[var_name]
+
+        adapter_type = self._config.credentials.type
+        lookup = FQNLookup(self._project_name)
+        active_vars = self._config.vars.vars_for(lookup, adapter_type)
+        all_vars = MultiDict([active_vars])
+
+        if self._config.project_name != my_config.project_name:
+            all_vars.add(my_config.vars.vars_for(lookup, adapter_type))
+
+        if var_name in all_vars:
+            return all_vars[var_name]
+
+        if default is not Var._VAR_NOTSET:
+            return default
+
+        return self.get_missing_var(var_name)
+
+
+class SchemaYamlContext(ConfiguredContext):
+    def __init__(self, config, project_name: str):
+        super().__init__(config)
+        self._project_name = project_name
+
+    @contextproperty
+    def var(self) -> ConfiguredVar:
+        return ConfiguredVar(
+            self._ctx, self.config, self._project_name
         )
 
-    def get_macros(self) -> Dict[str, Any]:
-        nsp = self._get_namespace()
-        nsp.add_macros(self.manifest.macros.values(), self._ctx)
-        return nsp.get_macro_dict()
 
-    def to_dict(self) -> Dict[str, Any]:
-        dct = super().to_dict()
-        dct.update(self.get_macros())
-        return dct
+class MacroResolvingContext(ConfiguredContext):
+    def __init__(self, config):
+        super().__init__(config)
+
+    @contextproperty
+    def var(self) -> ConfiguredVar:
+        return ConfiguredVar(
+            self._ctx, self.config, self.config.project_name
+        )
 
 
-class QueryHeaderContext(ManifestContext):
-    def __init__(
-        self, config: AdapterRequiredConfig, manifest: Manifest
-    ) -> None:
-        super().__init__(config, manifest, config.project_name)
-
-
-def generate_query_header_context(
-    config: AdapterRequiredConfig, manifest: Manifest
-):
-    ctx = QueryHeaderContext(config, manifest)
+def generate_schema_yml(
+    config: AdapterRequiredConfig, project_name: str
+) -> Dict[str, Any]:
+    ctx = SchemaYamlContext(config, project_name)
+    return ctx.to_dict()
+
+
+def generate_macro_context(
+    config: AdapterRequiredConfig,
+) -> Dict[str, Any]:
+    ctx = MacroResolvingContext(config)
     return ctx.to_dict()

core/dbt/context/context_config.py

@@ -0,0 +1,313 @@
+from abc import abstractmethod
+from copy import deepcopy
+from dataclasses import dataclass
+from typing import List, Iterator, Dict, Any, TypeVar, Generic
+
+from dbt.config import RuntimeConfig, Project, IsFQNResource
+from dbt.contracts.graph.model_config import BaseConfig, get_config_for
+from dbt.exceptions import InternalException
+from dbt.node_types import NodeType
+from dbt.utils import fqn_search
+
+
+@dataclass
+class ModelParts(IsFQNResource):
+    fqn: List[str]
+    resource_type: NodeType
+    package_name: str
+
+
+T = TypeVar('T')  # any old type
+C = TypeVar('C', bound=BaseConfig)
+
+
+class ConfigSource:
+    def __init__(self, project):
+        self.project = project
+
+    def get_config_dict(self, resource_type: NodeType):
+        ...
+
+
+class UnrenderedConfig(ConfigSource):
+    def __init__(self, project: Project):
+        self.project = project
+
+    def get_config_dict(self, resource_type: NodeType) -> Dict[str, Any]:
+        unrendered = self.project.unrendered.project_dict
+        if resource_type == NodeType.Seed:
+            model_configs = unrendered.get('seeds')
+        elif resource_type == NodeType.Snapshot:
+            model_configs = unrendered.get('snapshots')
+        elif resource_type == NodeType.Source:
+            model_configs = unrendered.get('sources')
+        elif resource_type == NodeType.Test:
+            model_configs = unrendered.get('tests')
+        else:
+            model_configs = unrendered.get('models')
+
+        if model_configs is None:
+            return {}
+        else:
+            return model_configs
+
+
+class RenderedConfig(ConfigSource):
+    def __init__(self, project: Project):
+        self.project = project
+
+    def get_config_dict(self, resource_type: NodeType) -> Dict[str, Any]:
+        if resource_type == NodeType.Seed:
+            model_configs = self.project.seeds
+        elif resource_type == NodeType.Snapshot:
+            model_configs = self.project.snapshots
+        elif resource_type == NodeType.Source:
+            model_configs = self.project.sources
+        elif resource_type == NodeType.Test:
+            model_configs = self.project.tests
+        else:
+            model_configs = self.project.models
+        return model_configs
+
+
+class BaseContextConfigGenerator(Generic[T]):
+    def __init__(self, active_project: RuntimeConfig):
+        self._active_project = active_project
+
+    def get_config_source(self, project: Project) -> ConfigSource:
+        return RenderedConfig(project)
+
+    def get_node_project(self, project_name: str):
+        if project_name == self._active_project.project_name:
+            return self._active_project
+        dependencies = self._active_project.load_dependencies()
+        if project_name not in dependencies:
+            raise InternalException(
+                f'Project name {project_name} not found in dependencies '
+                f'(found {list(dependencies)})'
+            )
+        return dependencies[project_name]
+
+    def _project_configs(
+        self, project: Project, fqn: List[str], resource_type: NodeType
+    ) -> Iterator[Dict[str, Any]]:
+        src = self.get_config_source(project)
+        model_configs = src.get_config_dict(resource_type)
+        for level_config in fqn_search(model_configs, fqn):
+            result = {}
+            for key, value in level_config.items():
+                if key.startswith('+'):
+                    result[key[1:].strip()] = deepcopy(value)
+                elif not isinstance(value, dict):
+                    result[key] = deepcopy(value)
+
+            yield result
+
+    def _active_project_configs(
+        self, fqn: List[str], resource_type: NodeType
+    ) -> Iterator[Dict[str, Any]]:
+        return self._project_configs(self._active_project, fqn, resource_type)
+
+    @abstractmethod
+    def _update_from_config(
+        self, result: T, partial: Dict[str, Any], validate: bool = False
+    ) -> T:
+        ...
+
+    @abstractmethod
+    def initial_result(self, resource_type: NodeType, base: bool) -> T:
+        ...
+
+    def calculate_node_config(
+        self,
+        config_call_dict: Dict[str, Any],
+        fqn: List[str],
+        resource_type: NodeType,
+        project_name: str,
+        base: bool,
+        patch_config_dict: Dict[str, Any] = None
+    ) -> BaseConfig:
+        own_config = self.get_node_project(project_name)
+
+        result = self.initial_result(resource_type=resource_type, base=base)
+
+        project_configs = self._project_configs(own_config, fqn, resource_type)
+        for fqn_config in project_configs:
+            result = self._update_from_config(result, fqn_config)
+
+        # When schema files patch config, it has lower precedence than
+        # config in the models (config_call_dict), so we add the patch_config_dict
+        # before the config_call_dict
+        if patch_config_dict:
+            result = self._update_from_config(result, patch_config_dict)
+
+        # config_calls are created in the 'experimental' model parser and
+        # the ParseConfigObject (via add_config_call)
+        result = self._update_from_config(result, config_call_dict)
+
+        if own_config.project_name != self._active_project.project_name:
+            for fqn_config in self._active_project_configs(fqn, resource_type):
+                result = self._update_from_config(result, fqn_config)
+
+        # this is mostly impactful in the snapshot config case
+        return result
+
+    @abstractmethod
+    def calculate_node_config_dict(
+        self,
+        config_call_dict: Dict[str, Any],
+        fqn: List[str],
+        resource_type: NodeType,
+        project_name: str,
+        base: bool,
+        patch_config_dict: Dict[str, Any],
+    ) -> Dict[str, Any]:
+        ...
+
+
+class ContextConfigGenerator(BaseContextConfigGenerator[C]):
+    def __init__(self, active_project: RuntimeConfig):
+        self._active_project = active_project
+
+    def get_config_source(self, project: Project) -> ConfigSource:
+        return RenderedConfig(project)
+
+    def initial_result(self, resource_type: NodeType, base: bool) -> C:
+        # defaults, own_config, config calls, active_config (if != own_config)
+        config_cls = get_config_for(resource_type, base=base)
+        # Calculate the defaults. We don't want to validate the defaults,
+        # because it might be invalid in the case of required config members
+        # (such as on snapshots!)
+        result = config_cls.from_dict({})
+        return result
+
+    def _update_from_config(
+        self, result: C, partial: Dict[str, Any], validate: bool = False
+    ) -> C:
+        translated = self._active_project.credentials.translate_aliases(
+            partial
+        )
+        return result.update_from(
+            translated,
+            self._active_project.credentials.type,
+            validate=validate
+        )
+
+    def calculate_node_config_dict(
+        self,
+        config_call_dict: Dict[str, Any],
+        fqn: List[str],
+        resource_type: NodeType,
+        project_name: str,
+        base: bool,
+        patch_config_dict: dict = None
+    ) -> Dict[str, Any]:
+        config = self.calculate_node_config(
+            config_call_dict=config_call_dict,
+            fqn=fqn,
+            resource_type=resource_type,
+            project_name=project_name,
+            base=base,
+            patch_config_dict=patch_config_dict
+        )
+        finalized = config.finalize_and_validate()
+        return finalized.to_dict(omit_none=True)
+
+
+class UnrenderedConfigGenerator(BaseContextConfigGenerator[Dict[str, Any]]):
+    def get_config_source(self, project: Project) -> ConfigSource:
+        return UnrenderedConfig(project)
+
+    def calculate_node_config_dict(
+        self,
+        config_call_dict: Dict[str, Any],
+        fqn: List[str],
+        resource_type: NodeType,
+        project_name: str,
+        base: bool,
+        patch_config_dict: dict = None
+    ) -> Dict[str, Any]:
+        return self.calculate_node_config(
+            config_call_dict=config_call_dict,
+            fqn=fqn,
+            resource_type=resource_type,
+            project_name=project_name,
+            base=base,
+            patch_config_dict=patch_config_dict
+        )
+
+    def initial_result(
+        self,
+        resource_type: NodeType,
+        base: bool
+    ) -> Dict[str, Any]:
+        return {}
+
+    def _update_from_config(
+        self,
+        result: Dict[str, Any],
+        partial: Dict[str, Any],
+        validate: bool = False,
+    ) -> Dict[str, Any]:
+        translated = self._active_project.credentials.translate_aliases(
+            partial
+        )
+        result.update(translated)
+        return result
+
+
+class ContextConfig:
+    def __init__(
+        self,
+        active_project: RuntimeConfig,
+        fqn: List[str],
+        resource_type: NodeType,
+        project_name: str,
+    ) -> None:
+        self._config_call_dict: Dict[str, Any] = {}
+        self._active_project = active_project
+        self._fqn = fqn
+        self._resource_type = resource_type
+        self._project_name = project_name
+
+    def add_config_call(self, opts: Dict[str, Any]) -> None:
+        dct = self._config_call_dict
+        self._add_config_call(dct, opts)
+
+    @classmethod
+    def _add_config_call(cls, config_call_dict, opts: Dict[str, Any]) -> None:
+        for k, v in opts.items():
+            # MergeBehavior for post-hook and pre-hook is to collect all
+            # values, instead of overwriting
+            if k in BaseConfig.mergebehavior['append']:
+                if not isinstance(v, list):
+                    v = [v]
+            if k in BaseConfig.mergebehavior['update'] and not isinstance(v, dict):
+                raise InternalException(f'expected dict, got {v}')
+            if k in config_call_dict and isinstance(config_call_dict[k], list):
+                config_call_dict[k].extend(v)
+            elif k in config_call_dict and isinstance(config_call_dict[k], dict):
+                config_call_dict[k].update(v)
+            else:
+                config_call_dict[k] = v
+
+    def build_config_dict(
+        self,
+        base: bool = False,
+        *,
+        rendered: bool = True,
+        patch_config_dict: dict = None
+    ) -> Dict[str, Any]:
+        if rendered:
+            src = ContextConfigGenerator(self._active_project)
+        else:
+            src = UnrenderedConfigGenerator(self._active_project)
+
+        return src.calculate_node_config_dict(
+            config_call_dict=self._config_call_dict,
+            fqn=self._fqn,
+            resource_type=self._resource_type,
+            project_name=self._project_name,
+            base=base,
+            patch_config_dict=patch_config_dict
+        )

core/dbt/context/docs.py

@@ -12,28 +12,10 @@ from dbt.contracts.graph.manifest import Manifest
 from dbt.contracts.graph.parsed import ParsedMacro
 
 from dbt.context.base import contextmember
-from dbt.context.configured import ConfiguredContext
+from dbt.context.configured import SchemaYamlContext
 
 
-class DocsParseContext(ConfiguredContext):
-    def __init__(
-        self,
-        config: RuntimeConfig,
-        node: Any,
-    ) -> None:
-        super().__init__(config)
-        self.node = node
-
-    @contextmember
-    def doc(self, *args: str) -> str:
-        # when you call doc(), this is what happens at parse time
-        if len(args) != 1 and len(args) != 2:
-            doc_invalid_args(self.node, args)
-        # At parse time, nothing should care about what doc() returns
-        return ''
-
-
-class DocsRuntimeContext(ConfiguredContext):
+class DocsRuntimeContext(SchemaYamlContext):
     def __init__(
         self,
         config: RuntimeConfig,
@@ -41,13 +23,31 @@ class DocsRuntimeContext(ConfiguredContext):
         manifest: Manifest,
         current_project: str,
     ) -> None:
-        super().__init__(config)
+        super().__init__(config, current_project)
         self.node = node
         self.manifest = manifest
-        self.current_project = current_project
 
     @contextmember
     def doc(self, *args: str) -> str:
+        """The `doc` function is used to reference docs blocks in schema.yml
+        files. It is analogous to the `ref` function. For more information,
+        consult the Documentation guide.
+
+        > orders.md:
+
+            {% docs orders %}
+            # docs
+            - go
+            - here
+            {% enddocs %}
+
+        > schema.yml
+
+            version: 2
+                models:
+                  - name: orders
+                    description: "{{ doc('orders') }}"
+        """
         # when you call doc(), this is what happens at runtime
         if len(args) == 1:
             doc_package_name = None
@@ -57,27 +57,24 @@ class DocsRuntimeContext(ConfiguredContext):
         else:
             doc_invalid_args(self.node, args)
 
+        # ParsedDocumentation
         target_doc = self.manifest.resolve_doc(
             doc_name,
             doc_package_name,
-            self.current_project,
+            self._project_name,
             self.node.package_name,
         )
-
-        if target_doc is None:
+        if target_doc:
+            file_id = target_doc.file_id
+            if file_id in self.manifest.files:
+                source_file = self.manifest.files[file_id]
+                source_file.add_node(self.node.unique_id)
+        else:
             doc_target_not_found(self.node, doc_name, doc_package_name)
 
         return target_doc.block_contents
 
 
-def generate_parser_docs(
-    config: RuntimeConfig,
-    unparsed: Any,
-) -> Dict[str, Any]:
-    ctx = DocsParseContext(config, unparsed)
-    return ctx.to_dict()
-
-
 def generate_runtime_docs(
     config: RuntimeConfig,
     target: Any,
@@ -85,4 +82,5 @@ def generate_runtime_docs(
     current_project: str,
 ) -> Dict[str, Any]:
     ctx = DocsRuntimeContext(config, target, manifest, current_project)
+    # This is not a Mashumaro to_dict call
     return ctx.to_dict()

core/dbt/context/macro_resolver.py

@@ -0,0 +1,199 @@
+from typing import (
+    Dict, MutableMapping, Optional
+)
+from dbt.contracts.graph.parsed import ParsedMacro
+from dbt.exceptions import raise_duplicate_macro_name, raise_compiler_error
+from dbt.include.global_project import PROJECT_NAME as GLOBAL_PROJECT_NAME
+from dbt.clients.jinja import MacroGenerator
+
+MacroNamespace = Dict[str, ParsedMacro]
+
+
+# This class builds the MacroResolver by adding macros
+# to various categories for finding macros in the right order,
+# so that higher precedence macros are found first.
+# This functionality is also provided by the MacroNamespace,
+# but the intention is to eventually replace that class.
+# This enables us to get the macro unique_id without
+# processing every macro in the project.
+# Note: the root project macros override everything in the
+# dbt internal projects. External projects (dependencies) will
+# use their own macros first, then pull from the root project
+# followed by dbt internal projects.
+class MacroResolver:
+    def __init__(
+        self,
+        macros: MutableMapping[str, ParsedMacro],
+        root_project_name: str,
+        internal_package_names,
+    ) -> None:
+        self.root_project_name = root_project_name
+        self.macros = macros
+        # internal packages comes from get_adapter_package_names
+        self.internal_package_names = internal_package_names
+
+        # To be filled in from macros.
+        self.internal_packages: Dict[str, MacroNamespace] = {}
+        self.packages: Dict[str, MacroNamespace] = {}
+        self.root_package_macros: MacroNamespace = {}
+
+        # add the macros to internal_packages, packages, and root packages
+        self.add_macros()
+        self._build_internal_packages_namespace()
+        self._build_macros_by_name()
+
+    def _build_internal_packages_namespace(self):
+        # Iterate in reverse-order and overwrite: the packages that are first
+        # in the list are the ones we want to "win".
+        self.internal_packages_namespace: MacroNamespace = {}
+        for pkg in reversed(self.internal_package_names):
+            if pkg in self.internal_packages:
+                # Turn the internal packages into a flat namespace
+                self.internal_packages_namespace.update(
+                    self.internal_packages[pkg])
+
+    # search order:
+    # local_namespace (package of particular node), not including
+    #    the internal packages or the root package
+    #    This means that within an extra package, it uses its own macros
+    # root package namespace
+    # non-internal packages (that aren't local or root)
+    # dbt internal packages
+    def _build_macros_by_name(self):
+        macros_by_name = {}
+
+        # all internal packages (already in the right order)
+        for macro in self.internal_packages_namespace.values():
+            macros_by_name[macro.name] = macro
+
+        # non-internal packages
+        for fnamespace in self.packages.values():
+            for macro in fnamespace.values():
+                macros_by_name[macro.name] = macro
+
+        # root package macros
+        for macro in self.root_package_macros.values():
+            macros_by_name[macro.name] = macro
+
+        self.macros_by_name = macros_by_name
+
+    def _add_macro_to(
+        self,
+        package_namespaces: Dict[str, MacroNamespace],
+        macro: ParsedMacro,
+    ):
+        if macro.package_name in package_namespaces:
+            namespace = package_namespaces[macro.package_name]
+        else:
+            namespace = {}
+            package_namespaces[macro.package_name] = namespace
+
+        if macro.name in namespace:
+            raise_duplicate_macro_name(
+                macro, macro, macro.package_name
+            )
+        package_namespaces[macro.package_name][macro.name] = macro
+
+    def add_macro(self, macro: ParsedMacro):
+        macro_name: str = macro.name
+
+        # internal macros (from plugins) will be processed separately from
+        # project macros, so store them in a different place
+        if macro.package_name in self.internal_package_names:
+            self._add_macro_to(self.internal_packages, macro)
+        else:
+            # if it's not an internal package
+            self._add_macro_to(self.packages, macro)
+            # add to root_package_macros if it's in the root package
+            if macro.package_name == self.root_project_name:
+                self.root_package_macros[macro_name] = macro
+
+    def add_macros(self):
+        for macro in self.macros.values():
+            self.add_macro(macro)
+
+    def get_macro(self, local_package, macro_name):
+        local_package_macros = {}
+        if (local_package not in self.internal_package_names and
+                local_package in self.packages):
+            local_package_macros = self.packages[local_package]
+        # First: search the local packages for this macro
+        if macro_name in local_package_macros:
+            return local_package_macros[macro_name]
+        # Now look up in the standard search order
+        if macro_name in self.macros_by_name:
+            return self.macros_by_name[macro_name]
+        return None
+
+    def get_macro_id(self, local_package, macro_name):
+        macro = self.get_macro(local_package, macro_name)
+        if macro is None:
+            return None
+        else:
+            return macro.unique_id
+
+
+# Currently this is just used by test processing in the schema
+# parser (in connection with the MacroResolver). Future work
+# will extend the use of these classes to other parsing areas.
+# One of the features of this class compared to the MacroNamespace
+# is that you can limit the number of macros provided to the
+# context dictionary in the 'to_dict' manifest method.
+class TestMacroNamespace:
+    def __init__(
+        self, macro_resolver, ctx, node, thread_ctx, depends_on_macros
+    ):
+        self.macro_resolver = macro_resolver
+        self.ctx = ctx
+        self.node = node  # can be none
+        self.thread_ctx = thread_ctx
+        self.local_namespace = {}
+        self.project_namespace = {}
+        if depends_on_macros:
+            dep_macros = []
+            self.recursively_get_depends_on_macros(depends_on_macros, dep_macros)
+            for macro_unique_id in dep_macros:
+                if macro_unique_id in self.macro_resolver.macros:
+                    # Split up the macro unique_id to get the project_name
+                    (_, project_name, macro_name) = macro_unique_id.split('.')
+                    # Save the plain macro_name in the local_namespace
+                    macro = self.macro_resolver.macros[macro_unique_id]
+                    macro_gen = MacroGenerator(
+                        macro, self.ctx, self.node, self.thread_ctx,
+                    )
+                    self.local_namespace[macro_name] = macro_gen
+                    # We also need the two part macro name
+                    if project_name not in self.project_namespace:
+                        self.project_namespace[project_name] = {}
+                    self.project_namespace[project_name][macro_name] = macro_gen
+
+    def recursively_get_depends_on_macros(self, depends_on_macros, dep_macros):
+        for macro_unique_id in depends_on_macros:
+            if macro_unique_id in dep_macros:
+                continue
+            dep_macros.append(macro_unique_id)
+            if macro_unique_id in self.macro_resolver.macros:
+                macro = self.macro_resolver.macros[macro_unique_id]
+                if macro.depends_on.macros:
+                    self.recursively_get_depends_on_macros(macro.depends_on.macros, dep_macros)
+
+    def get_from_package(
+        self, package_name: Optional[str], name: str
+    ) -> Optional[MacroGenerator]:
+        macro = None
+        if package_name is None:
+            macro = self.macro_resolver.macros_by_name.get(name)
+        elif package_name == GLOBAL_PROJECT_NAME:
+            macro = self.macro_resolver.internal_packages_namespace.get(name)
+        elif package_name in self.macro_resolver.packages:
+            macro = self.macro_resolver.packages[package_name].get(name)
+        else:
+            raise_compiler_error(
+                f"Could not find package '{package_name}'"
+            )
+        if not macro:
+            return None
+        macro_func = MacroGenerator(
+            macro, self.ctx, self.node, self.thread_ctx
+        )
+        return macro_func

core/dbt/context/macros.py

@@ -0,0 +1,182 @@
+from typing import (
+    Any, Dict, Iterable, Union, Optional, List, Iterator, Mapping, Set
+)
+
+from dbt.clients.jinja import MacroGenerator, MacroStack
+from dbt.contracts.graph.parsed import ParsedMacro
+from dbt.include.global_project import PROJECT_NAME as GLOBAL_PROJECT_NAME
+from dbt.exceptions import (
+    raise_duplicate_macro_name, raise_compiler_error
+)
+
+
+FlatNamespace = Dict[str, MacroGenerator]
+NamespaceMember = Union[FlatNamespace, MacroGenerator]
+FullNamespace = Dict[str, NamespaceMember]
+
+
+# The point of this class is to collect the various macros
+# and provide the ability to flatten them into the ManifestContexts
+# that are created for jinja, so that macro calls can be resolved.
+# Creates special iterators and _keys methods to flatten the lists.
+# When this class is created it has a static 'local_namespace' which
+# depends on the package of the node, so it only works for one
+# particular local package at a time for "flattening" into a context.
+# 'get_by_package' should work for any macro.
+class MacroNamespace(Mapping):
+    def __init__(
+        self,
+        global_namespace: FlatNamespace,  # root package macros
+        local_namespace: FlatNamespace,   # packages for *this* node
+        global_project_namespace: FlatNamespace,  # internal packages
+        packages: Dict[str, FlatNamespace],  # non-internal packages
+    ):
+        self.global_namespace: FlatNamespace = global_namespace
+        self.local_namespace: FlatNamespace = local_namespace
+        self.packages: Dict[str, FlatNamespace] = packages
+        self.global_project_namespace: FlatNamespace = global_project_namespace
+
+    def _search_order(self) -> Iterable[Union[FullNamespace, FlatNamespace]]:
+        yield self.local_namespace   # local package
+        yield self.global_namespace  # root package
+        yield self.packages          # non-internal packages
+        yield {
+            GLOBAL_PROJECT_NAME: self.global_project_namespace,  # dbt
+        }
+        yield self.global_project_namespace  # other internal project besides dbt
+
+    # provides special keys method for MacroNamespace iterator
+    # returns keys from local_namespace, global_namespace, packages,
+    # global_project_namespace
+    def _keys(self) -> Set[str]:
+        keys: Set[str] = set()
+        for search in self._search_order():
+            keys.update(search)
+        return keys
+
+    # special iterator using special keys
+    def __iter__(self) -> Iterator[str]:
+        for key in self._keys():
+            yield key
+
+    def __len__(self):
+        return len(self._keys())
+
+    def __getitem__(self, key: str) -> NamespaceMember:
+        for dct in self._search_order():
+            if key in dct:
+                return dct[key]
+        raise KeyError(key)
+
+    def get_from_package(
+        self, package_name: Optional[str], name: str
+    ) -> Optional[MacroGenerator]:
+        pkg: FlatNamespace
+        if package_name is None:
+            return self.get(name)
+        elif package_name == GLOBAL_PROJECT_NAME:
+            return self.global_project_namespace.get(name)
+        elif package_name in self.packages:
+            return self.packages[package_name].get(name)
+        else:
+            raise_compiler_error(
+                f"Could not find package '{package_name}'"
+            )
+
+
+# This class builds the MacroNamespace by adding macros to
+# internal_packages or packages, and locals/globals.
+# Call 'build_namespace' to return a MacroNamespace.
+# This is used by ManifestContext (and subclasses)
+class MacroNamespaceBuilder:
+    def __init__(
+        self,
+        root_package: str,
+        search_package: str,
+        thread_ctx: MacroStack,
+        internal_packages: List[str],
+        node: Optional[Any] = None,
+    ) -> None:
+        self.root_package = root_package
+        self.search_package = search_package
+        # internal packages comes from get_adapter_package_names
+        self.internal_package_names = set(internal_packages)
+        self.internal_package_names_order = internal_packages
+        # macro_func is added here if in root package, since
+        # the root package acts as a "global" namespace, overriding
+        # everything else except local external package macro calls
+        self.globals: FlatNamespace = {}
+        # macro_func is added here if it's the package for this node
+        self.locals: FlatNamespace = {}
+        # Create a dictionary of [package name][macro name] =
+        #     MacroGenerator object which acts like a function
+        self.internal_packages: Dict[str, FlatNamespace] = {}
+        self.packages: Dict[str, FlatNamespace] = {}
+        self.thread_ctx = thread_ctx
+        self.node = node
+
+    def _add_macro_to(
+        self,
+        hierarchy: Dict[str, FlatNamespace],
+        macro: ParsedMacro,
+        macro_func: MacroGenerator,
+    ):
+        if macro.package_name in hierarchy:
+            namespace = hierarchy[macro.package_name]
+        else:
+            namespace = {}
+            hierarchy[macro.package_name] = namespace
+
+        if macro.name in namespace:
+            raise_duplicate_macro_name(
+                macro_func.macro, macro, macro.package_name
+            )
+        hierarchy[macro.package_name][macro.name] = macro_func
+
+    def add_macro(self, macro: ParsedMacro, ctx: Dict[str, Any]):
+        macro_name: str = macro.name
+
+        # MacroGenerator is in clients/jinja.py
+        # a MacroGenerator object is a callable object that will
+        # execute the MacroGenerator.__call__ function
+        macro_func: MacroGenerator = MacroGenerator(
+            macro, ctx, self.node, self.thread_ctx
+        )
+
+        # internal macros (from plugins) will be processed separately from
+        # project macros, so store them in a different place
+        if macro.package_name in self.internal_package_names:
+            self._add_macro_to(self.internal_packages, macro, macro_func)
+        else:
+            # if it's not an internal package
+            self._add_macro_to(self.packages, macro, macro_func)
+            # add to locals if it's the package this node is in
+            if macro.package_name == self.search_package:
+                self.locals[macro_name] = macro_func
+            # add to globals if it's in the root package
+            elif macro.package_name == self.root_package:
+                self.globals[macro_name] = macro_func
+
+    def add_macros(self, macros: Iterable[ParsedMacro], ctx: Dict[str, Any]):
+        for macro in macros:
+            self.add_macro(macro, ctx)
+
+    def build_namespace(
+        self, macros: Iterable[ParsedMacro], ctx: Dict[str, Any]
+    ) -> MacroNamespace:
+        self.add_macros(macros, ctx)
+
+        # Iterate in reverse-order and overwrite: the packages that are first
+        # in the list are the ones we want to "win".
+        global_project_namespace: FlatNamespace = {}
+        for pkg in reversed(self.internal_package_names_order):
+            if pkg in self.internal_packages:
+                # add the macros pointed to by this package name
+                global_project_namespace.update(self.internal_packages[pkg])
+
+        return MacroNamespace(
+            global_namespace=self.globals,  # root package macros
+            local_namespace=self.locals,  # packages for *this* node
+            global_project_namespace=global_project_namespace,  # internal packages
+            packages=self.packages,  # non internal_packages
+        )

core/dbt/context/manifest.py

@@ -0,0 +1,82 @@
+from typing import List
+
+from dbt.clients.jinja import MacroStack
+from dbt.contracts.connection import AdapterRequiredConfig
+from dbt.contracts.graph.manifest import Manifest
+from dbt.context.macro_resolver import TestMacroNamespace
+
+
+from .configured import ConfiguredContext
+from .macros import MacroNamespaceBuilder
+
+
+class ManifestContext(ConfiguredContext):
+    """The Macro context has everything in the target context, plus the macros
+    in the manifest.
+
+    The given macros can override any previous context values, which will be
+    available as if they were accessed relative to the package name.
+    """
+    def __init__(
+        self,
+        config: AdapterRequiredConfig,
+        manifest: Manifest,
+        search_package: str,
+    ) -> None:
+        super().__init__(config)
+        self.manifest = manifest
+        # this is the package of the node for which this context was built
+        self.search_package = search_package
+        self.macro_stack = MacroStack()
+        # This namespace is used by the BaseDatabaseWrapper in jinja rendering.
+        # The namespace is passed to it when it's constructed. It expects
+        # to be able to do: namespace.get_from_package(..)
+        self.namespace = self._build_namespace()
+
+    def _build_namespace(self):
+        # this takes all the macros in the manifest and adds them
+        # to the MacroNamespaceBuilder stored in self.namespace
+        builder = self._get_namespace_builder()
+        return builder.build_namespace(
+            self.manifest.macros.values(), self._ctx
+        )
+
+    def _get_namespace_builder(self) -> MacroNamespaceBuilder:
+        # avoid an import loop
+        from dbt.adapters.factory import get_adapter_package_names
+        internal_packages: List[str] = get_adapter_package_names(
+            self.config.credentials.type
+        )
+        return MacroNamespaceBuilder(
+            self.config.project_name,
+            self.search_package,
+            self.macro_stack,
+            internal_packages,
+            None,
+        )
+
+    # This does not use the Mashumaro code
+    def to_dict(self):
+        dct = super().to_dict()
+        # This moves all of the macros in the 'namespace' into top level
+        # keys in the manifest dictionary
+        if isinstance(self.namespace, TestMacroNamespace):
+            dct.update(self.namespace.local_namespace)
+            dct.update(self.namespace.project_namespace)
+        else:
+            dct.update(self.namespace)
+        return dct
+
+
+class QueryHeaderContext(ManifestContext):
+    def __init__(
+        self, config: AdapterRequiredConfig, manifest: Manifest
+    ) -> None:
+        super().__init__(config, manifest, config.project_name)
+
+
+def generate_query_header_context(
+    config: AdapterRequiredConfig, manifest: Manifest
+):
+    ctx = QueryHeaderContext(config, manifest)
+    return ctx.to_dict()

core/dbt/context/target.py

@@ -14,19 +14,67 @@ class TargetContext(BaseContext):
 
     @contextproperty
     def target(self) -> Dict[str, Any]:
-        target = dict(
-            self.config.credentials.connection_info(with_aliases=True)
-        )
-        target.update({
-            'type': self.config.credentials.type,
-            'threads': self.config.threads,
-            'name': self.config.target_name,
-            # not specified, but present for compatibility
-            'target_name': self.config.target_name,
-            'profile_name': self.config.profile_name,
-            'config': self.config.config.to_dict(),
-        })
-        return target
+        """`target` contains information about your connection to the warehouse
+        (specified in profiles.yml). Some configs are shared between all
+        adapters, while others are adapter-specific.
+
+        Common:
+
+            |----------|-----------|------------------------------------------|
+            | Variable |  Example  |                Description               |
+            |----------|-----------|------------------------------------------|
+            |   name   |    dev    | Name of the active target                |
+            |----------|-----------|------------------------------------------|
+            |  schema  | dbt_alice | Name of the dbt schema (or, dataset on   |
+            |          |           | BigQuery)                                |
+            |----------|-----------|------------------------------------------|
+            |   type   |  postgres | The active adapter being used.           |
+            |----------|-----------|------------------------------------------|
+            | threads  |    4      | The number of threads in use by dbt      |
+            |----------|-----------|------------------------------------------|
+
+        Snowflake:
+
+            |----------|-----------|------------------------------------------|
+            | Variable |  Example  |                Description               |
+            |----------|-----------|------------------------------------------|
+            | database |    RAW    | The active target's database.            |
+            |----------|-----------|------------------------------------------|
+            | warehouse| TRANSFORM | The active target's warehouse.           |
+            |----------|-----------|------------------------------------------|
+            |   user   |  USERNAME | The active target's user                 |
+            |----------|-----------|------------------------------------------|
+            | role     |  ROLENAME | The active target's role                 |
+            |----------|-----------|------------------------------------------|
+            | account  |  abc123   | The active target's account              |
+            |----------|-----------|------------------------------------------|
+
+        Postgres/Redshift:
+
+            |----------|-------------------|----------------------------------|
+            | Variable |  Example          |        Description               |
+            |----------|-------------------|----------------------------------|
+            |  dbname  | analytics         | The active target's database.    |
+            |----------|-------------------|----------------------------------|
+            |  host    | abc123.us-west-2. | The active target's host.        |
+            |          | redshift.amazonaws|                                  |
+            |          |  .com             |                                  |
+            |----------|-------------------|----------------------------------|
+            |   user   |  dbt_user         |  The active target's user        |
+            |----------|-------------------|----------------------------------|
+            |   port   |    5439           | The active target's port         |
+            |----------|-------------------|----------------------------------|
+
+        BigQuery:
+
+            |----------|-----------|------------------------------------------|
+            | Variable |  Example  |                Description               |
+            |----------|-----------|------------------------------------------|
+            | project  |  abc-123  | The active target's project.             |
+            |----------|-----------|------------------------------------------|
+
+        """
+        return self.config.to_target_dict()
 
 
 def generate_target_context(

core/dbt/contracts/common.py

@@ -1,11 +0,0 @@
-
-
-def named_property(name, doc=None):
-    def get_prop(self):
-        return self._contents.get(name)
-
-    def set_prop(self, value):
-        self._contents[name] = value
-        self.validate()
-
-    return property(get_prop, set_prop, doc=doc)

core/dbt/contracts/connection.py

@@ -1,27 +1,39 @@
 import abc
 import itertools
+import hashlib
 from dataclasses import dataclass, field
 from typing import (
-    Any, ClassVar, Dict, Tuple, Iterable, Optional, NewType, List, Callable,
-    Union
+    Any, ClassVar, Dict, Tuple, Iterable, Optional, List, Callable,
 )
-from typing_extensions import Protocol
-
-from hologram import JsonSchemaMixin
-from hologram.helpers import (
-    StrEnum, register_pattern, ExtensibleJsonSchemaMixin
-)
-
-from dbt.contracts.util import Replaceable
 from dbt.exceptions import InternalException
-from dbt.helper_types import NoValue
 from dbt.utils import translate_aliases
+from dbt.logger import GLOBAL_LOGGER as logger
+from typing_extensions import Protocol
+from dbt.dataclass_schema import (
+    dbtClassMixin, StrEnum, ExtensibleDbtClassMixin, HyphenatedDbtClassMixin,
+    ValidatedStringMixin, register_pattern
+)
+from dbt.contracts.util import Replaceable
 
 
-Identifier = NewType('Identifier', str)
+class Identifier(ValidatedStringMixin):
+    ValidationRegex = r'^[A-Za-z_][A-Za-z0-9_]+$'
+
+
+# we need register_pattern for jsonschema validation
 register_pattern(Identifier, r'^[A-Za-z_][A-Za-z0-9_]+$')
 
 
+@dataclass
+class AdapterResponse(dbtClassMixin):
+    _message: str
+    code: Optional[str] = None
+    rows_affected: Optional[int] = None
+
+    def __str__(self):
+        return self._message
+
+
 class ConnectionState(StrEnum):
     INIT = 'init'
     OPEN = 'open'
@@ -30,20 +42,19 @@ class ConnectionState(StrEnum):
 
 
 @dataclass(init=False)
-class Connection(ExtensibleJsonSchemaMixin, Replaceable):
+class Connection(ExtensibleDbtClassMixin, Replaceable):
     type: Identifier
-    name: Optional[str]
+    name: Optional[str] = None
     state: ConnectionState = ConnectionState.INIT
     transaction_open: bool = False
-    # prevent serialization
     _handle: Optional[Any] = None
-    _credentials: JsonSchemaMixin = field(init=False)
+    _credentials: Optional[Any] = None
 
     def __init__(
         self,
         type: Identifier,
         name: Optional[str],
-        credentials: JsonSchemaMixin,
+        credentials: dbtClassMixin,
         state: ConnectionState = ConnectionState.INIT,
         transaction_open: bool = False,
         handle: Optional[Any] = None,
@@ -85,10 +96,15 @@ class LazyHandle:
     """Opener must be a callable that takes a Connection object and opens the
     connection, updating the handle on the Connection.
     """
+
     def __init__(self, opener: Callable[[Connection], Connection]):
         self.opener = opener
 
     def resolve(self, connection: Connection) -> Connection:
+        logger.debug(
+            'Opening a new connection, currently in state {}'
+            .format(connection.state)
+        )
         return self.opener(connection)
 
 
@@ -98,7 +114,7 @@ class LazyHandle:
 # will work.
 @dataclass  # type: ignore
 class Credentials(
-    ExtensibleJsonSchemaMixin,
+    ExtensibleDbtClassMixin,
     Replaceable,
     metaclass=abc.ABCMeta
 ):
@@ -112,12 +128,21 @@ class Credentials(
             'type not implemented for base credentials class'
         )
 
+    @abc.abstractproperty
+    def unique_field(self) -> str:
+        raise NotImplementedError(
+            'type not implemented for base credentials class'
+        )
+
+    def hashed_unique_field(self) -> str:
+        return hashlib.md5(self.unique_field.encode('utf-8')).hexdigest()
+
     def connection_info(
         self, *, with_aliases: bool = False
     ) -> Iterable[Tuple[str, Any]]:
         """Return an ordered iterator of key/value pairs for pretty-printing.
         """
-        as_dict = self.to_dict(omit_none=False, with_aliases=with_aliases)
+        as_dict = self.to_dict(omit_none=False)
         connection_keys = set(self._connection_keys())
         aliases: List[str] = []
         if with_aliases:
@@ -133,39 +158,37 @@ class Credentials(
         raise NotImplementedError
 
     @classmethod
-    def from_dict(cls, data):
+    def __pre_deserialize__(cls, data):
+        data = super().__pre_deserialize__(data)
         data = cls.translate_aliases(data)
-        return super().from_dict(data)
+        return data
 
     @classmethod
-    def translate_aliases(cls, kwargs: Dict[str, Any]) -> Dict[str, Any]:
-        return translate_aliases(kwargs, cls._ALIASES)
+    def translate_aliases(
+        cls, kwargs: Dict[str, Any], recurse: bool = False
+    ) -> Dict[str, Any]:
+        return translate_aliases(kwargs, cls._ALIASES, recurse)
 
-    def to_dict(self, omit_none=True, validate=False, *, with_aliases=False):
-        serialized = super().to_dict(omit_none=omit_none, validate=validate)
-        if with_aliases:
-            serialized.update({
-                new_name: serialized[canonical_name]
+    def __post_serialize__(self, dct):
+        # no super() -- do we need it?
+        if self._ALIASES:
+            dct.update({
+                new_name: dct[canonical_name]
                 for new_name, canonical_name in self._ALIASES.items()
-                if canonical_name in serialized
+                if canonical_name in dct
             })
-        return serialized
+        return dct
 
 
 class UserConfigContract(Protocol):
     send_anonymous_usage_stats: bool
-    use_colors: bool
-    partial_parse: Optional[bool]
-    printer_width: Optional[int]
+    use_colors: Optional[bool] = None
+    partial_parse: Optional[bool] = None
+    printer_width: Optional[int] = None
 
     def set_values(self, cookie_dir: str) -> None:
         ...
 
-    def to_dict(
-        self, omit_none: bool = True, validate: bool = False
-    ) -> Dict[str, Any]:
-        ...
-
 
 class HasCredentials(Protocol):
     credentials: Credentials
@@ -174,9 +197,39 @@ class HasCredentials(Protocol):
     target_name: str
     threads: int
 
+    def to_target_dict(self):
+        raise NotImplementedError('to_target_dict not implemented')
+
+
+DEFAULT_QUERY_COMMENT = '''
+{%- set comment_dict = {} -%}
+{%- do comment_dict.update(
+    app='dbt',
+    dbt_version=dbt_version,
+    profile_name=target.get('profile_name'),
+    target_name=target.get('target_name'),
+) -%}
+{%- if node is not none -%}
+  {%- do comment_dict.update(
+    node_id=node.unique_id,
+  ) -%}
+{% else %}
+  {# in the node context, the connection name is the node_id #}
+  {%- do comment_dict.update(connection_name=connection_name) -%}
+{%- endif -%}
+{{ return(tojson(comment_dict)) }}
+'''
+
+
+@dataclass
+class QueryComment(HyphenatedDbtClassMixin):
+    comment: str = DEFAULT_QUERY_COMMENT
+    append: bool = False
+    job_label: bool = False
+
 
 class AdapterRequiredConfig(HasCredentials, Protocol):
     project_name: str
-    query_comment: Optional[Union[str, NoValue]]
+    query_comment: QueryComment
     cli_vars: Dict[str, Any]
     target_path: str

core/dbt/contracts/files.py

@@ -0,0 +1,298 @@
+import hashlib
+import os
+from dataclasses import dataclass, field
+from mashumaro.types import SerializableType
+from typing import List, Optional, Union, Dict, Any
+
+from dbt.dataclass_schema import dbtClassMixin, StrEnum
+
+from .util import SourceKey
+
+
+MAXIMUM_SEED_SIZE = 1 * 1024 * 1024
+MAXIMUM_SEED_SIZE_NAME = '1MB'
+
+
+class ParseFileType(StrEnum):
+    Macro = 'macro'
+    Model = 'model'
+    Snapshot = 'snapshot'
+    Analysis = 'analysis'
+    Test = 'test'
+    Seed = 'seed'
+    Documentation = 'docs'
+    Schema = 'schema'
+    Hook = 'hook'   # not a real filetype, from dbt_project.yml
+
+
+parse_file_type_to_parser = {
+    ParseFileType.Macro: 'MacroParser',
+    ParseFileType.Model: 'ModelParser',
+    ParseFileType.Snapshot: 'SnapshotParser',
+    ParseFileType.Analysis: 'AnalysisParser',
+    ParseFileType.Test: 'DataTestParser',
+    ParseFileType.Seed: 'SeedParser',
+    ParseFileType.Documentation: 'DocumentationParser',
+    ParseFileType.Schema: 'SchemaParser',
+    ParseFileType.Hook: 'HookParser',
+}
+
+
+@dataclass
+class FilePath(dbtClassMixin):
+    searched_path: str
+    relative_path: str
+    project_root: str
+
+    @property
+    def search_key(self) -> str:
+        # TODO: should this be project name + path relative to project root?
+        return self.absolute_path
+
+    @property
+    def full_path(self) -> str:
+        # useful for symlink preservation
+        return os.path.join(
+            self.project_root, self.searched_path, self.relative_path
+        )
+
+    @property
+    def absolute_path(self) -> str:
+        return os.path.abspath(self.full_path)
+
+    @property
+    def original_file_path(self) -> str:
+        # this is mostly used for reporting errors. It doesn't show the project
+        # name, should it?
+        return os.path.join(
+            self.searched_path, self.relative_path
+        )
+
+    def seed_too_large(self) -> bool:
+        """Return whether the file this represents is over the seed size limit
+        """
+        return os.stat(self.full_path).st_size > MAXIMUM_SEED_SIZE
+
+
+@dataclass
+class FileHash(dbtClassMixin):
+    name: str  # the hash type name
+    checksum: str  # the hashlib.hash_type().hexdigest() of the file contents
+
+    @classmethod
+    def empty(cls):
+        return FileHash(name='none', checksum='')
+
+    @classmethod
+    def path(cls, path: str):
+        return FileHash(name='path', checksum=path)
+
+    def __eq__(self, other):
+        if not isinstance(other, FileHash):
+            return NotImplemented
+
+        if self.name == 'none' or self.name != other.name:
+            return False
+
+        return self.checksum == other.checksum
+
+    def compare(self, contents: str) -> bool:
+        """Compare the file contents with the given hash"""
+        if self.name == 'none':
+            return False
+
+        return self.from_contents(contents, name=self.name) == self.checksum
+
+    @classmethod
+    def from_contents(cls, contents: str, name='sha256') -> 'FileHash':
+        """Create a file hash from the given file contents. The hash is always
+        the utf-8 encoding of the contents given, because dbt only reads files
+        as utf-8.
+        """
+        data = contents.encode('utf-8')
+        checksum = hashlib.new(name, data).hexdigest()
+        return cls(name=name, checksum=checksum)
+
+
+@dataclass
+class RemoteFile(dbtClassMixin):
+    @property
+    def searched_path(self) -> str:
+        return 'from remote system'
+
+    @property
+    def relative_path(self) -> str:
+        return 'from remote system'
+
+    @property
+    def absolute_path(self) -> str:
+        return 'from remote system'
+
+    @property
+    def original_file_path(self):
+        return 'from remote system'
+
+
+@dataclass
+class BaseSourceFile(dbtClassMixin, SerializableType):
+    """Define a source file in dbt"""
+    path: Union[FilePath, RemoteFile]  # the path information
+    checksum: FileHash
+    # Seems like knowing which project the file came from would be useful
+    project_name: Optional[str] = None
+    # Parse file type: i.e. which parser will process this file
+    parse_file_type: Optional[ParseFileType] = None
+    # we don't want to serialize this
+    contents: Optional[str] = None
+    # the unique IDs contained in this file
+
+    @property
+    def file_id(self):
+        if isinstance(self.path, RemoteFile):
+            return None
+        if self.checksum.name == 'none':
+            return None
+        return f'{self.project_name}://{self.path.original_file_path}'
+
+    def _serialize(self):
+        dct = self.to_dict()
+        return dct
+
+    @classmethod
+    def _deserialize(cls, dct: Dict[str, int]):
+        if dct['parse_file_type'] == 'schema':
+            sf = SchemaSourceFile.from_dict(dct)
+        else:
+            sf = SourceFile.from_dict(dct)
+        return sf
+
+    def __post_serialize__(self, dct):
+        dct = super().__post_serialize__(dct)
+        # remove empty lists to save space
+        dct_keys = list(dct.keys())
+        for key in dct_keys:
+            if isinstance(dct[key], list) and not dct[key]:
+                del dct[key]
+        # remove contents. Schema files will still have 'dict_from_yaml'
+        # from the contents
+        if 'contents' in dct:
+            del dct['contents']
+        return dct
+
+
+@dataclass
+class SourceFile(BaseSourceFile):
+    nodes: List[str] = field(default_factory=list)
+    docs: List[str] = field(default_factory=list)
+    macros: List[str] = field(default_factory=list)
+
+    @classmethod
+    def big_seed(cls, path: FilePath) -> 'SourceFile':
+        """Parse seeds over the size limit with just the path"""
+        self = cls(path=path, checksum=FileHash.path(path.original_file_path))
+        self.contents = ''
+        return self
+
+    def add_node(self, value):
+        if value not in self.nodes:
+            self.nodes.append(value)
+
+    # TODO: do this a different way. This remote file kludge isn't going
+    # to work long term
+    @classmethod
+    def remote(cls, contents: str, project_name: str) -> 'SourceFile':
+        self = cls(
+            path=RemoteFile(),
+            checksum=FileHash.from_contents(contents),
+            project_name=project_name,
+            contents=contents,
+        )
+        return self
+
+
+@dataclass
+class SchemaSourceFile(BaseSourceFile):
+    dfy: Dict[str, Any] = field(default_factory=dict)
+    # these are in the manifest.nodes dictionary
+    tests: Dict[str, Any] = field(default_factory=dict)
+    sources: List[str] = field(default_factory=list)
+    exposures: List[str] = field(default_factory=list)
+    # node patches contain models, seeds, snapshots, analyses
+    ndp: List[str] = field(default_factory=list)
+    # any macro patches in this file by macro unique_id.
+    mcp: Dict[str, str] = field(default_factory=dict)
+    # any source patches in this file. The entries are package, name pairs
+    # Patches are only against external sources. Sources can be
+    # created too, but those are in 'sources'
+    sop: List[SourceKey] = field(default_factory=list)
+    pp_dict: Optional[Dict[str, Any]] = None
+    pp_test_index: Optional[Dict[str, Any]] = None
+
+    @property
+    def dict_from_yaml(self):
+        return self.dfy
+
+    @property
+    def node_patches(self):
+        return self.ndp
+
+    @property
+    def macro_patches(self):
+        return self.mcp
+
+    @property
+    def source_patches(self):
+        return self.sop
+
+    def __post_serialize__(self, dct):
+        dct = super().__post_serialize__(dct)
+        # Remove partial parsing specific data
+        for key in ('pp_files', 'pp_test_index', 'pp_dict'):
+            if key in dct:
+                del dct[key]
+        return dct
+
+    def append_patch(self, yaml_key, unique_id):
+        self.node_patches.append(unique_id)
+
+    def add_test(self, node_unique_id, test_from):
+        name = test_from['name']
+        key = test_from['key']
+        if key not in self.tests:
+            self.tests[key] = {}
+        if name not in self.tests[key]:
+            self.tests[key][name] = []
+        self.tests[key][name].append(node_unique_id)
+
+    def remove_tests(self, yaml_key, name):
+        if yaml_key in self.tests:
+            if name in self.tests[yaml_key]:
+                del self.tests[yaml_key][name]
+
+    def get_tests(self, yaml_key, name):
+        if yaml_key in self.tests:
+            if name in self.tests[yaml_key]:
+                return self.tests[yaml_key][name]
+        return []
+
+    def get_key_and_name_for_test(self, test_unique_id):
+        yaml_key = None
+        block_name = None
+        for key in self.tests.keys():
+            for name in self.tests[key]:
+                for unique_id in self.tests[key][name]:
+                    if unique_id == test_unique_id:
+                        yaml_key = key
+                        block_name = name
+                        break
+        return (yaml_key, block_name)
+
+    def get_all_test_ids(self):
+        test_ids = []
+        for key in self.tests.keys():
+            for name in self.tests[key]:
+                test_ids.extend(self.tests[key][name])
+        return test_ids
+
+
+AnySourceFile = Union[SchemaSourceFile, SourceFile]

core/dbt/contracts/graph/compiled.py

@@ -1,61 +1,49 @@
 from dbt.contracts.graph.parsed import (
+    HasTestMetadata,
     ParsedNode,
     ParsedAnalysisNode,
-    ParsedModelNode,
+    ParsedDataTestNode,
     ParsedHookNode,
+    ParsedModelNode,
+    ParsedExposure,
     ParsedResource,
     ParsedRPCNode,
+    ParsedSchemaTestNode,
     ParsedSeedNode,
     ParsedSnapshotNode,
     ParsedSourceDefinition,
-    ParsedTestNode,
     SeedConfig,
     TestConfig,
-    TestMetadata,
-    PARSED_TYPES,
+    same_seeds,
 )
 from dbt.node_types import NodeType
 from dbt.contracts.util import Replaceable
-from dbt.exceptions import InternalException, RuntimeException
 
-from hologram import JsonSchemaMixin
+from dbt.dataclass_schema import dbtClassMixin
 from dataclasses import dataclass, field
-import sqlparse  # type: ignore
 from typing import Optional, List, Union, Dict, Type
 
 
 @dataclass
-class InjectedCTE(JsonSchemaMixin, Replaceable):
+class InjectedCTE(dbtClassMixin, Replaceable):
     id: str
     sql: str
 
-# for some frustrating reason, we can't subclass from ParsedNode directly,
-# or typing.Union will flatten CompiledNode+ParsedNode into just ParsedNode.
-# TODO: understand that issue and come up with some way for these two to share
-# logic
+
+@dataclass
+class CompiledNodeMixin(dbtClassMixin):
+    # this is a special mixin class to provide a required argument. If a node
+    # is missing a `compiled` flag entirely, it must not be a CompiledNode.
+    compiled: bool
 
 
 @dataclass
-class CompiledNode(ParsedNode):
-    compiled: bool = False
+class CompiledNode(ParsedNode, CompiledNodeMixin):
     compiled_sql: Optional[str] = None
     extra_ctes_injected: bool = False
     extra_ctes: List[InjectedCTE] = field(default_factory=list)
-    injected_sql: Optional[str] = None
-    wrapped_sql: Optional[str] = None
-
-    def prepend_ctes(self, prepended_ctes: List[InjectedCTE]):
-        self.extra_ctes_injected = True
-        self.extra_ctes = prepended_ctes
-        if self.compiled_sql is None:
-            raise RuntimeException(
-                'Cannot prepend ctes to an unparsed node', self
-            )
-        self.injected_sql = _inject_ctes_into_sql(
-            self.compiled_sql,
-            prepended_ctes,
-        )
-        self.validate(self.to_dict())
+    relation_name: Optional[str] = None
+    _pre_injected_sql: Optional[str] = None
 
     def set_cte(self, cte_id: str, sql: str):
         """This is the equivalent of what self.extra_ctes[cte_id] = sql would
@@ -68,6 +56,12 @@ class CompiledNode(ParsedNode):
         else:
             self.extra_ctes.append(InjectedCTE(id=cte_id, sql=sql))
 
+    def __post_serialize__(self, dct):
+        dct = super().__post_serialize__(dct)
+        if '_pre_injected_sql' in dct:
+            del dct['_pre_injected_sql']
+        return dct
+
 
 @dataclass
 class CompiledAnalysisNode(CompiledNode):
@@ -94,21 +88,18 @@ class CompiledRPCNode(CompiledNode):
 
 @dataclass
 class CompiledSeedNode(CompiledNode):
+    # keep this in sync with ParsedSeedNode!
     resource_type: NodeType = field(metadata={'restrict': [NodeType.Seed]})
     config: SeedConfig = field(default_factory=SeedConfig)
-    seed_file_path: str = ''
-
-    def __post_init__(self):
-        if self.seed_file_path == '':
-            raise InternalException(
-                'Seeds should always have a seed_file_path'
-            )
 
     @property
     def empty(self):
         """ Seeds are never empty"""
         return False
 
+    def same_body(self, other) -> bool:
+        return same_seeds(self, other)
+
 
 @dataclass
 class CompiledSnapshotNode(CompiledNode):
@@ -116,124 +107,121 @@ class CompiledSnapshotNode(CompiledNode):
 
 
 @dataclass
-class CompiledTestNode(CompiledNode):
+class CompiledDataTestNode(CompiledNode):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Test]})
+    # Was not able to make mypy happy and keep the code working. We need to
+    # refactor the various configs.
+    config: TestConfig = field(default_factory=TestConfig)  # type:ignore
+
+
+@dataclass
+class CompiledSchemaTestNode(CompiledNode, HasTestMetadata):
+    # keep this in sync with ParsedSchemaTestNode!
     resource_type: NodeType = field(metadata={'restrict': [NodeType.Test]})
     column_name: Optional[str] = None
-    config: TestConfig = field(default_factory=TestConfig)
-    test_metadata: Optional[TestMetadata] = None
+    # Was not able to make mypy happy and keep the code working. We need to
+    # refactor the various configs.
+    config: TestConfig = field(default_factory=TestConfig)  # type:ignore
+
+    def same_contents(self, other) -> bool:
+        if other is None:
+            return False
+
+        return (
+            self.same_config(other) and
+            self.same_fqn(other) and
+            True
+        )
 
 
-def _inject_ctes_into_sql(sql: str, ctes: List[InjectedCTE]) -> str:
-    """
-    `ctes` is a list of InjectedCTEs like:
-
-        [
-            InjectedCTE(
-                id="cte_id_1",
-                sql="__dbt__CTE__ephemeral as (select * from table)",
-            ),
-            InjectedCTE(
-                id="cte_id_2",
-                sql="__dbt__CTE__events as (select id, type from events)",
-            ),
-        ]
-
-    Given `sql` like:
-
-      "with internal_cte as (select * from sessions)
-       select * from internal_cte"
-
-    This will spit out:
-
-      "with __dbt__CTE__ephemeral as (select * from table),
-            __dbt__CTE__events as (select id, type from events),
-            with internal_cte as (select * from sessions)
-       select * from internal_cte"
-
-    (Whitespace enhanced for readability.)
-    """
-    if len(ctes) == 0:
-        return sql
-
-    parsed_stmts = sqlparse.parse(sql)
-    parsed = parsed_stmts[0]
-
-    with_stmt = None
-    for token in parsed.tokens:
-        if token.is_keyword and token.normalized == 'WITH':
-            with_stmt = token
-            break
-
-    if with_stmt is None:
-        # no with stmt, add one, and inject CTEs right at the beginning
-        first_token = parsed.token_first()
-        with_stmt = sqlparse.sql.Token(sqlparse.tokens.Keyword, 'with')
-        parsed.insert_before(first_token, with_stmt)
-    else:
-        # stmt exists, add a comma (which will come after injected CTEs)
-        trailing_comma = sqlparse.sql.Token(sqlparse.tokens.Punctuation, ',')
-        parsed.insert_after(with_stmt, trailing_comma)
-
-    token = sqlparse.sql.Token(
-        sqlparse.tokens.Keyword,
-        ", ".join(c.sql for c in ctes)
-    )
-    parsed.insert_after(with_stmt, token)
-
-    return str(parsed)
+CompiledTestNode = Union[CompiledDataTestNode, CompiledSchemaTestNode]
 
 
-COMPILED_TYPES: Dict[NodeType, Type[CompiledNode]] = {
-    NodeType.Analysis: CompiledAnalysisNode,
-    NodeType.Model: CompiledModelNode,
-    NodeType.Operation: CompiledHookNode,
-    NodeType.RPCCall: CompiledRPCNode,
-    NodeType.Seed: CompiledSeedNode,
-    NodeType.Snapshot: CompiledSnapshotNode,
-    NodeType.Test: CompiledTestNode,
+PARSED_TYPES: Dict[Type[CompiledNode], Type[ParsedResource]] = {
+    CompiledAnalysisNode: ParsedAnalysisNode,
+    CompiledModelNode: ParsedModelNode,
+    CompiledHookNode: ParsedHookNode,
+    CompiledRPCNode: ParsedRPCNode,
+    CompiledSeedNode: ParsedSeedNode,
+    CompiledSnapshotNode: ParsedSnapshotNode,
+    CompiledDataTestNode: ParsedDataTestNode,
+    CompiledSchemaTestNode: ParsedSchemaTestNode,
 }
 
 
-def compiled_type_for(parsed: ParsedNode):
-    if parsed.resource_type in COMPILED_TYPES:
-        return COMPILED_TYPES[parsed.resource_type]
+COMPILED_TYPES: Dict[Type[ParsedResource], Type[CompiledNode]] = {
+    ParsedAnalysisNode: CompiledAnalysisNode,
+    ParsedModelNode: CompiledModelNode,
+    ParsedHookNode: CompiledHookNode,
+    ParsedRPCNode: CompiledRPCNode,
+    ParsedSeedNode: CompiledSeedNode,
+    ParsedSnapshotNode: CompiledSnapshotNode,
+    ParsedDataTestNode: CompiledDataTestNode,
+    ParsedSchemaTestNode: CompiledSchemaTestNode,
+}
+
+
+# for some types, the compiled type is the parsed type, so make this easy
+CompiledType = Union[Type[CompiledNode], Type[ParsedResource]]
+CompiledResource = Union[ParsedResource, CompiledNode]
+
+
+def compiled_type_for(parsed: ParsedNode) -> CompiledType:
+    if type(parsed) in COMPILED_TYPES:
+        return COMPILED_TYPES[type(parsed)]
     else:
         return type(parsed)
 
 
 def parsed_instance_for(compiled: CompiledNode) -> ParsedResource:
-    cls = PARSED_TYPES.get(compiled.resource_type)
+    cls = PARSED_TYPES.get(type(compiled))
     if cls is None:
         # how???
         raise ValueError('invalid resource_type: {}'
                          .format(compiled.resource_type))
 
-    # validate=False to allow extra keys from compiling
-    return cls.from_dict(compiled.to_dict(), validate=False)
+    return cls.from_dict(compiled.to_dict(omit_none=True))
 
 
-# This is anything that can be in manifest.nodes and isn't a Source.
-NonSourceNode = Union[
+NonSourceCompiledNode = Union[
     CompiledAnalysisNode,
+    CompiledDataTestNode,
     CompiledModelNode,
     CompiledHookNode,
     CompiledRPCNode,
+    CompiledSchemaTestNode,
     CompiledSeedNode,
     CompiledSnapshotNode,
-    CompiledTestNode,
+]
+
+NonSourceParsedNode = Union[
     ParsedAnalysisNode,
-    ParsedModelNode,
+    ParsedDataTestNode,
     ParsedHookNode,
+    ParsedModelNode,
     ParsedRPCNode,
+    ParsedSchemaTestNode,
     ParsedSeedNode,
     ParsedSnapshotNode,
-    ParsedTestNode,
+]
+
+
+# This is anything that can be in manifest.nodes.
+ManifestNode = Union[
+    NonSourceCompiledNode,
+    NonSourceParsedNode,
 ]
 
 # We allow either parsed or compiled nodes, or parsed sources, as some
 # 'compile()' calls in the runner actually just return the original parsed
 # node they were given.
 CompileResultNode = Union[
-    NonSourceNode,
+    ManifestNode,
     ParsedSourceDefinition,
 ]
+
+# anything that participates in the graph: sources, exposures, manifest nodes
+GraphMemberNode = Union[
+    CompileResultNode,
+    ParsedExposure,
+]

core/dbt/contracts/graph/model_config.py

@@ -0,0 +1,566 @@
+from dataclasses import field, Field, dataclass
+from enum import Enum
+from itertools import chain
+from typing import (
+    Any, List, Optional, Dict, Union, Type, TypeVar, Callable
+)
+from dbt.dataclass_schema import (
+    dbtClassMixin, ValidationError, register_pattern,
+)
+from dbt.contracts.graph.unparsed import AdditionalPropertiesAllowed
+from dbt.exceptions import InternalException, CompilationException
+from dbt.contracts.util import Replaceable, list_str
+from dbt import hooks
+from dbt.node_types import NodeType
+
+
+M = TypeVar('M', bound='Metadata')
+
+
+def _get_meta_value(cls: Type[M], fld: Field, key: str, default: Any) -> M:
+    # a metadata field might exist. If it does, it might have a matching key.
+    # If it has both, make sure the value is valid and return it. If it
+    # doesn't, return the default.
+    if fld.metadata:
+        value = fld.metadata.get(key, default)
+    else:
+        value = default
+
+    try:
+        return cls(value)
+    except ValueError as exc:
+        raise InternalException(
+            f'Invalid {cls} value: {value}'
+        ) from exc
+
+
+def _set_meta_value(
+    obj: M, key: str, existing: Optional[Dict[str, Any]] = None
+) -> Dict[str, Any]:
+    if existing is None:
+        result = {}
+    else:
+        result = existing.copy()
+    result.update({key: obj})
+    return result
+
+
+class Metadata(Enum):
+    @classmethod
+    def from_field(cls: Type[M], fld: Field) -> M:
+        default = cls.default_field()
+        key = cls.metadata_key()
+
+        return _get_meta_value(cls, fld, key, default)
+
+    def meta(
+        self, existing: Optional[Dict[str, Any]] = None
+    ) -> Dict[str, Any]:
+        key = self.metadata_key()
+        return _set_meta_value(self, key, existing)
+
+    @classmethod
+    def default_field(cls) -> 'Metadata':
+        raise NotImplementedError('Not implemented')
+
+    @classmethod
+    def metadata_key(cls) -> str:
+        raise NotImplementedError('Not implemented')
+
+
+class MergeBehavior(Metadata):
+    Append = 1
+    Update = 2
+    Clobber = 3
+
+    @classmethod
+    def default_field(cls) -> 'MergeBehavior':
+        return cls.Clobber
+
+    @classmethod
+    def metadata_key(cls) -> str:
+        return 'merge'
+
+
+class ShowBehavior(Metadata):
+    Show = 1
+    Hide = 2
+
+    @classmethod
+    def default_field(cls) -> 'ShowBehavior':
+        return cls.Show
+
+    @classmethod
+    def metadata_key(cls) -> str:
+        return 'show_hide'
+
+    @classmethod
+    def should_show(cls, fld: Field) -> bool:
+        return cls.from_field(fld) == cls.Show
+
+
+class CompareBehavior(Metadata):
+    Include = 1
+    Exclude = 2
+
+    @classmethod
+    def default_field(cls) -> 'CompareBehavior':
+        return cls.Include
+
+    @classmethod
+    def metadata_key(cls) -> str:
+        return 'compare'
+
+    @classmethod
+    def should_include(cls, fld: Field) -> bool:
+        return cls.from_field(fld) == cls.Include
+
+
+def metas(*metas: Metadata) -> Dict[str, Any]:
+    existing: Dict[str, Any] = {}
+    for m in metas:
+        existing = m.meta(existing)
+    return existing
+
+
+def _listify(value: Any) -> List:
+    if isinstance(value, list):
+        return value[:]
+    else:
+        return [value]
+
+
+def _merge_field_value(
+    merge_behavior: MergeBehavior,
+    self_value: Any,
+    other_value: Any,
+):
+    if merge_behavior == MergeBehavior.Clobber:
+        return other_value
+    elif merge_behavior == MergeBehavior.Append:
+        return _listify(self_value) + _listify(other_value)
+    elif merge_behavior == MergeBehavior.Update:
+        if not isinstance(self_value, dict):
+            raise InternalException(f'expected dict, got {self_value}')
+        if not isinstance(other_value, dict):
+            raise InternalException(f'expected dict, got {other_value}')
+        value = self_value.copy()
+        value.update(other_value)
+        return value
+    else:
+        raise InternalException(
+            f'Got an invalid merge_behavior: {merge_behavior}'
+        )
+
+
+def insensitive_patterns(*patterns: str):
+    lowercased = []
+    for pattern in patterns:
+        lowercased.append(
+            ''.join('[{}{}]'.format(s.upper(), s.lower()) for s in pattern)
+        )
+    return '^({})$'.format('|'.join(lowercased))
+
+
+class Severity(str):
+    pass
+
+
+register_pattern(Severity, insensitive_patterns('warn', 'error'))
+
+
+@dataclass
+class Hook(dbtClassMixin, Replaceable):
+    sql: str
+    transaction: bool = True
+    index: Optional[int] = None
+
+
+T = TypeVar('T', bound='BaseConfig')
+
+
+@dataclass
+class BaseConfig(
+    AdditionalPropertiesAllowed, Replaceable
+):
+
+    # enable syntax like: config['key']
+    def __getitem__(self, key):
+        return self.get(key)
+
+    # like doing 'get' on a dictionary
+    def get(self, key, default=None):
+        if hasattr(self, key):
+            return getattr(self, key)
+        elif key in self._extra:
+            return self._extra[key]
+        else:
+            return default
+
+    # enable syntax like: config['key'] = value
+    def __setitem__(self, key, value):
+        if hasattr(self, key):
+            setattr(self, key, value)
+        else:
+            self._extra[key] = value
+
+    def __delitem__(self, key):
+        if hasattr(self, key):
+            msg = (
+                'Error, tried to delete config key "{}": Cannot delete '
+                'built-in keys'
+            ).format(key)
+            raise CompilationException(msg)
+        else:
+            del self._extra[key]
+
+    def _content_iterator(self, include_condition: Callable[[Field], bool]):
+        seen = set()
+        for fld, _ in self._get_fields():
+            seen.add(fld.name)
+            if include_condition(fld):
+                yield fld.name
+
+        for key in self._extra:
+            if key not in seen:
+                seen.add(key)
+                yield key
+
+    def __iter__(self):
+        yield from self._content_iterator(include_condition=lambda f: True)
+
+    def __len__(self):
+        return len(self._get_fields()) + len(self._extra)
+
+    @staticmethod
+    def compare_key(
+        unrendered: Dict[str, Any],
+        other: Dict[str, Any],
+        key: str,
+    ) -> bool:
+        if key not in unrendered and key not in other:
+            return True
+        elif key not in unrendered and key in other:
+            return False
+        elif key in unrendered and key not in other:
+            return False
+        else:
+            return unrendered[key] == other[key]
+
+    @classmethod
+    def same_contents(
+        cls, unrendered: Dict[str, Any], other: Dict[str, Any]
+    ) -> bool:
+        """This is like __eq__, except it ignores some fields."""
+        seen = set()
+        for fld, target_name in cls._get_fields():
+            key = target_name
+            seen.add(key)
+            if CompareBehavior.should_include(fld):
+                if not cls.compare_key(unrendered, other, key):
+                    return False
+
+        for key in chain(unrendered, other):
+            if key not in seen:
+                seen.add(key)
+                if not cls.compare_key(unrendered, other, key):
+                    return False
+        return True
+
+    # This is used in 'add_config_call' to created the combined config_call_dict.
+    # 'meta' moved here from node
+    mergebehavior = {
+        "append": ['pre-hook', 'pre_hook', 'post-hook', 'post_hook', 'tags'],
+        "update": ['quoting', 'column_types', 'meta'],
+    }
+
+    @classmethod
+    def _merge_dicts(
+        cls, src: Dict[str, Any], data: Dict[str, Any]
+    ) -> Dict[str, Any]:
+        """Find all the items in data that match a target_field on this class,
+        and merge them with the data found in `src` for target_field, using the
+        field's specified merge behavior. Matching items will be removed from
+        `data` (but _not_ `src`!).
+
+        Returns a dict with the merge results.
+
+        That means this method mutates its input! Any remaining values in data
+        were not merged.
+        """
+        result = {}
+
+        for fld, target_field in cls._get_fields():
+            if target_field not in data:
+                continue
+
+            data_attr = data.pop(target_field)
+            if target_field not in src:
+                result[target_field] = data_attr
+                continue
+
+            merge_behavior = MergeBehavior.from_field(fld)
+            self_attr = src[target_field]
+
+            result[target_field] = _merge_field_value(
+                merge_behavior=merge_behavior,
+                self_value=self_attr,
+                other_value=data_attr,
+            )
+        return result
+
+    def update_from(
+        self: T, data: Dict[str, Any], adapter_type: str, validate: bool = True
+    ) -> T:
+        """Given a dict of keys, update the current config from them, validate
+        it, and return a new config with the updated values
+        """
+        # sadly, this is a circular import
+        from dbt.adapters.factory import get_config_class_by_name
+        dct = self.to_dict(omit_none=False)
+
+        adapter_config_cls = get_config_class_by_name(adapter_type)
+
+        self_merged = self._merge_dicts(dct, data)
+        dct.update(self_merged)
+
+        adapter_merged = adapter_config_cls._merge_dicts(dct, data)
+        dct.update(adapter_merged)
+
+        # any remaining fields must be "clobber"
+        dct.update(data)
+
+        # any validation failures must have come from the update
+        if validate:
+            self.validate(dct)
+        return self.from_dict(dct)
+
+    def finalize_and_validate(self: T) -> T:
+        dct = self.to_dict(omit_none=False)
+        self.validate(dct)
+        return self.from_dict(dct)
+
+    def replace(self, **kwargs):
+        dct = self.to_dict(omit_none=True)
+
+        mapping = self.field_mapping()
+        for key, value in kwargs.items():
+            new_key = mapping.get(key, key)
+            dct[new_key] = value
+        return self.from_dict(dct)
+
+
+@dataclass
+class SourceConfig(BaseConfig):
+    enabled: bool = True
+
+
+@dataclass
+class NodeAndTestConfig(BaseConfig):
+    enabled: bool = True
+    # these fields are included in serialized output, but are not part of
+    # config comparison (they are part of database_representation)
+    alias: Optional[str] = field(
+        default=None,
+        metadata=CompareBehavior.Exclude.meta(),
+    )
+    schema: Optional[str] = field(
+        default=None,
+        metadata=CompareBehavior.Exclude.meta(),
+    )
+    database: Optional[str] = field(
+        default=None,
+        metadata=CompareBehavior.Exclude.meta(),
+    )
+    tags: Union[List[str], str] = field(
+        default_factory=list_str,
+        metadata=metas(ShowBehavior.Hide,
+                       MergeBehavior.Append,
+                       CompareBehavior.Exclude),
+    )
+    meta: Dict[str, Any] = field(
+        default_factory=dict,
+        metadata=MergeBehavior.Update.meta(),
+    )
+
+
+@dataclass
+class NodeConfig(NodeAndTestConfig):
+    # Note: if any new fields are added with MergeBehavior, also update the
+    # 'mergebehavior' dictionary
+    materialized: str = 'view'
+    persist_docs: Dict[str, Any] = field(default_factory=dict)
+    post_hook: List[Hook] = field(
+        default_factory=list,
+        metadata=MergeBehavior.Append.meta(),
+    )
+    pre_hook: List[Hook] = field(
+        default_factory=list,
+        metadata=MergeBehavior.Append.meta(),
+    )
+    quoting: Dict[str, Any] = field(
+        default_factory=dict,
+        metadata=MergeBehavior.Update.meta(),
+    )
+    # This is actually only used by seeds. Should it be available to others?
+    # That would be a breaking change!
+    column_types: Dict[str, Any] = field(
+        default_factory=dict,
+        metadata=MergeBehavior.Update.meta(),
+    )
+    full_refresh: Optional[bool] = None
+    on_schema_change: Optional[str] = 'ignore'
+
+    @classmethod
+    def __pre_deserialize__(cls, data):
+        data = super().__pre_deserialize__(data)
+        field_map = {'post-hook': 'post_hook', 'pre-hook': 'pre_hook'}
+        # create a new dict because otherwise it gets overwritten in
+        # tests
+        new_dict = {}
+        for key in data:
+            new_dict[key] = data[key]
+        data = new_dict
+        for key in hooks.ModelHookType:
+            if key in data:
+                data[key] = [hooks.get_hook_dict(h) for h in data[key]]
+        for field_name in field_map:
+            if field_name in data:
+                new_name = field_map[field_name]
+                data[new_name] = data.pop(field_name)
+        return data
+
+    def __post_serialize__(self, dct):
+        dct = super().__post_serialize__(dct)
+        field_map = {'post_hook': 'post-hook', 'pre_hook': 'pre-hook'}
+        for field_name in field_map:
+            if field_name in dct:
+                dct[field_map[field_name]] = dct.pop(field_name)
+        return dct
+
+    # this is still used by jsonschema validation
+    @classmethod
+    def field_mapping(cls):
+        return {'post_hook': 'post-hook', 'pre_hook': 'pre-hook'}
+
+
+@dataclass
+class SeedConfig(NodeConfig):
+    materialized: str = 'seed'
+    quote_columns: Optional[bool] = None
+
+
+@dataclass
+class TestConfig(NodeAndTestConfig):
+    # this is repeated because of a different default
+    schema: Optional[str] = field(
+        default='dbt_test__audit',
+        metadata=CompareBehavior.Exclude.meta(),
+    )
+    materialized: str = 'test'
+    severity: Severity = Severity('ERROR')
+    store_failures: Optional[bool] = None
+    where: Optional[str] = None
+    limit: Optional[int] = None
+    fail_calc: str = 'count(*)'
+    warn_if: str = '!= 0'
+    error_if: str = '!= 0'
+
+    @classmethod
+    def same_contents(
+        cls, unrendered: Dict[str, Any], other: Dict[str, Any]
+    ) -> bool:
+        """This is like __eq__, except it explicitly checks certain fields."""
+        modifiers = [
+            'severity',
+            'where',
+            'limit',
+            'fail_calc',
+            'warn_if',
+            'error_if',
+            'store_failures'
+        ]
+
+        seen = set()
+        for _, target_name in cls._get_fields():
+            key = target_name
+            seen.add(key)
+            if key in modifiers:
+                if not cls.compare_key(unrendered, other, key):
+                    return False
+        return True
+
+
+@dataclass
+class EmptySnapshotConfig(NodeConfig):
+    materialized: str = 'snapshot'
+
+
+@dataclass
+class SnapshotConfig(EmptySnapshotConfig):
+    strategy: Optional[str] = None
+    unique_key: Optional[str] = None
+    target_schema: Optional[str] = None
+    target_database: Optional[str] = None
+    updated_at: Optional[str] = None
+    check_cols: Optional[Union[str, List[str]]] = None
+
+    @classmethod
+    def validate(cls, data):
+        super().validate(data)
+        if not data.get('strategy') or not data.get('unique_key') or not \
+                data.get('target_schema'):
+            raise ValidationError(
+                "Snapshots must be configured with a 'strategy', 'unique_key', "
+                "and 'target_schema'.")
+        if data.get('strategy') == 'check':
+            if not data.get('check_cols'):
+                raise ValidationError(
+                    "A snapshot configured with the check strategy must "
+                    "specify a check_cols configuration.")
+            if (isinstance(data['check_cols'], str) and
+                    data['check_cols'] != 'all'):
+                raise ValidationError(
+                    f"Invalid value for 'check_cols': {data['check_cols']}. "
+                    "Expected 'all' or a list of strings.")
+
+        elif data.get('strategy') == 'timestamp':
+            if not data.get('updated_at'):
+                raise ValidationError(
+                    "A snapshot configured with the timestamp strategy "
+                    "must specify an updated_at configuration.")
+            if data.get('check_cols'):
+                raise ValidationError(
+                    "A 'timestamp' snapshot should not have 'check_cols'")
+        # If the strategy is not 'check' or 'timestamp' it's a custom strategy,
+        # formerly supported with GenericSnapshotConfig
+
+    def finalize_and_validate(self):
+        data = self.to_dict(omit_none=True)
+        self.validate(data)
+        return self.from_dict(data)
+
+
+RESOURCE_TYPES: Dict[NodeType, Type[BaseConfig]] = {
+    NodeType.Source: SourceConfig,
+    NodeType.Seed: SeedConfig,
+    NodeType.Test: TestConfig,
+    NodeType.Model: NodeConfig,
+    NodeType.Snapshot: SnapshotConfig,
+}
+
+
+# base resource types are like resource types, except nothing has mandatory
+# configs.
+BASE_RESOURCE_TYPES: Dict[NodeType, Type[BaseConfig]] = RESOURCE_TYPES.copy()
+BASE_RESOURCE_TYPES.update({
+    NodeType.Snapshot: EmptySnapshotConfig
+})
+
+
+def get_config_for(resource_type: NodeType, base=False) -> Type[BaseConfig]:
+    if base:
+        lookup = BASE_RESOURCE_TYPES
+    else:
+        lookup = RESOURCE_TYPES
+    return lookup.get(resource_type, NodeConfig)

core/dbt/contracts/graph/parsed.py

@@ -1,141 +1,80 @@
 import os
-from dataclasses import dataclass, field, Field
+import time
+from dataclasses import dataclass, field
+from mashumaro.types import SerializableType
+from pathlib import Path
 from typing import (
     Optional,
     Union,
     List,
     Dict,
     Any,
-    Type,
+    Sequence,
     Tuple,
-    NewType,
-    MutableMapping,
+    Iterator,
+    TypeVar,
 )
 
-from hologram import JsonSchemaMixin
-from hologram.helpers import (
-    StrEnum, register_pattern
+from dbt.dataclass_schema import (
+    dbtClassMixin, ExtensibleDbtClassMixin
 )
 
 from dbt.clients.system import write_file
-import dbt.flags
+from dbt.contracts.files import FileHash, MAXIMUM_SEED_SIZE_NAME
 from dbt.contracts.graph.unparsed import (
-    UnparsedNode, UnparsedMacro, UnparsedDocumentationFile, Quoting, Docs,
+    UnparsedNode, UnparsedDocumentation, Quoting, Docs,
     UnparsedBaseNode, FreshnessThreshold, ExternalTable,
-    AdditionalPropertiesAllowed, HasYamlMetadata, MacroArgument
+    HasYamlMetadata, MacroArgument, UnparsedSourceDefinition,
+    UnparsedSourceTableDefinition, UnparsedColumn, TestDef,
+    ExposureOwner, ExposureType, MaturityType
 )
-from dbt.contracts.util import Replaceable, list_str
+from dbt.contracts.util import Replaceable, AdditionalPropertiesMixin
+from dbt.exceptions import warn_or_error
 from dbt.logger import GLOBAL_LOGGER as logger  # noqa
+from dbt import flags
 from dbt.node_types import NodeType
 
 
-class SnapshotStrategy(StrEnum):
-    Timestamp = 'timestamp'
-    Check = 'check'
-
-
-class All(StrEnum):
-    All = 'all'
+from .model_config import (
+    NodeConfig,
+    SeedConfig,
+    TestConfig,
+    SourceConfig,
+    EmptySnapshotConfig,
+    SnapshotConfig,
+)
 
 
 @dataclass
-class Hook(JsonSchemaMixin, Replaceable):
-    sql: str
-    transaction: bool = True
-    index: Optional[int] = None
-
-
-def insensitive_patterns(*patterns: str):
-    lowercased = []
-    for pattern in patterns:
-        lowercased.append(
-            ''.join('[{}{}]'.format(s.upper(), s.lower()) for s in pattern)
-        )
-    return '^({})$'.format('|'.join(lowercased))
-
-
-Severity = NewType('Severity', str)
-register_pattern(Severity, insensitive_patterns('warn', 'error'))
-
-
-@dataclass
-class NodeConfig(
-    AdditionalPropertiesAllowed, Replaceable, MutableMapping[str, Any]
+class ColumnInfo(
+    AdditionalPropertiesMixin,
+    ExtensibleDbtClassMixin,
+    Replaceable
 ):
-    enabled: bool = True
-    materialized: str = 'view'
-    persist_docs: Dict[str, Any] = field(default_factory=dict)
-    post_hook: List[Hook] = field(default_factory=list)
-    pre_hook: List[Hook] = field(default_factory=list)
-    vars: Dict[str, Any] = field(default_factory=dict)
-    quoting: Dict[str, Any] = field(default_factory=dict)
-    column_types: Dict[str, Any] = field(default_factory=dict)
-    tags: Union[List[str], str] = field(default_factory=list_str)
-
-    @classmethod
-    def field_mapping(cls):
-        return {'post_hook': 'post-hook', 'pre_hook': 'pre-hook'}
-
-    # Implement MutableMapping so this config will behave as some macros expect
-    # during parsing (notably, syntax like `{{ node.config['schema'] }}`)
-
-    def __getitem__(self, key):
-        """Handle parse-time use of `config` as a dictionary, making the extra
-        values available during parsing.
-        """
-        if hasattr(self, key):
-            return getattr(self, key)
-        else:
-            return self._extra[key]
-
-    def __setitem__(self, key, value):
-        if hasattr(self, key):
-            setattr(self, key, value)
-        else:
-            self._extra[key] = value
-
-    def __delitem__(self, key):
-        if hasattr(self, key):
-            msg = (
-                'Error, tried to delete config key "{}": Cannot delete '
-                'built-in keys'
-            ).format(key)
-            raise dbt.exceptions.CompilationException(msg)
-        else:
-            del self._extra[key]
-
-    def __iter__(self):
-        for fld, _ in self._get_fields():
-            yield fld.name
-
-        for key in self._extra:
-            yield key
-
-    def __len__(self):
-        return len(self._get_fields()) + len(self._extra)
-
-
-@dataclass
-class ColumnInfo(JsonSchemaMixin, Replaceable):
     name: str
     description: str = ''
     meta: Dict[str, Any] = field(default_factory=dict)
     data_type: Optional[str] = None
+    quote: Optional[bool] = None
     tags: List[str] = field(default_factory=list)
+    _extra: Dict[str, Any] = field(default_factory=dict)
 
 
 @dataclass
-class HasFqn(JsonSchemaMixin, Replaceable):
+class HasFqn(dbtClassMixin, Replaceable):
     fqn: List[str]
 
+    def same_fqn(self, other: 'HasFqn') -> bool:
+        return self.fqn == other.fqn
+
 
 @dataclass
-class HasUniqueID(JsonSchemaMixin, Replaceable):
+class HasUniqueID(dbtClassMixin, Replaceable):
     unique_id: str
 
 
 @dataclass
-class MacroDependsOn(JsonSchemaMixin, Replaceable):
+class MacroDependsOn(dbtClassMixin, Replaceable):
     macros: List[str] = field(default_factory=list)
 
     # 'in' on lists is O(n) so this is O(n^2) for # of macros
@@ -154,12 +93,22 @@ class DependsOn(MacroDependsOn):
 
 
 @dataclass
-class HasRelationMetadata(JsonSchemaMixin, Replaceable):
-    database: str
+class HasRelationMetadata(dbtClassMixin, Replaceable):
+    database: Optional[str]
     schema: str
 
+    # Can't set database to None like it ought to be
+    # because it messes up the subclasses and default parameters
+    # so hack it here
+    @classmethod
+    def __pre_deserialize__(cls, data):
+        data = super().__pre_deserialize__(data)
+        if 'database' not in data:
+            data['database'] = None
+        return data
 
-class ParsedNodeMixins(JsonSchemaMixin):
+
+class ParsedNodeMixins(dbtClassMixin):
     resource_type: NodeType
     depends_on: DependsOn
     config: NodeConfig
@@ -168,6 +117,21 @@ class ParsedNodeMixins(JsonSchemaMixin):
     def is_refable(self):
         return self.resource_type in NodeType.refable()
 
+    @property
+    def should_store_failures(self):
+        return self.resource_type == NodeType.Test and (
+            self.config.store_failures if self.config.store_failures is not None
+            else flags.STORE_FAILURES
+        )
+
+    # will this node map to an object in the database?
+    @property
+    def is_relational(self):
+        return (
+            self.resource_type in NodeType.refable() or
+            self.should_store_failures
+        )
+
     @property
     def is_ephemeral(self):
         return self.config.materialized == 'ephemeral'
@@ -184,21 +148,25 @@ class ParsedNodeMixins(JsonSchemaMixin):
         """Given a ParsedNodePatch, add the new information to the node."""
         # explicitly pick out the parts to update so we don't inadvertently
         # step on the model name or anything
-        self.patch_path: Optional[str] = patch.original_file_path
+        # Note: config should already be updated
+        self.patch_path: Optional[str] = patch.file_id
+        # update created_at so process_docs will run in partial parsing
+        self.created_at = int(time.time())
         self.description = patch.description
         self.columns = patch.columns
         self.meta = patch.meta
         self.docs = patch.docs
-        if dbt.flags.STRICT_MODE:
-            assert isinstance(self, JsonSchemaMixin)
-            self.to_dict(validate=True)
+        if flags.STRICT_MODE:
+            # It seems odd that an instance can be invalid
+            # Maybe there should be validation or restrictions
+            # elsewhere?
+            assert isinstance(self, dbtClassMixin)
+            dct = self.to_dict(omit_none=False)
+            self.validate(dct)
 
     def get_materialization(self):
         return self.config.materialized
 
-    def local_vars(self):
-        return self.config.vars
-
 
 @dataclass
 class ParsedNodeMandatory(
@@ -209,6 +177,8 @@ class ParsedNodeMandatory(
     Replaceable
 ):
     alias: str
+    checksum: FileHash
+    config: NodeConfig = field(default_factory=NodeConfig)
 
     @property
     def identifier(self):
@@ -217,7 +187,6 @@ class ParsedNodeMandatory(
 
 @dataclass
 class ParsedNodeDefaults(ParsedNodeMandatory):
-    config: NodeConfig = field(default_factory=NodeConfig)
     tags: List[str] = field(default_factory=list)
     refs: List[List[str]] = field(default_factory=list)
     sources: List[List[Any]] = field(default_factory=list)
@@ -227,20 +196,138 @@ class ParsedNodeDefaults(ParsedNodeMandatory):
     meta: Dict[str, Any] = field(default_factory=dict)
     docs: Docs = field(default_factory=Docs)
     patch_path: Optional[str] = None
+    compiled_path: Optional[str] = None
     build_path: Optional[str] = None
+    deferred: bool = False
+    unrendered_config: Dict[str, Any] = field(default_factory=dict)
+    created_at: int = field(default_factory=lambda: int(time.time()))
+    config_call_dict: Dict[str, Any] = field(default_factory=dict)
 
     def write_node(self, target_path: str, subdirectory: str, payload: str):
+        if (os.path.basename(self.path) ==
+                os.path.basename(self.original_file_path)):
+            # One-to-one relationship of nodes to files.
+            path = self.original_file_path
+        else:
+            #  Many-to-one relationship of nodes to files.
+            path = os.path.join(self.original_file_path, self.path)
         full_path = os.path.join(
-            target_path, subdirectory, self.package_name, self.path
+            target_path, subdirectory, self.package_name, path
         )
 
         write_file(full_path, payload)
         return full_path
 
 
+T = TypeVar('T', bound='ParsedNode')
+
+
 @dataclass
-class ParsedNode(ParsedNodeDefaults, ParsedNodeMixins):
-    pass
+class ParsedNode(ParsedNodeDefaults, ParsedNodeMixins, SerializableType):
+
+    def _serialize(self):
+        return self.to_dict()
+
+    def __post_serialize__(self, dct):
+        if 'config_call_dict' in dct:
+            del dct['config_call_dict']
+        return dct
+
+    @classmethod
+    def _deserialize(cls, dct: Dict[str, int]):
+        # The serialized ParsedNodes do not differ from each other
+        # in fields that would allow 'from_dict' to distinguis
+        # between them.
+        resource_type = dct['resource_type']
+        if resource_type == 'model':
+            return ParsedModelNode.from_dict(dct)
+        elif resource_type == 'analysis':
+            return ParsedAnalysisNode.from_dict(dct)
+        elif resource_type == 'seed':
+            return ParsedSeedNode.from_dict(dct)
+        elif resource_type == 'rpc':
+            return ParsedRPCNode.from_dict(dct)
+        elif resource_type == 'test':
+            if 'test_metadata' in dct:
+                return ParsedSchemaTestNode.from_dict(dct)
+            else:
+                return ParsedDataTestNode.from_dict(dct)
+        elif resource_type == 'operation':
+            return ParsedHookNode.from_dict(dct)
+        elif resource_type == 'seed':
+            return ParsedSeedNode.from_dict(dct)
+        elif resource_type == 'snapshot':
+            return ParsedSnapshotNode.from_dict(dct)
+        else:
+            return cls.from_dict(dct)
+
+    def _persist_column_docs(self) -> bool:
+        if hasattr(self.config, 'persist_docs'):
+            assert isinstance(self.config, NodeConfig)
+            return bool(self.config.persist_docs.get('columns'))
+        return False
+
+    def _persist_relation_docs(self) -> bool:
+        if hasattr(self.config, 'persist_docs'):
+            assert isinstance(self.config, NodeConfig)
+            return bool(self.config.persist_docs.get('relation'))
+        return False
+
+    def same_body(self: T, other: T) -> bool:
+        return self.raw_sql == other.raw_sql
+
+    def same_persisted_description(self: T, other: T) -> bool:
+        # the check on configs will handle the case where we have different
+        # persist settings, so we only have to care about the cases where they
+        # are the same..
+        if self._persist_relation_docs():
+            if self.description != other.description:
+                return False
+
+        if self._persist_column_docs():
+            # assert other._persist_column_docs()
+            column_descriptions = {
+                k: v.description for k, v in self.columns.items()
+            }
+            other_column_descriptions = {
+                k: v.description for k, v in other.columns.items()
+            }
+            if column_descriptions != other_column_descriptions:
+                return False
+
+        return True
+
+    def same_database_representation(self, other: T) -> bool:
+        # compare the config representation, not the node's config value. This
+        # compares the configured value, rather than the ultimate value (so
+        # generate_*_name and unset values derived from the target are
+        # ignored)
+        keys = ('database', 'schema', 'alias')
+        for key in keys:
+            mine = self.unrendered_config.get(key)
+            others = other.unrendered_config.get(key)
+            if mine != others:
+                return False
+        return True
+
+    def same_config(self, old: T) -> bool:
+        return self.config.same_contents(
+            self.unrendered_config,
+            old.unrendered_config,
+        )
+
+    def same_contents(self: T, old: Optional[T]) -> bool:
+        if old is None:
+            return False
+
+        return (
+            self.same_body(old) and
+            self.same_config(old) and
+            self.same_persisted_description(old) and
+            self.same_fqn(old) and
+            self.same_database_representation(old) and
+            True
+        )
 
 
 @dataclass
@@ -266,138 +353,97 @@ class ParsedRPCNode(ParsedNode):
     resource_type: NodeType = field(metadata={'restrict': [NodeType.RPCCall]})
 
 
-class SeedConfig(NodeConfig):
-    quote_columns: Optional[bool] = None
+def same_seeds(first: ParsedNode, second: ParsedNode) -> bool:
+    # for seeds, we check the hashes. If the hashes are different types,
+    # no match. If the hashes are both the same 'path', log a warning and
+    # assume they are the same
+    # if the current checksum is a path, we want to log a warning.
+    result = first.checksum == second.checksum
+
+    if first.checksum.name == 'path':
+        msg: str
+        if second.checksum.name != 'path':
+            msg = (
+                f'Found a seed ({first.package_name}.{first.name}) '
+                f'>{MAXIMUM_SEED_SIZE_NAME} in size. The previous file was '
+                f'<={MAXIMUM_SEED_SIZE_NAME}, so it has changed'
+            )
+        elif result:
+            msg = (
+                f'Found a seed ({first.package_name}.{first.name}) '
+                f'>{MAXIMUM_SEED_SIZE_NAME} in size at the same path, dbt '
+                f'cannot tell if it has changed: assuming they are the same'
+            )
+        elif not result:
+            msg = (
+                f'Found a seed ({first.package_name}.{first.name}) '
+                f'>{MAXIMUM_SEED_SIZE_NAME} in size. The previous file was in '
+                f'a different location, assuming it has changed'
+            )
+        else:
+            msg = (
+                f'Found a seed ({first.package_name}.{first.name}) '
+                f'>{MAXIMUM_SEED_SIZE_NAME} in size. The previous file had a '
+                f'checksum type of {second.checksum.name}, so it has changed'
+            )
+        warn_or_error(msg, node=first)
+
+    return result
 
 
 @dataclass
 class ParsedSeedNode(ParsedNode):
+    # keep this in sync with CompiledSeedNode!
     resource_type: NodeType = field(metadata={'restrict': [NodeType.Seed]})
     config: SeedConfig = field(default_factory=SeedConfig)
-    seed_file_path: str = ''
-
-    def __post_init__(self):
-        if self.seed_file_path == '':
-            raise dbt.exceptions.InternalException(
-                'Seeds should always have a seed_file_path'
-            )
 
     @property
     def empty(self):
         """ Seeds are never empty"""
         return False
 
-
-@dataclass
-class TestConfig(NodeConfig):
-    severity: Severity = Severity('error')
+    def same_body(self: T, other: T) -> bool:
+        return same_seeds(self, other)
 
 
 @dataclass
-class TestMetadata(JsonSchemaMixin):
-    namespace: Optional[str]
+class TestMetadata(dbtClassMixin, Replaceable):
     name: str
-    kwargs: Dict[str, Any]
+    kwargs: Dict[str, Any] = field(default_factory=dict)
+    namespace: Optional[str] = None
 
 
 @dataclass
-class ParsedTestNode(ParsedNode):
+class HasTestMetadata(dbtClassMixin):
+    test_metadata: TestMetadata
+
+
+@dataclass
+class ParsedDataTestNode(ParsedNode):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Test]})
+    # Was not able to make mypy happy and keep the code working. We need to
+    # refactor the various configs.
+    config: TestConfig = field(default_factory=TestConfig)  # type: ignore
+
+
+@dataclass
+class ParsedSchemaTestNode(ParsedNode, HasTestMetadata):
+    # keep this in sync with CompiledSchemaTestNode!
     resource_type: NodeType = field(metadata={'restrict': [NodeType.Test]})
     column_name: Optional[str] = None
-    config: TestConfig = field(default_factory=TestConfig)
-    test_metadata: Optional[TestMetadata] = None
+    # Was not able to make mypy happy and keep the code working. We need to
+    # refactor the various configs.
+    config: TestConfig = field(default_factory=TestConfig)  # type: ignore
 
+    def same_contents(self, other) -> bool:
+        if other is None:
+            return False
 
-@dataclass(init=False)
-class _SnapshotConfig(NodeConfig):
-    unique_key: str = field(init=False, metadata=dict(init_required=True))
-    target_schema: str = field(init=False, metadata=dict(init_required=True))
-    target_database: Optional[str] = None
-
-    def __init__(
-        self,
-        unique_key: str,
-        target_schema: str,
-        target_database: Optional[str] = None,
-        **kwargs
-    ) -> None:
-        self.unique_key = unique_key
-        self.target_schema = target_schema
-        self.target_database = target_database
-        super().__init__(**kwargs)
-
-    # type hacks...
-    @classmethod
-    def _get_fields(cls) -> List[Tuple[Field, str]]:  # type: ignore
-        fields: List[Tuple[Field, str]] = []
-        for old_field, name in super()._get_fields():
-            new_field = old_field
-            # tell hologram we're really an initvar
-            if old_field.metadata and old_field.metadata.get('init_required'):
-                new_field = field(init=True, metadata=old_field.metadata)
-                new_field.name = old_field.name
-                new_field.type = old_field.type
-                new_field._field_type = old_field._field_type  # type: ignore
-            fields.append((new_field, name))
-        return fields
-
-
-@dataclass(init=False)
-class GenericSnapshotConfig(_SnapshotConfig):
-    strategy: str = field(init=False, metadata=dict(init_required=True))
-
-    def __init__(self, strategy: str, **kwargs) -> None:
-        self.strategy = strategy
-        super().__init__(**kwargs)
-
-
-@dataclass(init=False)
-class TimestampSnapshotConfig(_SnapshotConfig):
-    strategy: str = field(
-        init=False,
-        metadata=dict(
-            restrict=[str(SnapshotStrategy.Timestamp)],
-            init_required=True,
-        ),
-    )
-    updated_at: str = field(init=False, metadata=dict(init_required=True))
-
-    def __init__(
-        self, strategy: str, updated_at: str, **kwargs
-    ) -> None:
-        self.strategy = strategy
-        self.updated_at = updated_at
-        super().__init__(**kwargs)
-
-
-@dataclass(init=False)
-class CheckSnapshotConfig(_SnapshotConfig):
-    strategy: str = field(
-        init=False,
-        metadata=dict(
-            restrict=[str(SnapshotStrategy.Check)],
-            init_required=True,
-        ),
-    )
-    # TODO: is there a way to get this to accept tuples of strings? Adding
-    # `Tuple[str, ...]` to the list of types results in this:
-    # ['email'] is valid under each of {'type': 'array', 'items':
-    # {'type': 'string'}}, {'type': 'array', 'items': {'type': 'string'}}
-    # but without it, parsing gets upset about values like `('email',)`
-    # maybe hologram itself should support this behavior? It's not like tuples
-    # are meaningful in json
-    check_cols: Union[All, List[str]] = field(
-        init=False,
-        metadata=dict(init_required=True),
-    )
-
-    def __init__(
-        self, strategy: str, check_cols: Union[All, List[str]],
-        **kwargs
-    ) -> None:
-        self.strategy = strategy
-        self.check_cols = check_cols
-        super().__init__(**kwargs)
+        return (
+            self.same_config(other) and
+            self.same_fqn(other) and
+            True
+        )
 
 
 @dataclass
@@ -409,59 +455,13 @@ class IntermediateSnapshotNode(ParsedNode):
     # uses a regular node config, which the snapshot parser will then convert
     # into a full ParsedSnapshotNode after rendering.
     resource_type: NodeType = field(metadata={'restrict': [NodeType.Snapshot]})
-
-
-def _create_if_else_chain(
-    key: str,
-    criteria: List[Tuple[str, Type[JsonSchemaMixin]]],
-    default: Type[JsonSchemaMixin]
-) -> Dict[str, Any]:
-    """Mutate a given schema key that contains a 'oneOf' to instead be an
-    'if-then-else' chain. This results is much better/more consistent errors
-    from jsonschema.
-    """
-    schema: Dict[str, Any] = {}
-    result: Dict[str, Any] = {}
-    criteria = criteria[:]
-    while criteria:
-        if_clause, then_clause = criteria.pop()
-        schema['if'] = {'properties': {
-            key: {'enum': [if_clause]}
-        }}
-        schema['then'] = then_clause.json_schema()
-        schema['else'] = {}
-        schema = schema['else']
-    schema.update(default.json_schema())
-    return result
+    config: EmptySnapshotConfig = field(default_factory=EmptySnapshotConfig)
 
 
 @dataclass
 class ParsedSnapshotNode(ParsedNode):
     resource_type: NodeType = field(metadata={'restrict': [NodeType.Snapshot]})
-    config: Union[
-        CheckSnapshotConfig,
-        TimestampSnapshotConfig,
-        GenericSnapshotConfig,
-    ]
-
-    @classmethod
-    def json_schema(cls, embeddable: bool = False) -> Dict[str, Any]:
-        schema = super().json_schema(embeddable)
-
-        # mess with config
-        configs: List[Tuple[str, Type[JsonSchemaMixin]]] = [
-            (str(SnapshotStrategy.Check), CheckSnapshotConfig),
-            (str(SnapshotStrategy.Timestamp), TimestampSnapshotConfig),
-        ]
-
-        if embeddable:
-            dest = schema[cls.__name__]['properties']
-        else:
-            dest = schema['properties']
-        dest['config'] = _create_if_else_chain(
-            'strategy', configs, GenericSnapshotConfig
-        )
-        return schema
+    config: SnapshotConfig
 
 
 @dataclass
@@ -470,6 +470,7 @@ class ParsedPatch(HasYamlMetadata, Replaceable):
     description: str
     meta: Dict[str, Any]
     docs: Docs
+    config: Dict[str, Any]
 
 
 # The parsed node update is only the 'patch', not the test. The test became a
@@ -486,7 +487,7 @@ class ParsedMacroPatch(ParsedPatch):
 
 
 @dataclass
-class ParsedMacro(UnparsedMacro, HasUniqueID):
+class ParsedMacro(UnparsedBaseNode, HasUniqueID):
     name: str
     macro_sql: str
     resource_type: NodeType = field(metadata={'restrict': [NodeType.Macro]})
@@ -496,24 +497,34 @@ class ParsedMacro(UnparsedMacro, HasUniqueID):
     depends_on: MacroDependsOn = field(default_factory=MacroDependsOn)
     description: str = ''
     meta: Dict[str, Any] = field(default_factory=dict)
+    docs: Docs = field(default_factory=Docs)
     patch_path: Optional[str] = None
     arguments: List[MacroArgument] = field(default_factory=list)
-
-    def local_vars(self):
-        return {}
+    created_at: int = field(default_factory=lambda: int(time.time()))
 
     def patch(self, patch: ParsedMacroPatch):
-        self.patch_path: Optional[str] = patch.original_file_path
+        self.patch_path: Optional[str] = patch.file_id
         self.description = patch.description
+        self.created_at = int(time.time())
         self.meta = patch.meta
+        self.docs = patch.docs
         self.arguments = patch.arguments
-        if dbt.flags.STRICT_MODE:
-            assert isinstance(self, JsonSchemaMixin)
-            self.to_dict(validate=True)
+        if flags.STRICT_MODE:
+            # What does this actually validate?
+            assert isinstance(self, dbtClassMixin)
+            dct = self.to_dict(omit_none=False)
+            self.validate(dct)
+
+    def same_contents(self, other: Optional['ParsedMacro']) -> bool:
+        if other is None:
+            return False
+        # the only thing that makes one macro different from another with the
+        # same name/package is its content
+        return self.macro_sql == other.macro_sql
 
 
 @dataclass
-class ParsedDocumentation(UnparsedDocumentationFile, HasUniqueID):
+class ParsedDocumentation(UnparsedDocumentation, HasUniqueID):
     name: str
     block_contents: str
 
@@ -521,13 +532,81 @@ class ParsedDocumentation(UnparsedDocumentationFile, HasUniqueID):
     def search_name(self):
         return self.name
 
+    def same_contents(self, other: Optional['ParsedDocumentation']) -> bool:
+        if other is None:
+            return False
+        # the only thing that makes one doc different from another with the
+        # same name/package is its content
+        return self.block_contents == other.block_contents
+
+
+def normalize_test(testdef: TestDef) -> Dict[str, Any]:
+    if isinstance(testdef, str):
+        return {testdef: {}}
+    else:
+        return testdef
+
+
+@dataclass
+class UnpatchedSourceDefinition(UnparsedBaseNode, HasUniqueID, HasFqn):
+    source: UnparsedSourceDefinition
+    table: UnparsedSourceTableDefinition
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Source]})
+    patch_path: Optional[Path] = None
+
+    def get_full_source_name(self):
+        return f'{self.source.name}_{self.table.name}'
+
+    def get_source_representation(self):
+        return f'source("{self.source.name}", "{self.table.name}")'
+
+    @property
+    def name(self) -> str:
+        return self.get_full_source_name()
+
+    @property
+    def quote_columns(self) -> Optional[bool]:
+        result = None
+        if self.source.quoting.column is not None:
+            result = self.source.quoting.column
+        if self.table.quoting.column is not None:
+            result = self.table.quoting.column
+        return result
+
+    @property
+    def columns(self) -> Sequence[UnparsedColumn]:
+        if self.table.columns is None:
+            return []
+        else:
+            return self.table.columns
+
+    def get_tests(
+        self
+    ) -> Iterator[Tuple[Dict[str, Any], Optional[UnparsedColumn]]]:
+        for test in self.tests:
+            yield normalize_test(test), None
+
+        for column in self.columns:
+            if column.tests is not None:
+                for test in column.tests:
+                    yield normalize_test(test), column
+
+    @property
+    def tests(self) -> List[TestDef]:
+        if self.table.tests is None:
+            return []
+        else:
+            return self.table.tests
+
 
 @dataclass
 class ParsedSourceDefinition(
-        UnparsedBaseNode,
-        HasUniqueID,
-        HasRelationMetadata,
-        HasFqn):
+    UnparsedBaseNode,
+    HasUniqueID,
+    HasRelationMetadata,
+    HasFqn,
+
+):
     name: str
     source_name: str
     source_description: str
@@ -543,6 +622,69 @@ class ParsedSourceDefinition(
     meta: Dict[str, Any] = field(default_factory=dict)
     source_meta: Dict[str, Any] = field(default_factory=dict)
     tags: List[str] = field(default_factory=list)
+    config: SourceConfig = field(default_factory=SourceConfig)
+    patch_path: Optional[Path] = None
+    unrendered_config: Dict[str, Any] = field(default_factory=dict)
+    relation_name: Optional[str] = None
+    created_at: int = field(default_factory=lambda: int(time.time()))
+
+    def same_database_representation(
+        self, other: 'ParsedSourceDefinition'
+    ) -> bool:
+        return (
+            self.database == other.database and
+            self.schema == other.schema and
+            self.identifier == other.identifier and
+            True
+        )
+
+    def same_quoting(self, other: 'ParsedSourceDefinition') -> bool:
+        return self.quoting == other.quoting
+
+    def same_freshness(self, other: 'ParsedSourceDefinition') -> bool:
+        return (
+            self.freshness == other.freshness and
+            self.loaded_at_field == other.loaded_at_field and
+            True
+        )
+
+    def same_external(self, other: 'ParsedSourceDefinition') -> bool:
+        return self.external == other.external
+
+    def same_config(self, old: 'ParsedSourceDefinition') -> bool:
+        return self.config.same_contents(
+            self.unrendered_config,
+            old.unrendered_config,
+        )
+
+    def same_contents(self, old: Optional['ParsedSourceDefinition']) -> bool:
+        # existing when it didn't before is a change!
+        if old is None:
+            return True
+
+        # config changes are changes (because the only config is "enabled", and
+        # enabling a source is a change!)
+        # changing the database/schema/identifier is a change
+        # messing around with external stuff is a change (uh, right?)
+        # quoting changes are changes
+        # freshness changes are changes, I guess
+        # metadata/tags changes are not "changes"
+        # patching/description changes are not "changes"
+        return (
+            self.same_database_representation(old) and
+            self.same_fqn(old) and
+            self.same_config(old) and
+            self.same_quoting(old) and
+            self.same_freshness(old) and
+            self.same_external(old) and
+            True
+        )
+
+    def get_full_source_name(self):
+        return f'{self.source_name}_{self.name}'
+
+    def get_source_representation(self):
+        return f'source("{self.source.name}", "{self.table.name}")'
 
     @property
     def is_refable(self):
@@ -560,6 +702,10 @@ class ParsedSourceDefinition(
     def depends_on_nodes(self):
         return []
 
+    @property
+    def depends_on(self):
+        return DependsOn(macros=[], nodes=[])
+
     @property
     def refs(self):
         return []
@@ -577,20 +723,82 @@ class ParsedSourceDefinition(
         return f'{self.source_name}.{self.name}'
 
 
-ParsedResource = Union[
-    ParsedMacro, ParsedNode, ParsedDocumentation, ParsedSourceDefinition
+@dataclass
+class ParsedExposure(UnparsedBaseNode, HasUniqueID, HasFqn):
+    name: str
+    type: ExposureType
+    owner: ExposureOwner
+    resource_type: NodeType = NodeType.Exposure
+    description: str = ''
+    maturity: Optional[MaturityType] = None
+    meta: Dict[str, Any] = field(default_factory=dict)
+    tags: List[str] = field(default_factory=list)
+    url: Optional[str] = None
+    depends_on: DependsOn = field(default_factory=DependsOn)
+    refs: List[List[str]] = field(default_factory=list)
+    sources: List[List[str]] = field(default_factory=list)
+    created_at: int = field(default_factory=lambda: int(time.time()))
+
+    @property
+    def depends_on_nodes(self):
+        return self.depends_on.nodes
+
+    @property
+    def search_name(self):
+        return self.name
+
+    def same_depends_on(self, old: 'ParsedExposure') -> bool:
+        return set(self.depends_on.nodes) == set(old.depends_on.nodes)
+
+    def same_description(self, old: 'ParsedExposure') -> bool:
+        return self.description == old.description
+
+    def same_maturity(self, old: 'ParsedExposure') -> bool:
+        return self.maturity == old.maturity
+
+    def same_owner(self, old: 'ParsedExposure') -> bool:
+        return self.owner == old.owner
+
+    def same_exposure_type(self, old: 'ParsedExposure') -> bool:
+        return self.type == old.type
+
+    def same_url(self, old: 'ParsedExposure') -> bool:
+        return self.url == old.url
+
+    def same_contents(self, old: Optional['ParsedExposure']) -> bool:
+        # existing when it didn't before is a change!
+        # metadata/tags changes are not "changes"
+        if old is None:
+            return True
+
+        return (
+            self.same_fqn(old) and
+            self.same_exposure_type(old) and
+            self.same_owner(old) and
+            self.same_maturity(old) and
+            self.same_url(old) and
+            self.same_description(old) and
+            self.same_depends_on(old) and
+            True
+        )
+
+
+ManifestNodes = Union[
+    ParsedAnalysisNode,
+    ParsedDataTestNode,
+    ParsedHookNode,
+    ParsedModelNode,
+    ParsedRPCNode,
+    ParsedSchemaTestNode,
+    ParsedSeedNode,
+    ParsedSnapshotNode,
 ]
 
 
-PARSED_TYPES: Dict[NodeType, Type[ParsedResource]] = {
-    NodeType.Analysis: ParsedAnalysisNode,
-    NodeType.Documentation: ParsedDocumentation,
-    NodeType.Macro: ParsedMacro,
-    NodeType.Model: ParsedModelNode,
-    NodeType.Operation: ParsedHookNode,
-    NodeType.RPCCall: ParsedRPCNode,
-    NodeType.Seed: ParsedSeedNode,
-    NodeType.Snapshot: ParsedSnapshotNode,
-    NodeType.Source: ParsedSourceDefinition,
-    NodeType.Test: ParsedTestNode,
-}
+ParsedResource = Union[
+    ParsedDocumentation,
+    ParsedMacro,
+    ParsedNode,
+    ParsedExposure,
+    ParsedSourceDefinition,
+]

core/dbt/contracts/graph/unparsed.py

@@ -1,22 +1,34 @@
 from dbt.node_types import NodeType
-from dbt.contracts.util import Replaceable, Mergeable
+from dbt.contracts.util import (
+    AdditionalPropertiesMixin,
+    Mergeable,
+    Replaceable,
+)
+# trigger the PathEncoder
+import dbt.helper_types  # noqa:F401
 from dbt.exceptions import CompilationException
 
-from hologram import JsonSchemaMixin
-from hologram.helpers import StrEnum, ExtensibleJsonSchemaMixin
+from dbt.dataclass_schema import (
+    dbtClassMixin, StrEnum, ExtensibleDbtClassMixin
+)
 
 from dataclasses import dataclass, field
 from datetime import timedelta
+from pathlib import Path
 from typing import Optional, List, Union, Dict, Any, Sequence
 
 
 @dataclass
-class UnparsedBaseNode(JsonSchemaMixin, Replaceable):
+class UnparsedBaseNode(dbtClassMixin, Replaceable):
     package_name: str
     root_path: str
     path: str
     original_file_path: str
 
+    @property
+    def file_id(self):
+        return f'{self.package_name}://{self.original_file_path}'
+
 
 @dataclass
 class HasSQL:
@@ -59,17 +71,19 @@ class UnparsedRunHook(UnparsedNode):
 
 
 @dataclass
-class Docs(JsonSchemaMixin, Replaceable):
+class Docs(dbtClassMixin, Replaceable):
     show: bool = True
 
 
 @dataclass
-class HasDocs(JsonSchemaMixin, Replaceable):
+class HasDocs(AdditionalPropertiesMixin, ExtensibleDbtClassMixin,
+              Replaceable):
     name: str
     description: str = ''
     meta: Dict[str, Any] = field(default_factory=dict)
     data_type: Optional[str] = None
     docs: Docs = field(default_factory=Docs)
+    _extra: Dict[str, Any] = field(default_factory=dict)
 
 
 TestDef = Union[Dict[str, Any], str]
@@ -91,7 +105,7 @@ class UnparsedColumn(HasTests):
 
 
 @dataclass
-class HasColumnDocs(JsonSchemaMixin, Replaceable):
+class HasColumnDocs(dbtClassMixin, Replaceable):
     columns: Sequence[HasDocs] = field(default_factory=list)
 
 
@@ -101,31 +115,40 @@ class HasColumnTests(HasColumnDocs):
 
 
 @dataclass
-class HasYamlMetadata(JsonSchemaMixin):
+class HasYamlMetadata(dbtClassMixin):
     original_file_path: str
     yaml_key: str
     package_name: str
 
+    @property
+    def file_id(self):
+        return f'{self.package_name}://{self.original_file_path}'
+
 
 @dataclass
-class UnparsedAnalysisUpdate(HasColumnDocs, HasDocs, HasYamlMetadata):
+class HasConfig():
+    config: Dict[str, Any] = field(default_factory=dict)
+
+
+@dataclass
+class UnparsedAnalysisUpdate(HasConfig, HasColumnDocs, HasDocs, HasYamlMetadata):
     pass
 
 
 @dataclass
-class UnparsedNodeUpdate(HasColumnTests, HasTests, HasYamlMetadata):
+class UnparsedNodeUpdate(HasConfig, HasColumnTests, HasTests, HasYamlMetadata):
     quote_columns: Optional[bool] = None
 
 
 @dataclass
-class MacroArgument(JsonSchemaMixin):
+class MacroArgument(dbtClassMixin):
     name: str
     type: Optional[str] = None
     description: str = ''
 
 
 @dataclass
-class UnparsedMacroUpdate(HasDocs, HasYamlMetadata):
+class UnparsedMacroUpdate(HasConfig, HasDocs, HasYamlMetadata):
     arguments: List[MacroArgument] = field(default_factory=list)
 
 
@@ -139,7 +162,7 @@ class TimePeriod(StrEnum):
 
 
 @dataclass
-class Time(JsonSchemaMixin, Replaceable):
+class Time(dbtClassMixin, Replaceable):
     count: int
     period: TimePeriod
 
@@ -149,19 +172,14 @@ class Time(JsonSchemaMixin, Replaceable):
         return actual_age > difference
 
 
-class FreshnessStatus(StrEnum):
-    Pass = 'pass'
-    Warn = 'warn'
-    Error = 'error'
-
-
 @dataclass
-class FreshnessThreshold(JsonSchemaMixin, Mergeable):
+class FreshnessThreshold(dbtClassMixin, Mergeable):
     warn_after: Optional[Time] = None
     error_after: Optional[Time] = None
     filter: Optional[str] = None
 
-    def status(self, age: float) -> FreshnessStatus:
+    def status(self, age: float) -> "dbt.contracts.results.FreshnessStatus":
+        from dbt.contracts.results import FreshnessStatus
         if self.error_after and self.error_after.exceeded(age):
             return FreshnessStatus.Error
         elif self.warn_after and self.warn_after.exceeded(age):
@@ -174,32 +192,12 @@ class FreshnessThreshold(JsonSchemaMixin, Mergeable):
 
 
 @dataclass
-class AdditionalPropertiesAllowed(ExtensibleJsonSchemaMixin):
+class AdditionalPropertiesAllowed(
+    AdditionalPropertiesMixin,
+    ExtensibleDbtClassMixin
+):
     _extra: Dict[str, Any] = field(default_factory=dict)
 
-    @property
-    def extra(self):
-        return self._extra
-
-    @classmethod
-    def from_dict(cls, data, validate=True):
-        self = super().from_dict(data=data, validate=validate)
-        keys = self.to_dict(validate=False, omit_none=False)
-        for key, value in data.items():
-            if key not in keys:
-                self._extra[key] = value
-        return self
-
-    def to_dict(self, omit_none=True, validate=False):
-        data = super().to_dict(omit_none=omit_none, validate=validate)
-        data.update(self._extra)
-        return data
-
-    def replace(self, **kwargs):
-        dct = self.to_dict(omit_none=False, validate=False)
-        dct.update(kwargs)
-        return self.from_dict(dct)
-
 
 @dataclass
 class ExternalPartition(AdditionalPropertiesAllowed, Replaceable):
@@ -228,7 +226,7 @@ class ExternalTable(AdditionalPropertiesAllowed, Mergeable):
 
 
 @dataclass
-class Quoting(JsonSchemaMixin, Mergeable):
+class Quoting(dbtClassMixin, Mergeable):
     database: Optional[bool] = None
     schema: Optional[bool] = None
     identifier: Optional[bool] = None
@@ -243,14 +241,18 @@ class UnparsedSourceTableDefinition(HasColumnTests, HasTests):
     freshness: Optional[FreshnessThreshold] = field(
         default_factory=FreshnessThreshold
     )
-    external: Optional[ExternalTable] = field(
-        default_factory=ExternalTable
-    )
+    external: Optional[ExternalTable] = None
     tags: List[str] = field(default_factory=list)
 
+    def __post_serialize__(self, dct):
+        dct = super().__post_serialize__(dct)
+        if 'freshness' not in dct and self.freshness is None:
+            dct['freshness'] = None
+        return dct
+
 
 @dataclass
-class UnparsedSourceDefinition(JsonSchemaMixin, Replaceable):
+class UnparsedSourceDefinition(dbtClassMixin, Replaceable):
     name: str
     description: str = ''
     meta: Dict[str, Any] = field(default_factory=dict)
@@ -264,20 +266,172 @@ class UnparsedSourceDefinition(JsonSchemaMixin, Replaceable):
     loaded_at_field: Optional[str] = None
     tables: List[UnparsedSourceTableDefinition] = field(default_factory=list)
     tags: List[str] = field(default_factory=list)
+    config: Dict[str, Any] = field(default_factory=dict)
 
     @property
     def yaml_key(self) -> 'str':
         return 'sources'
 
+    def __post_serialize__(self, dct):
+        dct = super().__post_serialize__(dct)
+        if 'freshnewss' not in dct and self.freshness is None:
+            dct['freshness'] = None
+        return dct
+
 
 @dataclass
-class UnparsedDocumentationFile(JsonSchemaMixin, Replaceable):
+class SourceTablePatch(dbtClassMixin):
+    name: str
+    description: Optional[str] = None
+    meta: Optional[Dict[str, Any]] = None
+    data_type: Optional[str] = None
+    docs: Optional[Docs] = None
+    loaded_at_field: Optional[str] = None
+    identifier: Optional[str] = None
+    quoting: Quoting = field(default_factory=Quoting)
+    freshness: Optional[FreshnessThreshold] = field(
+        default_factory=FreshnessThreshold
+    )
+    external: Optional[ExternalTable] = None
+    tags: Optional[List[str]] = None
+    tests: Optional[List[TestDef]] = None
+    columns: Optional[Sequence[UnparsedColumn]] = None
+
+    def to_patch_dict(self) -> Dict[str, Any]:
+        dct = self.to_dict(omit_none=True)
+        remove_keys = ('name')
+        for key in remove_keys:
+            if key in dct:
+                del dct[key]
+
+        if self.freshness is None:
+            dct['freshness'] = None
+
+        return dct
+
+
+@dataclass
+class SourcePatch(dbtClassMixin, Replaceable):
+    name: str = field(
+        metadata=dict(description='The name of the source to override'),
+    )
+    overrides: str = field(
+        metadata=dict(description='The package of the source to override'),
+    )
+    path: Path = field(
+        metadata=dict(description='The path to the patch-defining yml file'),
+    )
+    description: Optional[str] = None
+    meta: Optional[Dict[str, Any]] = None
+    database: Optional[str] = None
+    schema: Optional[str] = None
+    loader: Optional[str] = None
+    quoting: Optional[Quoting] = None
+    freshness: Optional[Optional[FreshnessThreshold]] = field(
+        default_factory=FreshnessThreshold
+    )
+    loaded_at_field: Optional[str] = None
+    tables: Optional[List[SourceTablePatch]] = None
+    tags: Optional[List[str]] = None
+
+    def to_patch_dict(self) -> Dict[str, Any]:
+        dct = self.to_dict(omit_none=True)
+        remove_keys = ('name', 'overrides', 'tables', 'path')
+        for key in remove_keys:
+            if key in dct:
+                del dct[key]
+
+        if self.freshness is None:
+            dct['freshness'] = None
+
+        return dct
+
+    def get_table_named(self, name: str) -> Optional[SourceTablePatch]:
+        if self.tables is not None:
+            for table in self.tables:
+                if table.name == name:
+                    return table
+        return None
+
+
+@dataclass
+class UnparsedDocumentation(dbtClassMixin, Replaceable):
     package_name: str
     root_path: str
     path: str
     original_file_path: str
-    file_contents: str
+
+    @property
+    def file_id(self):
+        return f'{self.package_name}://{self.original_file_path}'
 
     @property
     def resource_type(self):
         return NodeType.Documentation
+
+
+@dataclass
+class UnparsedDocumentationFile(UnparsedDocumentation):
+    file_contents: str
+
+
+# can't use total_ordering decorator here, as str provides an ordering already
+# and it's not the one we want.
+class Maturity(StrEnum):
+    low = 'low'
+    medium = 'medium'
+    high = 'high'
+
+    def __lt__(self, other):
+        if not isinstance(other, Maturity):
+            return NotImplemented
+        order = (Maturity.low, Maturity.medium, Maturity.high)
+        return order.index(self) < order.index(other)
+
+    def __gt__(self, other):
+        if not isinstance(other, Maturity):
+            return NotImplemented
+        return self != other and not (self < other)
+
+    def __ge__(self, other):
+        if not isinstance(other, Maturity):
+            return NotImplemented
+        return self == other or not (self < other)
+
+    def __le__(self, other):
+        if not isinstance(other, Maturity):
+            return NotImplemented
+        return self == other or self < other
+
+
+class ExposureType(StrEnum):
+    Dashboard = 'dashboard'
+    Notebook = 'notebook'
+    Analysis = 'analysis'
+    ML = 'ml'
+    Application = 'application'
+
+
+class MaturityType(StrEnum):
+    Low = 'low'
+    Medium = 'medium'
+    High = 'high'
+
+
+@dataclass
+class ExposureOwner(dbtClassMixin, Replaceable):
+    email: str
+    name: Optional[str] = None
+
+
+@dataclass
+class UnparsedExposure(dbtClassMixin, Replaceable):
+    name: str
+    type: ExposureType
+    owner: ExposureOwner
+    description: str = ''
+    maturity: Optional[MaturityType] = None
+    meta: Dict[str, Any] = field(default_factory=dict)
+    tags: List[str] = field(default_factory=list)
+    url: Optional[str] = None
+    depends_on: List[str] = field(default_factory=list)

core/dbt/contracts/project.py

@@ -1,29 +1,42 @@
 from dbt.contracts.util import Replaceable, Mergeable, list_str
-from dbt.contracts.connection import UserConfigContract
+from dbt.contracts.connection import UserConfigContract, QueryComment
+from dbt.helper_types import NoValue
 from dbt.logger import GLOBAL_LOGGER as logger  # noqa
 from dbt import tracking
-from dbt.ui import printer
-from dbt.helper_types import NoValue
-
-from hologram import JsonSchemaMixin, ValidationError
-from hologram.helpers import HyphenatedJsonSchemaMixin, register_pattern, \
-    ExtensibleJsonSchemaMixin
-
+from dbt import ui
+from dbt.dataclass_schema import (
+    dbtClassMixin, ValidationError,
+    HyphenatedDbtClassMixin,
+    ExtensibleDbtClassMixin,
+    register_pattern, ValidatedStringMixin
+)
 from dataclasses import dataclass, field
-from typing import Optional, List, Dict, Union, Any, NewType
+from typing import Optional, List, Dict, Union, Any
+from mashumaro.types import SerializableType
 
-PIN_PACKAGE_URL = 'https://docs.getdbt.com/docs/package-management#section-specifying-package-versions' # noqa
+PIN_PACKAGE_URL = 'https://docs.getdbt.com/docs/package-management#section-specifying-package-versions'  # noqa
 DEFAULT_SEND_ANONYMOUS_USAGE_STATS = True
-DEFAULT_USE_COLORS = True
 
 
-Name = NewType('Name', str)
+class Name(ValidatedStringMixin):
+    ValidationRegex = r'^[^\d\W]\w*$'
+
+
 register_pattern(Name, r'^[^\d\W]\w*$')
 
+
+class SemverString(str, SerializableType):
+    def _serialize(self) -> str:
+        return self
+
+    @classmethod
+    def _deserialize(cls, value: str) -> 'SemverString':
+        return SemverString(value)
+
+
 # this does not support the full semver (does not allow a trailing -fooXYZ) and
 # is not restrictive enough for full semver, (allows '1.0'). But it's like
 # 'semver lite'.
-SemverString = NewType('SemverString', str)
 register_pattern(
     SemverString,
     r'^(?:0|[1-9]\d*)\.(?:0|[1-9]\d*)(\.(?:0|[1-9]\d*))?$',
@@ -31,15 +44,15 @@ register_pattern(
 
 
 @dataclass
-class Quoting(JsonSchemaMixin, Mergeable):
-    identifier: Optional[bool]
-    schema: Optional[bool]
-    database: Optional[bool]
-    project: Optional[bool]
+class Quoting(dbtClassMixin, Mergeable):
+    schema: Optional[bool] = None
+    database: Optional[bool] = None
+    project: Optional[bool] = None
+    identifier: Optional[bool] = None
 
 
 @dataclass
-class Package(Replaceable, HyphenatedJsonSchemaMixin):
+class Package(Replaceable, HyphenatedDbtClassMixin):
     pass
 
 
@@ -48,24 +61,42 @@ class LocalPackage(Package):
     local: str
 
 
+# `float` also allows `int`, according to PEP484 (and jsonschema!)
+RawVersion = Union[str, float]
+
+
 @dataclass
 class GitPackage(Package):
     git: str
-    revision: Optional[str]
+    revision: Optional[RawVersion] = None
     warn_unpinned: Optional[bool] = None
+    subdirectory: Optional[str] = None
+
+    def get_revisions(self) -> List[str]:
+        if self.revision is None:
+            return []
+        else:
+            return [str(self.revision)]
 
 
 @dataclass
 class RegistryPackage(Package):
     package: str
-    version: Union[str, List[str]]
+    version: Union[RawVersion, List[RawVersion]]
+    install_prerelease: Optional[bool] = False
+
+    def get_versions(self) -> List[str]:
+        if isinstance(self.version, list):
+            return [str(v) for v in self.version]
+        else:
+            return [str(self.version)]
 
 
 PackageSpec = Union[LocalPackage, GitPackage, RegistryPackage]
 
 
 @dataclass
-class PackageConfig(JsonSchemaMixin, Replaceable):
+class PackageConfig(dbtClassMixin, Replaceable):
     packages: List[PackageSpec]
 
 
@@ -81,13 +112,13 @@ class ProjectPackageMetadata:
 
 
 @dataclass
-class Downloads(ExtensibleJsonSchemaMixin, Replaceable):
+class Downloads(ExtensibleDbtClassMixin, Replaceable):
     tarball: str
 
 
 @dataclass
 class RegistryPackageMetadata(
-    ExtensibleJsonSchemaMixin,
+    ExtensibleDbtClassMixin,
     ProjectPackageMetadata,
 ):
     downloads: Downloads
@@ -127,6 +158,7 @@ BANNED_PROJECT_NAMES = {
     'sql',
     'sql_now',
     'store_result',
+    'store_raw_result',
     'target',
     'this',
     'tojson',
@@ -138,9 +170,10 @@ BANNED_PROJECT_NAMES = {
 
 
 @dataclass
-class Project(HyphenatedJsonSchemaMixin, Replaceable):
+class Project(HyphenatedDbtClassMixin, Replaceable):
     name: Name
     version: Union[SemverString, float]
+    config_version: int
     project_root: Optional[str] = None
     source_paths: Optional[List[str]] = None
     macro_paths: Optional[List[str]] = None
@@ -148,6 +181,7 @@ class Project(HyphenatedJsonSchemaMixin, Replaceable):
     test_paths: Optional[List[str]] = None
     analysis_paths: Optional[List[str]] = None
     docs_paths: Optional[List[str]] = None
+    asset_paths: Optional[List[str]] = None
     target_path: Optional[str] = None
     snapshot_paths: Optional[List[str]] = None
     clean_targets: Optional[List[str]] = None
@@ -158,27 +192,42 @@ class Project(HyphenatedJsonSchemaMixin, Replaceable):
     on_run_start: Optional[List[str]] = field(default_factory=list_str)
     on_run_end: Optional[List[str]] = field(default_factory=list_str)
     require_dbt_version: Optional[Union[List[str], str]] = None
+    dispatch: List[Dict[str, Any]] = field(default_factory=list)
     models: Dict[str, Any] = field(default_factory=dict)
     seeds: Dict[str, Any] = field(default_factory=dict)
     snapshots: Dict[str, Any] = field(default_factory=dict)
+    analyses: Dict[str, Any] = field(default_factory=dict)
+    sources: Dict[str, Any] = field(default_factory=dict)
+    tests: Dict[str, Any] = field(default_factory=dict)
+    vars: Optional[Dict[str, Any]] = field(
+        default=None,
+        metadata=dict(
+            description='map project names to their vars override dicts',
+        ),
+    )
     packages: List[PackageSpec] = field(default_factory=list)
-    query_comment: Optional[Union[str, NoValue]] = NoValue()
+    query_comment: Optional[Union[QueryComment, NoValue, str]] = NoValue()
 
     @classmethod
-    def from_dict(cls, data, validate=True):
-        result = super().from_dict(data, validate=validate)
-        if result.name in BANNED_PROJECT_NAMES:
+    def validate(cls, data):
+        super().validate(data)
+        if data['name'] in BANNED_PROJECT_NAMES:
             raise ValidationError(
-                'Invalid project name: {} is a reserved word'
-                .format(result.name)
+                f"Invalid project name: {data['name']} is a reserved word"
             )
-        return result
+        # validate dispatch config
+        if 'dispatch' in data and data['dispatch']:
+            entries = data['dispatch']
+            for entry in entries:
+                if ('macro_namespace' not in entry or 'search_order' not in entry or
+                        not isinstance(entry['search_order'], list)):
+                    raise ValidationError(f"Invalid project dispatch config: {entry}")
 
 
 @dataclass
-class UserConfig(ExtensibleJsonSchemaMixin, Replaceable, UserConfigContract):
+class UserConfig(ExtensibleDbtClassMixin, Replaceable, UserConfigContract):
     send_anonymous_usage_stats: bool = DEFAULT_SEND_ANONYMOUS_USAGE_STATS
-    use_colors: bool = DEFAULT_USE_COLORS
+    use_colors: Optional[bool] = None
     partial_parse: Optional[bool] = None
     printer_width: Optional[int] = None
 
@@ -188,15 +237,15 @@ class UserConfig(ExtensibleJsonSchemaMixin, Replaceable, UserConfigContract):
         else:
             tracking.do_not_track()
 
-        if self.use_colors:
-            printer.use_colors()
+        if self.use_colors is not None:
+            ui.use_colors(self.use_colors)
 
         if self.printer_width:
-            printer.printer_width(self.printer_width)
+            ui.printer_width(self.printer_width)
 
 
 @dataclass
-class ProfileConfig(HyphenatedJsonSchemaMixin, Replaceable):
+class ProfileConfig(HyphenatedDbtClassMixin, Replaceable):
     profile_name: str = field(metadata={'preserve_underscore': True})
     target_name: str = field(metadata={'preserve_underscore': True})
     config: UserConfig
@@ -207,10 +256,10 @@ class ProfileConfig(HyphenatedJsonSchemaMixin, Replaceable):
 
 @dataclass
 class ConfiguredQuoting(Quoting, Replaceable):
-    identifier: bool
-    schema: bool
-    database: Optional[bool]
-    project: Optional[bool]
+    identifier: bool = True
+    schema: bool = True
+    database: Optional[bool] = None
+    project: Optional[bool] = None
 
 
 @dataclass
@@ -223,5 +272,5 @@ class Configuration(Project, ProfileConfig):
 
 
 @dataclass
-class ProjectList(JsonSchemaMixin):
+class ProjectList(dbtClassMixin):
     projects: Dict[str, Project]

core/dbt/contracts/relation.py

@@ -0,0 +1,130 @@
+from collections.abc import Mapping
+from dataclasses import dataclass, fields
+from typing import (
+    Optional, Dict,
+)
+from typing_extensions import Protocol
+
+from dbt.dataclass_schema import dbtClassMixin, StrEnum
+
+from dbt import deprecations
+from dbt.contracts.util import Replaceable
+from dbt.exceptions import CompilationException
+from dbt.utils import deep_merge
+
+
+class RelationType(StrEnum):
+    Table = 'table'
+    View = 'view'
+    CTE = 'cte'
+    MaterializedView = 'materializedview'
+    External = 'external'
+
+
+class ComponentName(StrEnum):
+    Database = 'database'
+    Schema = 'schema'
+    Identifier = 'identifier'
+
+
+class HasQuoting(Protocol):
+    quoting: Dict[str, bool]
+
+
+class FakeAPIObject(dbtClassMixin, Replaceable, Mapping):
+    # override the mapping truthiness, len is always >1
+    def __bool__(self):
+        return True
+
+    def __getitem__(self, key):
+        try:
+            return getattr(self, key)
+        except AttributeError:
+            raise KeyError(key) from None
+
+    def __iter__(self):
+        deprecations.warn('not-a-dictionary', obj=self)
+        for _, name in self._get_fields():
+            yield name
+
+    def __len__(self):
+        deprecations.warn('not-a-dictionary', obj=self)
+        return len(fields(self.__class__))
+
+    def incorporate(self, **kwargs):
+        value = self.to_dict(omit_none=True)
+        value = deep_merge(value, kwargs)
+        return self.from_dict(value)
+
+
+@dataclass
+class Policy(FakeAPIObject):
+    database: bool = True
+    schema: bool = True
+    identifier: bool = True
+
+    def get_part(self, key: ComponentName) -> bool:
+        if key == ComponentName.Database:
+            return self.database
+        elif key == ComponentName.Schema:
+            return self.schema
+        elif key == ComponentName.Identifier:
+            return self.identifier
+        else:
+            raise ValueError(
+                'Got a key of {}, expected one of {}'
+                .format(key, list(ComponentName))
+            )
+
+    def replace_dict(self, dct: Dict[ComponentName, bool]):
+        kwargs: Dict[str, bool] = {}
+        for k, v in dct.items():
+            kwargs[str(k)] = v
+        return self.replace(**kwargs)
+
+
+@dataclass
+class Path(FakeAPIObject):
+    database: Optional[str] = None
+    schema: Optional[str] = None
+    identifier: Optional[str] = None
+
+    def __post_init__(self):
+        # handle pesky jinja2.Undefined sneaking in here and messing up rende
+        if not isinstance(self.database, (type(None), str)):
+            raise CompilationException(
+                'Got an invalid path database: {}'.format(self.database)
+            )
+        if not isinstance(self.schema, (type(None), str)):
+            raise CompilationException(
+                'Got an invalid path schema: {}'.format(self.schema)
+            )
+        if not isinstance(self.identifier, (type(None), str)):
+            raise CompilationException(
+                'Got an invalid path identifier: {}'.format(self.identifier)
+            )
+
+    def get_lowered_part(self, key: ComponentName) -> Optional[str]:
+        part = self.get_part(key)
+        if part is not None:
+            part = part.lower()
+        return part
+
+    def get_part(self, key: ComponentName) -> Optional[str]:
+        if key == ComponentName.Database:
+            return self.database
+        elif key == ComponentName.Schema:
+            return self.schema
+        elif key == ComponentName.Identifier:
+            return self.identifier
+        else:
+            raise ValueError(
+                'Got a key of {}, expected one of {}'
+                .format(key, list(ComponentName))
+            )
+
+    def replace_dict(self, dct: Dict[ComponentName, str]):
+        kwargs: Dict[str, str] = {}
+        for k, v in dct.items():
+            kwargs[str(k)] = v
+        return self.replace(**kwargs)

core/dbt/contracts/results.py

@@ -1,27 +1,37 @@
 from dbt.contracts.graph.manifest import CompileResultNode
 from dbt.contracts.graph.unparsed import (
-    Time, FreshnessStatus, FreshnessThreshold
+    FreshnessThreshold
 )
 from dbt.contracts.graph.parsed import ParsedSourceDefinition
-from dbt.contracts.util import Writable, Replaceable
+from dbt.contracts.util import (
+    BaseArtifactMetadata,
+    ArtifactMixin,
+    VersionedSchema,
+    Replaceable,
+    schema_version,
+)
 from dbt.exceptions import InternalException
 from dbt.logger import (
     TimingProcessor,
     JsonOnly,
     GLOBAL_LOGGER as logger,
 )
-from hologram.helpers import StrEnum
-from hologram import JsonSchemaMixin
+from dbt.utils import lowercase
+from dbt.dataclass_schema import dbtClassMixin, StrEnum
 
 import agate
 
 from dataclasses import dataclass, field
 from datetime import datetime
-from typing import Union, Dict, List, Optional, Any, NamedTuple
+from typing import (
+    Union, Dict, List, Optional, Any, NamedTuple, Sequence,
+)
+
+from dbt.clients.system import write_json
 
 
 @dataclass
-class TimingInfo(JsonSchemaMixin):
+class TimingInfo(dbtClassMixin):
     name: str
     started_at: Optional[datetime] = None
     completed_at: Optional[datetime] = None
@@ -47,43 +57,78 @@ class collect_timing_info:
             logger.debug('finished collecting timing info')
 
 
-@dataclass
-class PartialResult(JsonSchemaMixin, Writable):
-    node: CompileResultNode
-    error: Optional[str] = None
-    status: Union[None, str, int, bool] = None
-    execution_time: Union[str, int] = 0
-    thread_id: Optional[str] = None
-    timing: List[TimingInfo] = field(default_factory=list)
-    fail: Optional[bool] = None
-    warn: Optional[bool] = None
+class NodeStatus(StrEnum):
+    Success = "success"
+    Error = "error"
+    Fail = "fail"
+    Warn = "warn"
+    Skipped = "skipped"
+    Pass = "pass"
+    RuntimeErr = "runtime error"
+
+
+class RunStatus(StrEnum):
+    Success = NodeStatus.Success
+    Error = NodeStatus.Error
+    Skipped = NodeStatus.Skipped
+
+
+class TestStatus(StrEnum):
+    Pass = NodeStatus.Pass
+    Error = NodeStatus.Error
+    Fail = NodeStatus.Fail
+    Warn = NodeStatus.Warn
+    Skipped = NodeStatus.Skipped
+
+
+class FreshnessStatus(StrEnum):
+    Pass = NodeStatus.Pass
+    Warn = NodeStatus.Warn
+    Error = NodeStatus.Error
+    RuntimeErr = NodeStatus.RuntimeErr
+
+
+@dataclass
+class BaseResult(dbtClassMixin):
+    status: Union[RunStatus, TestStatus, FreshnessStatus]
+    timing: List[TimingInfo]
+    thread_id: str
+    execution_time: float
+    adapter_response: Dict[str, Any]
+    message: Optional[str]
+    failures: Optional[int]
+
+    @classmethod
+    def __pre_deserialize__(cls, data):
+        data = super().__pre_deserialize__(data)
+        if 'message' not in data:
+            data['message'] = None
+        if 'failures' not in data:
+            data['failures'] = None
+        return data
+
+
+@dataclass
+class NodeResult(BaseResult):
+    node: CompileResultNode
+
+
+@dataclass
+class RunResult(NodeResult):
+    agate_table: Optional[agate.Table] = field(
+        default=None, metadata={
+            'serialize': lambda x: None, 'deserialize': lambda x: None
+        }
+    )
 
-    # if the result got to the point where it could be skipped/failed, we would
-    # be returning a real result, not a partial.
     @property
     def skipped(self):
-        return False
+        return self.status == RunStatus.Skipped
 
 
 @dataclass
-class WritableRunModelResult(PartialResult):
-    skip: bool = False
-
-
-@dataclass
-class RunModelResult(WritableRunModelResult):
-    agate_table: Optional[agate.Table] = None
-
-    def to_dict(self, *args, **kwargs):
-        dct = super().to_dict(*args, **kwargs)
-        dct.pop('agate_table', None)
-        return dct
-
-
-@dataclass
-class ExecutionResult(JsonSchemaMixin, Writable):
-    results: List[Union[WritableRunModelResult, PartialResult]]
-    generated_at: datetime
+class ExecutionResult(dbtClassMixin):
+    results: Sequence[BaseResult]
     elapsed_time: float
 
     def __len__(self):
@@ -96,183 +141,291 @@ class ExecutionResult(JsonSchemaMixin, Writable):
         return self.results[idx]
 
 
+@dataclass
+class RunResultsMetadata(BaseArtifactMetadata):
+    dbt_schema_version: str = field(
+        default_factory=lambda: str(RunResultsArtifact.dbt_schema_version)
+    )
+
+
+@dataclass
+class RunResultOutput(BaseResult):
+    unique_id: str
+
+
+def process_run_result(result: RunResult) -> RunResultOutput:
+    return RunResultOutput(
+        unique_id=result.node.unique_id,
+        status=result.status,
+        timing=result.timing,
+        thread_id=result.thread_id,
+        execution_time=result.execution_time,
+        message=result.message,
+        adapter_response=result.adapter_response,
+        failures=result.failures
+    )
+
+
+@dataclass
+class RunExecutionResult(
+    ExecutionResult,
+):
+    results: Sequence[RunResult]
+    args: Dict[str, Any] = field(default_factory=dict)
+    generated_at: datetime = field(default_factory=datetime.utcnow)
+
+    def write(self, path: str):
+        writable = RunResultsArtifact.from_execution_results(
+            results=self.results,
+            elapsed_time=self.elapsed_time,
+            generated_at=self.generated_at,
+            args=self.args,
+        )
+        writable.write(path)
+
+
+@dataclass
+@schema_version('run-results', 2)
+class RunResultsArtifact(ExecutionResult, ArtifactMixin):
+    results: Sequence[RunResultOutput]
+    args: Dict[str, Any] = field(default_factory=dict)
+
+    @classmethod
+    def from_execution_results(
+        cls,
+        results: Sequence[RunResult],
+        elapsed_time: float,
+        generated_at: datetime,
+        args: Dict,
+    ):
+        processed_results = [process_run_result(result) for result in results]
+        meta = RunResultsMetadata(
+            dbt_schema_version=str(cls.dbt_schema_version),
+            generated_at=generated_at,
+        )
+        return cls(
+            metadata=meta,
+            results=processed_results,
+            elapsed_time=elapsed_time,
+            args=args
+        )
+
+    def write(self, path: str):
+        write_json(path, self.to_dict(omit_none=False))
+
+
 @dataclass
 class RunOperationResult(ExecutionResult):
     success: bool
 
 
+@dataclass
+class RunOperationResultMetadata(BaseArtifactMetadata):
+    dbt_schema_version: str = field(default_factory=lambda: str(
+        RunOperationResultsArtifact.dbt_schema_version
+    ))
+
+
+@dataclass
+@schema_version('run-operation-result', 1)
+class RunOperationResultsArtifact(RunOperationResult, ArtifactMixin):
+
+    @classmethod
+    def from_success(
+        cls,
+        success: bool,
+        elapsed_time: float,
+        generated_at: datetime,
+    ):
+        meta = RunOperationResultMetadata(
+            dbt_schema_version=str(cls.dbt_schema_version),
+            generated_at=generated_at,
+        )
+        return cls(
+            metadata=meta,
+            results=[],
+            elapsed_time=elapsed_time,
+            success=success,
+        )
+
 # due to issues with typing.Union collapsing subclasses, this can't subclass
 # PartialResult
+
+
 @dataclass
-class SourceFreshnessResult(JsonSchemaMixin, Writable):
+class SourceFreshnessResult(NodeResult):
     node: ParsedSourceDefinition
+    status: FreshnessStatus
     max_loaded_at: datetime
     snapshotted_at: datetime
     age: float
-    status: FreshnessStatus
-    error: Optional[str] = None
-    execution_time: Union[str, int] = 0
-    thread_id: Optional[str] = None
-    timing: List[TimingInfo] = field(default_factory=list)
-    fail: Optional[bool] = None
-
-    def __post_init__(self):
-        self.fail = self.status == 'error'
-
-    @property
-    def warned(self):
-        return self.status == 'warn'
 
     @property
     def skipped(self):
         return False
 
 
-@dataclass
-class FreshnessMetadata(JsonSchemaMixin):
-    generated_at: datetime
-    elapsed_time: float
-
-
-@dataclass
-class FreshnessExecutionResult(FreshnessMetadata):
-    results: List[Union[PartialResult, SourceFreshnessResult]]
-
-    def write(self, path, omit_none=True):
-        """Create a new object with the desired output schema and write it."""
-        meta = FreshnessMetadata(
-            generated_at=self.generated_at,
-            elapsed_time=self.elapsed_time,
-        )
-        sources = {}
-        for result in self.results:
-            result_value: Union[
-                SourceFreshnessRuntimeError, SourceFreshnessOutput
-            ]
-            unique_id = result.node.unique_id
-            if result.error is not None:
-                result_value = SourceFreshnessRuntimeError(
-                    error=result.error,
-                    state=FreshnessErrorEnum.runtime_error,
-                )
-            else:
-                # we know that this must be a SourceFreshnessResult
-                if not isinstance(result, SourceFreshnessResult):
-                    raise InternalException(
-                        'Got {} instead of a SourceFreshnessResult for a '
-                        'non-error result in freshness execution!'
-                        .format(type(result))
-                    )
-                # if we're here, we must have a non-None freshness threshold
-                criteria = result.node.freshness
-                if criteria is None:
-                    raise InternalException(
-                        'Somehow evaluated a freshness result for a source '
-                        'that has no freshness criteria!'
-                    )
-                result_value = SourceFreshnessOutput(
-                    max_loaded_at=result.max_loaded_at,
-                    snapshotted_at=result.snapshotted_at,
-                    max_loaded_at_time_ago_in_s=result.age,
-                    state=result.status,
-                    criteria=criteria,
-                )
-            sources[unique_id] = result_value
-        output = FreshnessRunOutput(meta=meta, sources=sources)
-        output.write(path, omit_none=omit_none)
-
-    def __len__(self):
-        return len(self.results)
-
-    def __iter__(self):
-        return iter(self.results)
-
-    def __getitem__(self, idx):
-        return self.results[idx]
-
-
-def _copykeys(src, keys, **updates):
-    return {k: getattr(src, k) for k in keys}
-
-
-@dataclass
-class FreshnessCriteria(JsonSchemaMixin):
-    warn_after: Time
-    error_after: Time
-
-
 class FreshnessErrorEnum(StrEnum):
     runtime_error = 'runtime error'
 
 
 @dataclass
-class SourceFreshnessRuntimeError(JsonSchemaMixin):
-    error: str
-    state: FreshnessErrorEnum
+class SourceFreshnessRuntimeError(dbtClassMixin):
+    unique_id: str
+    error: Optional[Union[str, int]]
+    status: FreshnessErrorEnum
 
 
 @dataclass
-class SourceFreshnessOutput(JsonSchemaMixin):
+class SourceFreshnessOutput(dbtClassMixin):
+    unique_id: str
     max_loaded_at: datetime
     snapshotted_at: datetime
     max_loaded_at_time_ago_in_s: float
-    state: FreshnessStatus
+    status: FreshnessStatus
     criteria: FreshnessThreshold
-
-
-SourceFreshnessRunResult = Union[SourceFreshnessOutput,
-                                 SourceFreshnessRuntimeError]
+    adapter_response: Dict[str, Any]
 
 
 @dataclass
-class FreshnessRunOutput(JsonSchemaMixin, Writable):
-    meta: FreshnessMetadata
-    sources: Dict[str, SourceFreshnessRunResult]
+class PartialSourceFreshnessResult(NodeResult):
+    status: FreshnessStatus
+
+    @property
+    def skipped(self):
+        return False
+
+
+FreshnessNodeResult = Union[PartialSourceFreshnessResult,
+                            SourceFreshnessResult]
+FreshnessNodeOutput = Union[SourceFreshnessRuntimeError, SourceFreshnessOutput]
+
+
+def process_freshness_result(
+    result: FreshnessNodeResult
+) -> FreshnessNodeOutput:
+    unique_id = result.node.unique_id
+    if result.status == FreshnessStatus.RuntimeErr:
+        return SourceFreshnessRuntimeError(
+            unique_id=unique_id,
+            error=result.message,
+            status=FreshnessErrorEnum.runtime_error,
+        )
+
+    # we know that this must be a SourceFreshnessResult
+    if not isinstance(result, SourceFreshnessResult):
+        raise InternalException(
+            'Got {} instead of a SourceFreshnessResult for a '
+            'non-error result in freshness execution!'
+            .format(type(result))
+        )
+    # if we're here, we must have a non-None freshness threshold
+    criteria = result.node.freshness
+    if criteria is None:
+        raise InternalException(
+            'Somehow evaluated a freshness result for a source '
+            'that has no freshness criteria!'
+        )
+    return SourceFreshnessOutput(
+        unique_id=unique_id,
+        max_loaded_at=result.max_loaded_at,
+        snapshotted_at=result.snapshotted_at,
+        max_loaded_at_time_ago_in_s=result.age,
+        status=result.status,
+        criteria=criteria,
+        adapter_response=result.adapter_response
+    )
+
+
+@dataclass
+class FreshnessMetadata(BaseArtifactMetadata):
+    dbt_schema_version: str = field(
+        default_factory=lambda: str(
+            FreshnessExecutionResultArtifact.dbt_schema_version
+        )
+    )
+
+
+@dataclass
+class FreshnessResult(ExecutionResult):
+    metadata: FreshnessMetadata
+    results: Sequence[FreshnessNodeResult]
+
+    @classmethod
+    def from_node_results(
+        cls,
+        results: List[FreshnessNodeResult],
+        elapsed_time: float,
+        generated_at: datetime,
+    ):
+        meta = FreshnessMetadata(generated_at=generated_at)
+        return cls(metadata=meta, results=results, elapsed_time=elapsed_time)
+
+
+@dataclass
+@schema_version('sources', 1)
+class FreshnessExecutionResultArtifact(
+    ArtifactMixin,
+    VersionedSchema,
+):
+    metadata: FreshnessMetadata
+    results: Sequence[FreshnessNodeOutput]
+    elapsed_time: float
+
+    @classmethod
+    def from_result(cls, base: FreshnessResult):
+        processed = [process_freshness_result(r) for r in base.results]
+        return cls(
+            metadata=base.metadata,
+            results=processed,
+            elapsed_time=base.elapsed_time,
+        )
 
 
 Primitive = Union[bool, str, float, None]
+PrimitiveDict = Dict[str, Primitive]
 
 CatalogKey = NamedTuple(
     'CatalogKey',
-    [('database', str), ('schema', str), ('name', str)]
+    [('database', Optional[str]), ('schema', str), ('name', str)]
 )
 
 
 @dataclass
-class StatsItem(JsonSchemaMixin):
+class StatsItem(dbtClassMixin):
     id: str
     label: str
     value: Primitive
-    description: str
     include: bool
+    description: Optional[str] = None
 
 
 StatsDict = Dict[str, StatsItem]
 
 
 @dataclass
-class ColumnMetadata(JsonSchemaMixin):
+class ColumnMetadata(dbtClassMixin):
     type: str
-    comment: Optional[str]
     index: int
     name: str
+    comment: Optional[str] = None
 
 
 ColumnMap = Dict[str, ColumnMetadata]
 
 
 @dataclass
-class TableMetadata(JsonSchemaMixin):
+class TableMetadata(dbtClassMixin):
     type: str
-    database: str
     schema: str
     name: str
-    comment: Optional[str]
-    owner: Optional[str]
+    database: Optional[str] = None
+    comment: Optional[str] = None
+    owner: Optional[str] = None
 
 
 @dataclass
-class CatalogTable(JsonSchemaMixin, Replaceable):
+class CatalogTable(dbtClassMixin, Replaceable):
     metadata: TableMetadata
     columns: ColumnMap
     stats: StatsDict
@@ -281,15 +434,52 @@ class CatalogTable(JsonSchemaMixin, Replaceable):
 
     def key(self) -> CatalogKey:
         return CatalogKey(
-            self.metadata.database.lower(),
+            lowercase(self.metadata.database),
             self.metadata.schema.lower(),
             self.metadata.name.lower(),
         )
 
 
 @dataclass
-class CatalogResults(JsonSchemaMixin, Writable):
+class CatalogMetadata(BaseArtifactMetadata):
+    dbt_schema_version: str = field(
+        default_factory=lambda: str(CatalogArtifact.dbt_schema_version)
+    )
+
+
+@dataclass
+class CatalogResults(dbtClassMixin):
     nodes: Dict[str, CatalogTable]
-    generated_at: datetime
-    errors: Optional[List[str]]
+    sources: Dict[str, CatalogTable]
+    errors: Optional[List[str]] = None
     _compile_results: Optional[Any] = None
+
+    def __post_serialize__(self, dct):
+        dct = super().__post_serialize__(dct)
+        if '_compile_results' in dct:
+            del dct['_compile_results']
+        return dct
+
+
+@dataclass
+@schema_version('catalog', 1)
+class CatalogArtifact(CatalogResults, ArtifactMixin):
+    metadata: CatalogMetadata
+
+    @classmethod
+    def from_results(
+        cls,
+        generated_at: datetime,
+        nodes: Dict[str, CatalogTable],
+        sources: Dict[str, CatalogTable],
+        compile_results: Optional[Any],
+        errors: Optional[List[str]]
+    ) -> 'CatalogArtifact':
+        meta = CatalogMetadata(generated_at=generated_at)
+        return cls(
+            metadata=meta,
+            nodes=nodes,
+            sources=sources,
+            errors=errors,
+            _compile_results=compile_results,
+        )

core/dbt/contracts/rpc.py

@@ -3,17 +3,24 @@ import os
 import uuid
 from dataclasses import dataclass, field
 from datetime import datetime, timedelta
-from typing import Optional, Union, List, Any, Dict, Type
+from typing import Optional, Union, List, Any, Dict, Type, Sequence
 
-from hologram import JsonSchemaMixin
-from hologram.helpers import StrEnum
+from dbt.dataclass_schema import dbtClassMixin, StrEnum
 
 from dbt.contracts.graph.compiled import CompileResultNode
+from dbt.contracts.graph.manifest import WritableManifest
 from dbt.contracts.results import (
-    TimingInfo,
+    RunResult, RunResultsArtifact, TimingInfo,
+    CatalogArtifact,
     CatalogResults,
     ExecutionResult,
+    FreshnessExecutionResultArtifact,
+    FreshnessResult,
+    RunOperationResult,
+    RunOperationResultsArtifact,
+    RunExecutionResult,
 )
+from dbt.contracts.util import VersionedSchema, schema_version
 from dbt.exceptions import InternalException
 from dbt.logger import LogMessage
 from dbt.utils import restrict_to
@@ -26,23 +33,57 @@ TaskID = uuid.UUID
 
 
 @dataclass
-class RPCParameters(JsonSchemaMixin):
-    timeout: Optional[float]
+class RPCParameters(dbtClassMixin):
     task_tags: TaskTags
+    timeout: Optional[float]
+
+    @classmethod
+    def __pre_deserialize__(cls, data, omit_none=True):
+        data = super().__pre_deserialize__(data)
+        if 'timeout' not in data:
+            data['timeout'] = None
+        if 'task_tags' not in data:
+            data['task_tags'] = None
+        return data
 
 
 @dataclass
 class RPCExecParameters(RPCParameters):
     name: str
     sql: str
-    macros: Optional[str]
+    macros: Optional[str] = None
 
 
 @dataclass
 class RPCCompileParameters(RPCParameters):
     threads: Optional[int] = None
     models: Union[None, str, List[str]] = None
+    select: Union[None, str, List[str]] = None
     exclude: Union[None, str, List[str]] = None
+    selector: Optional[str] = None
+    state: Optional[str] = None
+
+
+@dataclass
+class RPCListParameters(RPCParameters):
+    resource_types: Optional[List[str]] = None
+    models: Union[None, str, List[str]] = None
+    exclude: Union[None, str, List[str]] = None
+    select: Union[None, str, List[str]] = None
+    selector: Optional[str] = None
+    output: Optional[str] = 'json'
+    output_keys: Optional[List[str]] = None
+
+
+@dataclass
+class RPCRunParameters(RPCParameters):
+    threads: Optional[int] = None
+    models: Union[None, str, List[str]] = None
+    select: Union[None, str, List[str]] = None
+    exclude: Union[None, str, List[str]] = None
+    selector: Optional[str] = None
+    state: Optional[str] = None
+    defer: Optional[bool] = None
 
 
 @dataclass
@@ -50,12 +91,16 @@ class RPCSnapshotParameters(RPCParameters):
     threads: Optional[int] = None
     select: Union[None, str, List[str]] = None
     exclude: Union[None, str, List[str]] = None
+    selector: Optional[str] = None
+    state: Optional[str] = None
 
 
 @dataclass
 class RPCTestParameters(RPCCompileParameters):
     data: bool = False
     schema: bool = False
+    state: Optional[str] = None
+    defer: Optional[bool] = None
 
 
 @dataclass
@@ -63,12 +108,26 @@ class RPCSeedParameters(RPCParameters):
     threads: Optional[int] = None
     select: Union[None, str, List[str]] = None
     exclude: Union[None, str, List[str]] = None
+    selector: Optional[str] = None
     show: bool = False
+    state: Optional[str] = None
 
 
 @dataclass
 class RPCDocsGenerateParameters(RPCParameters):
     compile: bool = True
+    state: Optional[str] = None
+
+
+@dataclass
+class RPCBuildParameters(RPCParameters):
+    threads: Optional[int] = None
+    models: Union[None, str, List[str]] = None
+    select: Union[None, str, List[str]] = None
+    exclude: Union[None, str, List[str]] = None
+    selector: Optional[str] = None
+    state: Optional[str] = None
+    defer: Optional[bool] = None
 
 
 @dataclass
@@ -77,7 +136,7 @@ class RPCCliParameters(RPCParameters):
 
 
 @dataclass
-class RPCNoParameters(RPCParameters):
+class RPCDepsParameters(RPCParameters):
     pass
 
 
@@ -105,7 +164,7 @@ class StatusParameters(RPCParameters):
 
 
 @dataclass
-class GCSettings(JsonSchemaMixin):
+class GCSettings(dbtClassMixin):
     # start evicting the longest-ago-ended tasks here
     maxsize: int
     # start evicting all tasks before now - auto_reap_age when we have this
@@ -141,70 +200,175 @@ class RPCRunOperationParameters(RPCParameters):
 class RPCSourceFreshnessParameters(RPCParameters):
     threads: Optional[int] = None
     select: Union[None, str, List[str]] = None
+    exclude: Union[None, str, List[str]] = None
+    selector: Optional[str] = None
 
 
+@dataclass
+class GetManifestParameters(RPCParameters):
+    pass
+
 # Outputs
 
+
 @dataclass
-class RemoteResult(JsonSchemaMixin):
+class RemoteResult(VersionedSchema):
     logs: List[LogMessage]
 
 
 @dataclass
-class RemoteEmptyResult(RemoteResult):
-    pass
+@schema_version('remote-list-results', 1)
+class RemoteListResults(RemoteResult):
+    output: List[Any]
+    generated_at: datetime = field(default_factory=datetime.utcnow)
 
 
 @dataclass
+@schema_version('remote-deps-result', 1)
+class RemoteDepsResult(RemoteResult):
+    generated_at: datetime = field(default_factory=datetime.utcnow)
+
+
+@dataclass
+@schema_version('remote-catalog-result', 1)
 class RemoteCatalogResults(CatalogResults, RemoteResult):
-    pass
+    generated_at: datetime = field(default_factory=datetime.utcnow)
+
+    def write(self, path: str):
+        artifact = CatalogArtifact.from_results(
+            generated_at=self.generated_at,
+            nodes=self.nodes,
+            sources=self.sources,
+            compile_results=self._compile_results,
+            errors=self.errors,
+        )
+        artifact.write(path)
 
 
 @dataclass
-class RemoteCompileResult(RemoteResult):
+class RemoteCompileResultMixin(RemoteResult):
     raw_sql: str
     compiled_sql: str
     node: CompileResultNode
     timing: List[TimingInfo]
 
+
+@dataclass
+@schema_version('remote-compile-result', 1)
+class RemoteCompileResult(RemoteCompileResultMixin):
+    generated_at: datetime = field(default_factory=datetime.utcnow)
+
     @property
     def error(self):
         return None
 
 
 @dataclass
+@schema_version('remote-execution-result', 1)
 class RemoteExecutionResult(ExecutionResult, RemoteResult):
-    pass
+    results: Sequence[RunResult]
+    args: Dict[str, Any] = field(default_factory=dict)
+    generated_at: datetime = field(default_factory=datetime.utcnow)
+
+    def write(self, path: str):
+        writable = RunResultsArtifact.from_execution_results(
+            generated_at=self.generated_at,
+            results=self.results,
+            elapsed_time=self.elapsed_time,
+            args=self.args,
+        )
+        writable.write(path)
+
+    @classmethod
+    def from_local_result(
+        cls,
+        base: RunExecutionResult,
+        logs: List[LogMessage],
+    ) -> 'RemoteExecutionResult':
+        return cls(
+            generated_at=base.generated_at,
+            results=base.results,
+            elapsed_time=base.elapsed_time,
+            args=base.args,
+            logs=logs,
+        )
 
 
 @dataclass
-class ResultTable(JsonSchemaMixin):
+class ResultTable(dbtClassMixin):
     column_names: List[str]
     rows: List[Any]
 
 
 @dataclass
-class RemoteRunOperationResult(ExecutionResult, RemoteResult):
-    success: bool
+@schema_version('remote-run-operation-result', 1)
+class RemoteRunOperationResult(RunOperationResult, RemoteResult):
+    generated_at: datetime = field(default_factory=datetime.utcnow)
+
+    @classmethod
+    def from_local_result(
+        cls,
+        base: RunOperationResultsArtifact,
+        logs: List[LogMessage],
+    ) -> 'RemoteRunOperationResult':
+        return cls(
+            generated_at=base.metadata.generated_at,
+            results=base.results,
+            elapsed_time=base.elapsed_time,
+            success=base.success,
+            logs=logs,
+        )
+
+    def write(self, path: str):
+        writable = RunOperationResultsArtifact.from_success(
+            success=self.success,
+            generated_at=self.generated_at,
+            elapsed_time=self.elapsed_time,
+        )
+        writable.write(path)
 
 
 @dataclass
-class RemoteRunResult(RemoteCompileResult):
+@schema_version('remote-freshness-result', 1)
+class RemoteFreshnessResult(FreshnessResult, RemoteResult):
+
+    @classmethod
+    def from_local_result(
+        cls,
+        base: FreshnessResult,
+        logs: List[LogMessage],
+    ) -> 'RemoteFreshnessResult':
+        return cls(
+            metadata=base.metadata,
+            results=base.results,
+            elapsed_time=base.elapsed_time,
+            logs=logs,
+        )
+
+    def write(self, path: str):
+        writable = FreshnessExecutionResultArtifact.from_result(base=self)
+        writable.write(path)
+
+
+@dataclass
+@schema_version('remote-run-result', 1)
+class RemoteRunResult(RemoteCompileResultMixin):
     table: ResultTable
+    generated_at: datetime = field(default_factory=datetime.utcnow)
 
 
 RPCResult = Union[
     RemoteCompileResult,
     RemoteExecutionResult,
+    RemoteFreshnessResult,
     RemoteCatalogResults,
-    RemoteEmptyResult,
+    RemoteDepsResult,
     RemoteRunOperationResult,
 ]
 
 
 # GC types
 
-
 class GCResultState(StrEnum):
     Deleted = 'deleted'  # successful GC
     Missing = 'missing'  # nothing to GC
@@ -212,6 +376,7 @@ class GCResultState(StrEnum):
 
 
 @dataclass
+@schema_version('remote-gc-result', 1)
 class GCResult(RemoteResult):
     logs: List[LogMessage] = field(default_factory=list)
     deleted: List[TaskID] = field(default_factory=list)
@@ -287,24 +452,35 @@ class TaskHandlerState(StrEnum):
 
 
 @dataclass
-class TaskTiming(JsonSchemaMixin):
+class TaskTiming(dbtClassMixin):
     state: TaskHandlerState
     start: Optional[datetime]
     end: Optional[datetime]
     elapsed: Optional[float]
 
+    # These ought to be defaults but superclass order doesn't
+    # allow that to work
+    @classmethod
+    def __pre_deserialize__(cls, data):
+        data = super().__pre_deserialize__(data)
+        for field_name in ('start', 'end', 'elapsed'):
+            if field_name not in data:
+                data[field_name] = None
+        return data
+
 
 @dataclass
 class TaskRow(TaskTiming):
     task_id: TaskID
-    request_id: Union[str, int]
     request_source: str
     method: str
-    timeout: Optional[float]
-    tags: TaskTags
+    request_id: Union[str, int]
+    tags: TaskTags = None
+    timeout: Optional[float] = None
 
 
 @dataclass
+@schema_version('remote-ps-result', 1)
 class PSResult(RemoteResult):
     rows: List[TaskRow]
 
@@ -317,11 +493,18 @@ class KillResultStatus(StrEnum):
 
 
 @dataclass
+@schema_version('remote-kill-result', 1)
 class KillResult(RemoteResult):
     state: KillResultStatus = KillResultStatus.Missing
     logs: List[LogMessage] = field(default_factory=list)
 
 
+@dataclass
+@schema_version('remote-manifest-result', 1)
+class GetManifestResult(RemoteResult):
+    manifest: Optional[WritableManifest] = None
+
+
 # this is kind of carefuly structured: BlocksManifestTasks is implied by
 # RequiresConfigReloadBefore and RequiresManifestReloadAfter
 class RemoteMethodFlags(enum.Flag):
@@ -343,18 +526,30 @@ class PollResult(RemoteResult, TaskTiming):
     end: Optional[datetime]
     elapsed: Optional[float]
 
+    # These ought to be defaults but superclass order doesn't
+    # allow that to work
+    @classmethod
+    def __pre_deserialize__(cls, data):
+        data = super().__pre_deserialize__(data)
+        for field_name in ('start', 'end', 'elapsed'):
+            if field_name not in data:
+                data[field_name] = None
+        return data
+
 
 @dataclass
-class PollRemoteEmptyCompleteResult(PollResult, RemoteEmptyResult):
+@schema_version('poll-remote-deps-result', 1)
+class PollRemoteEmptyCompleteResult(PollResult, RemoteResult):
     state: TaskHandlerState = field(
         metadata=restrict_to(TaskHandlerState.Success,
                              TaskHandlerState.Failed),
     )
+    generated_at: datetime = field(default_factory=datetime.utcnow)
 
     @classmethod
     def from_result(
         cls: Type['PollRemoteEmptyCompleteResult'],
-        base: RemoteEmptyResult,
+        base: RemoteDepsResult,
         tags: TaskTags,
         timing: TaskTiming,
         logs: List[LogMessage],
@@ -366,10 +561,12 @@ class PollRemoteEmptyCompleteResult(PollResult, RemoteEmptyResult):
             start=timing.start,
             end=timing.end,
             elapsed=timing.elapsed,
+            generated_at=base.generated_at
         )
 
 
 @dataclass
+@schema_version('poll-remote-killed-result', 1)
 class PollKilledResult(PollResult):
     state: TaskHandlerState = field(
         metadata=restrict_to(TaskHandlerState.Killed),
@@ -377,7 +574,11 @@ class PollKilledResult(PollResult):
 
 
 @dataclass
-class PollExecuteCompleteResult(RemoteExecutionResult, PollResult):
+@schema_version('poll-remote-execution-result', 1)
+class PollExecuteCompleteResult(
+    RemoteExecutionResult,
+    PollResult,
+):
     state: TaskHandlerState = field(
         metadata=restrict_to(TaskHandlerState.Success,
                              TaskHandlerState.Failed),
@@ -393,7 +594,6 @@ class PollExecuteCompleteResult(RemoteExecutionResult, PollResult):
     ) -> 'PollExecuteCompleteResult':
         return cls(
             results=base.results,
-            generated_at=base.generated_at,
             elapsed_time=base.elapsed_time,
             logs=logs,
             tags=tags,
@@ -401,11 +601,16 @@ class PollExecuteCompleteResult(RemoteExecutionResult, PollResult):
             start=timing.start,
             end=timing.end,
             elapsed=timing.elapsed,
+            generated_at=base.generated_at,
         )
 
 
 @dataclass
-class PollCompileCompleteResult(RemoteCompileResult, PollResult):
+@schema_version('poll-remote-compile-result', 1)
+class PollCompileCompleteResult(
+    RemoteCompileResult,
+    PollResult,
+):
     state: TaskHandlerState = field(
         metadata=restrict_to(TaskHandlerState.Success,
                              TaskHandlerState.Failed),
@@ -430,11 +635,16 @@ class PollCompileCompleteResult(RemoteCompileResult, PollResult):
             start=timing.start,
             end=timing.end,
             elapsed=timing.elapsed,
+            generated_at=base.generated_at
         )
 
 
 @dataclass
-class PollRunCompleteResult(RemoteRunResult, PollResult):
+@schema_version('poll-remote-run-result', 1)
+class PollRunCompleteResult(
+    RemoteRunResult,
+    PollResult,
+):
     state: TaskHandlerState = field(
         metadata=restrict_to(TaskHandlerState.Success,
                              TaskHandlerState.Failed),
@@ -460,11 +670,16 @@ class PollRunCompleteResult(RemoteRunResult, PollResult):
             start=timing.start,
             end=timing.end,
             elapsed=timing.elapsed,
+            generated_at=base.generated_at
         )
 
 
 @dataclass
-class PollRunOperationCompleteResult(RemoteRunOperationResult, PollResult):
+@schema_version('poll-remote-run-operation-result', 1)
+class PollRunOperationCompleteResult(
+    RemoteRunOperationResult,
+    PollResult,
+):
     state: TaskHandlerState = field(
         metadata=restrict_to(TaskHandlerState.Success,
                              TaskHandlerState.Failed),
@@ -493,6 +708,7 @@ class PollRunOperationCompleteResult(RemoteRunOperationResult, PollResult):
 
 
 @dataclass
+@schema_version('poll-remote-catalog-result', 1)
 class PollCatalogCompleteResult(RemoteCatalogResults, PollResult):
     state: TaskHandlerState = field(
         metadata=restrict_to(TaskHandlerState.Success,
@@ -509,6 +725,7 @@ class PollCatalogCompleteResult(RemoteCatalogResults, PollResult):
     ) -> 'PollCatalogCompleteResult':
         return cls(
             nodes=base.nodes,
+            sources=base.sources,
             generated_at=base.generated_at,
             errors=base.errors,
             _compile_results=base._compile_results,
@@ -522,12 +739,69 @@ class PollCatalogCompleteResult(RemoteCatalogResults, PollResult):
 
 
 @dataclass
+@schema_version('poll-remote-in-progress-result', 1)
 class PollInProgressResult(PollResult):
     pass
 
 
+@dataclass
+@schema_version('poll-remote-get-manifest-result', 1)
+class PollGetManifestResult(GetManifestResult, PollResult):
+    state: TaskHandlerState = field(
+        metadata=restrict_to(TaskHandlerState.Success,
+                             TaskHandlerState.Failed),
+    )
+
+    @classmethod
+    def from_result(
+        cls: Type['PollGetManifestResult'],
+        base: GetManifestResult,
+        tags: TaskTags,
+        timing: TaskTiming,
+        logs: List[LogMessage],
+    ) -> 'PollGetManifestResult':
+        return cls(
+            manifest=base.manifest,
+            logs=logs,
+            tags=tags,
+            state=timing.state,
+            start=timing.start,
+            end=timing.end,
+            elapsed=timing.elapsed,
+        )
+
+
+@dataclass
+@schema_version('poll-remote-freshness-result', 1)
+class PollFreshnessResult(RemoteFreshnessResult, PollResult):
+    state: TaskHandlerState = field(
+        metadata=restrict_to(TaskHandlerState.Success,
+                             TaskHandlerState.Failed),
+    )
+
+    @classmethod
+    def from_result(
+        cls: Type['PollFreshnessResult'],
+        base: RemoteFreshnessResult,
+        tags: TaskTags,
+        timing: TaskTiming,
+        logs: List[LogMessage],
+    ) -> 'PollFreshnessResult':
+        return cls(
+            logs=logs,
+            tags=tags,
+            state=timing.state,
+            start=timing.start,
+            end=timing.end,
+            elapsed=timing.elapsed,
+            metadata=base.metadata,
+            results=base.results,
+            elapsed_time=base.elapsed_time,
+        )
+
 # Manifest parsing types
 
+
 class ManifestStatus(StrEnum):
     Init = 'init'
     Compiling = 'compiling'
@@ -536,6 +810,7 @@ class ManifestStatus(StrEnum):
 
 
 @dataclass
+@schema_version('remote-status-result', 1)
 class LastParse(RemoteResult):
     state: ManifestStatus = ManifestStatus.Init
     logs: List[LogMessage] = field(default_factory=list)

core/dbt/contracts/selection.py

@@ -0,0 +1,22 @@
+from dataclasses import dataclass
+from dbt.dataclass_schema import dbtClassMixin
+
+from typing import List, Dict, Any, Union
+
+
+@dataclass
+class SelectorDefinition(dbtClassMixin):
+    name: str
+    definition: Union[str, Dict[str, Any]]
+    description: str = ''
+
+
+@dataclass
+class SelectorFile(dbtClassMixin):
+    selectors: List[SelectorDefinition]
+    version: int = 2
+
+
+# @dataclass
+# class SelectorCollection:
+#     packages: Dict[str, List[SelectorFile]] = field(default_factory=dict)

core/dbt/contracts/state.py

@@ -0,0 +1,18 @@
+from pathlib import Path
+from .graph.manifest import WritableManifest
+from typing import Optional
+from dbt.exceptions import IncompatibleSchemaException
+
+
+class PreviousState:
+    def __init__(self, path: Path):
+        self.path: Path = path
+        self.manifest: Optional[WritableManifest] = None
+
+        manifest_path = self.path / 'manifest.json'
+        if manifest_path.exists() and manifest_path.is_file():
+            try:
+                self.manifest = WritableManifest.read(str(manifest_path))
+            except IncompatibleSchemaException as exc:
+                exc.add_filename(str(manifest_path))
+                raise

core/dbt/contracts/util.py

@@ -1,11 +1,24 @@
 import dataclasses
-from typing import List
+import os
+from datetime import datetime
+from typing import (
+    List, Tuple, ClassVar, Type, TypeVar, Dict, Any, Optional
+)
 
-from dbt.clients.system import write_json
+from dbt.clients.system import write_json, read_json
+from dbt.exceptions import (
+    InternalException,
+    RuntimeException,
+)
+from dbt.version import __version__
+from dbt.tracking import get_invocation_id
+from dbt.dataclass_schema import dbtClassMixin
+
+SourceKey = Tuple[str, str]
 
 
 def list_str() -> List[str]:
-    """Mypy gets upset about¸stuff like:
+    """Mypy gets upset about stuff like:
 
     from dataclasses import dataclass, field
     from typing import Optional, List
@@ -42,5 +55,158 @@ class Mergeable(Replaceable):
 
 
 class Writable:
-    def write(self, path: str, omit_none: bool = False):
-        write_json(path, self.to_dict(omit_none=omit_none))  # type: ignore
+    def write(self, path: str):
+        write_json(
+            path, self.to_dict(omit_none=False)  # type: ignore
+        )
+
+
+class AdditionalPropertiesMixin:
+    """Make this class an extensible property.
+
+    The underlying class definition must include a type definition for a field
+    named '_extra' that is of type `Dict[str, Any]`.
+    """
+    ADDITIONAL_PROPERTIES = True
+
+    # This takes attributes in the dictionary that are
+    # not in the class definitions and puts them in an
+    # _extra dict in the class
+    @classmethod
+    def __pre_deserialize__(cls, data):
+        # dir() did not work because fields with
+        # metadata settings are not found
+        # The original version of this would create the
+        # object first and then update extra with the
+        # extra keys, but that won't work here, so
+        # we're copying the dict so we don't insert the
+        # _extra in the original data. This also requires
+        # that Mashumaro actually build the '_extra' field
+        cls_keys = cls._get_field_names()
+        new_dict = {}
+        for key, value in data.items():
+            if key not in cls_keys and key != '_extra':
+                if '_extra' not in new_dict:
+                    new_dict['_extra'] = {}
+                new_dict['_extra'][key] = value
+            else:
+                new_dict[key] = value
+        data = new_dict
+        data = super().__pre_deserialize__(data)
+        return data
+
+    def __post_serialize__(self, dct):
+        data = super().__post_serialize__(dct)
+        data.update(self.extra)
+        if '_extra' in data:
+            del data['_extra']
+        return data
+
+    def replace(self, **kwargs):
+        dct = self.to_dict(omit_none=False)
+        dct.update(kwargs)
+        return self.from_dict(dct)
+
+    @property
+    def extra(self):
+        return self._extra
+
+
+class Readable:
+    @classmethod
+    def read(cls, path: str):
+        try:
+            data = read_json(path)
+        except (EnvironmentError, ValueError) as exc:
+            raise RuntimeException(
+                f'Could not read {cls.__name__} at "{path}" as JSON: {exc}'
+            ) from exc
+
+        return cls.from_dict(data)  # type: ignore
+
+
+BASE_SCHEMAS_URL = 'https://schemas.getdbt.com/'
+SCHEMA_PATH = 'dbt/{name}/v{version}.json'
+
+
+@dataclasses.dataclass
+class SchemaVersion:
+    name: str
+    version: int
+
+    @property
+    def path(self) -> str:
+        return SCHEMA_PATH.format(
+            name=self.name,
+            version=self.version
+        )
+
+    def __str__(self) -> str:
+        return BASE_SCHEMAS_URL + self.path
+
+
+SCHEMA_VERSION_KEY = 'dbt_schema_version'
+
+
+METADATA_ENV_PREFIX = 'DBT_ENV_CUSTOM_ENV_'
+
+
+def get_metadata_env() -> Dict[str, str]:
+    return {
+        k[len(METADATA_ENV_PREFIX):]: v for k, v in os.environ.items()
+        if k.startswith(METADATA_ENV_PREFIX)
+    }
+
+
+@dataclasses.dataclass
+class BaseArtifactMetadata(dbtClassMixin):
+    dbt_schema_version: str
+    dbt_version: str = __version__
+    generated_at: datetime = dataclasses.field(
+        default_factory=datetime.utcnow
+    )
+    invocation_id: Optional[str] = dataclasses.field(
+        default_factory=get_invocation_id
+    )
+    env: Dict[str, str] = dataclasses.field(default_factory=get_metadata_env)
+
+
+def schema_version(name: str, version: int):
+    def inner(cls: Type[VersionedSchema]):
+        cls.dbt_schema_version = SchemaVersion(
+            name=name,
+            version=version,
+        )
+        return cls
+    return inner
+
+
+@dataclasses.dataclass
+class VersionedSchema(dbtClassMixin):
+    dbt_schema_version: ClassVar[SchemaVersion]
+
+    @classmethod
+    def json_schema(cls, embeddable: bool = False) -> Dict[str, Any]:
+        result = super().json_schema(embeddable=embeddable)
+        if not embeddable:
+            result['$id'] = str(cls.dbt_schema_version)
+        return result
+
+
+T = TypeVar('T', bound='ArtifactMixin')
+
+
+# metadata should really be a Generic[T_M] where T_M is a TypeVar bound to
+# BaseArtifactMetadata. Unfortunately this isn't possible due to a mypy issue:
+# https://github.com/python/mypy/issues/7520
+@dataclasses.dataclass(init=False)
+class ArtifactMixin(VersionedSchema, Writable, Readable):
+    metadata: BaseArtifactMetadata
+
+    @classmethod
+    def validate(cls, data):
+        super().validate(data)
+        if cls.dbt_schema_version is None:
+            raise InternalException(
+                'Cannot call from_dict with no schema version!'
+            )

core/dbt/dataclass_schema.py

@@ -0,0 +1,169 @@
+from typing import (
+    Type, ClassVar, cast,
+)
+import re
+from dataclasses import fields
+from enum import Enum
+from datetime import datetime
+from dateutil.parser import parse
+
+from hologram import JsonSchemaMixin, FieldEncoder, ValidationError
+
+# type: ignore
+from mashumaro import DataClassDictMixin
+from mashumaro.config import (
+    TO_DICT_ADD_OMIT_NONE_FLAG, BaseConfig as MashBaseConfig
+)
+from mashumaro.types import SerializableType, SerializationStrategy
+
+
+class DateTimeSerialization(SerializationStrategy):
+    def serialize(self, value):
+        out = value.isoformat()
+        # Assume UTC if timezone is missing
+        if value.tzinfo is None:
+            out = out + "Z"
+        return out
+
+    def deserialize(self, value):
+        return (
+            value if isinstance(value, datetime) else parse(cast(str, value))
+        )
+
+
+# This class pulls in both JsonSchemaMixin from Hologram and
+# DataClassDictMixin from our fork of Mashumaro. The 'to_dict'
+# and 'from_dict' methods come from Mashumaro. Building
+# jsonschemas for every class and the 'validate' method
+# come from Hologram.
+class dbtClassMixin(DataClassDictMixin, JsonSchemaMixin):
+    """Mixin which adds methods to generate a JSON schema and
+       convert to and from JSON encodable dicts with validation
+       against the schema
+    """
+
+    class Config(MashBaseConfig):
+        code_generation_options = [
+            TO_DICT_ADD_OMIT_NONE_FLAG,
+        ]
+        serialization_strategy = {
+            datetime: DateTimeSerialization(),
+        }
+
+    _hyphenated: ClassVar[bool] = False
+    ADDITIONAL_PROPERTIES: ClassVar[bool] = False
+
+    # This is called by the mashumaro to_dict in order to handle
+    # nested classes.
+    # Munges the dict that's returned.
+    def __post_serialize__(self, dct):
+        if self._hyphenated:
+            new_dict = {}
+            for key in dct:
+                if '_' in key:
+                    new_key = key.replace('_', '-')
+                    new_dict[new_key] = dct[key]
+                else:
+                    new_dict[key] = dct[key]
+            dct = new_dict
+
+        return dct
+
+    # This is called by the mashumaro _from_dict method, before
+    # performing the conversion to a dict
+    @classmethod
+    def __pre_deserialize__(cls, data):
+        # `data` might not be a dict, e.g. for `query_comment`, which accepts
+        # a dict or a string; only snake-case for dict values.
+        if cls._hyphenated and isinstance(data, dict):
+            new_dict = {}
+            for key in data:
+                if '-' in key:
+                    new_key = key.replace('-', '_')
+                    new_dict[new_key] = data[key]
+                else:
+                    new_dict[key] = data[key]
+            data = new_dict
+        return data
+
+    # This is used in the hologram._encode_field method, which calls
+    # a 'to_dict' method which does not have the same parameters in
+    # hologram and in mashumaro.
+    def _local_to_dict(self, **kwargs):
+        args = {}
+        if 'omit_none' in kwargs:
+            args['omit_none'] = kwargs['omit_none']
+        return self.to_dict(**args)
+
+
+class ValidatedStringMixin(str, SerializableType):
+    ValidationRegex = ''
+
+    @classmethod
+    def _deserialize(cls, value: str) -> 'ValidatedStringMixin':
+        cls.validate(value)
+        return ValidatedStringMixin(value)
+
+    def _serialize(self) -> str:
+        return str(self)
+
+    @classmethod
+    def validate(cls, value):
+        res = re.match(cls.ValidationRegex, value)
+
+        if res is None:
+            raise ValidationError(f"Invalid value: {value}")  # TODO
+
+
+# These classes must be in this order or it doesn't work
+class StrEnum(str, SerializableType, Enum):
+    def __str__(self):
+        return self.value
+
+    # https://docs.python.org/3.6/library/enum.html#using-automatic-values
+    def _generate_next_value_(name, *_):
+        return name
+
+    def _serialize(self) -> str:
+        return self.value
+
+    @classmethod
+    def _deserialize(cls, value: str):
+        return cls(value)
+
+
+class HyphenatedDbtClassMixin(dbtClassMixin):
+    # used by from_dict/to_dict
+    _hyphenated: ClassVar[bool] = True
+
+    # used by jsonschema validation, _get_fields
+    @classmethod
+    def field_mapping(cls):
+        result = {}
+        for field in fields(cls):
+            skip = field.metadata.get("preserve_underscore")
+            if skip:
+                continue
+
+            if "_" in field.name:
+                result[field.name] = field.name.replace("_", "-")
+        return result
+
+
+class ExtensibleDbtClassMixin(dbtClassMixin):
+    ADDITIONAL_PROPERTIES = True
+
+
+# This is used by Hologram in jsonschema validation
+def register_pattern(base_type: Type, pattern: str) -> None:
+    """base_type should be a typing.NewType that should always have the given
+    regex pattern. That means that its underlying type ('__supertype__') had
+    better be a str!
+    """
+
+    class PatternEncoder(FieldEncoder):
+        @property
+        def json_schema(self):
+            return {"type": "string", "pattern": pattern}
+
+    dbtClassMixin.register_field_encoders({base_type: PatternEncoder()})

core/dbt/deprecations.py

@@ -1,9 +1,9 @@
 from typing import Optional, Set, List, Dict, ClassVar
 
-import dbt.links
 import dbt.exceptions
-import dbt.flags
-from dbt.ui import printer
+from dbt import ui
+
+import dbt.tracking
 
 
 class DBTDeprecation:
@@ -18,6 +18,12 @@ class DBTDeprecation:
             'name not implemented for {}'.format(self)
         )
 
+    def track_deprecation_warn(self) -> None:
+        if dbt.tracking.active_user is not None:
+            dbt.tracking.track_deprecation_warn({
+                "deprecation_name": self.name
+            })
+
     @property
     def description(self) -> str:
         if self._description is not None:
@@ -29,26 +35,25 @@ class DBTDeprecation:
     def show(self, *args, **kwargs) -> None:
         if self.name not in active_deprecations:
             desc = self.description.format(**kwargs)
-            msg = printer.line_wrap_message(
+            msg = ui.line_wrap_message(
                 desc, prefix='* Deprecation Warning: '
             )
             dbt.exceptions.warn_or_error(msg)
+            self.track_deprecation_warn()
             active_deprecations.add(self.name)
 
 
-class GenerateSchemaNameSingleArgDeprecated(DBTDeprecation):
-    _name = 'generate-schema-name-single-arg'
-
+class DispatchPackagesDeprecation(DBTDeprecation):
+    _name = 'dispatch-packages'
     _description = '''\
-    As of dbt v0.14.0, the `generate_schema_name` macro accepts a second "node"
-    argument. The one-argument form of `generate_schema_name` is deprecated,
-    and will become unsupported in a future release.
-
+    The "packages" argument of adapter.dispatch() has been deprecated.
+    Use the "macro_namespace" argument instead.
 
+    Raised during dispatch for: {macro_name}
 
     For more information, see:
 
-    https://docs.getdbt.com/v0.14/docs/upgrading-to-014
+    https://docs.getdbt.com/reference/dbt-jinja-functions/dispatch
     '''
 
 
@@ -108,23 +113,29 @@ class ModelsKeyNonModelDeprecation(DBTDeprecation):
     '''
 
 
-class BigQueryPartitionByStringDeprecation(DBTDeprecation):
-    _name = 'bq-partition-by-string'
-
-    _description = '''
-    As of dbt v0.16.0, the `partition_by` config in BigQuery accepts a
-    dictionary containing `field` and `data_type`.
+class ExecuteMacrosReleaseDeprecation(DBTDeprecation):
+    _name = 'execute-macro-release'
+    _description = '''\
+    The "release" argument to execute_macro is now ignored, and will be removed
+    in a future relase of dbt. At that time, providing a `release` argument
+    will result in an error.
+    '''
 
 
-    - Provided partition_by: {raw_partition_by}
+class AdapterMacroDeprecation(DBTDeprecation):
+    _name = 'adapter-macro'
+    _description = '''\
+    The "adapter_macro" macro has been deprecated. Instead, use the
+    `adapter.dispatch` method to find a macro and call the result.
+    adapter_macro was called for: {macro_name}
+    '''
 
 
-    - dbt inferred: {inferred_partition_by}
-
-
-
-    For more information, see:
-    https://docs.getdbt.com/docs/upgrading-to-0-16-0
+class PackageRedirectDeprecation(DBTDeprecation):
+    _name = 'package-redirect'
+    _description = '''\
+    The `{old_name}` package is deprecated in favor of `{new_name}`. Please update
+    your `packages.yml` configuration to use `{new_name}` instead.
     '''
 
 
@@ -166,12 +177,14 @@ def warn(name, *args, **kwargs):
 active_deprecations: Set[str] = set()
 
 deprecations_list: List[DBTDeprecation] = [
-    GenerateSchemaNameSingleArgDeprecated(),
+    DispatchPackagesDeprecation(),
     MaterializationReturnDeprecation(),
     NotADictionaryDeprecation(),
     ColumnQuotingDeprecation(),
     ModelsKeyNonModelDeprecation(),
-    BigQueryPartitionByStringDeprecation(),
+    ExecuteMacrosReleaseDeprecation(),
+    AdapterMacroDeprecation(),
+    PackageRedirectDeprecation()
 ]
 
 deprecations: Dict[str, DBTDeprecation] = {

core/dbt/deps/base.py

@@ -93,6 +93,9 @@ class PinnedPackage(BasePackage):
         dest_dirname = self.get_project_name(project, renderer)
         return os.path.join(project.modules_path, dest_dirname)
 
+    def get_subdirectory(self):
+        return None
+
 
 SomePinned = TypeVar('SomePinned', bound=PinnedPackage)
 SomeUnpinned = TypeVar('SomeUnpinned', bound='UnpinnedPackage')

core/dbt/deps/git.py

@@ -1,6 +1,6 @@
 import os
 import hashlib
-from typing import List
+from typing import List, Optional
 
 from dbt.clients import git, system
 from dbt.config import Project
@@ -13,7 +13,7 @@ from dbt.exceptions import (
     ExecutableError, warn_or_error, raise_dependency_error
 )
 from dbt.logger import GLOBAL_LOGGER as logger
-from dbt.ui import printer
+from dbt import ui
 
 PIN_PACKAGE_URL = 'https://docs.getdbt.com/docs/package-management#section-specifying-package-versions' # noqa
 
@@ -37,18 +37,37 @@ class GitPackageMixin:
 
 class GitPinnedPackage(GitPackageMixin, PinnedPackage):
     def __init__(
-        self, git: str, revision: str, warn_unpinned: bool = True
+        self,
+        git: str,
+        revision: str,
+        warn_unpinned: bool = True,
+        subdirectory: Optional[str] = None,
     ) -> None:
         super().__init__(git)
         self.revision = revision
         self.warn_unpinned = warn_unpinned
+        self.subdirectory = subdirectory
         self._checkout_name = md5sum(self.git)
 
     def get_version(self):
         return self.revision
 
+    def get_subdirectory(self):
+        return self.subdirectory
+
     def nice_version_name(self):
-        return 'revision {}'.format(self.revision)
+        if self.revision == 'HEAD':
+            return 'HEAD (default revision)'
+        else:
+            return 'revision {}'.format(self.revision)
+
+    def unpinned_msg(self):
+        if self.revision == 'HEAD':
+            return 'not pinned, using HEAD (default branch)'
+        elif self.revision in ('main', 'master'):
+            return f'pinned to the "{self.revision}" branch'
+        else:
+            return None
 
     def _checkout(self):
         """Performs a shallow clone of the repository into the downloads
@@ -57,8 +76,8 @@ class GitPinnedPackage(GitPackageMixin, PinnedPackage):
         the path to the checked out directory."""
         try:
             dir_ = git.clone_and_checkout(
-                self.git, get_downloads_path(), branch=self.revision,
-                dirname=self._checkout_name
+                self.git, get_downloads_path(), revision=self.revision,
+                dirname=self._checkout_name, subdirectory=self.subdirectory
             )
         except ExecutableError as exc:
             if exc.cmd and exc.cmd[0] == 'git':
@@ -72,12 +91,13 @@ class GitPinnedPackage(GitPackageMixin, PinnedPackage):
 
     def _fetch_metadata(self, project, renderer) -> ProjectPackageMetadata:
         path = self._checkout()
-        if self.revision == 'master' and self.warn_unpinned:
+
+        if self.unpinned_msg() and self.warn_unpinned:
             warn_or_error(
-                'The git package "{}" is not pinned.\n\tThis can introduce '
+                'The git package "{}" \n\tis {}.\n\tThis can introduce '
                 'breaking changes into your project without warning!\n\nSee {}'
-                .format(self.git, PIN_PACKAGE_URL),
-                log_fmt=printer.yellow('WARNING: {}')
+                .format(self.git, self.unpinned_msg(), PIN_PACKAGE_URL),
+                log_fmt=ui.yellow('WARNING: {}')
             )
         loaded = Project.from_project_root(path, renderer)
         return ProjectPackageMetadata.from_project(loaded)
@@ -95,22 +115,27 @@ class GitPinnedPackage(GitPackageMixin, PinnedPackage):
 
 class GitUnpinnedPackage(GitPackageMixin, UnpinnedPackage[GitPinnedPackage]):
     def __init__(
-        self, git: str, revisions: List[str], warn_unpinned: bool = True
+        self,
+        git: str,
+        revisions: List[str],
+        warn_unpinned: bool = True,
+        subdirectory: Optional[str] = None,
     ) -> None:
         super().__init__(git)
         self.revisions = revisions
         self.warn_unpinned = warn_unpinned
+        self.subdirectory = subdirectory
 
     @classmethod
     def from_contract(
         cls, contract: GitPackage
     ) -> 'GitUnpinnedPackage':
-        revisions = [contract.revision] if contract.revision else []
+        revisions = contract.get_revisions()
 
         # we want to map None -> True
         warn_unpinned = contract.warn_unpinned is not False
         return cls(git=contract.git, revisions=revisions,
-                   warn_unpinned=warn_unpinned)
+                   warn_unpinned=warn_unpinned, subdirectory=contract.subdirectory)
 
     def all_names(self) -> List[str]:
         if self.git.endswith('.git'):
@@ -128,12 +153,13 @@ class GitUnpinnedPackage(GitPackageMixin, UnpinnedPackage[GitPinnedPackage]):
             git=self.git,
             revisions=self.revisions + other.revisions,
             warn_unpinned=warn_unpinned,
+            subdirectory=self.subdirectory,
         )
 
     def resolved(self) -> GitPinnedPackage:
         requested = set(self.revisions)
         if len(requested) == 0:
-            requested = {'master'}
+            requested = {'HEAD'}
         elif len(requested) > 1:
             raise_dependency_error(
                 'git dependencies should contain exactly one version. '
@@ -141,5 +167,5 @@ class GitUnpinnedPackage(GitPackageMixin, UnpinnedPackage[GitPinnedPackage]):
 
         return GitPinnedPackage(
             git=self.git, revision=requested.pop(),
-            warn_unpinned=self.warn_unpinned
+            warn_unpinned=self.warn_unpinned, subdirectory=self.subdirectory
         )

core/dbt/deps/registry.py

@@ -30,9 +30,13 @@ class RegistryPackageMixin:
 
 
 class RegistryPinnedPackage(RegistryPackageMixin, PinnedPackage):
-    def __init__(self, package: str, version: str) -> None:
+    def __init__(self,
+                 package: str,
+                 version: str,
+                 version_latest: str) -> None:
         super().__init__(package)
         self.version = version
+        self.version_latest = version_latest
 
     @property
     def name(self):
@@ -44,6 +48,9 @@ class RegistryPinnedPackage(RegistryPackageMixin, PinnedPackage):
     def get_version(self):
         return self.version
 
+    def get_version_latest(self):
+        return self.version_latest
+
     def nice_version_name(self):
         return 'version {}'.format(self.version)
 
@@ -61,7 +68,7 @@ class RegistryPinnedPackage(RegistryPackageMixin, PinnedPackage):
         system.make_directory(os.path.dirname(tar_path))
 
         download_url = metadata.downloads.tarball
-        system.download(download_url, tar_path)
+        system.download_with_retries(download_url, tar_path)
         deps_path = project.modules_path
         package_name = self.get_project_name(project, renderer)
         system.untar_package(tar_path, deps_path, package_name)
@@ -71,10 +78,14 @@ class RegistryUnpinnedPackage(
     RegistryPackageMixin, UnpinnedPackage[RegistryPinnedPackage]
 ):
     def __init__(
-        self, package: str, versions: List[semver.VersionSpecifier]
+        self,
+        package: str,
+        versions: List[semver.VersionSpecifier],
+        install_prerelease: bool
     ) -> None:
         super().__init__(package)
         self.versions = versions
+        self.install_prerelease = install_prerelease
 
     def _check_in_index(self):
         index = registry.index_cached()
@@ -85,21 +96,24 @@ class RegistryUnpinnedPackage(
     def from_contract(
         cls, contract: RegistryPackage
     ) -> 'RegistryUnpinnedPackage':
-        raw_version = contract.version
-        if isinstance(raw_version, str):
-            raw_version = [raw_version]
+        raw_version = contract.get_versions()
 
         versions = [
             semver.VersionSpecifier.from_version_string(v)
             for v in raw_version
         ]
-        return cls(package=contract.package, versions=versions)
+        return cls(
+            package=contract.package,
+            versions=versions,
+            install_prerelease=contract.install_prerelease
+        )
 
     def incorporate(
         self, other: 'RegistryUnpinnedPackage'
     ) -> 'RegistryUnpinnedPackage':
         return RegistryUnpinnedPackage(
             package=self.package,
+            install_prerelease=self.install_prerelease,
             versions=self.versions + other.versions,
         )
 
@@ -113,12 +127,18 @@ class RegistryUnpinnedPackage(
             raise DependencyException(new_msg) from e
 
         available = registry.get_available_versions(self.package)
+        installable = semver.filter_installable(
+            available,
+            self.install_prerelease
+        )
+        available_latest = installable[-1]
 
         # for now, pick a version and then recurse. later on,
         # we'll probably want to traverse multiple options
         # so we can match packages. not going to make a difference
         # right now.
-        target = semver.resolve_to_specific_version(range_, available)
+        target = semver.resolve_to_specific_version(range_, installable)
         if not target:
-            package_version_not_found(self.package, range_, available)
-        return RegistryPinnedPackage(package=self.package, version=target)
+            package_version_not_found(self.package, range_, installable)
+        return RegistryPinnedPackage(package=self.package, version=target,
+                                     version_latest=available_latest)

core/dbt/deps/resolver.py

@@ -4,7 +4,8 @@ from typing import Dict, List, NoReturn, Union, Type, Iterator, Set
 from dbt.exceptions import raise_dependency_error, InternalException
 
 from dbt.context.target import generate_target_context
-from dbt.config import Project, ConfigRenderer, RuntimeConfig
+from dbt.config import Project, RuntimeConfig
+from dbt.config.renderer import DbtProjectYamlRenderer
 from dbt.deps.base import BasePackage, PinnedPackage, UnpinnedPackage
 from dbt.deps.local import LocalUnpinnedPackage
 from dbt.deps.git import GitUnpinnedPackage
@@ -97,7 +98,9 @@ class PackageListing:
 
 
 def _check_for_duplicate_project_names(
-    final_deps: List[PinnedPackage], config: Project, renderer: ConfigRenderer
+    final_deps: List[PinnedPackage],
+    config: Project,
+    renderer: DbtProjectYamlRenderer,
 ):
     seen: Set[str] = set()
     for package in final_deps:
@@ -123,7 +126,8 @@ def resolve_packages(
     pending = PackageListing.from_contracts(packages)
     final = PackageListing()
 
-    renderer = ConfigRenderer(generate_target_context(config, config.cli_vars))
+    ctx = generate_target_context(config, config.cli_vars)
+    renderer = DbtProjectYamlRenderer(ctx)
 
     while pending:
         next_pending = PackageListing()

core/dbt/exceptions.py

@@ -1,19 +1,20 @@
 import builtins
 import functools
-from typing import NoReturn, Optional
+from typing import NoReturn, Optional, Mapping, Any
 
 from dbt.logger import GLOBAL_LOGGER as logger
 from dbt.node_types import NodeType
-import dbt.flags
+from dbt import flags
+from dbt.ui import line_wrap_message
 
-import hologram
+import dbt.dataclass_schema
 
 
 def validator_error_message(exc):
-    """Given a hologram.ValidationError (which is basically a
+    """Given a dbt.dataclass_schema.ValidationError (which is basically a
     jsonschema.ValidationError), return the relevant parts as a string
     """
-    if not isinstance(exc, hologram.ValidationError):
+    if not isinstance(exc, dbt.dataclass_schema.ValidationError):
         return str(exc)
     path = "[%s]" % "][".join(map(repr, exc.relative_path))
     return 'at path {}: {}'.format(path, exc.message)
@@ -53,6 +54,12 @@ class RuntimeException(RuntimeError, Exception):
         self.node = node
         self.msg = msg
 
+    def add_node(self, node=None):
+        if node is not None and node is not self.node:
+            if self.node is not None:
+                self.stack.append(self.node)
+            self.node = node
+
     @property
     def type(self):
         return 'Runtime'
@@ -125,7 +132,7 @@ class RuntimeException(RuntimeError, Exception):
         result.update({
             'raw_sql': self.node.raw_sql,
             # the node isn't always compiled, but if it is, include that!
-            'compiled_sql': getattr(self.node, 'injected_sql', None),
+            'compiled_sql': getattr(self.node, 'compiled_sql', None),
         })
         return result
 
@@ -250,6 +257,47 @@ class JSONValidationException(ValidationException):
         return (JSONValidationException, (self.typename, self.errors))
 
 
+class IncompatibleSchemaException(RuntimeException):
+    def __init__(self, expected: str, found: Optional[str]):
+        self.expected = expected
+        self.found = found
+        self.filename = 'input file'
+
+        super().__init__(self.get_message())
+
+    def add_filename(self, filename: str):
+        self.filename = filename
+        self.msg = self.get_message()
+
+    def get_message(self) -> str:
+        found_str = 'nothing'
+        if self.found is not None:
+            found_str = f'"{self.found}"'
+
+        msg = (
+            f'Expected a schema version of "{self.expected}" in '
+            f'{self.filename}, but found {found_str}. Are you running with a '
+            f'different version of dbt?'
+        )
+        return msg
+
+    CODE = 10014
+    MESSAGE = "Incompatible Schema"
+
+
+class JinjaRenderingException(CompilationException):
+    pass
+
+
+class UndefinedMacroException(CompilationException):
+
+    def __str__(self, prefix='! ') -> str:
+        msg = super().__str__(prefix)
+        return f'{msg}. This can happen when calling a macro that does ' \
+            'not exist. Check for typos and/or install package dependencies ' \
+            'with "dbt deps".'
+
+
 class UnknownAsyncIDException(Exception):
     CODE = 10012
     MESSAGE = 'RPC server got an unknown async ID'
@@ -275,16 +323,43 @@ class DbtConfigError(RuntimeException):
     CODE = 10007
     MESSAGE = "DBT Configuration Error"
 
-    def __init__(self, message, project=None, result_type='invalid_project'):
+    def __init__(
+        self, message, project=None, result_type='invalid_project', path=None
+    ):
         self.project = project
         super().__init__(message)
         self.result_type = result_type
+        self.path = path
+
+    def __str__(self, prefix='! ') -> str:
+        msg = super().__str__(prefix)
+        if self.path is None:
+            return msg
+        else:
+            return f'{msg}\n\nError encountered in {self.path}'
+
+
+class FailFastException(RuntimeException):
+    CODE = 10013
+    MESSAGE = 'FailFast Error'
+
+    def __init__(self, message, result=None, node=None):
+        super().__init__(msg=message, node=node)
+        self.result = result
+
+    @property
+    def type(self):
+        return 'FailFast'
 
 
 class DbtProjectError(DbtConfigError):
     pass
 
 
+class DbtSelectorsError(DbtConfigError):
+    pass
+
+
 class DbtProfileError(DbtConfigError):
     pass
 
@@ -359,6 +434,12 @@ class InvalidConnectionException(RuntimeException):
         )
 
 
+class InvalidSelectorException(RuntimeException):
+    def __init__(self, name: str):
+        self.name = name
+        super().__init__(name)
+
+
 def raise_compiler_error(msg, node=None) -> NoReturn:
     raise CompilationException(msg, node)
 
@@ -443,8 +524,10 @@ def doc_target_not_found(
     raise_compiler_error(msg, model)
 
 
-def _get_target_failure_msg(model, target_model_name, target_model_package,
-                            include_path, reason):
+def _get_target_failure_msg(
+    model, target_name: str, target_model_package: Optional[str],
+    include_path: bool, reason: str, target_kind: str
+) -> str:
     target_package_string = ''
     if target_model_package is not None:
         target_package_string = "in package '{}' ".format(target_model_package)
@@ -453,52 +536,73 @@ def _get_target_failure_msg(model, target_model_name, target_model_package,
     if include_path:
         source_path_string = ' ({})'.format(model.original_file_path)
 
-    return "{} '{}'{} depends on a node named '{}' {}which {}".format(
+    return "{} '{}'{} depends on a {} named '{}' {}which {}".format(
         model.resource_type.title(),
         model.unique_id,
         source_path_string,
-        target_model_name,
+        target_kind,
+        target_name,
         target_package_string,
         reason
     )
 
 
-def get_target_disabled_msg(model, target_model_name, target_model_package):
-    return _get_target_failure_msg(model, target_model_name,
-                                   target_model_package, include_path=True,
-                                   reason='is disabled')
+def get_target_not_found_or_disabled_msg(
+    model, target_model_name: str, target_model_package: Optional[str],
+    disabled: Optional[bool] = None,
+) -> str:
+    if disabled is None:
+        reason = 'was not found or is disabled'
+    elif disabled is True:
+        reason = 'is disabled'
+    else:
+        reason = 'was not found'
+    return _get_target_failure_msg(
+        model, target_model_name, target_model_package, include_path=True,
+        reason=reason, target_kind='node'
+    )
 
 
-def get_target_not_found_msg(model, target_model_name, target_model_package):
-    return _get_target_failure_msg(model, target_model_name,
-                                   target_model_package, include_path=True,
-                                   reason='was not found')
-
-
-def get_target_not_found_or_disabled_msg(model, target_model_name,
-                                         target_model_package):
-    return _get_target_failure_msg(model, target_model_name,
-                                   target_model_package, include_path=False,
-                                   reason='was not found or is disabled')
-
-
-def ref_target_not_found(model, target_model_name, target_model_package):
-    msg = get_target_not_found_or_disabled_msg(model, target_model_name,
-                                               target_model_package)
+def ref_target_not_found(
+    model,
+    target_model_name: str,
+    target_model_package: Optional[str],
+    disabled: Optional[bool] = None,
+) -> NoReturn:
+    msg = get_target_not_found_or_disabled_msg(
+        model, target_model_name, target_model_package, disabled
+    )
     raise_compiler_error(msg, model)
 
 
-def source_disabled_message(model, target_name, target_table_name):
-    return ("{} '{}' ({}) depends on source '{}.{}' which was not found"
-            .format(model.resource_type.title(),
-                    model.unique_id,
-                    model.original_file_path,
-                    target_name,
-                    target_table_name))
+def get_source_not_found_or_disabled_msg(
+    model,
+    target_name: str,
+    target_table_name: str,
+    disabled: Optional[bool] = None,
+) -> str:
+    full_name = f'{target_name}.{target_table_name}'
+    if disabled is None:
+        reason = 'was not found or is disabled'
+    elif disabled is True:
+        reason = 'is disabled'
+    else:
+        reason = 'was not found'
+    return _get_target_failure_msg(
+        model, full_name, None, include_path=True,
+        reason=reason, target_kind='source'
+    )
 
 
-def source_target_not_found(model, target_name, target_table_name) -> NoReturn:
-    msg = source_disabled_message(model, target_name, target_table_name)
+def source_target_not_found(
+    model,
+    target_name: str,
+    target_table_name: str,
+    disabled: Optional[bool] = None
+) -> NoReturn:
+    msg = get_source_not_found_or_disabled_msg(
+        model, target_name, target_table_name, disabled
+    )
     raise_compiler_error(msg, model)
 
 
@@ -606,11 +710,11 @@ def system_error(operation_name):
     raise_compiler_error(
         "dbt encountered an error when attempting to {}. "
         "If this error persists, please create an issue at: \n\n"
-        "https://github.com/fishtown-analytics/dbt"
+        "https://github.com/dbt-labs/dbt"
         .format(operation_name))
 
 
-class RegistryException(Exception):
+class ConnectionException(Exception):
     pass
 
 
@@ -644,10 +748,10 @@ def approximate_relation_match(target, relation):
 
 
 def raise_duplicate_macro_name(node_1, node_2, namespace) -> NoReturn:
-    duped_name = node_1.namespace
+    duped_name = node_1.name
     if node_1.package_name != node_2.package_name:
         extra = (
-            ' ({} and {} are both in the {} namespace)'
+            ' ("{}" and "{}" are both in the "{}" namespace)'
             .format(node_1.package_name, node_2.package_name, namespace)
         )
     else:
@@ -672,11 +776,14 @@ def raise_duplicate_resource_name(node_1, node_2):
     if node_1.resource_type in NodeType.refable():
         get_func = 'ref("{}")'.format(duped_name)
     elif node_1.resource_type == NodeType.Source:
-        get_func = 'source("{}", "{}")'.format(node_1.source_name, duped_name)
+        duped_name = node_1.get_full_source_name()
+        get_func = node_1.get_source_representation()
     elif node_1.resource_type == NodeType.Documentation:
         get_func = 'doc("{}")'.format(duped_name)
     elif node_1.resource_type == NodeType.Test and 'schema' in node_1.tags:
         return
+    else:
+        get_func = '"{}"'.format(duped_name)
 
     raise_compiler_error(
         'dbt found two resources with the name "{}". Since these resources '
@@ -689,14 +796,15 @@ def raise_duplicate_resource_name(node_1, node_2):
             node_2.unique_id, node_2.original_file_path))
 
 
-def raise_ambiguous_alias(node_1, node_2):
-    duped_name = "{}.{}".format(node_1.schema, node_1.alias)
+def raise_ambiguous_alias(node_1, node_2, duped_name=None):
+    if duped_name is None:
+        duped_name = f"{node_1.database}.{node_1.schema}.{node_1.alias}"
 
     raise_compiler_error(
         'dbt found two resources with the database representation "{}".\ndbt '
         'cannot create two resources with identical database representations. '
-        'To fix this,\nchange the "schema" or "alias" configuration of one of '
-        'these resources:\n- {} ({})\n- {} ({})'.format(
+        'To fix this,\nchange the configuration of one of these resources:'
+        '\n- {} ({})\n- {} ({})'.format(
             duped_name,
             node_1.unique_id, node_1.original_file_path,
             node_2.unique_id, node_2.original_file_path))
@@ -733,26 +841,62 @@ def raise_patch_targets_not_found(patches):
     )
 
 
-def raise_duplicate_patch_name(patch_1, patch_2):
+def _fix_dupe_msg(path_1: str, path_2: str, name: str, type_name: str) -> str:
+    if path_1 == path_2:
+        return (
+            f'remove one of the {type_name} entries for {name} in this file:\n'
+            f' - {path_1!s}\n'
+        )
+    else:
+        return (
+            f'remove the {type_name} entry for {name} in one of these files:\n'
+            f' - {path_1!s}\n{path_2!s}'
+        )
+
+
+def raise_duplicate_patch_name(patch_1, existing_patch_path):
     name = patch_1.name
+    fix = _fix_dupe_msg(
+        patch_1.original_file_path,
+        existing_patch_path,
+        name,
+        'resource',
+    )
     raise_compiler_error(
         f'dbt found two schema.yml entries for the same resource named '
         f'{name}. Resources and their associated columns may only be '
-        f'described a single time. To fix this, remove the resource entry '
-        f'for {name} in one of these files:\n  - '
-        f'{patch_1.original_file_path}\n  - {patch_2.original_file_path}'
+        f'described a single time. To fix this, {fix}'
     )
 
 
-def raise_duplicate_macro_patch_name(patch_1, patch_2):
+def raise_duplicate_macro_patch_name(patch_1, existing_patch_path):
     package_name = patch_1.package_name
     name = patch_1.name
+    fix = _fix_dupe_msg(
+        patch_1.original_file_path,
+        existing_patch_path,
+        name,
+        'macros'
+    )
     raise_compiler_error(
         f'dbt found two schema.yml entries for the same macro in package '
         f'{package_name} named {name}. Macros may only be described a single '
-        f'time. To fix this, remove the macros entry for {name} in one '
-        f'of these files:'
-        f'\n  - {patch_1.original_file_path}\n  - {patch_2.original_file_path}'
+        f'time. To fix this, {fix}'
+    )
+
+
+def raise_duplicate_source_patch_name(patch_1, patch_2):
+    name = f'{patch_1.overrides}.{patch_1.name}'
+    fix = _fix_dupe_msg(
+        patch_1.path,
+        patch_2.path,
+        name,
+        'sources',
+    )
+    raise_compiler_error(
+        f'dbt found two schema.yml entries for the same source named '
+        f'{patch_1.name} in package {patch_1.overrides}. Sources may only be '
+        f'overridden a single time. To fix this, {fix}'
     )
 
 
@@ -772,14 +916,46 @@ def raise_unrecognized_credentials_type(typename, supported_types):
     )
 
 
+def raise_invalid_patch(
+    node, patch_section: str, patch_path: str,
+) -> NoReturn:
+    msg = line_wrap_message(
+        f'''\
+        '{node.name}' is a {node.resource_type} node, but it is
+        specified in the {patch_section} section of
+        {patch_path}.
+
+
+
+        To fix this error, place the `{node.name}`
+        specification under the {node.resource_type.pluralize()} key instead.
+        '''
+    )
+    raise_compiler_error(msg, node)
+
+
 def raise_not_implemented(msg):
     raise NotImplementedException(
         "ERROR: {}"
         .format(msg))
 
 
+def raise_duplicate_alias(
+    kwargs: Mapping[str, Any], aliases: Mapping[str, str], canonical_key: str
+) -> NoReturn:
+    # dupe found: go through the dict so we can have a nice-ish error
+    key_names = ', '.join(
+        "{}".format(k) for k in kwargs if
+        aliases.get(k) == canonical_key
+    )
+
+    raise AliasException(
+        f'Got duplicate keys: ({key_names}) all map to "{canonical_key}"'
+    )
+
+
 def warn_or_error(msg, node=None, log_fmt=None):
-    if dbt.flags.WARN_ERROR:
+    if flags.WARN_ERROR:
         raise_compiler_error(msg, node)
     else:
         if log_fmt is not None:
@@ -788,7 +964,7 @@ def warn_or_error(msg, node=None, log_fmt=None):
 
 
 def warn_or_raise(exc, log_fmt=None):
-    if dbt.flags.WARN_ERROR:
+    if flags.WARN_ERROR:
         raise exc
     else:
         msg = str(exc)
@@ -800,7 +976,8 @@ def warn_or_raise(exc, log_fmt=None):
 def warn(msg, node=None):
     # there's no reason to expose log_fmt to macros - it's only useful for
     # handling colors
-    return warn_or_error(msg, node=node)
+    warn_or_error(msg, node=node)
+    return ""
 
 
 # Update this when a new function should be added to the
@@ -836,8 +1013,7 @@ def wrapper(model):
             try:
                 return func(*args, **kwargs)
             except RuntimeException as exc:
-                if exc.node is None:
-                    exc.node = model
+                exc.add_node(model)
                 raise exc
         return inner
     return wrap

core/dbt/flags.py

@@ -1,6 +1,11 @@
 import os
 import multiprocessing
+if os.name != 'nt':
+    # https://bugs.python.org/issue41567
+    import multiprocessing.popen_spawn_posix  # type: ignore
+from pathlib import Path
 from typing import Optional
+
 # initially all flags are set to None, the on-load call of reset() will set
 # them for their first time.
 STRICT_MODE = None
@@ -8,8 +13,11 @@ FULL_REFRESH = None
 USE_CACHE = None
 WARN_ERROR = None
 TEST_NEW_PARSER = None
+USE_EXPERIMENTAL_PARSER = None
 WRITE_JSON = None
 PARTIAL_PARSE = None
+USE_COLORS = None
+STORE_FAILURES = None
 
 
 def env_set_truthy(key: str) -> Optional[str]:
@@ -22,19 +30,24 @@ def env_set_truthy(key: str) -> Optional[str]:
     return value
 
 
+def env_set_path(key: str) -> Optional[Path]:
+    value = os.getenv(key)
+    if value is None:
+        return value
+    else:
+        return Path(value)
+
+
 SINGLE_THREADED_WEBSERVER = env_set_truthy('DBT_SINGLE_THREADED_WEBSERVER')
 SINGLE_THREADED_HANDLER = env_set_truthy('DBT_SINGLE_THREADED_HANDLER')
 MACRO_DEBUGGING = env_set_truthy('DBT_MACRO_DEBUGGING')
+DEFER_MODE = env_set_truthy('DBT_DEFER_TO_STATE')
+ARTIFACT_STATE_PATH = env_set_path('DBT_ARTIFACT_STATE_PATH')
 
 
 def _get_context():
-    if os.name == 'posix' and os.uname().sysname.lower() != 'darwin':
-        # on linux fork is available and it's fast
-        return multiprocessing.get_context('fork')
-    else:
-        # on windows, spawn is the only choice.
-        # On osx, fork is buggy: https://bugs.python.org/issue33725
-        return multiprocessing.get_context('spawn')
+    # TODO: change this back to use fork() on linux when we have made that safe
+    return multiprocessing.get_context('spawn')
 
 
 MP_CONTEXT = _get_context()
@@ -42,21 +55,26 @@ MP_CONTEXT = _get_context()
 
 def reset():
     global STRICT_MODE, FULL_REFRESH, USE_CACHE, WARN_ERROR, TEST_NEW_PARSER, \
-        WRITE_JSON, PARTIAL_PARSE, MP_CONTEXT
+        USE_EXPERIMENTAL_PARSER, WRITE_JSON, PARTIAL_PARSE, MP_CONTEXT, USE_COLORS, \
+        STORE_FAILURES
 
     STRICT_MODE = False
     FULL_REFRESH = False
     USE_CACHE = True
     WARN_ERROR = False
     TEST_NEW_PARSER = False
+    USE_EXPERIMENTAL_PARSER = False
     WRITE_JSON = True
     PARTIAL_PARSE = False
     MP_CONTEXT = _get_context()
+    USE_COLORS = True
+    STORE_FAILURES = False
 
 
 def set_from_args(args):
     global STRICT_MODE, FULL_REFRESH, USE_CACHE, WARN_ERROR, TEST_NEW_PARSER, \
-        WRITE_JSON, PARTIAL_PARSE, MP_CONTEXT
+        USE_EXPERIMENTAL_PARSER, WRITE_JSON, PARTIAL_PARSE, MP_CONTEXT, USE_COLORS, \
+        STORE_FAILURES
 
     USE_CACHE = getattr(args, 'use_cache', USE_CACHE)
 
@@ -68,10 +86,20 @@ def set_from_args(args):
     )
 
     TEST_NEW_PARSER = getattr(args, 'test_new_parser', TEST_NEW_PARSER)
+    USE_EXPERIMENTAL_PARSER = getattr(args, 'use_experimental_parser', USE_EXPERIMENTAL_PARSER)
     WRITE_JSON = getattr(args, 'write_json', WRITE_JSON)
     PARTIAL_PARSE = getattr(args, 'partial_parse', None)
     MP_CONTEXT = _get_context()
 
+    # The use_colors attribute will always have a value because it is assigned
+    # None by default from the add_mutually_exclusive_group function
+    use_colors_override = getattr(args, 'use_colors')
+
+    if use_colors_override is not None:
+        USE_COLORS = use_colors_override
+
+    STORE_FAILURES = getattr(args, 'store_failures', STORE_FAILURES)
+
 
 # initialize everything to the defaults on module load
 reset()

core/dbt/graph/__init__.py

@@ -0,0 +1,18 @@
+from .selector_spec import (  # noqa: F401
+    SelectionUnion,
+    SelectionSpec,
+    SelectionIntersection,
+    SelectionDifference,
+    SelectionCriteria,
+)
+from .selector import (  # noqa: F401
+    ResourceTypeSelector,
+    NodeSelector,
+)
+from .cli import (  # noqa: F401
+    parse_difference,
+    parse_test_selectors,
+    parse_from_selectors_definition,
+)
+from .queue import GraphQueue  # noqa: F401
+from .graph import Graph, UniqueId  # noqa: F401

core/dbt/graph/cli.py

@@ -0,0 +1,280 @@
+# special support for CLI argument parsing.
+import itertools
+from dbt.clients.yaml_helper import yaml, Loader, Dumper  # noqa: F401
+
+from typing import (
+    Dict, List, Optional, Tuple, Any, Union
+)
+
+from dbt.contracts.selection import SelectorDefinition, SelectorFile
+from dbt.exceptions import InternalException, ValidationException
+
+from .selector_spec import (
+    SelectionUnion,
+    SelectionSpec,
+    SelectionIntersection,
+    SelectionDifference,
+    SelectionCriteria,
+)
+
+INTERSECTION_DELIMITER = ','
+
+DEFAULT_INCLUDES: List[str] = ['fqn:*', 'source:*', 'exposure:*']
+DEFAULT_EXCLUDES: List[str] = []
+DATA_TEST_SELECTOR: str = 'test_type:data'
+SCHEMA_TEST_SELECTOR: str = 'test_type:schema'
+
+
+def parse_union(
+    components: List[str], expect_exists: bool, greedy: bool = False
+) -> SelectionUnion:
+    # turn ['a b', 'c'] -> ['a', 'b', 'c']
+    raw_specs = itertools.chain.from_iterable(
+        r.split(' ') for r in components
+    )
+    union_components: List[SelectionSpec] = []
+
+    # ['a', 'b', 'c,d'] -> union('a', 'b', intersection('c', 'd'))
+    for raw_spec in raw_specs:
+        intersection_components: List[SelectionSpec] = [
+            SelectionCriteria.from_single_spec(part, greedy=greedy)
+            for part in raw_spec.split(INTERSECTION_DELIMITER)
+        ]
+        union_components.append(SelectionIntersection(
+            components=intersection_components,
+            expect_exists=expect_exists,
+            raw=raw_spec,
+        ))
+    return SelectionUnion(
+        components=union_components,
+        expect_exists=False,
+        raw=components,
+    )
+
+
+def parse_union_from_default(
+    raw: Optional[List[str]], default: List[str], greedy: bool = False
+) -> SelectionUnion:
+    components: List[str]
+    expect_exists: bool
+    if raw is None:
+        return parse_union(components=default, expect_exists=False, greedy=greedy)
+    else:
+        return parse_union(components=raw, expect_exists=True, greedy=greedy)
+
+
+def parse_difference(
+    include: Optional[List[str]], exclude: Optional[List[str]]
+) -> SelectionDifference:
+    included = parse_union_from_default(include, DEFAULT_INCLUDES)
+    excluded = parse_union_from_default(exclude, DEFAULT_EXCLUDES, greedy=True)
+    return SelectionDifference(components=[included, excluded])
+
+
+def parse_test_selectors(
+    data: bool, schema: bool, base: SelectionSpec
+) -> SelectionSpec:
+    union_components = []
+
+    if data:
+        union_components.append(
+            SelectionCriteria.from_single_spec(DATA_TEST_SELECTOR)
+        )
+    if schema:
+        union_components.append(
+            SelectionCriteria.from_single_spec(SCHEMA_TEST_SELECTOR)
+        )
+
+    intersect_with: SelectionSpec
+    if not union_components:
+        return base
+    elif len(union_components) == 1:
+        intersect_with = union_components[0]
+    else:  # data and schema tests
+        intersect_with = SelectionUnion(
+            components=union_components,
+            expect_exists=True,
+            raw=[DATA_TEST_SELECTOR, SCHEMA_TEST_SELECTOR],
+        )
+
+    return SelectionIntersection(
+        components=[base, intersect_with], expect_exists=True
+    )
+
+
+RawDefinition = Union[str, Dict[str, Any]]
+
+
+def _get_list_dicts(
+    dct: Dict[str, Any], key: str
+) -> List[RawDefinition]:
+    result: List[RawDefinition] = []
+    if key not in dct:
+        raise InternalException(
+            f'Expected to find key {key} in dict, only found {list(dct)}'
+        )
+    values = dct[key]
+    if not isinstance(values, list):
+        raise ValidationException(
+            f'Invalid value for key "{key}". Expected a list.'
+        )
+    for value in values:
+        if isinstance(value, dict):
+            for value_key in value:
+                if not isinstance(value_key, str):
+                    raise ValidationException(
+                        f'Expected all keys to "{key}" dict to be strings, '
+                        f'but "{value_key}" is a "{type(value_key)}"'
+                    )
+            result.append(value)
+        elif isinstance(value, str):
+            result.append(value)
+        else:
+            raise ValidationException(
+                f'Invalid value type {type(value)} in key "{key}", expected '
+                f'dict or str (value: {value}).'
+            )
+
+    return result
+
+
+def _parse_exclusions(definition) -> Optional[SelectionSpec]:
+    exclusions = _get_list_dicts(definition, 'exclude')
+    parsed_exclusions = [
+        parse_from_definition(excl) for excl in exclusions
+    ]
+    if len(parsed_exclusions) == 1:
+        return parsed_exclusions[0]
+    elif len(parsed_exclusions) > 1:
+        return SelectionUnion(
+            components=parsed_exclusions,
+            raw=exclusions
+        )
+    else:
+        return None
+
+
+def _parse_include_exclude_subdefs(
+    definitions: List[RawDefinition]
+) -> Tuple[List[SelectionSpec], Optional[SelectionSpec]]:
+    include_parts: List[SelectionSpec] = []
+    diff_arg: Optional[SelectionSpec] = None
+
+    for definition in definitions:
+        if isinstance(definition, dict) and 'exclude' in definition:
+            # do not allow multiple exclude: defs at the same level
+            if diff_arg is not None:
+                yaml_sel_cfg = yaml.dump(definition)
+                raise ValidationException(
+                    f"You cannot provide multiple exclude arguments to the "
+                    f"same selector set operator:\n{yaml_sel_cfg}"
+                )
+            diff_arg = _parse_exclusions(definition)
+        else:
+            include_parts.append(parse_from_definition(definition))
+
+    return (include_parts, diff_arg)
+
+
+def parse_union_definition(definition: Dict[str, Any]) -> SelectionSpec:
+    union_def_parts = _get_list_dicts(definition, 'union')
+    include, exclude = _parse_include_exclude_subdefs(union_def_parts)
+
+    union = SelectionUnion(components=include)
+
+    if exclude is None:
+        union.raw = definition
+        return union
+    else:
+        return SelectionDifference(
+            components=[union, exclude],
+            raw=definition
+        )
+
+
+def parse_intersection_definition(
+    definition: Dict[str, Any]
+) -> SelectionSpec:
+    intersection_def_parts = _get_list_dicts(definition, 'intersection')
+    include, exclude = _parse_include_exclude_subdefs(intersection_def_parts)
+    intersection = SelectionIntersection(components=include)
+
+    if exclude is None:
+        intersection.raw = definition
+        return intersection
+    else:
+        return SelectionDifference(
+            components=[intersection, exclude],
+            raw=definition
+        )
+
+
+def parse_dict_definition(definition: Dict[str, Any]) -> SelectionSpec:
+    diff_arg: Optional[SelectionSpec] = None
+    if len(definition) == 1:
+        key = list(definition)[0]
+        value = definition[key]
+        if not isinstance(key, str):
+            raise ValidationException(
+                f'Expected definition key to be a "str", got one of type '
+                f'"{type(key)}" ({key})'
+            )
+        dct = {
+            'method': key,
+            'value': value,
+        }
+    elif 'method' in definition and 'value' in definition:
+        dct = definition
+        if 'exclude' in definition:
+            diff_arg = _parse_exclusions(definition)
+            dct = {k: v for k, v in dct.items() if k != 'exclude'}
+    else:
+        raise ValidationException(
+            f'Expected either 1 key or else "method" '
+            f'and "value" keys, but got {list(definition)}'
+        )
+
+    # if key isn't a valid method name, this will raise
+    base = SelectionCriteria.selection_criteria_from_dict(definition, dct)
+    if diff_arg is None:
+        return base
+    else:
+        return SelectionDifference(components=[base, diff_arg])
+
+
+def parse_from_definition(
+    definition: RawDefinition, rootlevel=False
+) -> SelectionSpec:
+
+    if (isinstance(definition, dict) and
+            ('union' in definition or 'intersection' in definition) and
+            rootlevel and len(definition) > 1):
+        keys = ",".join(definition.keys())
+        raise ValidationException(
+            f"Only a single 'union' or 'intersection' key is allowed "
+            f"in a root level selector definition; found {keys}."
+        )
+    if isinstance(definition, str):
+        return SelectionCriteria.from_single_spec(definition)
+    elif 'union' in definition:
+        return parse_union_definition(definition)
+    elif 'intersection' in definition:
+        return parse_intersection_definition(definition)
+    elif isinstance(definition, dict):
+        return parse_dict_definition(definition)
+    else:
+        raise ValidationException(
+            f'Expected to find union, intersection, str or dict, instead '
+            f'found {type(definition)}: {definition}'
+        )
+
+
+def parse_from_selectors_definition(
+    source: SelectorFile
+) -> Dict[str, SelectionSpec]:
+    result: Dict[str, SelectionSpec] = {}
+    selector: SelectorDefinition
+    for selector in source.selectors:
+        result[selector.name] = parse_from_definition(selector.definition,
+                                                      rootlevel=True)
+    return result

core/dbt/graph/graph.py

@@ -0,0 +1,105 @@
+from typing import (
+    Set, Iterable, Iterator, Optional, NewType
+)
+import networkx as nx  # type: ignore
+
+from dbt.exceptions import InternalException
+
+UniqueId = NewType('UniqueId', str)
+
+
+class Graph:
+    """A wrapper around the networkx graph that understands SelectionCriteria
+    and how they interact with the graph.
+    """
+    def __init__(self, graph):
+        self.graph = graph
+
+    def nodes(self) -> Set[UniqueId]:
+        return set(self.graph.nodes())
+
+    def edges(self):
+        return self.graph.edges()
+
+    def __iter__(self) -> Iterator[UniqueId]:
+        return iter(self.graph.nodes())
+
+    def ancestors(
+        self, node: UniqueId, max_depth: Optional[int] = None
+    ) -> Set[UniqueId]:
+        """Returns all nodes having a path to `node` in `graph`"""
+        if not self.graph.has_node(node):
+            raise InternalException(f'Node {node} not found in the graph!')
+        with nx.utils.reversed(self.graph):
+            anc = nx.single_source_shortest_path_length(G=self.graph,
+                                                        source=node,
+                                                        cutoff=max_depth)\
+                .keys()
+        return anc - {node}
+
+    def descendants(
+        self, node: UniqueId, max_depth: Optional[int] = None
+    ) -> Set[UniqueId]:
+        """Returns all nodes reachable from `node` in `graph`"""
+        if not self.graph.has_node(node):
+            raise InternalException(f'Node {node} not found in the graph!')
+        des = nx.single_source_shortest_path_length(G=self.graph,
+                                                    source=node,
+                                                    cutoff=max_depth)\
+            .keys()
+        return des - {node}
+
+    def select_childrens_parents(
+        self, selected: Set[UniqueId]
+    ) -> Set[UniqueId]:
+        ancestors_for = self.select_children(selected) | selected
+        return self.select_parents(ancestors_for) | ancestors_for
+
+    def select_children(
+        self, selected: Set[UniqueId], max_depth: Optional[int] = None
+    ) -> Set[UniqueId]:
+        descendants: Set[UniqueId] = set()
+        for node in selected:
+            descendants.update(self.descendants(node, max_depth))
+        return descendants
+
+    def select_parents(
+        self, selected: Set[UniqueId], max_depth: Optional[int] = None
+    ) -> Set[UniqueId]:
+        ancestors: Set[UniqueId] = set()
+        for node in selected:
+            ancestors.update(self.ancestors(node, max_depth))
+        return ancestors
+
+    def select_successors(self, selected: Set[UniqueId]) -> Set[UniqueId]:
+        successors: Set[UniqueId] = set()
+        for node in selected:
+            successors.update(self.graph.successors(node))
+        return successors
+
+    def get_subset_graph(self, selected: Iterable[UniqueId]) -> 'Graph':
+        """Create and return a new graph that is a shallow copy of the graph,
+        but with only the nodes in include_nodes. Transitive edges across
+        removed nodes are preserved as explicit new edges.
+        """
+        new_graph = nx.algorithms.transitive_closure(self.graph)
+
+        include_nodes = set(selected)
+
+        for node in self:
+            if node not in include_nodes:
+                new_graph.remove_node(node)
+
+        for node in include_nodes:
+            if node not in new_graph:
+                raise ValueError(
+                    "Couldn't find model '{}' -- does it exist or is "
+                    "it disabled?".format(node)
+                )
+        return Graph(new_graph)
+
+    def subgraph(self, nodes: Iterable[UniqueId]) -> 'Graph':
+        return Graph(self.graph.subgraph(nodes))
+
+    def get_dependent_nodes(self, node: UniqueId):
+        return nx.descendants(self.graph, node)

core/dbt/graph/queue.py

@@ -0,0 +1,205 @@
+import networkx as nx  # type: ignore
+import threading
+
+from queue import PriorityQueue
+from typing import Dict, Set, List, Generator, Optional
+
+from .graph import UniqueId
+from dbt.contracts.graph.parsed import ParsedSourceDefinition, ParsedExposure
+from dbt.contracts.graph.compiled import GraphMemberNode
+from dbt.contracts.graph.manifest import Manifest
+from dbt.node_types import NodeType
+
+
+class GraphQueue:
+    """A fancy queue that is backed by the dependency graph.
+    Note: this will mutate input!
+
+    This queue is thread-safe for `mark_done` calls, though you must ensure
+    that separate threads do not call `.empty()` or `__len__()` and `.get()` at
+    the same time, as there is an unlocked race!
+    """
+
+    def __init__(self, graph: nx.DiGraph, manifest: Manifest, selected: Set[UniqueId]):
+        self.graph = graph
+        self.manifest = manifest
+        self._selected = selected
+        # store the queue as a priority queue.
+        self.inner: PriorityQueue = PriorityQueue()
+        # things that have been popped off the queue but not finished
+        # and worker thread reservations
+        self.in_progress: Set[UniqueId] = set()
+        # things that are in the queue
+        self.queued: Set[UniqueId] = set()
+        # this lock controls most things
+        self.lock = threading.Lock()
+        # store the 'score' of each node as a number. Lower is higher priority.
+        self._scores = self._get_scores(self.graph)
+        # populate the initial queue
+        self._find_new_additions()
+        # awaits after task end
+        self.some_task_done = threading.Condition(self.lock)
+
+    def get_selected_nodes(self) -> Set[UniqueId]:
+        return self._selected.copy()
+
+    def _include_in_cost(self, node_id: UniqueId) -> bool:
+        node = self.manifest.expect(node_id)
+        if node.resource_type != NodeType.Model:
+            return False
+        # must be a Model - tell mypy this won't be a Source or Exposure
+        assert not isinstance(node, (ParsedSourceDefinition, ParsedExposure))
+        if node.is_ephemeral:
+            return False
+        return True
+
+    @staticmethod
+    def _grouped_topological_sort(
+        graph: nx.DiGraph,
+    ) -> Generator[List[str], None, None]:
+        """Topological sort of given graph that groups ties.
+
+        Adapted from `nx.topological_sort`, this function returns a topo sort of a graph however
+        instead of arbitrarily ordering ties in the sort order, ties are grouped together in
+        lists.
+
+        Args:
+            graph: The graph to be sorted.
+
+        Returns:
+            A generator that yields lists of nodes, one list per graph depth level.
+        """
+        indegree_map = {v: d for v, d in graph.in_degree() if d > 0}
+        zero_indegree = [v for v, d in graph.in_degree() if d == 0]
+
+        while zero_indegree:
+            yield zero_indegree
+            new_zero_indegree = []
+            for v in zero_indegree:
+                for _, child in graph.edges(v):
+                    indegree_map[child] -= 1
+                    if not indegree_map[child]:
+                        new_zero_indegree.append(child)
+            zero_indegree = new_zero_indegree
+
+    def _get_scores(self, graph: nx.DiGraph) -> Dict[str, int]:
+        """Scoring nodes for processing order.
+
+        Scores are calculated by the graph depth level. Lowest score (0) should be processed first.
+
+        Args:
+            graph: The graph to be scored.
+
+        Returns:
+            A dictionary consisting of `node name`:`score` pairs.
+        """
+        # split graph by connected subgraphs
+        subgraphs = (
+            graph.subgraph(x) for x in nx.connected_components(nx.Graph(graph))
+        )
+
+        # score all nodes in all subgraphs
+        scores = {}
+        for subgraph in subgraphs:
+            grouped_nodes = self._grouped_topological_sort(subgraph)
+            for level, group in enumerate(grouped_nodes):
+                for node in group:
+                    scores[node] = level
+
+        return scores
+
+    def get(
+        self, block: bool = True, timeout: Optional[float] = None
+    ) -> GraphMemberNode:
+        """Get a node off the inner priority queue. By default, this blocks.
+
+        This takes the lock, but only for part of it.
+
+        :param block: If True, block until the inner queue has data
+        :param timeout: If set, block for timeout seconds waiting for data.
+        :return: The node as present in the manifest.
+
+        See `queue.PriorityQueue` for more information on `get()` behavior and
+        exceptions.
+        """
+        _, node_id = self.inner.get(block=block, timeout=timeout)
+        with self.lock:
+            self._mark_in_progress(node_id)
+        return self.manifest.expect(node_id)
+
+    def __len__(self) -> int:
+        """The length of the queue is the number of tasks left for the queue to
+        give out, regardless of where they are. Incomplete tasks are not part
+        of the length.
+
+        This takes the lock.
+        """
+        with self.lock:
+            return len(self.graph) - len(self.in_progress)
+
+    def empty(self) -> bool:
+        """The graph queue is 'empty' if it all remaining nodes in the graph
+        are in progress.
+
+        This takes the lock.
+        """
+        return len(self) == 0
+
+    def _already_known(self, node: UniqueId) -> bool:
+        """Decide if a node is already known (either handed out as a task, or
+        in the queue).
+
+        Callers must hold the lock.
+
+        :param str node: The node ID to check
+        :returns bool: If the node is in progress/queued.
+        """
+        return node in self.in_progress or node in self.queued
+
+    def _find_new_additions(self) -> None:
+        """Find any nodes in the graph that need to be added to the internal
+        queue and add them.
+        """
+        for node, in_degree in self.graph.in_degree():
+            if not self._already_known(node) and in_degree == 0:
+                self.inner.put((self._scores[node], node))
+                self.queued.add(node)
+
+    def mark_done(self, node_id: UniqueId) -> None:
+        """Given a node's unique ID, mark it as done.
+
+        This method takes the lock.
+
+        :param str node_id: The node ID to mark as complete.
+        """
+        with self.lock:
+            self.in_progress.remove(node_id)
+            self.graph.remove_node(node_id)
+            self._find_new_additions()
+            self.inner.task_done()
+            self.some_task_done.notify_all()
+
+    def _mark_in_progress(self, node_id: UniqueId) -> None:
+        """Mark the node as 'in progress'.
+
+        Callers must hold the lock.
+
+        :param str node_id: The node ID to mark as in progress.
+        """
+        self.queued.remove(node_id)
+        self.in_progress.add(node_id)
+
+    def join(self) -> None:
+        """Join the queue. Blocks until all tasks are marked as done.
+
+        Make sure not to call this before the queue reports that it is empty.
+        """
+        self.inner.join()
+
+    def wait_until_something_was_done(self) -> int:
+        """Block until a task is done, then return the number of unfinished
+        tasks.
+        """
+        with self.lock:
+            self.some_task_done.wait()
+            return self.inner.unfinished_tasks

core/dbt/graph/selector.py

@@ -1,418 +1,291 @@
-from enum import Enum
-from itertools import chain
-from typing import Set, Iterable, Union, List, Container, Tuple, Optional
 
-import networkx as nx  # type: ignore
+from typing import Set, List, Optional, Tuple
+
+from .graph import Graph, UniqueId
+from .queue import GraphQueue
+from .selector_methods import MethodManager
+from .selector_spec import SelectionCriteria, SelectionSpec
 
 from dbt.logger import GLOBAL_LOGGER as logger
-from dbt.utils import coalesce
 from dbt.node_types import NodeType
-import dbt.exceptions
-
-SELECTOR_PARENTS = '+'
-SELECTOR_CHILDREN = '+'
-SELECTOR_GLOB = '*'
-SELECTOR_CHILDREN_AND_ANCESTORS = '@'
-SELECTOR_DELIMITER = ':'
+from dbt.exceptions import (
+    InternalException,
+    InvalidSelectorException,
+    warn_or_error,
+)
+from dbt.contracts.graph.compiled import GraphMemberNode
+from dbt.contracts.graph.manifest import Manifest
+from dbt.contracts.state import PreviousState
 
 
-class SelectionCriteria:
-    def __init__(self, node_spec: str):
-        self.raw = node_spec
-        self.select_children = False
-        self.select_parents = False
-        self.select_childrens_parents = False
-        self.selector_type = SELECTOR_FILTERS.FQN
-
-        if node_spec.startswith(SELECTOR_CHILDREN_AND_ANCESTORS):
-            self.select_childrens_parents = True
-            node_spec = node_spec[1:]
-
-        if node_spec.startswith(SELECTOR_PARENTS):
-            self.select_parents = True
-            node_spec = node_spec[1:]
-
-        if node_spec.endswith(SELECTOR_CHILDREN):
-            self.select_children = True
-            node_spec = node_spec[:-1]
-
-        if self.select_children and self.select_childrens_parents:
-            raise dbt.exceptions.RuntimeException(
-                'Invalid node spec {} - "@" prefix and "+" suffix are '
-                'incompatible'.format(self.raw)
-            )
-
-        if SELECTOR_DELIMITER in node_spec:
-            selector_parts = node_spec.split(SELECTOR_DELIMITER, 1)
-            selector_type, self.selector_value = selector_parts
-            self.selector_type = SELECTOR_FILTERS(selector_type)
-        else:
-            self.selector_value = node_spec
+def get_package_names(nodes):
+    return set([node.split(".")[1] for node in nodes])
 
 
-class SELECTOR_FILTERS(str, Enum):
-    FQN = 'fqn'
-    TAG = 'tag'
-    SOURCE = 'source'
-
-    def __str__(self):
-        return self._value_
+def alert_non_existence(raw_spec, nodes):
+    if len(nodes) == 0:
+        warn_or_error(
+            f"The selection criterion '{str(raw_spec)}' does not match"
+            f" any nodes"
+        )
 
 
-def split_specs(node_specs: Iterable[str]):
-    specs: Set[str] = set()
-    for spec in node_specs:
-        parts = spec.split(" ")
-        specs.update(parts)
-
-    return specs
-
-
-def get_package_names(graph):
-    return set([node.split(".")[1] for node in graph.nodes()])
-
-
-def is_selected_node(real_node, node_selector):
-    for i, selector_part in enumerate(node_selector):
-
-        is_last = (i == len(node_selector) - 1)
-
-        # if we hit a GLOB, then this node is selected
-        if selector_part == SELECTOR_GLOB:
-            return True
-
-        # match package.node_name or package.dir.node_name
-        elif is_last and selector_part == real_node[-1]:
-            return True
-
-        elif len(real_node) <= i:
-            return False
-
-        elif real_node[i] == selector_part:
-            continue
-
-        else:
-            return False
-
-    # if we get all the way down here, then the node is a match
-    return True
-
-
-def _node_is_match(
-    qualified_name: List[str], package_names: Set[str], fqn: List[str]
-) -> bool:
-    """Determine if a qualfied name matches an fqn, given the set of package
-    names in the graph.
-
-    :param List[str] qualified_name: The components of the selector or node
-        name, split on '.'.
-    :param Set[str] package_names: The set of pacakge names in the graph.
-    :param List[str] fqn: The node's fully qualified name in the graph.
+def can_select_indirectly(node):
+    """If a node is not selected itself, but its parent(s) are, it may qualify
+    for indirect selection.
+    Today, only Test nodes can be indirectly selected. In the future,
+    other node types or invocation flags might qualify.
     """
-    if len(qualified_name) == 1 and fqn[-1] == qualified_name[0]:
+    if node.resource_type == NodeType.Test:
         return True
-
-    if qualified_name[0] in package_names:
-        if is_selected_node(fqn, qualified_name):
-            return True
-
-    for package_name in package_names:
-        local_qualified_node_name = [package_name] + qualified_name
-        if is_selected_node(fqn, local_qualified_node_name):
-            return True
-
-    return False
+    else:
+        return False
 
 
-class ManifestSelector:
-    FILTER: str
-
-    def __init__(self, manifest):
-        self.manifest = manifest
-
-    def _node_iterator(
+class NodeSelector(MethodManager):
+    """The node selector is aware of the graph and manifest,
+    """
+    def __init__(
         self,
-        included_nodes: Set[str],
-        exclude: Optional[Container[str]],
-        include: Optional[Container[str]],
-    ) -> Iterable[Tuple[str, str]]:
-        for unique_id, node in self.manifest.nodes.items():
-            if unique_id not in included_nodes:
-                continue
-            if include is not None and node.resource_type not in include:
-                continue
-            if exclude is not None and node.resource_type in exclude:
-                continue
-            yield unique_id, node
-
-    def parsed_nodes(self, included_nodes):
-        return self._node_iterator(
-            included_nodes,
-            exclude=(NodeType.Source,),
-            include=None)
-
-    def source_nodes(self, included_nodes):
-        return self._node_iterator(
-            included_nodes,
-            exclude=None,
-            include=(NodeType.Source,))
-
-    def search(self, included_nodes, selector):
-        raise NotImplementedError('subclasses should implement this')
-
-
-class QualifiedNameSelector(ManifestSelector):
-    FILTER = SELECTOR_FILTERS.FQN
-
-    def search(self, included_nodes, selector):
-        """Yield all nodes in the graph that match the selector.
-
-        :param str selector: The selector or node name
-        """
-        qualified_name = selector.split(".")
-        package_names = {node.split(".")[1] for node in included_nodes}
-        for node, real_node in self.parsed_nodes(included_nodes):
-            if _node_is_match(qualified_name, package_names, real_node.fqn):
-                yield node
-
-
-class TagSelector(ManifestSelector):
-    FILTER = SELECTOR_FILTERS.TAG
-
-    def search(self, included_nodes, selector):
-        """ yields nodes from graph that have the specified tag """
-        search = chain(self.parsed_nodes(included_nodes),
-                       self.source_nodes(included_nodes))
-        for node, real_node in search:
-            if selector in real_node.tags:
-                yield node
-
-
-class SourceSelector(ManifestSelector):
-    FILTER = SELECTOR_FILTERS.SOURCE
-
-    def search(self, included_nodes, selector):
-        """yields nodes from graph are the specified source."""
-        parts = selector.split('.')
-        target_package = SELECTOR_GLOB
-        if len(parts) == 1:
-            target_source, target_table = parts[0], None
-        elif len(parts) == 2:
-            target_source, target_table = parts
-        elif len(parts) == 3:
-            target_package, target_source, target_table = parts
-        else:  # len(parts) > 3 or len(parts) == 0
-            msg = (
-                'Invalid source selector value "{}". Sources must be of the '
-                'form `${{source_name}}`, '
-                '`${{source_name}}.${{target_name}}`, or '
-                '`${{package_name}}.${{source_name}}.${{target_name}}'
-            ).format(selector)
-            raise dbt.exceptions.RuntimeException(msg)
-
-        for node, real_node in self.source_nodes(included_nodes):
-            if target_package not in (real_node.package_name, SELECTOR_GLOB):
-                continue
-            if target_source not in (real_node.source_name, SELECTOR_GLOB):
-                continue
-            if target_table in (None, real_node.name, SELECTOR_GLOB):
-                yield node
-
-
-class InvalidSelectorError(Exception):
-    pass
-
-
-ValidSelector = Union[QualifiedNameSelector, TagSelector, SourceSelector]
-
-
-class MultiSelector:
-    """The base class of the node selector. It only about the manifest and
-    selector types, including the glob operator, but does not handle any graph
-    related behavior.
-    """
-    SELECTORS = [QualifiedNameSelector, TagSelector, SourceSelector]
-
-    def __init__(self, manifest):
-        self.manifest = manifest
-
-    def get_selector(
-        self, selector_type: str
+        graph: Graph,
+        manifest: Manifest,
+        previous_state: Optional[PreviousState] = None,
     ):
-        for cls in self.SELECTORS:
-            if cls.FILTER == selector_type:
-                return cls(self.manifest)
+        super().__init__(manifest, previous_state)
+        self.full_graph = graph
 
-        raise InvalidSelectorError(selector_type)
+        # build a subgraph containing only non-empty, enabled nodes and enabled
+        # sources.
+        graph_members = {
+            unique_id for unique_id in self.full_graph.nodes()
+            if self._is_graph_member(unique_id)
+        }
+        self.graph = self.full_graph.subgraph(graph_members)
 
-    def select_included(self, included_nodes, selector_type, selector_value):
-        selector = self.get_selector(selector_type)
-        return set(selector.search(included_nodes, selector_value))
+    def select_included(
+        self, included_nodes: Set[UniqueId], spec: SelectionCriteria,
+    ) -> Set[UniqueId]:
+        """Select the explicitly included nodes, using the given spec. Return
+        the selected set of unique IDs.
+        """
+        method = self.get_method(spec.method, spec.method_arguments)
+        return set(method.search(included_nodes, spec.value))
 
+    def get_nodes_from_criteria(
+        self,
+        spec: SelectionCriteria
+    ) -> Tuple[Set[UniqueId], Set[UniqueId]]:
+        """Get all nodes specified by the single selection criteria.
 
-class Graph:
-    """A wrapper around the networkx graph that understands SelectionCriteria
-    and how they interact with the graph.
-    """
-    def __init__(self, graph):
-        self.graph = graph
+        - collect the directly included nodes
+        - find their specified relatives
+        - perform any selector-specific expansion
+        """
 
-    def nodes(self):
-        return set(self.graph.nodes())
+        nodes = self.graph.nodes()
+        try:
+            collected = self.select_included(nodes, spec)
+        except InvalidSelectorException:
+            valid_selectors = ", ".join(self.SELECTOR_METHODS)
+            logger.info(
+                f"The '{spec.method}' selector specified in {spec.raw} is "
+                f"invalid. Must be one of [{valid_selectors}]"
+            )
+            return set(), set()
 
-    def __iter__(self):
-        return iter(self.graph.nodes())
+        neighbors = self.collect_specified_neighbors(spec, collected)
+        direct_nodes, indirect_nodes = self.expand_selection(
+            selected=(collected | neighbors),
+            greedy=spec.greedy
+        )
+        return direct_nodes, indirect_nodes
 
-    def select_childrens_parents(self, selected: Set[str]) -> Set[str]:
-        ancestors_for = self.select_children(selected) | selected
-        return self.select_parents(ancestors_for) | ancestors_for
+    def collect_specified_neighbors(
+        self, spec: SelectionCriteria, selected: Set[UniqueId]
+    ) -> Set[UniqueId]:
+        """Given the set of models selected by the explicit part of the
+        selector (like "tag:foo"), apply the modifiers on the spec ("+"/"@").
+        Return the set of additional nodes that should be collected (which may
+        overlap with the selected set).
+        """
+        additional: Set[UniqueId] = set()
+        if spec.childrens_parents:
+            additional.update(self.graph.select_childrens_parents(selected))
 
-    def select_children(self, selected: Set[str]) -> Set[str]:
-        descendants: Set[str] = set()
-        for node in selected:
-            descendants.update(nx.descendants(self.graph, node))
-        return descendants
+        if spec.parents:
+            depth = spec.parents_depth
+            additional.update(self.graph.select_parents(selected, depth))
 
-    def select_parents(self, selected: Set[str]) -> Set[str]:
-        ancestors: Set[str] = set()
-        for node in selected:
-            ancestors.update(nx.ancestors(self.graph, node))
-        return ancestors
-
-    def select_successors(self, selected: Set[str]) -> Set[str]:
-        successors: Set[str] = set()
-        for node in selected:
-            successors.update(self.graph.successors(node))
-        return successors
-
-    def collect_models(
-        self, selected: Set[str], spec: SelectionCriteria,
-    ) -> Set[str]:
-        additional: Set[str] = set()
-        if spec.select_childrens_parents:
-            additional.update(self.select_childrens_parents(selected))
-        if spec.select_parents:
-            additional.update(self.select_parents(selected))
-        if spec.select_children:
-            additional.update(self.select_children(selected))
+        if spec.children:
+            depth = spec.children_depth
+            additional.update(self.graph.select_children(selected, depth))
         return additional
 
-    def subgraph(self, nodes: Iterable[str]) -> 'Graph':
-        cls = type(self)
-        return cls(self.graph.subgraph(nodes))
+    def select_nodes_recursively(self, spec: SelectionSpec) -> Tuple[Set[UniqueId], Set[UniqueId]]:
+        """If the spec is a composite spec (a union, difference, or intersection),
+        recurse into its selections and combine them. If the spec is a concrete
+        selection criteria, resolve that using the given graph.
+        """
+        if isinstance(spec, SelectionCriteria):
+            direct_nodes, indirect_nodes = self.get_nodes_from_criteria(spec)
+        else:
+            bundles = [
+                self.select_nodes_recursively(component)
+                for component in spec
+            ]
 
+            direct_sets = []
+            indirect_sets = []
 
-class NodeSelector(MultiSelector):
-    def __init__(self, graph, manifest):
-        self.full_graph = Graph(graph)
-        super().__init__(manifest)
+            for direct, indirect in bundles:
+                direct_sets.append(direct)
+                indirect_sets.append(direct | indirect)
 
-    def get_nodes_from_spec(self, graph, spec):
-        try:
-            collected = self.select_included(graph.nodes(),
-                                             spec.selector_type,
-                                             spec.selector_value)
-        except InvalidSelectorError:
-            valid_selectors = ", ".join(s.FILTER for s in self.SELECTORS)
-            logger.info("The '{}' selector specified in {} is invalid. Must "
-                        "be one of [{}]".format(
-                            spec.selector_type,
-                            spec.raw,
-                            valid_selectors))
-            return set()
+            initial_direct = spec.combined(direct_sets)
+            indirect_nodes = spec.combined(indirect_sets)
 
-        specified = graph.collect_models(collected, spec)
-        collected.update(specified)
+            direct_nodes = self.incorporate_indirect_nodes(initial_direct, indirect_nodes)
 
-        tests = {
-            n for n in graph.select_successors(collected)
-            if self.manifest.nodes[n].resource_type == NodeType.Test
-        }
-        collected.update(tests)
+            if spec.expect_exists:
+                alert_non_existence(spec.raw, direct_nodes)
 
-        return collected
+        return direct_nodes, indirect_nodes
 
-    def select_nodes(self, graph, raw_include_specs, raw_exclude_specs):
-        selected_nodes: Set[str] = set()
+    def select_nodes(self, spec: SelectionSpec) -> Set[UniqueId]:
+        """Select the nodes in the graph according to the spec.
 
-        for raw_spec in split_specs(raw_include_specs):
-            spec = SelectionCriteria(raw_spec)
-            included_nodes = self.get_nodes_from_spec(graph, spec)
-            selected_nodes.update(included_nodes)
+        This is the main point of entry for turning a spec into a set of nodes:
+        - Recurse through spec, select by criteria, combine by set operation
+        - Return final (unfiltered) selection set
+        """
 
-        for raw_spec in split_specs(raw_exclude_specs):
-            spec = SelectionCriteria(raw_spec)
-            excluded_nodes = self.get_nodes_from_spec(graph, spec)
-            selected_nodes.difference_update(excluded_nodes)
+        direct_nodes, indirect_nodes = self.select_nodes_recursively(spec)
+        return direct_nodes
 
-        return selected_nodes
-
-    def _is_graph_member(self, node_name):
-        node = self.manifest.nodes[node_name]
-        if node.resource_type == NodeType.Source:
+    def _is_graph_member(self, unique_id: UniqueId) -> bool:
+        if unique_id in self.manifest.sources:
+            source = self.manifest.sources[unique_id]
+            return source.config.enabled
+        elif unique_id in self.manifest.exposures:
             return True
+        node = self.manifest.nodes[unique_id]
         return not node.empty and node.config.enabled
 
-    def _is_match(self, node_name, resource_types, tags, required):
-        node = self.manifest.nodes[node_name]
-        if node.resource_type not in resource_types:
-            return False
-        tags = set(tags)
-        if tags and not bool(set(node.tags) & tags):
-            # there are tags specified but none match
-            return False
-        for attr in required:
-            if not getattr(node, attr):
-                return False
+    def node_is_match(self, node: GraphMemberNode) -> bool:
+        """Determine if a node is a match for the selector. Non-match nodes
+        will be excluded from results during filtering.
+        """
         return True
 
-    def get_selected(self, include, exclude, resource_types, tags, required):
-        include = coalesce(include, ['fqn:*', 'source:*'])
-        exclude = coalesce(exclude, [])
-        tags = coalesce(tags, [])
+    def _is_match(self, unique_id: UniqueId) -> bool:
+        node: GraphMemberNode
+        if unique_id in self.manifest.nodes:
+            node = self.manifest.nodes[unique_id]
+        elif unique_id in self.manifest.sources:
+            node = self.manifest.sources[unique_id]
+        elif unique_id in self.manifest.exposures:
+            node = self.manifest.exposures[unique_id]
+        else:
+            raise InternalException(
+                f'Node {unique_id} not found in the manifest!'
+            )
+        return self.node_is_match(node)
 
-        graph_members = {
-            node_name for node_name in self.full_graph.nodes()
-            if self._is_graph_member(node_name)
+    def filter_selection(self, selected: Set[UniqueId]) -> Set[UniqueId]:
+        """Return the subset of selected nodes that is a match for this
+        selector.
+        """
+        return {
+            unique_id for unique_id in selected if self._is_match(unique_id)
         }
-        filtered_graph = self.full_graph.subgraph(graph_members)
-        selected_nodes = self.select_nodes(filtered_graph, include, exclude)
 
-        filtered_nodes = set()
-        for node_name in selected_nodes:
-            if self._is_match(node_name, resource_types, tags, required):
-                filtered_nodes.add(node_name)
+    def expand_selection(
+        self, selected: Set[UniqueId], greedy: bool = False
+    ) -> Tuple[Set[UniqueId], Set[UniqueId]]:
+        # Test selection can expand to include an implicitly/indirectly selected test.
+        # In this way, `dbt test -m model_a` also includes tests that directly depend on `model_a`.
+        # Expansion has two modes, GREEDY and NOT GREEDY.
+        #
+        # GREEDY mode: If ANY parent is selected, select the test. We use this for EXCLUSION.
+        #
+        # NOT GREEDY mode:
+        #  - If ALL parents are selected, select the test.
+        #  - If ANY parent is missing, return it separately. We'll keep it around
+        #    for later and see if its other parents show up.
+        # We use this for INCLUSION.
 
+        direct_nodes = set(selected)
+        indirect_nodes = set()
+
+        for unique_id in self.graph.select_successors(selected):
+            if unique_id in self.manifest.nodes:
+                node = self.manifest.nodes[unique_id]
+                if can_select_indirectly(node):
+                    # should we add it in directly?
+                    if greedy or set(node.depends_on.nodes) <= set(selected):
+                        direct_nodes.add(unique_id)
+                    # if not:
+                    else:
+                        indirect_nodes.add(unique_id)
+
+        return direct_nodes, indirect_nodes
+
+    def incorporate_indirect_nodes(
+        self, direct_nodes: Set[UniqueId], indirect_nodes: Set[UniqueId] = set()
+    ) -> Set[UniqueId]:
+        # Check tests previously selected indirectly to see if ALL their
+        # parents are now present.
+
+        selected = set(direct_nodes)
+
+        for unique_id in indirect_nodes:
+            if unique_id in self.manifest.nodes:
+                node = self.manifest.nodes[unique_id]
+                if set(node.depends_on.nodes) <= set(selected):
+                    selected.add(unique_id)
+
+        return selected
+
+    def get_selected(self, spec: SelectionSpec) -> Set[UniqueId]:
+        """get_selected runs through the node selection process:
+
+            - node selection. Based on the include/exclude sets, the set
+                of matched unique IDs is returned
+                - expand the graph at each leaf node, before combination
+                    - selectors might override this. for example, this is where
+                        tests are added
+            - filtering:
+                - selectors can filter the nodes after all of them have been
+                  selected
+        """
+        selected_nodes = self.select_nodes(spec)
+        filtered_nodes = self.filter_selection(selected_nodes)
         return filtered_nodes
 
-    def select(self, query):
-        include = query.get('include')
-        exclude = query.get('exclude')
-        resource_types = query.get('resource_types')
-        tags = query.get('tags')
-        required = query.get('required', ())
-        addin_ephemeral_nodes = query.get('addin_ephemeral_nodes', True)
+    def get_graph_queue(self, spec: SelectionSpec) -> GraphQueue:
+        """Returns a queue over nodes in the graph that tracks progress of
+        dependecies.
+        """
+        selected_nodes = self.get_selected(spec)
+        new_graph = self.full_graph.get_subset_graph(selected_nodes)
+        # should we give a way here for consumers to mutate the graph?
+        return GraphQueue(new_graph.graph, self.manifest, selected_nodes)
 
-        selected = self.get_selected(include, exclude, resource_types, tags,
-                                     required)
 
-        # if you haven't selected any nodes, return that so we can give the
-        # nice "no models selected" message.
-        if not selected:
-            return selected
+class ResourceTypeSelector(NodeSelector):
+    def __init__(
+        self,
+        graph: Graph,
+        manifest: Manifest,
+        previous_state: Optional[PreviousState],
+        resource_types: List[NodeType],
+    ):
+        super().__init__(
+            graph=graph,
+            manifest=manifest,
+            previous_state=previous_state,
+        )
+        self.resource_types: Set[NodeType] = set(resource_types)
 
-        # we used to carefully go through all node ancestors and add those if
-        # they were ephemeral. Sadly, the algorithm we used ended up being
-        # O(n^2). Instead, since ephemeral nodes are almost free, just add all
-        # ephemeral nodes in the graph.
-        # someday at large enough scale we might want to prune it to only be
-        # ancestors of the selected nodes so we can skip the compile.
-        if addin_ephemeral_nodes:
-            addins = {
-                uid for uid, node in self.manifest.nodes.items()
-                if node.is_ephemeral_model
-            }
-        else:
-            addins = set()
-
-        return selected | addins
+    def node_is_match(self, node):
+        return node.resource_type in self.resource_types

core/dbt/graph/selector_methods.py

@@ -0,0 +1,539 @@
+import abc
+from itertools import chain
+from pathlib import Path
+from typing import Set, List, Dict, Iterator, Tuple, Any, Union, Type, Optional
+
+from dbt.dataclass_schema import StrEnum
+
+from .graph import UniqueId
+
+from dbt.contracts.graph.compiled import (
+    CompiledDataTestNode,
+    CompiledSchemaTestNode,
+    CompileResultNode,
+    ManifestNode,
+)
+from dbt.contracts.graph.manifest import Manifest, WritableManifest
+from dbt.contracts.graph.parsed import (
+    HasTestMetadata,
+    ParsedDataTestNode,
+    ParsedExposure,
+    ParsedSchemaTestNode,
+    ParsedSourceDefinition,
+)
+from dbt.contracts.state import PreviousState
+from dbt.exceptions import (
+    InternalException,
+    RuntimeException,
+)
+from dbt.node_types import NodeType
+
+
+SELECTOR_GLOB = '*'
+SELECTOR_DELIMITER = ':'
+
+
+class MethodName(StrEnum):
+    FQN = 'fqn'
+    Tag = 'tag'
+    Source = 'source'
+    Path = 'path'
+    Package = 'package'
+    Config = 'config'
+    TestName = 'test_name'
+    TestType = 'test_type'
+    ResourceType = 'resource_type'
+    State = 'state'
+    Exposure = 'exposure'
+
+
+def is_selected_node(fqn: List[str], node_selector: str):
+
+    # If qualified_name exactly matches model name (fqn's leaf), return True
+    if fqn[-1] == node_selector:
+        return True
+    # Flatten node parts. Dots in model names act as namespace separators
+    flat_fqn = [item for segment in fqn for item in segment.split('.')]
+    # Selector components cannot be more than fqn's
+    if len(flat_fqn) < len(node_selector.split('.')):
+        return False
+
+    for i, selector_part in enumerate(node_selector.split('.')):
+        # if we hit a GLOB, then this node is selected
+        if selector_part == SELECTOR_GLOB:
+            return True
+        elif flat_fqn[i] == selector_part:
+            continue
+        else:
+            return False
+
+    # if we get all the way down here, then the node is a match
+    return True
+
+
+SelectorTarget = Union[ParsedSourceDefinition, ManifestNode, ParsedExposure]
+
+
+class SelectorMethod(metaclass=abc.ABCMeta):
+    def __init__(
+        self,
+        manifest: Manifest,
+        previous_state: Optional[PreviousState],
+        arguments: List[str]
+    ):
+        self.manifest: Manifest = manifest
+        self.previous_state = previous_state
+        self.arguments: List[str] = arguments
+
+    def parsed_nodes(
+        self,
+        included_nodes: Set[UniqueId]
+    ) -> Iterator[Tuple[UniqueId, ManifestNode]]:
+
+        for key, node in self.manifest.nodes.items():
+            unique_id = UniqueId(key)
+            if unique_id not in included_nodes:
+                continue
+            yield unique_id, node
+
+    def source_nodes(
+        self,
+        included_nodes: Set[UniqueId]
+    ) -> Iterator[Tuple[UniqueId, ParsedSourceDefinition]]:
+
+        for key, source in self.manifest.sources.items():
+            unique_id = UniqueId(key)
+            if unique_id not in included_nodes:
+                continue
+            yield unique_id, source
+
+    def exposure_nodes(
+        self,
+        included_nodes: Set[UniqueId]
+    ) -> Iterator[Tuple[UniqueId, ParsedExposure]]:
+
+        for key, exposure in self.manifest.exposures.items():
+            unique_id = UniqueId(key)
+            if unique_id not in included_nodes:
+                continue
+            yield unique_id, exposure
+
+    def all_nodes(
+        self,
+        included_nodes: Set[UniqueId]
+    ) -> Iterator[Tuple[UniqueId, SelectorTarget]]:
+        yield from chain(self.parsed_nodes(included_nodes),
+                         self.source_nodes(included_nodes),
+                         self.exposure_nodes(included_nodes))
+
+    def configurable_nodes(
+        self,
+        included_nodes: Set[UniqueId]
+    ) -> Iterator[Tuple[UniqueId, CompileResultNode]]:
+        yield from chain(self.parsed_nodes(included_nodes),
+                         self.source_nodes(included_nodes))
+
+    def non_source_nodes(
+        self,
+        included_nodes: Set[UniqueId],
+    ) -> Iterator[Tuple[UniqueId, Union[ParsedExposure, ManifestNode]]]:
+        yield from chain(self.parsed_nodes(included_nodes),
+                         self.exposure_nodes(included_nodes))
+
+    @abc.abstractmethod
+    def search(
+        self,
+        included_nodes: Set[UniqueId],
+        selector: str,
+    ) -> Iterator[UniqueId]:
+        raise NotImplementedError('subclasses should implement this')
+
+
+class QualifiedNameSelectorMethod(SelectorMethod):
+    def node_is_match(self, qualified_name: str, fqn: List[str]) -> bool:
+        """Determine if a qualified name matches an fqn for all package
+        names in the graph.
+
+        :param str qualified_name: The qualified name to match the nodes with
+        :param List[str] fqn: The node's fully qualified name in the graph.
+        """
+        unscoped_fqn = fqn[1:]
+
+        if is_selected_node(fqn, qualified_name):
+            return True
+        # Match nodes across different packages
+        elif is_selected_node(unscoped_fqn, qualified_name):
+            return True
+
+        return False
+
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        """Yield all nodes in the graph that match the selector.
+
+        :param str selector: The selector or node name
+        """
+        parsed_nodes = list(self.parsed_nodes(included_nodes))
+        for node, real_node in parsed_nodes:
+            if self.node_is_match(selector, real_node.fqn):
+                yield node
+
+
+class TagSelectorMethod(SelectorMethod):
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        """ yields nodes from included that have the specified tag """
+        for node, real_node in self.all_nodes(included_nodes):
+            if selector in real_node.tags:
+                yield node
+
+
+class SourceSelectorMethod(SelectorMethod):
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        """yields nodes from included are the specified source."""
+        parts = selector.split('.')
+        target_package = SELECTOR_GLOB
+        if len(parts) == 1:
+            target_source, target_table = parts[0], None
+        elif len(parts) == 2:
+            target_source, target_table = parts
+        elif len(parts) == 3:
+            target_package, target_source, target_table = parts
+        else:  # len(parts) > 3 or len(parts) == 0
+            msg = (
+                'Invalid source selector value "{}". Sources must be of the '
+                'form `${{source_name}}`, '
+                '`${{source_name}}.${{target_name}}`, or '
+                '`${{package_name}}.${{source_name}}.${{target_name}}'
+            ).format(selector)
+            raise RuntimeException(msg)
+
+        for node, real_node in self.source_nodes(included_nodes):
+            if target_package not in (real_node.package_name, SELECTOR_GLOB):
+                continue
+            if target_source not in (real_node.source_name, SELECTOR_GLOB):
+                continue
+            if target_table not in (None, real_node.name, SELECTOR_GLOB):
+                continue
+
+            yield node
+
+
+class ExposureSelectorMethod(SelectorMethod):
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        parts = selector.split('.')
+        target_package = SELECTOR_GLOB
+        if len(parts) == 1:
+            target_name = parts[0]
+        elif len(parts) == 2:
+            target_package, target_name = parts
+        else:
+            msg = (
+                'Invalid exposure selector value "{}". Exposures must be of '
+                'the form ${{exposure_name}} or '
+                '${{exposure_package.exposure_name}}'
+            ).format(selector)
+            raise RuntimeException(msg)
+
+        for node, real_node in self.exposure_nodes(included_nodes):
+            if target_package not in (real_node.package_name, SELECTOR_GLOB):
+                continue
+            if target_name not in (real_node.name, SELECTOR_GLOB):
+                continue
+
+            yield node
+
+
+class PathSelectorMethod(SelectorMethod):
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        """Yields nodes from inclucded that match the given path.
+
+        """
+        # use '.' and not 'root' for easy comparison
+        root = Path.cwd()
+        paths = set(p.relative_to(root) for p in root.glob(selector))
+        for node, real_node in self.all_nodes(included_nodes):
+            if Path(real_node.root_path) != root:
+                continue
+            ofp = Path(real_node.original_file_path)
+            if ofp in paths:
+                yield node
+            elif any(parent in paths for parent in ofp.parents):
+                yield node
+
+
+class PackageSelectorMethod(SelectorMethod):
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        """Yields nodes from included that have the specified package"""
+        for node, real_node in self.all_nodes(included_nodes):
+            if real_node.package_name == selector:
+                yield node
+
+
+def _getattr_descend(obj: Any, attrs: List[str]) -> Any:
+    value = obj
+    for attr in attrs:
+        try:
+            value = getattr(value, attr)
+        except AttributeError:
+            # if it implements getitem (dict, list, ...), use that. On failure,
+            # raise an attribute error instead of the KeyError, TypeError, etc.
+            # that arbitrary getitem calls might raise
+            try:
+                value = value[attr]
+            except Exception as exc:
+                raise AttributeError(
+                    f"'{type(value)}' object has no attribute '{attr}'"
+                ) from exc
+    return value
+
+
+class CaseInsensitive(str):
+    def __eq__(self, other):
+        if isinstance(other, str):
+            return self.upper() == other.upper()
+        else:
+            return self.upper() == other
+
+
+class ConfigSelectorMethod(SelectorMethod):
+    def search(
+        self,
+        included_nodes: Set[UniqueId],
+        selector: Any,
+    ) -> Iterator[UniqueId]:
+        parts = self.arguments
+        # special case: if the user wanted to compare test severity,
+        # make the comparison case-insensitive
+        if parts == ['severity']:
+            selector = CaseInsensitive(selector)
+
+        # search sources is kind of useless now source configs only have
+        # 'enabled', which you can't really filter on anyway, but maybe we'll
+        # add more someday, so search them anyway.
+        for node, real_node in self.configurable_nodes(included_nodes):
+            try:
+                value = _getattr_descend(real_node.config, parts)
+            except AttributeError:
+                continue
+            else:
+                if selector == value:
+                    yield node
+
+
+class ResourceTypeSelectorMethod(SelectorMethod):
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        try:
+            resource_type = NodeType(selector)
+        except ValueError as exc:
+            raise RuntimeException(
+                f'Invalid resource_type selector "{selector}"'
+            ) from exc
+        for node, real_node in self.parsed_nodes(included_nodes):
+            if real_node.resource_type == resource_type:
+                yield node
+
+
+class TestNameSelectorMethod(SelectorMethod):
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        for node, real_node in self.parsed_nodes(included_nodes):
+            if isinstance(real_node, HasTestMetadata):
+                if real_node.test_metadata.name == selector:
+                    yield node
+
+
+class TestTypeSelectorMethod(SelectorMethod):
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        search_types: Tuple[Type, ...]
+        if selector == 'schema':
+            search_types = (ParsedSchemaTestNode, CompiledSchemaTestNode)
+        elif selector == 'data':
+            search_types = (ParsedDataTestNode, CompiledDataTestNode)
+        else:
+            raise RuntimeException(
+                f'Invalid test type selector {selector}: expected "data" or '
+                '"schema"'
+            )
+
+        for node, real_node in self.parsed_nodes(included_nodes):
+            if isinstance(real_node, search_types):
+                yield node
+
+
+class StateSelectorMethod(SelectorMethod):
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.modified_macros: Optional[List[str]] = None
+
+    def _macros_modified(self) -> List[str]:
+        # we checked in the caller!
+        if self.previous_state is None or self.previous_state.manifest is None:
+            raise InternalException(
+                'No comparison manifest in _macros_modified'
+            )
+        old_macros = self.previous_state.manifest.macros
+        new_macros = self.manifest.macros
+
+        modified = []
+        for uid, macro in new_macros.items():
+            if uid in old_macros:
+                old_macro = old_macros[uid]
+                if macro.macro_sql != old_macro.macro_sql:
+                    modified.append(uid)
+            else:
+                modified.append(uid)
+
+        for uid, macro in old_macros.items():
+            if uid not in new_macros:
+                modified.append(uid)
+
+        return modified
+
+    def recursively_check_macros_modified(self, node):
+        # check if there are any changes in macros the first time
+        if self.modified_macros is None:
+            self.modified_macros = self._macros_modified()
+
+        # loop through all macros that this node depends on
+        for macro_uid in node.depends_on.macros:
+            # is this macro one of the modified macros?
+            if macro_uid in self.modified_macros:
+                return True
+            # if not, and this macro depends on other macros, keep looping
+            macro = self.manifest.macros[macro_uid]
+            if len(macro.depends_on.macros) > 0:
+                return self.recursively_check_macros_modified(macro)
+            else:
+                return False
+        return False
+
+    def check_modified(self, old: Optional[SelectorTarget], new: SelectorTarget) -> bool:
+        different_contents = not new.same_contents(old)  # type: ignore
+        upstream_macro_change = self.recursively_check_macros_modified(new)
+        return different_contents or upstream_macro_change
+
+    def check_modified_body(self, old: Optional[SelectorTarget], new: SelectorTarget) -> bool:
+        if hasattr(new, "same_body"):
+            return not new.same_body(old)  # type: ignore
+        else:
+            return False
+
+    def check_modified_configs(self, old: Optional[SelectorTarget], new: SelectorTarget) -> bool:
+        if hasattr(new, "same_config"):
+            return not new.same_config(old)  # type: ignore
+        else:
+            return False
+
+    def check_modified_persisted_descriptions(
+        self, old: Optional[SelectorTarget], new: SelectorTarget
+    ) -> bool:
+        if hasattr(new, "same_persisted_description"):
+            return not new.same_persisted_description(old)  # type: ignore
+        else:
+            return False
+
+    def check_modified_relation(
+        self, old: Optional[SelectorTarget], new: SelectorTarget
+    ) -> bool:
+        if hasattr(new, "same_database_representation"):
+            return not new.same_database_representation(old)  # type: ignore
+        else:
+            return False
+
+    def check_modified_macros(self, _, new: SelectorTarget) -> bool:
+        return self.recursively_check_macros_modified(new)
+
+    def check_new(self, old: Optional[SelectorTarget], new: SelectorTarget) -> bool:
+        return old is None
+
+    def search(
+        self, included_nodes: Set[UniqueId], selector: str
+    ) -> Iterator[UniqueId]:
+        if self.previous_state is None or self.previous_state.manifest is None:
+            raise RuntimeException(
+                'Got a state selector method, but no comparison manifest'
+            )
+
+        state_checks = {
+            # it's new if there is no old version
+            'new': lambda old, _: old is None,
+            # use methods defined above to compare properties of old + new
+            'modified': self.check_modified,
+            'modified.body': self.check_modified_body,
+            'modified.configs': self.check_modified_configs,
+            'modified.persisted_descriptions': self.check_modified_persisted_descriptions,
+            'modified.relation': self.check_modified_relation,
+            'modified.macros': self.check_modified_macros,
+        }
+        if selector in state_checks:
+            checker = state_checks[selector]
+        else:
+            raise RuntimeException(
+                f'Got an invalid selector "{selector}", expected one of '
+                f'"{list(state_checks)}"'
+            )
+
+        manifest: WritableManifest = self.previous_state.manifest
+
+        for node, real_node in self.all_nodes(included_nodes):
+            previous_node: Optional[SelectorTarget] = None
+            if node in manifest.nodes:
+                previous_node = manifest.nodes[node]
+            elif node in manifest.sources:
+                previous_node = manifest.sources[node]
+            elif node in manifest.exposures:
+                previous_node = manifest.exposures[node]
+
+            if checker(previous_node, real_node):
+                yield node
+
+
+class MethodManager:
+    SELECTOR_METHODS: Dict[MethodName, Type[SelectorMethod]] = {
+        MethodName.FQN: QualifiedNameSelectorMethod,
+        MethodName.Tag: TagSelectorMethod,
+        MethodName.Source: SourceSelectorMethod,
+        MethodName.Path: PathSelectorMethod,
+        MethodName.Package: PackageSelectorMethod,
+        MethodName.Config: ConfigSelectorMethod,
+        MethodName.TestName: TestNameSelectorMethod,
+        MethodName.TestType: TestTypeSelectorMethod,
+        MethodName.State: StateSelectorMethod,
+        MethodName.Exposure: ExposureSelectorMethod,
+    }
+
+    def __init__(
+        self,
+        manifest: Manifest,
+        previous_state: Optional[PreviousState],
+    ):
+        self.manifest = manifest
+        self.previous_state = previous_state
+
+    def get_method(
+        self, method: MethodName, method_arguments: List[str]
+    ) -> SelectorMethod:
+
+        if method not in self.SELECTOR_METHODS:
+            raise InternalException(
+                f'Method name "{method}" is a valid node selection '
+                f'method name, but it is not handled'
+            )
+        cls: Type[SelectorMethod] = self.SELECTOR_METHODS[method]
+        return cls(self.manifest, self.previous_state, method_arguments)

core/dbt/graph/selector_spec.py

@@ -0,0 +1,212 @@
+import os
+import re
+from abc import ABCMeta, abstractmethod
+from dataclasses import dataclass
+
+from typing import (
+    Set, Iterator, List, Optional, Dict, Union, Any, Iterable, Tuple
+)
+from .graph import UniqueId
+from .selector_methods import MethodName
+from dbt.exceptions import RuntimeException, InvalidSelectorException
+
+
+RAW_SELECTOR_PATTERN = re.compile(
+    r'\A'
+    r'(?P<childrens_parents>(\@))?'
+    r'(?P<parents>((?P<parents_depth>(\d*))\+))?'
+    r'((?P<method>([\w.]+)):)?(?P<value>(.*?))'
+    r'(?P<children>(\+(?P<children_depth>(\d*))))?'
+    r'\Z'
+)
+SELECTOR_METHOD_SEPARATOR = '.'
+
+
+def _probably_path(value: str):
+    """Decide if value is probably a path. Windows has two path separators, so
+    we should check both sep ('\\') and altsep ('/') there.
+    """
+    if os.path.sep in value:
+        return True
+    elif os.path.altsep is not None and os.path.altsep in value:
+        return True
+    else:
+        return False
+
+
+def _match_to_int(match: Dict[str, str], key: str) -> Optional[int]:
+    raw = match.get(key)
+    # turn the empty string into None, too.
+    if not raw:
+        return None
+    try:
+        return int(raw)
+    except ValueError as exc:
+        raise RuntimeException(
+            f'Invalid node spec - could not handle parent depth {raw}'
+        ) from exc
+
+
+SelectionSpec = Union[
+    'SelectionCriteria',
+    'SelectionIntersection',
+    'SelectionDifference',
+    'SelectionUnion',
+]
+
+
+@dataclass
+class SelectionCriteria:
+    raw: Any
+    method: MethodName
+    method_arguments: List[str]
+    value: Any
+    childrens_parents: bool
+    parents: bool
+    parents_depth: Optional[int]
+    children: bool
+    children_depth: Optional[int]
+    greedy: bool = False
+
+    def __post_init__(self):
+        if self.children and self.childrens_parents:
+            raise RuntimeException(
+                f'Invalid node spec {self.raw} - "@" prefix and "+" suffix '
+                'are incompatible'
+            )
+
+    @classmethod
+    def default_method(cls, value: str) -> MethodName:
+        if _probably_path(value):
+            return MethodName.Path
+        else:
+            return MethodName.FQN
+
+    @classmethod
+    def parse_method(
+        cls, groupdict: Dict[str, Any]
+    ) -> Tuple[MethodName, List[str]]:
+        raw_method = groupdict.get('method')
+        if raw_method is None:
+            return cls.default_method(groupdict['value']), []
+
+        method_parts: List[str] = raw_method.split(SELECTOR_METHOD_SEPARATOR)
+        try:
+            method_name = MethodName(method_parts[0])
+        except ValueError as exc:
+            raise InvalidSelectorException(
+                f"'{method_parts[0]}' is not a valid method name"
+            ) from exc
+
+        method_arguments: List[str] = method_parts[1:]
+
+        return method_name, method_arguments
+
+    @classmethod
+    def selection_criteria_from_dict(
+        cls, raw: Any, dct: Dict[str, Any], greedy: bool = False
+    ) -> 'SelectionCriteria':
+        if 'value' not in dct:
+            raise RuntimeException(
+                f'Invalid node spec "{raw}" - no search value!'
+            )
+        method_name, method_arguments = cls.parse_method(dct)
+
+        parents_depth = _match_to_int(dct, 'parents_depth')
+        children_depth = _match_to_int(dct, 'children_depth')
+        return cls(
+            raw=raw,
+            method=method_name,
+            method_arguments=method_arguments,
+            value=dct['value'],
+            childrens_parents=bool(dct.get('childrens_parents')),
+            parents=bool(dct.get('parents')),
+            parents_depth=parents_depth,
+            children=bool(dct.get('children')),
+            children_depth=children_depth,
+            greedy=greedy
+        )
+
+    @classmethod
+    def dict_from_single_spec(cls, raw: str, greedy: bool = False):
+        result = RAW_SELECTOR_PATTERN.match(raw)
+        if result is None:
+            return {'error': 'Invalid selector spec'}
+        dct: Dict[str, Any] = result.groupdict()
+        method_name, method_arguments = cls.parse_method(dct)
+        meth_name = str(method_name)
+        if method_arguments:
+            meth_name = meth_name + '.' + '.'.join(method_arguments)
+        dct['method'] = meth_name
+        dct = {k: v for k, v in dct.items() if (v is not None and v != '')}
+        if 'childrens_parents' in dct:
+            dct['childrens_parents'] = bool(dct.get('childrens_parents'))
+        if 'parents' in dct:
+            dct['parents'] = bool(dct.get('parents'))
+        if 'children' in dct:
+            dct['children'] = bool(dct.get('children'))
+        return dct
+
+    @classmethod
+    def from_single_spec(cls, raw: str, greedy: bool = False) -> 'SelectionCriteria':
+        result = RAW_SELECTOR_PATTERN.match(raw)
+        if result is None:
+            # bad spec!
+            raise RuntimeException(f'Invalid selector spec "{raw}"')
+
+        return cls.selection_criteria_from_dict(raw, result.groupdict(), greedy=greedy)
+
+
+class BaseSelectionGroup(Iterable[SelectionSpec], metaclass=ABCMeta):
+    def __init__(
+        self,
+        components: Iterable[SelectionSpec],
+        expect_exists: bool = False,
+        raw: Any = None,
+    ):
+        self.components: List[SelectionSpec] = list(components)
+        self.expect_exists = expect_exists
+        self.raw = raw
+
+    def __iter__(self) -> Iterator[SelectionSpec]:
+        for component in self.components:
+            yield component
+
+    @abstractmethod
+    def combine_selections(
+        self,
+        selections: List[Set[UniqueId]],
+    ) -> Set[UniqueId]:
+        raise NotImplementedError(
+            '_combine_selections not implemented!'
+        )
+
+    def combined(self, selections: List[Set[UniqueId]]) -> Set[UniqueId]:
+        if not selections:
+            return set()
+
+        return self.combine_selections(selections)
+
+
+class SelectionIntersection(BaseSelectionGroup):
+    def combine_selections(
+        self,
+        selections: List[Set[UniqueId]],
+    ) -> Set[UniqueId]:
+        return set.intersection(*selections)
+
+
+class SelectionDifference(BaseSelectionGroup):
+    def combine_selections(
+        self,
+        selections: List[Set[UniqueId]],
+    ) -> Set[UniqueId]:
+        return set.difference(*selections)
+
+
+class SelectionUnion(BaseSelectionGroup):
+    def combine_selections(
+        self,
+        selections: List[Set[UniqueId]],
+    ) -> Set[UniqueId]:
+        return set.union(*selections)

core/dbt/helper_types.py

@@ -1,13 +1,28 @@
 # never name this package "types", or mypy will crash in ugly ways
+from dataclasses import dataclass
 from datetime import timedelta
-from typing import NewType, Dict
+from pathlib import Path
+from typing import Tuple, AbstractSet, Union
 
-from hologram import (
-    FieldEncoder, JsonSchemaMixin, JsonDict, ValidationError
+from dbt.dataclass_schema import (
+    dbtClassMixin, ValidationError, StrEnum,
 )
+from hologram import FieldEncoder, JsonDict
+from mashumaro.types import SerializableType
 
 
-Port = NewType('Port', int)
+class Port(int, SerializableType):
+    @classmethod
+    def _deserialize(cls, value: Union[int, str]) -> 'Port':
+        try:
+            value = int(value)
+        except ValueError:
+            raise ValidationError(f'Cannot encode {value} into port number')
+
+        return Port(value)
+
+    def _serialize(self) -> int:
+        return self
 
 
 class PortEncoder(FieldEncoder):
@@ -37,41 +52,44 @@ class TimeDeltaFieldEncoder(FieldEncoder[timedelta]):
         return {'type': 'number'}
 
 
-class NoValue:
-    """Sometimes, you want a way to say none that isn't None"""
-    def __eq__(self, other):
-        return isinstance(other, NoValue)
+class PathEncoder(FieldEncoder):
+    def to_wire(self, value: Path) -> str:
+        return str(value)
 
-
-class NoValueEncoder(FieldEncoder):
-    # the FieldEncoder class specifies a narrow range that only includes value
-    # types (str, float, None) but we want to support something extra
-    def to_wire(self, value: NoValue) -> Dict[str, str]:  # type: ignore
-        return {'novalue': 'novalue'}
-
-    def to_python(self, value) -> NoValue:
-        if (
-            not isinstance(value, dict) or
-            'novalue' not in value or
-            value['novalue'] != 'novalue'
-        ):
-            raise ValidationError('Got invalid NoValue: {}'.format(value))
-        return NoValue()
+    def to_python(self, value) -> Path:
+        if isinstance(value, Path):
+            return value
+        try:
+            return Path(value)
+        except TypeError:
+            raise ValidationError(
+                'cannot encode {} into timedelta'.format(value)
+            ) from None
 
     @property
-    def json_schema(self):
-        return {
-            'type': 'object',
-            'properties': {
-                'novalue': {
-                    'enum': ['novalue'],
-                }
-            }
-        }
+    def json_schema(self) -> JsonDict:
+        return {'type': 'string'}
 
 
-JsonSchemaMixin.register_field_encoders({
+class NVEnum(StrEnum):
+    novalue = 'novalue'
+
+    def __eq__(self, other):
+        return isinstance(other, NVEnum)
+
+
+@dataclass
+class NoValue(dbtClassMixin):
+    """Sometimes, you want a way to say none that isn't None"""
+    novalue: NVEnum = NVEnum.novalue
+
+
+dbtClassMixin.register_field_encoders({
     Port: PortEncoder(),
     timedelta: TimeDeltaFieldEncoder(),
-    NoValue: NoValueEncoder(),
+    Path: PathEncoder(),
 })
+
+
+FQNPath = Tuple[str, ...]
+PathSet = AbstractSet[FQNPath]