add exception level

* updated context dir to new structured logging

.bumpversion.cfg

@@ -1,9 +1,37 @@
 [bumpversion]
-current_version = 0.9.0a2
-commit = True
-tag = True
+current_version = 1.0.0b2
+parse = (?P<major>\d+)
+	\.(?P<minor>\d+)
+	\.(?P<patch>\d+)
+	((?P<prekind>a|b|rc)
+	(?P<pre>\d+)  # pre-release version num
+	)?
+serialize = 
+	{major}.{minor}.{patch}{prekind}{pre}
+	{major}.{minor}.{patch}
+commit = False
+tag = False
+
+[bumpversion:part:prekind]
+first_value = a
+optional_value = final
+values = 
+	a
+	b
+	rc
+	final
+
+[bumpversion:part:pre]
+first_value = 1
 
 [bumpversion:file:setup.py]
 
-[bumpversion:file:dbt/version.py]
+[bumpversion:file:core/setup.py]
 
+[bumpversion:file:core/dbt/version.py]
+
+[bumpversion:file:core/scripts/create_adapter_plugins.py]
+
+[bumpversion:file:plugins/postgres/setup.py]
+
+[bumpversion:file:plugins/postgres/dbt/adapters/postgres/__version__.py]

.coveragerc

@@ -1,3 +0,0 @@
-[report]
-include =
-  dbt/*

.dockerignore

@@ -0,0 +1,3 @@
+*
+!docker/requirements/*.txt
+!dist

.github/ISSUE_TEMPLATE/bug-report.yml

@@ -0,0 +1,85 @@
+name: 🐞 Bug
+description: Report a bug or an issue you've found with dbt
+title: "[Bug] <title>"
+labels: ["bug", "triage"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for taking the time to fill out this bug report!
+  - type: checkboxes
+    attributes:
+      label: Is there an existing issue for this?
+      description: Please search to see if an issue already exists for the bug you encountered.
+      options:
+        - label: I have searched the existing issues
+          required: true
+  - type: textarea
+    attributes:
+      label: Current Behavior
+      description: A concise description of what you're experiencing.
+    validations:
+      required: false
+  - type: textarea
+    attributes:
+      label: Expected Behavior
+      description: A concise description of what you expected to happen.
+    validations:
+      required: false
+  - type: textarea
+    attributes:
+      label: Steps To Reproduce
+      description: Steps to reproduce the behavior.
+      placeholder: |
+        1. In this environment...
+        2. With this config...
+        3. Run '...'
+        4. See error...
+    validations:
+      required: false
+  - type: textarea
+    id: logs
+    attributes:
+      label: Relevant log output
+      description: |
+        If applicable, log output to help explain your problem.
+      render: shell
+    validations:
+      required: false
+  - type: textarea
+    attributes:
+      label: Environment
+      description: |
+        examples:
+          - **OS**: Ubuntu 20.04
+          - **Python**: 3.7.2 (`python --version`)
+          - **dbt**: 0.21.0 (`dbt --version`)
+      value: |
+        - OS:
+        - Python:
+        - dbt:
+      render: markdown
+    validations:
+      required: false
+  - type: dropdown
+    id: database
+    attributes:
+      label: What database are you using dbt with?
+      multiple: true
+      options:
+        - postgres
+        - redshift
+        - snowflake
+        - bigquery
+        - other (mention it in "Additional Context")
+    validations:
+      required: false
+  - type: textarea
+    attributes:
+      label: Additional Context
+      description: |
+        Links? References? Anything that will give us more context about the issue you are encountering!
+
+        Tip: You can attach images or log files by clicking this area to highlight it and then dragging files in.
+    validations:
+      required: false

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,16 @@
+contact_links:
+  - name: Create an issue for dbt-redshift
+    url: https://github.com/dbt-labs/dbt-redshift/issues/new/choose
+    about: Report a bug or request a feature for dbt-redshift
+  - name: Create an issue for dbt-bigquery
+    url: https://github.com/dbt-labs/dbt-bigquery/issues/new/choose
+    about: Report a bug or request a feature for dbt-bigquery
+  - name: Create an issue for dbt-snowflake
+    url: https://github.com/dbt-labs/dbt-snowflake/issues/new/choose
+    about: Report a bug or request a feature for dbt-snowflake
+  - name: Ask a question or get support
+    url: https://docs.getdbt.com/docs/guides/getting-help
+    about: Ask a question or request support
+  - name: Questions on Stack Overflow
+    url: https://stackoverflow.com/questions/tagged/dbt
+    about: Look at questions/answers at Stack Overflow

.github/ISSUE_TEMPLATE/feature-request.yml

@@ -0,0 +1,49 @@
+name: ✨ Feature
+description: Suggest an idea for dbt
+title: "[Feature] <title>"
+labels: ["enhancement", "triage"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Thanks for taking the time to fill out this feature requests!
+  - type: checkboxes
+    attributes:
+      label: Is there an existing feature request for this?
+      description: Please search to see if an issue already exists for the feature you would like.
+      options:
+        - label: I have searched the existing issues
+          required: true
+  - type: textarea
+    attributes:
+      label: Describe the Feature
+      description: A clear and concise description of what you want to happen.
+    validations:
+      required: true
+  - type: textarea
+    attributes:
+      label: Describe alternatives you've considered
+      description: |
+        A clear and concise description of any alternative solutions or features you've considered.
+    validations:
+      required: false
+  - type: textarea
+    attributes:
+      label: Who will this benefit?
+      description: |
+        What kind of use case will this feature be useful for? Please be specific and provide examples, this will help us prioritize properly.
+    validations:
+      required: false
+  - type: input
+    attributes:
+      label: Are you interested in contributing this feature?
+      description: Let us know if you want to write some code, and how we can help.
+    validations:
+      required: false
+  - type: textarea
+    attributes:
+      label: Anything else?
+      description: |
+        Links? References? Anything that will give us more context about the feature you are suggesting!
+    validations:
+      required: false

.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 +1,30 @@
+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/postgres"
+    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

@@ -0,0 +1,21 @@
+resolves #
+
+<!---
+  Include the number of the issue addressed by this PR above if applicable.
+  PRs for code changes without an associated issue *will not be merged*.
+  See CONTRIBUTING.md for more information.
+-->
+
+### Description
+
+<!---
+  Describe the Pull Request here. Add any references and info to help reviewers
+  understand your changes. Include any tradeoffs you considered.
+-->
+
+### 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

.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 = ["postgres"];
+
+  // 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 +1,222 @@
+# **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'
+
+      - 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: ${{ matrix.python-version }}
+
+      - name: Set up postgres (linux)
+        if: |
+          matrix.adapter == 'postgres' &&
+          runner.os == 'Linux'
+        uses: ./.github/actions/setup-postgres-linux
+
+      - name: Set up postgres (macos)
+        if: |
+          matrix.adapter == 'postgres' &&
+          runner.os == 'macOS'
+        uses: ./.github/actions/setup-postgres-macos
+
+      - name: Set up postgres (windows)
+        if: |
+          matrix.adapter == 'postgres' &&
+          runner.os == 'Windows'
+        uses: ./.github/actions/setup-postgres-windows
+
+      - name: Install python dependencies
+        run: |
+          pip install --user --upgrade pip
+          pip install tox
+          pip --version
+          tox --version
+
+      - name: Run tox (postgres)
+        if: matrix.adapter == 'postgres'
+        run: tox
+
+      - uses: actions/upload-artifact@v2
+        if: always()
+        with:
+          name: logs
+          path: ./logs
+
+      - name: Get current date
+        if: always()
+        id: date
+        run: echo "::set-output name=date::$(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 --user --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: ${{ matrix.python-version }}
+
+      - name: Install python dependencies
+        run: |
+          pip install --user --upgrade pip
+          pip install tox
+          pip --version
+          tox --version
+
+      - name: Run tox
+        run: tox
+
+      - name: Get current date
+        if: always()
+        id: date
+        run: echo "::set-output name=date::$(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 --user --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 --user --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 +1,176 @@
+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: make results directory
+        run: mkdir ./final-output/
+      - name: run calculation
+        run: ./runner calculate -r ./ -o ./final-output/
+        # 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-output/*

.github/workflows/schema-check.yml

@@ -0,0 +1,87 @@
+# **what?**
+# Compares the schema of the dbt version of the given ref vs 
+# the latest official schema releases found in schemas.getdbt.com.
+# If there are differences, the workflow will fail and upload the
+# diff as an artifact. The metadata team should be alerted to the change.
+#
+# **why?**
+# Reaction work may need to be done if artifact schema changes
+# occur so we want to proactively alert to it.
+#
+# **when?**
+# On pushes to `develop` and release branches. Manual runs are also enabled.
+name: Artifact Schema Check
+
+on:
+  workflow_dispatch:
+  pull_request: #TODO: remove before merging
+  push:
+    branches:
+      - "develop"
+      - "*.latest"
+      - "releases/*"
+
+env:
+  LATEST_SCHEMA_PATH: ${{ github.workspace }}/new_schemas
+  SCHEMA_DIFF_ARTIFACT: ${{ github.workspace }}//schema_schanges.txt
+  DBT_REPO_DIRECTORY: ${{ github.workspace }}/dbt
+  SCHEMA_REPO_DIRECTORY: ${{ github.workspace }}/schemas.getdbt.com
+
+jobs:
+  checking-schemas:
+    name: "Checking schemas"
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Set up Python
+        uses: actions/setup-python@v2
+        with:
+          python-version: 3.8
+                
+      - name: Checkout dbt repo
+        uses: actions/checkout@v2.3.4
+        with:
+            path: ${{ env.DBT_REPO_DIRECTORY }}
+                       
+      - name: Checkout schemas.getdbt.com repo
+        uses: actions/checkout@v2.3.4    
+        with: 
+          repository: dbt-labs/schemas.getdbt.com
+          ref: 'main'
+          ssh-key: ${{ secrets.SCHEMA_SSH_PRIVATE_KEY }}
+          path: ${{ env.SCHEMA_REPO_DIRECTORY }}
+        
+      - name: Generate current schema
+        run: |
+          cd ${{ env.DBT_REPO_DIRECTORY }}
+          python3 -m venv env
+          source env/bin/activate
+          pip install --upgrade pip
+          pip install -r dev-requirements.txt -r editable-requirements.txt
+          python scripts/collect-artifact-schema.py --path ${{ env.LATEST_SCHEMA_PATH }}
+          
+      # Copy generated schema files into the schemas.getdbt.com repo
+      # Do a git diff to find any changes
+      # Ignore any date or version changes though
+      - name: Compare schemas
+        run: |
+          cp -r ${{ env.LATEST_SCHEMA_PATH }}/dbt ${{ env.SCHEMA_REPO_DIRECTORY }}
+          cd ${{ env.SCHEMA_REPO_DIRECTORY }}
+          diff_results=$(git diff -I='*[0-9]{4}-(0[1-9]|1[0-2])-(0[1-9]|[1-2][0-9]|3[0-1])T' \
+          -I='*[0-9]{1}.[0-9]{2}.[0-9]{1}(rc[0-9]|b[0-9]| )' --compact-summary)
+          if [[ $(echo diff_results) ]]; then
+            echo $diff_results
+            echo "Schema changes detected!"
+            git diff -I='*[0-9]{4}-(0[1-9]|1[0-2])-(0[1-9]|[1-2][0-9]|3[0-1])T' \
+              -I='*[0-9]{1}.[0-9]{2}.[0-9]{1}(rc[0-9]|b[0-9]| )' > ${{ env.SCHEMA_DIFF_ARTIFACT }}
+            exit 1
+          else
+            echo "No schema changes detected"
+          fi
+
+      - name: Upload schema diff
+        uses: actions/upload-artifact@v2.2.4
+        if: ${{ failure() }}
+        with:
+          name: 'schema_schanges.txt'
+          path: '${{ env.SCHEMA_DIFF_ARTIFACT }}'

.github/workflows/stale.yml

@@ -0,0 +1,18 @@
+name: "Close stale issues and PRs"
+on:
+  schedule:
+    - cron: "30 1 * * *"
+
+jobs:
+  stale:
+    runs-on: ubuntu-latest
+    steps:
+      # pinned at v4 (https://github.com/actions/stale/releases/tag/v4.0.0)
+      - uses: actions/stale@cdf15f641adb27a71842045a94023bef6945e3aa
+        with:
+          stale-issue-message: "This issue has been marked as Stale because it has been open for 180 days with no activity. If you would like the issue to remain open, please remove the stale label or comment on the issue, or it will be closed in 7 days."
+          stale-pr-message: "This PR has been marked as Stale because it has been open for 180 days with no activity. If you would like the PR to remain open, please remove the stale label or comment on the PR, or it will be closed in 7 days."
+          # mark issues/PRs stale when they haven't seen activity in 180 days
+          days-before-stale: 180
+          # ignore checking issues with the following labels
+          exempt-issue-labels: "epic,discussion"

.github/workflows/version-bump.yml

@@ -0,0 +1,109 @@
+# **what?**
+# This workflow will take a version number and a dry run flag. With that
+# it will run versionbump to update the version number everywhere in the 
+# code base and then generate an update Docker requirements file. If this
+# is a dry run, a draft PR will open with the changes. If this isn't a dry
+# run, the changes will be committed to the branch this is run on.
+
+# **why?**
+# This is to aid in releasing dbt and making sure we have updated 
+# the versions and Docker requirements in all places.
+
+# **when?**
+# This is triggered either manually OR 
+# from the repository_dispatch event "version-bump" which is sent from
+# the dbt-release repo Action
+
+name: Version Bump
+
+on:
+  workflow_dispatch:
+    inputs:
+      version_number:
+       description: 'The version number to bump to'
+       required: true
+      is_dry_run:
+       description: 'Creates a draft PR to allow testing instead of committing to a branch'
+       required: true
+       default: 'true'  
+  repository_dispatch:
+    types: [version-bump]
+  
+jobs:
+  bump:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check out the repository
+        uses: actions/checkout@v2
+
+      - name: Set version and dry run values
+        id: variables
+        env:
+          VERSION_NUMBER: "${{ github.event.client_payload.version_number == '' && github.event.inputs.version_number || github.event.client_payload.version_number }}"
+          IS_DRY_RUN: "${{ github.event.client_payload.is_dry_run == '' && github.event.inputs.is_dry_run || github.event.client_payload.is_dry_run }}"
+        run: |
+          echo Repository dispatch event version: ${{ github.event.client_payload.version_number }}
+          echo Repository dispatch event dry run: ${{ github.event.client_payload.is_dry_run }}
+          echo Workflow dispatch event version: ${{ github.event.inputs.version_number }}
+          echo Workflow dispatch event dry run: ${{ github.event.inputs.is_dry_run }}
+          echo ::set-output name=VERSION_NUMBER::$VERSION_NUMBER
+          echo ::set-output name=IS_DRY_RUN::$IS_DRY_RUN
+
+      - uses: actions/setup-python@v2
+        with:
+          python-version: "3.8"
+
+      - name: Install python dependencies
+        run: |
+          python3 -m venv env
+          source env/bin/activate
+          pip install --upgrade pip     
+          
+      - name: Create PR branch
+        if: ${{ steps.variables.outputs.IS_DRY_RUN == 'true' }}
+        run: |
+          git checkout -b bumping-version/${{steps.variables.outputs.VERSION_NUMBER}}_$GITHUB_RUN_ID
+          git push origin bumping-version/${{steps.variables.outputs.VERSION_NUMBER}}_$GITHUB_RUN_ID
+          git branch --set-upstream-to=origin/bumping-version/${{steps.variables.outputs.VERSION_NUMBER}}_$GITHUB_RUN_ID bumping-version/${{steps.variables.outputs.VERSION_NUMBER}}_$GITHUB_RUN_ID
+      
+      - name: Generate Docker requirements
+        run: |
+          source env/bin/activate
+          pip install -r requirements.txt 
+          pip freeze -l > docker/requirements/requirements.txt
+          git status
+
+      - name: Bump version
+        run: |
+          source env/bin/activate
+          pip install -r dev-requirements.txt 
+          env/bin/bumpversion --allow-dirty --new-version ${{steps.variables.outputs.VERSION_NUMBER}} major
+          git status
+
+      - name: Commit version bump directly
+        uses: EndBug/add-and-commit@v7
+        if: ${{ steps.variables.outputs.IS_DRY_RUN == 'false' }}
+        with:
+          author_name: 'Github Build Bot'
+          author_email: 'buildbot@fishtownanalytics.com'
+          message: 'Bumping version to ${{steps.variables.outputs.VERSION_NUMBER}}'
+
+      - name: Commit version bump to branch
+        uses: EndBug/add-and-commit@v7
+        if: ${{ steps.variables.outputs.IS_DRY_RUN == 'true' }}
+        with:
+          author_name: 'Github Build Bot'
+          author_email: 'buildbot@fishtownanalytics.com'
+          message: 'Bumping version to ${{steps.variables.outputs.VERSION_NUMBER}}'
+          branch: 'bumping-version/${{steps.variables.outputs.VERSION_NUMBER}}_${{GITHUB.RUN_ID}}'
+          push: 'origin origin/bumping-version/${{steps.variables.outputs.VERSION_NUMBER}}_${{GITHUB.RUN_ID}}'
+
+      - name: Create Pull Request
+        uses: peter-evans/create-pull-request@v3
+        if: ${{ steps.variables.outputs.IS_DRY_RUN == 'true' }}
+        with:
+          author: 'Github Build Bot <buildbot@fishtownanalytics.com>'
+          draft: true
+          base: ${{github.ref}}
+          title: 'Bumping version to ${{steps.variables.outputs.VERSION_NUMBER}}'
+          branch: 'bumping-version/${{steps.variables.outputs.VERSION_NUMBER}}_${{GITHUB.RUN_ID}}'

.gitignore

@@ -8,7 +8,8 @@ __pycache__/
 
 # Distribution / packaging
 .Python
-env/
+env*/
+dbt_env/
 build/
 develop-eggs/
 dist/
@@ -23,6 +24,7 @@ var/
 *.egg-info/
 .installed.cfg
 *.egg
+*.mypy_cache/
 logs/
 
 # PyInstaller
@@ -41,12 +43,16 @@ htmlcov/
 .coverage
 .coverage.*
 .cache
+.env
 nosetests.xml
 coverage.xml
 *,cover
 .hypothesis/
 test.env
 
+# Mypy
+.mypy_cache/
+
 # Translations
 *.mo
 *.pot
@@ -65,3 +71,26 @@ target/
 
 #Emacs
 *~
+
+# Sublime Text
+*.sublime-*
+
+# Vim
+*.sw*
+
+.python-version
+
+# Vim
+*.sw*
+
+# pycharm
+.idea/
+venv/
+
+# AWS credentials
+.aws/
+
+.DS_Store
+
+# vscode
+.vscode/

ARCHITECTURE.md

@@ -0,0 +1,47 @@
+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
+- [`task`](core/dbt/task): Set forth the actions that dbt can perform when invoked
+
+### Invoking dbt
+
+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 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. For testing and development purposes, the dbt-postgres plugin lives alongside the dbt-core codebase, in the [`plugins`](plugins) subdirectory. Like other adapter plugins, it is a self-contained codebase and package that builds on top of dbt-core.
+
+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

@@ -0,0 +1,235 @@
+# 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`. 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.
+
+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).
+
+#### Adapters
+
+If you have an issue or code change suggestion related to a specific database [adapter](https://docs.getdbt.com/docs/available-adapters); please refer to that supported databases seperate repo for those contributions.
+
+### Signing the CLA
+
+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-core/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-core/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-core/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-core/labels/bug) | This issue represents a defect or regression in `dbt` |
+| [enhancement](https://github.com/dbt-labs/dbt-core/labels/enhancement) | This issue represents net-new functionality in `dbt` |
+| [good first issue](https://github.com/dbt-labs/dbt-core/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-core/labels/help%20wanted) / [discussion](https://github.com/dbt-labs/dbt-core/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-core/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-core/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-core/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-core/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 branch 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](https://developer.apple.com/support/xcode/).
+
+### External contributors
+
+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 locally
+3. check out a new branch for your proposed changes
+4. push changes to your fork
+5. open a pull request against `dbt-labs/dbt` from your forked repository
+
+### Core contributors
+
+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
+
+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.
+
+### Tools
+
+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
+- [Github Actions](https://github.com/features/actions)
+
+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 are both used in testing. Specific instructions for you OS can be found [here](https://docs.docker.com/get-docker/).
+
+
+#### 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
+
+### Installation
+
+First make sure that you set up your `virtualenv` as described in [Setting up an environment](#setting-up-an-environment).  Also ensure you have the latest version of pip installed with `pip install --upgrade pip`. Next, install `dbt` (and its dependencies) with:
+
+```sh
+make dev
+# or
+pip install -r dev-requirements.txt -r editable-requirements.txt
+```
+
+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`
+
+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.
+
+Configure your [profile](https://docs.getdbt.com/docs/configure-your-profile) as necessary to connect to your target databases. It may be a good idea to add a new profile pointing to a local postgres instance, or a specific test sandbox within your data warehouse if appropriate.
+
+## 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.
+
+Although `dbt` works with a number of different databases, you won't need to supply credentials for every one of these databases in your test environment. Instead you can test all dbt-core code changes with Python and Postgres.
+
+### Initial setup
+
+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
+```
+
+`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.
+
+```
+cp test.env.sample test.env
+$EDITOR test.env
+```
+
+> 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
+```
+> 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.
+
+Check out the other targets in the Makefile to see other commonly used test
+suites.
+
+#### `tox`
+
+[`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
+
+dbt Labs provides a CI environment to test changes to specific adapters, and periodic maintenance checks of `dbt-core` through Github Actions. For example, if you submit a pull request to the `dbt-redshift` repo, GitHub will trigger automated code checks and tests against Redshift.
+
+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.
+
+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

@@ -1,14 +0,0 @@
-FROM python:3.6
-
-RUN apt-get update
-
-RUN apt-get install -y python-pip netcat
-RUN apt-get install -y python-dev python3-dev
-
-RUN pip install pip --upgrade
-RUN pip install virtualenv
-RUN pip install virtualenvwrapper
-RUN pip install tox
-
-WORKDIR /usr/src/app
-RUN cd /usr/src/app

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 \
+    python3-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.

MANIFEST.in

@@ -1 +0,0 @@
-recursive-include dbt/include *.py *.sql *.yml

Makefile

@@ -1,24 +1,88 @@
-.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-py27,unit-py35,pep8
+.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-py27,integration-postgres-py35,integration-snowflake-py27,integration-snowflake-py35,integration-bigquery-py27,integration-bigquery-py35
+.PHONY: lint
+lint: .env ## Runs all code checks in parallel.
+	$(DOCKER_CMD) tox -p -e flake8,mypy
+
+.PHONY: unit
+unit: .env ## Runs unit tests with py38.
+	$(DOCKER_CMD) tox -e py38
+
+.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.
+	$(DOCKER_CMD) tox -e py38-postgres -- -x -nauto
+
+.PHONY: setup-db
+setup-db: ## Setup Postgres database with docker-compose for system testing.
+	docker-compose up -d database
+	PGHOST=localhost PGUSER=root PGPASSWORD=password PGDATABASE=postgres bash test/setup_db.sh
+
+# This rule creates a file named .env that is used by docker-compose for passing
+# the USER_ID and GROUP_ID arguments to the Docker image.
+.env: ## Setup step for using using docker-compose with make target.
+	@touch .env
+ifneq ($(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/
+	rm -f dbt_project.yml
+	rm -rf dist/
+	rm -f htmlcov/*.{css,html,js,json,png}
+	rm -rf logs/
+	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'
 
-test-new:
-	@echo "Test run starting..."
-	@echo "Changed test files:"
-	@echo "${changed_tests}"
-	@docker-compose run test /usr/src/app/test/runner.sh ${changed_tests}

README.md

@@ -1,49 +1,42 @@
-# dbt
+<p align="center">
+  <img src="https://raw.githubusercontent.com/dbt-labs/dbt-core/fa1ea14ddfb1d5ae319d5141844910dd53ab2834/etc/dbt-core.svg" alt="dbt logo" width="750"/>
+</p>
+<p align="center">
+  <a href="https://github.com/dbt-labs/dbt-core/actions/workflows/main.yml">
+    <img src="https://github.com/dbt-labs/dbt-core/actions/workflows/main.yml/badge.svg?event=push" alt="Unit Tests Badge"/>
+  </a>
+  <a href="https://github.com/dbt-labs/dbt-core/actions/workflows/integration.yml">
+    <img src="https://github.com/dbt-labs/dbt-core/actions/workflows/integration.yml/badge.svg?event=push" alt="Integration Tests Badge"/>
+  </a>
+</p>
 
-dbt (data build tool) helps analysts write reliable, modular code using a workflow that closely mirrors software development.
+**[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.
 
-A dbt project primarily consists of "models". These models are SQL `select` statements that filter, aggregate, and otherwise transform data to facilitate analytics. Analysts use dbt 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-core/6c6649f9129d5d108aa3b0526f634cd8f3a9d1ed/etc/dbt-arch.png)
 
-These models frequently build on top of one another. Fortunately, dbt makes it easy to [manage relationships](https://dbt.readme.io/reference#ref) between models, [test](https://dbt.readme.io/docs/testing) your assumptions, and [visualize](https://graph.sinterdata.com/) your projects.
+## Understanding dbt
 
-Still reading? Check out the [docs](https://dbt.readme.io/docs/overview) for more information.
+Analysts using dbt can transform their data by simply writing select statements, while dbt handles turning these statements into tables and views in a data warehouse.
 
-![dbt dag](/etc/dag.png?raw=true)
+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).
 
----
-### Getting Started
+![dbt dag](https://raw.githubusercontent.com/dbt-labs/dbt-core/6c6649f9129d5d108aa3b0526f634cd8f3a9d1ed/etc/dbt-dag.png)
 
-- [What is dbt]?
-- Read the [dbt viewpoint]
-- [Installation]
-- Join the [chat][slack-url] on Slack for live questions and support.
+## Getting started
 
----
-### The dbt ecosystem
-- Visualize your dbt graph [here](https://graph.sinterdata.com/)
-- Run your dbt projects on a schedule [here](http://sinterdata.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/)
 
----
+## Join the dbt Community
 
-[![Code Climate](https://codeclimate.com/github/fishtown-analytics/dbt/badges/gpa.svg)](https://codeclimate.com/github/fishtown-analytics/dbt) [![Slack](https://ac-slackin.herokuapp.com/badge.svg)](https://ac-slackin.herokuapp.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)
 
-### Testing
+## Reporting bugs and contributing code
 
-| service | development | master |
-| --- | --- | --- |
-| CircleCI| [![CircleCI](https://circleci.com/gh/fishtown-analytics/dbt/tree/development.svg?style=svg)](https://circleci.com/gh/fishtown-analytics/dbt/tree/development) | [![CircleCI](https://circleci.com/gh/fishtown-analytics/dbt/tree/master.svg?style=svg)](https://circleci.com/gh/fishtown-analytics/dbt/tree/master) |
-| AppVeyor | [![AppVeyor](https://ci.appveyor.com/api/projects/status/v01rwd3q91jnwp9m/branch/development?svg=true)](https://ci.appveyor.com/project/DrewBanin/dbt/branch/development) | [![AppVeyor](https://ci.appveyor.com/api/projects/status/v01rwd3q91jnwp9m/branch/master?svg=true)](https://ci.appveyor.com/project/DrewBanin/dbt/branch/master) |
-
-[Coverage](https://circleci.com/api/v1/project/fishtown-analytics/dbt/latest/artifacts/0/$CIRCLE_ARTIFACTS/htmlcov/index.html?branch=development)
+- 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-core/issues/new)
+- Want to help us build dbt? Check out the [Contributing Guide](https://github.com/dbt-labs/dbt-core/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].
-
-
-
-[PyPA Code of Conduct]: https://www.pypa.io/en/latest/code-of-conduct/
-[slack-url]: http://ac-slackin.herokuapp.com/
-[Installation]: https://dbt.readme.io/docs/installation
-[What is dbt]: https://dbt.readme.io/docs/overview
-[dbt viewpoint]: https://dbt.readme.io/docs/viewpoint
+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

@@ -1,78 +0,0 @@
-### Release Procedure :shipit:
-
-1. Update changelog
-1. Bumpversion
-1. Merge to master
-  - (on master) git pull origin development
-1. Deploy to pypi
-  - python setup.py sdist upload -r pypi
-1. Deploy to homebrew
-  - Make a pull request against homebrew-core
-1. Deploy to conda-forge
-  - Make a pull request against dbt-feedstock
-1. Git release notes (points to changelog)
-1. Post to slack (point to changelog)
-
-#### Homebrew Release Process
-
-1. fork homebrew and add a remote:
-
-```
-cd $(brew --repo homebrew/core)
-git remote add origin <your-github-username> <your-fork-url>
-```
-
-2. edit the formula.
-
-```bash
-brew update
-mkvirtualenv --python="$(which python3)" brew
-pip install homebrew-pypi-poet dbt
-diff "$(brew --repo homebrew/core)"/Formula/dbt.rb <(poet -f dbt)
-```
-
-find any differences in resource stanzas, and incorporate them into the formula
-
-```
-brew edit dbt
-...
-diff "$(brew --repo homebrew/core)"/Formula/dbt.rb <(poet -f dbt)
-```
-
-3. reinstall, test, and audit dbt. if the test or audit fails, fix the formula with step 1.
-
-```bash
-brew uninstall --force dbt
-brew install --build-from-source dbt
-brew test dbt
-brew audit --strict dbt
-```
-
-4. make a pull request for the change.
-
-```bash
-cd $(brew --repo homebrew/core)
-git pull origin master
-git checkout -b dbt-<version> origin/master
-git add . -p
-git commit -m 'dbt <version>'
-git push -u <your-github-username> dbt-<version>
-```
-
-#### 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

appveyor.yml

@@ -1,63 +0,0 @@
-version: 1.0.{build}-{branch}
-
-environment:
-  # SDK v7.0 MSVC Express 2008's SetEnv.cmd script will fail if the
-  # /E:ON and /V:ON options are not enabled in the batch script intepreter
-  # See: http://stackoverflow.com/a/13751649/163740
-  CMD_IN_ENV: "cmd /E:ON /V:ON /C .\\appveyor\\run_with_env.cmd"
-  TOX_ENV: "pywin"
-
-  matrix:
-    - PYTHON: "C:\\Python35"
-      PYTHON_VERSION: "3.5.2"
-      PYTHON_ARCH: "32"
-
-    #- PYTHON: "C:\\Python35"
-    #  PYTHON_VERSION: "3.5.2"
-    #  PYTHON_ARCH: "32"
-
-  PGUSER: postgres
-  PGPASSWORD: Password12!
-
-services:
-  - postgresql94
-
-hosts:
-  database: 127.0.0.1
-
-init:
-  - PATH=C:\Program Files\PostgreSQL\9.4\bin\;%PATH%
-  - ps: Set-Content "c:\program files\postgresql\9.4\data\pg_hba.conf" "host   all   all   ::1/128        trust"
-  - ps: Add-Content "c:\program files\postgresql\9.4\data\pg_hba.conf" "host   all   all   127.0.0.1/32   trust"
-
-install:
-  # Download setup scripts and unzip
-  - ps: "wget https://github.com/cloudify-cosmo/appveyor-utils/archive/master.zip -OutFile ./master.zip"
-  - "7z e master.zip */appveyor/* -oappveyor"
-
-  # Install Python (from the official .msi of http://python.org) and pip when
-  # not already installed.
-  - "powershell ./appveyor/install.ps1"
-
-  # Prepend newly installed Python to the PATH of this build (this cannot be
-  # done from inside the powershell script as it would require to restart
-  # the parent CMD process).
-  - "SET PATH=%PYTHON%;%PYTHON%\\Scripts;%PATH%"
-
-  # Check that we have the expected version and architecture for Python
-  - "python --version"
-  - "python -c \"import struct; print(struct.calcsize('P') * 8)\""
-
-build: false  # Not a C# project, build stuff at the test step instead.
-
-before_test:
-  - "%CMD_IN_ENV% pip install psycopg2==2.6.2"
-  - "%CMD_IN_ENV% pip install tox"
-
-test_script:
-  - "bash test/setup_db.sh"
-
-  # this is generally a bad idea TODO
-  - git config --system http.sslverify false
-
-  - "%CMD_IN_ENV% tox -e %TOX_ENV%"

circle.yml

@@ -1,25 +0,0 @@
-machine:
-  python:
-    version: 3.6.0
-  post:
-    - pyenv global 2.7.12 3.6.0
-  hosts:
-    database: 127.0.0.1
-
-database:
-  override:
-    - bash test/setup_db.sh
-
-dependencies:
-  pre:
-    - pip install --upgrade pip setuptools || true
-    - pip install --upgrade tox tox-pyenv
-  override:
-    - pyenv local 2.7.12 3.6.0
-
-test:
-  override:
-    - sudo chown -R ubuntu:ubuntu /root/
-    - /bin/bash -c 'cd /home/ubuntu/dbt && tox'
-  post:
-    - mv htmlcov $CIRCLE_ARTIFACTS/

core/MANIFEST.in

@@ -0,0 +1 @@
+recursive-include dbt/include *.py *.sql *.yml *.html *.md .gitkeep .gitignore

core/dbt/adapters/base/__init__.py

@@ -0,0 +1,14 @@
+# these are all just exports, #noqa them so flake8 will be happy
+
+# TODO: Should we still include this in the `adapters` namespace?
+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 (  # noqa
+    BaseRelation,
+    RelationType,
+    SchemaSearchMap,
+)
+from dbt.adapters.base.column import Column  # noqa
+from dbt.adapters.base.impl import AdapterConfig, BaseAdapter  # noqa
+from dbt.adapters.base.plugin import AdapterPlugin  # noqa

core/dbt/adapters/base/column.py

@@ -0,0 +1,153 @@
+from dataclasses import dataclass
+import re
+from typing import Dict, ClassVar, Any, Optional
+
+from dbt.exceptions import RuntimeException
+
+
+@dataclass
+class Column:
+    TYPE_LABELS: ClassVar[Dict[str, str]] = {
+        'STRING': 'TEXT',
+        'TIMESTAMP': 'TIMESTAMP',
+        'FLOAT': 'FLOAT',
+        'INTEGER': 'INT'
+    }
+    column: str
+    dtype: str
+    char_size: Optional[int] = None
+    numeric_precision: Optional[Any] = None
+    numeric_scale: Optional[Any] = None
+
+    @classmethod
+    def translate_type(cls, dtype: str) -> str:
+        return cls.TYPE_LABELS.get(dtype.upper(), dtype)
+
+    @classmethod
+    def create(cls, name, label_or_dtype: str) -> 'Column':
+        column_type = cls.translate_type(label_or_dtype)
+        return cls(name, column_type)
+
+    @property
+    def name(self) -> str:
+        return self.column
+
+    @property
+    def quoted(self) -> str:
+        return '"{}"'.format(self.column)
+
+    @property
+    def data_type(self) -> str:
+        if self.is_string():
+            return Column.string_type(self.string_size())
+        elif self.is_numeric():
+            return Column.numeric_type(self.dtype, self.numeric_precision,
+                                       self.numeric_scale)
+        else:
+            return self.dtype
+
+    def is_string(self) -> bool:
+        return self.dtype.lower() in ['text', 'character varying', 'character',
+                                      'varchar']
+
+    def is_number(self):
+        return any([self.is_integer(), self.is_numeric(), self.is_float()])
+
+    def is_float(self):
+        return self.dtype.lower() in [
+            # floats
+            'real', 'float4', 'float', 'double precision', 'float8'
+        ]
+
+    def is_integer(self) -> bool:
+        return self.dtype.lower() in [
+            # real types
+            'smallint', 'integer', 'bigint',
+            'smallserial', 'serial', 'bigserial',
+            # aliases
+            'int2', 'int4', 'int8',
+            'serial2', 'serial4', 'serial8',
+        ]
+
+    def is_numeric(self) -> bool:
+        return self.dtype.lower() in ['numeric', 'decimal']
+
+    def string_size(self) -> int:
+        if not self.is_string():
+            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
+            return 256
+        else:
+            return int(self.char_size)
+
+    def can_expand_to(self, other_column: 'Column') -> bool:
+        """returns True if this column can be expanded to the size of the
+        other column"""
+        if not self.is_string() or not other_column.is_string():
+            return False
+
+        return other_column.string_size() > self.string_size()
+
+    def literal(self, value: Any) -> str:
+        return "{}::{}".format(value, self.data_type)
+
+    @classmethod
+    def string_type(cls, size: int) -> str:
+        return "character varying({})".format(size)
+
+    @classmethod
+    def numeric_type(cls, dtype: str, precision: Any, scale: Any) -> str:
+        # This could be decimal(...), numeric(...), number(...)
+        # Just use whatever was fed in here -- don't try to get too clever
+        if precision is None or scale is None:
+            return dtype
+        else:
+            return "{}({},{})".format(dtype, precision, scale)
+
+    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

@@ -0,0 +1,294 @@
+import abc
+import os
+# multiprocessing.RLock is a function returning this type
+from multiprocessing.synchronize import RLock
+from threading import get_ident
+from typing import (
+    Dict, Tuple, Hashable, Optional, ContextManager, List, Union
+)
+
+import agate
+
+import dbt.exceptions
+from dbt.contracts.connection import (
+    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):
+    """Methods to implement:
+        - exception_handler
+        - cancel_open
+        - open
+        - begin
+        - commit
+        - clear_transaction
+        - execute
+
+    You must also set the 'TYPE' class attribute with a class-unique constant
+    string.
+    """
+    TYPE: str = NotImplemented
+
+    def __init__(self, profile: AdapterRequiredConfig):
+        self.profile = profile
+        self.thread_connections: Dict[Hashable, Connection] = {}
+        self.lock: RLock = flags.MP_CONTEXT.RLock()
+        self.query_header: Optional[MacroQueryStringSetter] = None
+
+    def set_query_header(self, manifest: Manifest) -> None:
+        self.query_header = MacroQueryStringSetter(self.profile, manifest)
+
+    @staticmethod
+    def get_thread_identifier() -> Hashable:
+        # note that get_ident() may be re-used, but we should never experience
+        # that within a single process
+        return (os.getpid(), get_ident())
+
+    def get_thread_connection(self) -> Connection:
+        key = self.get_thread_identifier()
+        with self.lock:
+            if key not in self.thread_connections:
+                raise dbt.exceptions.InvalidConnectionException(
+                    key, list(self.thread_connections)
+                )
+            return self.thread_connections[key]
+
+    def set_thread_connection(self, conn: Connection) -> None:
+        key = self.get_thread_identifier()
+        if key in self.thread_connections:
+            raise dbt.exceptions.InternalException(
+                'In set_thread_connection, existing connection exists for {}'
+            )
+        self.thread_connections[key] = conn
+
+    def get_if_exists(self) -> Optional[Connection]:
+        key = self.get_thread_identifier()
+        with self.lock:
+            return self.thread_connections.get(key)
+
+    def clear_thread_connection(self) -> None:
+        key = self.get_thread_identifier()
+        with self.lock:
+            if key in self.thread_connections:
+                del self.thread_connections[key]
+
+    def clear_transaction(self) -> None:
+        """Clear any existing transactions."""
+        conn = self.get_thread_connection()
+        if conn is not None:
+            if conn.transaction_open:
+                self._rollback(conn)
+            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
+        interactions.
+
+        :param str sql: The SQL string that the block inside the context
+            manager is executing.
+        :return: A context manager that handles exceptions raised by the
+            underlying database.
+        """
+        raise dbt.exceptions.NotImplementedException(
+            '`exception_handler` is not implemented for this adapter!')
+
+    def set_connection_name(self, name: Optional[str] = None) -> Connection:
+        conn_name: str
+        if name is None:
+            # if a name isn't specified, we'll re-use a single handle
+            # named 'master'
+            conn_name = 'master'
+        else:
+            if not isinstance(name, str):
+                raise dbt.exceptions.CompilerException(
+                    f'For connection name, got {name} - not a string!'
+                )
+            assert isinstance(name, str)
+            conn_name = name
+
+        conn = self.get_if_exists()
+        if conn is None:
+            conn = Connection(
+                type=Identifier(self.TYPE),
+                name=None,
+                state=ConnectionState.INIT,
+                transaction_open=False,
+                handle=None,
+                credentials=self.profile.credentials
+            )
+            self.set_thread_connection(conn)
+
+        if conn.name == conn_name and conn.state == 'open':
+            return conn
+
+        logger.debug(
+            'Acquiring new {} connection "{}".'.format(self.TYPE, conn_name))
+
+        if conn.state == 'open':
+            logger.debug(
+                'Re-using an available connection from the pool (formerly {}).'
+                .format(conn.name)
+            )
+        else:
+            conn.handle = LazyHandle(self.open)
+
+        conn.name = conn_name
+        return conn
+
+    @abc.abstractmethod
+    def cancel_open(self) -> Optional[List[str]]:
+        """Cancel all open connections on the adapter. (passable)"""
+        raise dbt.exceptions.NotImplementedException(
+            '`cancel_open` is not implemented for this adapter!'
+        )
+
+    @abc.abstractclassmethod
+    def open(cls, connection: Connection) -> Connection:
+        """Open the given connection on the adapter and return it.
+
+        This may mutate the given connection (in particular, its state and its
+        handle).
+
+        This should be thread-safe, or hold the lock if necessary. The given
+        connection should not be in either in_use or available.
+        """
+        raise dbt.exceptions.NotImplementedException(
+            '`open` is not implemented for this adapter!'
+        )
+
+    def release(self) -> None:
+        with self.lock:
+            conn = self.get_if_exists()
+            if conn is None:
+                return
+
+        try:
+            # 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()
+            raise
+
+    def cleanup_all(self) -> None:
+        with self.lock:
+            for connection in self.thread_connections.values():
+                if connection.state not in {'closed', 'init'}:
+                    logger.debug("Connection '{}' was left open."
+                                 .format(connection.name))
+                else:
+                    logger.debug("Connection '{}' was properly closed."
+                                 .format(connection.name))
+                self.close(connection)
+
+            # garbage collect these connections
+            self.thread_connections.clear()
+
+    @abc.abstractmethod
+    def begin(self) -> None:
+        """Begin a transaction. (passable)"""
+        raise dbt.exceptions.NotImplementedException(
+            '`begin` is not implemented for this adapter!'
+        )
+
+    @abc.abstractmethod
+    def commit(self) -> None:
+        """Commit a transaction. (passable)"""
+        raise dbt.exceptions.NotImplementedException(
+            '`commit` is not implemented for this adapter!'
+        )
+
+    @classmethod
+    def _rollback_handle(cls, connection: Connection) -> None:
+        """Perform the actual rollback operation."""
+        try:
+            connection.handle.rollback()
+        except Exception:
+            logger.debug(
+                'Failed to rollback {}'.format(connection.name),
+                exc_info=True
+            )
+
+    @classmethod
+    def _close_handle(cls, connection: Connection) -> None:
+        """Perform the actual close operation."""
+        # On windows, sometimes connection handles don't have a close() attr.
+        if hasattr(connection.handle, 'close'):
+            logger.debug(f'On {connection.name}: Close')
+            connection.handle.close()
+        else:
+            logger.debug(f'On {connection.name}: No close available on handle')
+
+    @classmethod
+    def _rollback(cls, connection: Connection) -> None:
+        """Roll back the given connection."""
+        if connection.transaction_open is False:
+            raise dbt.exceptions.InternalException(
+                f'Tried to rollback transaction on connection '
+                f'"{connection.name}", but it does not have one open!'
+            )
+
+        logger.debug(f'On {connection.name}: ROLLBACK')
+        cls._rollback_handle(connection)
+
+        connection.transaction_open = False
+
+    @classmethod
+    def close(cls, connection: Connection) -> Connection:
+        # if the connection is in closed or init, there's nothing to do
+        if connection.state in {ConnectionState.CLOSED, ConnectionState.INIT}:
+            return connection
+
+        if connection.transaction_open and connection.handle:
+            logger.debug('On {}: ROLLBACK'.format(connection.name))
+            cls._rollback_handle(connection)
+        connection.transaction_open = False
+
+        cls._close_handle(connection)
+        connection.state = ConnectionState.CLOSED
+
+        return connection
+
+    def commit_if_has_connection(self) -> None:
+        """If the named connection exists, commit the current transaction."""
+        connection = self.get_if_exists()
+        if connection:
+            self.commit()
+
+    def _add_query_comment(self, sql: str) -> str:
+        if self.query_header is None:
+            return sql
+        return self.query_header.add(sql)
+
+    @abc.abstractmethod
+    def execute(
+        self, sql: str, auto_begin: bool = False, fetch: bool = False
+    ) -> Tuple[Union[str, AdapterResponse], agate.Table]:
+        """Execute the given SQL.
+
+        :param str sql: The sql to execute.
+        :param bool auto_begin: If set, and dbt is not currently inside a
+            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[Union[str, AdapterResponse], agate.Table]
+        """
+        raise dbt.exceptions.NotImplementedException(
+            '`execute` is not implemented for this adapter!'
+        )

core/dbt/adapters/base/meta.py

@@ -0,0 +1,126 @@
+import abc
+from functools import wraps
+from typing import Callable, Optional, Any, FrozenSet, Dict, Set
+
+from dbt.deprecations import warn, renamed_method
+
+
+Decorator = Callable[[Any], Callable]
+
+
+class _Available:
+    def __call__(self, func: Callable) -> Callable:
+        func._is_available_ = True  # type: ignore
+        return func
+
+    def parse(self, parse_replacement: Callable) -> Decorator:
+        """A decorator factory to indicate that a method on the adapter will be
+        exposed to the database wrapper, and will be stubbed out at parse time
+        with the given function.
+
+        @available.parse()
+        def my_method(self, a, b):
+            if something:
+                return None
+            return big_expensive_db_query()
+
+        @available.parse(lambda *args, **args: {})
+        def my_other_method(self, a, b):
+            x = {}
+            x.update(big_expensive_db_query())
+            return x
+        """
+        def inner(func):
+            func._parse_replacement_ = parse_replacement
+            return self(func)
+        return inner
+
+    def deprecated(
+        self, supported_name: str, parse_replacement: Optional[Callable] = None
+    ) -> Decorator:
+        """A decorator that marks a function as available, but also prints a
+        deprecation warning. Use like
+
+        @available.deprecated('my_new_method')
+        def my_old_method(self, arg):
+            args = compatability_shim(arg)
+            return self.my_new_method(*args)
+
+        @available.deprecated('my_new_slow_method', lambda *a, **k: (0, ''))
+        def my_old_slow_method(self, arg):
+            args = compatibility_shim(arg)
+            return self.my_new_slow_method(*args)
+
+        To make `adapter.my_old_method` available but also print out a warning
+        on use directing users to `my_new_method`.
+
+        The optional parse_replacement, if provided, will provide a parse-time
+        replacement for the actual method (see `available.parse`).
+        """
+        def wrapper(func):
+            func_name = func.__name__
+            renamed_method(func_name, supported_name)
+
+            @wraps(func)
+            def inner(*args, **kwargs):
+                warn('adapter:{}'.format(func_name))
+                return func(*args, **kwargs)
+
+            if parse_replacement:
+                available_function = self.parse(parse_replacement)
+            else:
+                available_function = self
+            return available_function(inner)
+        return wrapper
+
+    def parse_none(self, func: Callable) -> Callable:
+        wrapper = self.parse(lambda *a, **k: None)
+        return wrapper(func)
+
+    def parse_list(self, func: Callable) -> Callable:
+        wrapper = self.parse(lambda *a, **k: [])
+        return wrapper(func)
+
+
+available = _Available()
+
+
+class AdapterMeta(abc.ABCMeta):
+    _available_: FrozenSet[str]
+    _parse_replacements_: Dict[str, Callable]
+
+    def __new__(mcls, name, bases, namespace, **kwargs):
+        # mypy does not like the `**kwargs`. But `ABCMeta` itself takes
+        # `**kwargs` in its argspec here (and passes them to `type.__new__`.
+        # I'm not sure there is any benefit to it after poking around a bit,
+        # but having it doesn't hurt on the python side (and omitting it could
+        # hurt for obscure metaclass reasons, for all I know)
+        cls = abc.ABCMeta.__new__(  # type: ignore
+            mcls, name, bases, namespace, **kwargs
+        )
+
+        # this is very much inspired by ABCMeta's own implementation
+
+        # dict mapping the method name to whether the model name should be
+        # injected into the arguments. All methods in here are exposed to the
+        # context.
+        available: Set[str] = set()
+        replacements: Dict[str, Any] = {}
+
+        # collect base class data first
+        for base in bases:
+            available.update(getattr(base, '_available_', set()))
+            replacements.update(getattr(base, '_parse_replacements_', set()))
+
+        # override with local data if it exists
+        for name, value in namespace.items():
+            if getattr(value, '_is_available_', False):
+                available.add(name)
+            parse_replacement = getattr(value, '_parse_replacement_', None)
+            if parse_replacement is not None:
+                replacements[name] = parse_replacement
+
+        cls._available_ = frozenset(available)
+        # should this be a namedtuple so it will be immutable like _available_?
+        cls._parse_replacements_ = replacements
+        return cls

core/dbt/adapters/base/plugin.py

@@ -0,0 +1,42 @@
+from typing import List, Optional, Type
+
+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:
+    """Defines the basic requirements for a dbt adapter plugin.
+
+    :param include_path: The path to this adapter plugin's root
+    :param dependencies: A list of adapter names that this adapter depends
+        upon.
+    """
+    def __init__(
+        self,
+        adapter: Type[AdapterProtocol],
+        credentials: Type[Credentials],
+        include_path: str,
+        dependencies: Optional[List[str]] = None
+    ):
+
+        self.adapter: Type[AdapterProtocol] = adapter
+        self.credentials: Type[Credentials] = credentials
+        self.include_path: str = include_path
+        self.project_name: str = project_name_from_path(include_path)
+        self.dependencies: List[str]
+        if dependencies is None:
+            self.dependencies = []
+        else:
+            self.dependencies = dependencies

core/dbt/adapters/base/query_headers.py

@@ -0,0 +1,101 @@
+from threading import local
+from typing import Optional, Callable, Dict, Any
+
+from dbt.clients.jinja import QueryStringGenerator
+
+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
+
+
+class NodeWrapper:
+    def __init__(self, node):
+        self._inner_node = node
+
+    def __getattr__(self, name):
+        return getattr(self._inner_node, name, '')
+
+
+class _QueryComment(local):
+    """A thread-local class storing thread-specific state information for
+    connection management, namely:
+        - the current thread's query comment.
+        - a source_name indicating what set the current thread's query comment
+    """
+    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
+
+        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
+            raise RuntimeException(
+                f'query comment contains illegal value "*/": {comment}'
+            )
+        self.query_comment = comment
+        self.append = append
+
+
+QueryStringFunc = Callable[[str, Optional[NodeWrapper]], str]
+
+
+class MacroQueryStringSetter:
+    def __init__(self, config: AdapterRequiredConfig, manifest: Manifest):
+        self.manifest = manifest
+        self.config = config
+
+        comment_macro = self._get_comment_macro()
+        self.generator: QueryStringFunc = lambda name, model: ''
+        # if the comment value was None or the empty string, just skip it
+        if comment_macro:
+            assert isinstance(comment_macro, str)
+            macro = '\n'.join((
+                '{%- macro query_comment_macro(connection_name, node) -%}',
+                comment_macro,
+                '{% endmacro %}'
+            ))
+            ctx = self._get_context()
+            self.generator = QueryStringGenerator(macro, ctx)
+        self.comment = _QueryComment(None)
+        self.reset()
+
+    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)
+
+    def add(self, sql: str) -> str:
+        return self.comment.add(sql)
+
+    def reset(self):
+        self.set('master', None)
+
+    def set(self, name: str, node: Optional[CompileResultNode]):
+        wrapped: Optional[NodeWrapper] = None
+        if node is not None:
+            wrapped = NodeWrapper(node)
+        comment_str = self.generator(name, wrapped)
+
+        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

@@ -0,0 +1,456 @@
+from collections.abc import Hashable
+from dataclasses import dataclass
+from typing import (
+    Optional, TypeVar, Any, Type, Dict, Union, Iterator, Tuple, Set
+)
+
+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.node_types import NodeType
+from dbt.utils import filter_null_values, deep_merge, classproperty
+
+import dbt.exceptions
+
+
+Self = TypeVar('Self', bound='BaseRelation')
+
+
+@dataclass(frozen=True, eq=False, repr=False)
+class BaseRelation(FakeAPIObject, Hashable):
+    path: Path
+    type: Optional[RelationType] = None
+    quote_character: str = '"'
+    include_policy: Policy = Policy()
+    quote_policy: Policy = Policy()
+    dbt_created: bool = False
+
+    def _is_exactish_match(self, field: ComponentName, value: str) -> bool:
+        if self.dbt_created and self.quote_policy.get_part(field) is False:
+            return self.path.get_lowered_part(field) == value.lower()
+        else:
+            return self.path.get_part(field) == value
+
+    @classmethod
+    def _get_field_named(cls, field_name):
+        for field, _ in cls._get_fields():
+            if field.name == field_name:
+                return field
+        # this should be unreachable
+        raise ValueError(f'BaseRelation has no {field_name} field!')
+
+    def __eq__(self, other):
+        if not isinstance(other, self.__class__):
+            return False
+        return self.to_dict(omit_none=True) == other.to_dict(omit_none=True)
+
+    @classmethod
+    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.
+        """
+        if key == 'metadata':
+            return {
+                'type': self.__class__.__name__
+            }
+        return super().get(key, default)
+
+    def matches(
+        self,
+        database: Optional[str] = None,
+        schema: Optional[str] = None,
+        identifier: Optional[str] = None,
+    ) -> bool:
+        search = filter_null_values({
+            ComponentName.Database: database,
+            ComponentName.Schema: schema,
+            ComponentName.Identifier: identifier
+        })
+
+        if not search:
+            # nothing was passed in
+            raise dbt.exceptions.RuntimeException(
+                "Tried to match relation, but no search path was passed!")
+
+        exact_match = True
+        approximate_match = True
+
+        for k, v in search.items():
+            if not self._is_exactish_match(k, v):
+                exact_match = False
+
+            if self.path.get_lowered_part(k) != v.lower():
+                approximate_match = False
+
+        if approximate_match and not exact_match:
+            target = self.create(
+                database=database, schema=schema, identifier=identifier
+            )
+            dbt.exceptions.approximate_relation_match(target, self)
+
+        return exact_match
+
+    def replace_path(self, **kwargs):
+        return self.replace(path=self.path.replace(**kwargs))
+
+    def quote(
+        self: Self,
+        database: Optional[bool] = None,
+        schema: Optional[bool] = None,
+        identifier: Optional[bool] = None,
+    ) -> Self:
+        policy = filter_null_values({
+            ComponentName.Database: database,
+            ComponentName.Schema: schema,
+            ComponentName.Identifier: identifier
+        })
+
+        new_quote_policy = self.quote_policy.replace_dict(policy)
+        return self.replace(quote_policy=new_quote_policy)
+
+    def include(
+        self: Self,
+        database: Optional[bool] = None,
+        schema: Optional[bool] = None,
+        identifier: Optional[bool] = None,
+    ) -> Self:
+        policy = filter_null_values({
+            ComponentName.Database: database,
+            ComponentName.Schema: schema,
+            ComponentName.Identifier: identifier
+        })
+
+        new_include_policy = self.include_policy.replace_dict(policy)
+        return self.replace(include_policy=new_include_policy)
+
+    def information_schema(self, view_name=None) -> 'InformationSchema':
+        # some of our data comes from jinja, where things can be `Undefined`.
+        if not isinstance(view_name, str):
+            view_name = None
+
+        # Kick the user-supplied schema out of the information schema relation
+        # Instead address this as <database>.information_schema by default
+        info_schema = InformationSchema.from_relation(self, view_name)
+        return info_schema.incorporate(path={"schema": None})
+
+    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]]]:
+
+        for key in ComponentName:
+            path_part: Optional[str] = None
+            if self.include_policy.get_part(key):
+                path_part = self.path.get_part(key)
+                if path_part is not None and self.quote_policy.get_part(key):
+                    path_part = self.quoted(path_part)
+            yield key, path_part
+
+    def render(self) -> str:
+        # 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(
+            quote_char=self.quote_character,
+            identifier=identifier,
+        )
+
+    @classmethod
+    def create_from_source(
+        cls: Type[Self], source: ParsedSourceDefinition, **kwargs: Any
+    ) -> Self:
+        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(omit_none=True),
+            source_quoting,
+            kwargs.get('quote_policy', {}),
+        )
+
+        return cls.create(
+            database=source.database,
+            schema=source.schema,
+            identifier=source.identifier,
+            quote_policy=quote_policy,
+            **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],
+        config: HasQuoting,
+        node: Union[ParsedNode, CompiledNode],
+        quote_policy: Optional[Dict[str, bool]] = None,
+        **kwargs: Any,
+    ) -> Self:
+        if quote_policy is None:
+            quote_policy = {}
+
+        quote_policy = dbt.utils.merge(config.quoting, quote_policy)
+
+        return cls.create(
+            database=node.database,
+            schema=node.schema,
+            identifier=node.alias,
+            quote_policy=quote_policy,
+            **kwargs)
+
+    @classmethod
+    def create_from(
+        cls: Type[Self],
+        config: HasQuoting,
+        node: Union[CompiledNode, ParsedNode, ParsedSourceDefinition],
+        **kwargs: Any,
+    ) -> Self:
+        if node.resource_type == NodeType.Source:
+            if not isinstance(node, ParsedSourceDefinition):
+                raise InternalException(
+                    'type mismatch, expected ParsedSourceDefinition but got {}'
+                    .format(type(node))
+                )
+            return cls.create_from_source(node, **kwargs)
+        else:
+            if not isinstance(node, (ParsedNode, CompiledNode)):
+                raise InternalException(
+                    'type mismatch, expected ParsedNode or CompiledNode but '
+                    'got {}'.format(type(node))
+                )
+            return cls.create_from_node(config, node, **kwargs)
+
+    @classmethod
+    def create(
+        cls: Type[Self],
+        database: Optional[str] = None,
+        schema: Optional[str] = None,
+        identifier: Optional[str] = None,
+        type: Optional[RelationType] = None,
+        **kwargs,
+    ) -> Self:
+        kwargs.update({
+            'path': {
+                'database': database,
+                'schema': schema,
+                'identifier': identifier,
+            },
+            'type': type,
+        })
+        return cls.from_dict(kwargs)
+
+    def __repr__(self) -> str:
+        return "<{} {}>".format(self.__class__.__name__, self.render())
+
+    def __hash__(self) -> int:
+        return hash(self.render())
+
+    def __str__(self) -> str:
+        return self.render()
+
+    @property
+    def database(self) -> Optional[str]:
+        return self.path.database
+
+    @property
+    def schema(self) -> Optional[str]:
+        return self.path.schema
+
+    @property
+    def identifier(self) -> Optional[str]:
+        return self.path.identifier
+
+    @property
+    def table(self) -> Optional[str]:
+        return self.path.identifier
+
+    # Here for compatibility with old Relation interface
+    @property
+    def name(self) -> Optional[str]:
+        return self.identifier
+
+    @property
+    def is_table(self) -> bool:
+        return self.type == RelationType.Table
+
+    @property
+    def is_cte(self) -> bool:
+        return self.type == RelationType.CTE
+
+    @property
+    def is_view(self) -> bool:
+        return self.type == RelationType.View
+
+    @classproperty
+    def Table(cls) -> str:
+        return str(RelationType.Table)
+
+    @classproperty
+    def CTE(cls) -> str:
+        return str(RelationType.CTE)
+
+    @classproperty
+    def View(cls) -> str:
+        return str(RelationType.View)
+
+    @classproperty
+    def External(cls) -> str:
+        return str(RelationType.External)
+
+    @classproperty
+    def get_relation_type(cls) -> Type[RelationType]:
+        return RelationType
+
+
+Info = TypeVar('Info', bound='InformationSchema')
+
+
+@dataclass(frozen=True, eq=False, repr=False)
+class InformationSchema(BaseRelation):
+    information_schema_view: Optional[str] = None
+
+    def __post_init__(self):
+        if not isinstance(self.information_schema_view, (type(None), str)):
+            raise dbt.exceptions.CompilationException(
+                'Got an invalid name: {}'.format(self.information_schema_view)
+            )
+
+    @classmethod
+    def get_path(
+        cls, relation: BaseRelation, information_schema_view: Optional[str]
+    ) -> Path:
+        return Path(
+            database=relation.database,
+            schema=relation.schema,
+            identifier='INFORMATION_SCHEMA',
+        )
+
+    @classmethod
+    def get_include_policy(
+        cls,
+        relation,
+        information_schema_view: Optional[str],
+    ) -> Policy:
+        return relation.include_policy.replace(
+            database=relation.database is not None,
+            schema=False,
+            identifier=True,
+        )
+
+    @classmethod
+    def get_quote_policy(
+        cls,
+        relation,
+        information_schema_view: Optional[str],
+    ) -> Policy:
+        return relation.quote_policy.replace(
+            identifier=False,
+        )
+
+    @classmethod
+    def from_relation(
+        cls: Type[Info],
+        relation: BaseRelation,
+        information_schema_view: Optional[str],
+    ) -> Info:
+        include_policy = cls.get_include_policy(
+            relation, information_schema_view
+        )
+        quote_policy = cls.get_quote_policy(relation, information_schema_view)
+        path = cls.get_path(relation, information_schema_view)
+        return cls(
+            type=RelationType.View,
+            path=path,
+            include_policy=include_policy,
+            quote_policy=quote_policy,
+            information_schema_view=information_schema_view,
+        )
+
+    def _render_iterator(self):
+        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

@@ -0,0 +1,523 @@
+from collections import namedtuple
+from copy import deepcopy
+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 _make_key(relation) -> _ReferenceKey:
+    """Make _ReferenceKeys with lowercase values for the cache so we don't have
+    to keep track of quoting
+    """
+    # databases and schemas can both be None
+    return _ReferenceKey(lowercase(relation.database),
+                         lowercase(relation.schema),
+                         lowercase(relation.identifier))
+
+
+def dot_separated(key: _ReferenceKey) -> str:
+    """Return the key in dot-separated string form.
+
+    :param _ReferenceKey key: The key to stringify.
+    """
+    return '.'.join(map(str, key))
+
+
+class _CachedRelation:
+    """Nothing about _CachedRelation is guaranteed to be thread-safe!
+
+    :attr str schema: The schema of this relation.
+    :attr str identifier: The identifier of this relation.
+    :attr Dict[_ReferenceKey, _CachedRelation] referenced_by: The relations
+        that refer to this relation.
+    :attr BaseRelation inner: The underlying dbt relation.
+    """
+    def __init__(self, inner):
+        self.referenced_by = {}
+        self.inner = inner
+
+    def __str__(self) -> str:
+        return (
+            '_CachedRelation(database={}, schema={}, identifier={}, inner={})'
+        ).format(self.database, self.schema, self.identifier, self.inner)
+
+    @property
+    def database(self) -> Optional[str]:
+        return lowercase(self.inner.database)
+
+    @property
+    def schema(self) -> Optional[str]:
+        return lowercase(self.inner.schema)
+
+    @property
+    def identifier(self) -> Optional[str]:
+        return lowercase(self.inner.identifier)
+
+    def __copy__(self):
+        new = self.__class__(self.inner)
+        new.__dict__.update(self.__dict__)
+        return new
+
+    def __deepcopy__(self, memo):
+        new = self.__class__(self.inner.incorporate())
+        new.__dict__.update(self.__dict__)
+        new.referenced_by = deepcopy(self.referenced_by, memo)
+
+    def is_referenced_by(self, key):
+        return key in self.referenced_by
+
+    def key(self):
+        """Get the _ReferenceKey that represents this relation
+
+        :return _ReferenceKey: A key for this relation.
+        """
+        return _make_key(self)
+
+    def add_reference(self, referrer: '_CachedRelation'):
+        """Add a reference from referrer to self, indicating that if this node
+        were drop...cascaded, the referrer would be dropped as well.
+
+        :param _CachedRelation referrer: The node that refers to this node.
+        """
+        self.referenced_by[referrer.key()] = referrer
+
+    def collect_consequences(self):
+        """Recursively collect a set of _ReferenceKeys that would
+        consequentially get dropped if this were dropped via
+        "drop ... cascade".
+
+        :return Set[_ReferenceKey]: All the relations that would be dropped
+        """
+        consequences = {self.key()}
+        for relation in self.referenced_by.values():
+            consequences.update(relation.collect_consequences())
+        return consequences
+
+    def release_references(self, keys):
+        """Non-recursively indicate that an iterable of _ReferenceKey no longer
+        exist. Unknown keys are ignored.
+
+        :param Iterable[_ReferenceKey] keys: The keys to drop.
+        """
+        keys = set(self.referenced_by) & set(keys)
+        for key in keys:
+            self.referenced_by.pop(key)
+
+    def rename(self, new_relation):
+        """Rename this cached relation to new_relation.
+        Note that this will change the output of key(), all refs must be
+        updated!
+
+        :param _CachedRelation new_relation: The new name to apply to the
+            relation
+        """
+        # Relations store this stuff inside their `path` dict. But they
+        # also store a table_name, and usually use it in their  .render(),
+        # so we need to update that as well. It doesn't appear that
+        # table_name is ever anything but the identifier (via .create())
+        self.inner = self.inner.incorporate(
+            path={
+                'database': new_relation.inner.database,
+                'schema': new_relation.inner.schema,
+                'identifier': new_relation.inner.identifier
+            },
+        )
+
+    def rename_key(self, old_key, new_key):
+        """Rename a reference that may or may not exist. Only handles the
+        reference itself, so this is the other half of what `rename` does.
+
+        If old_key is not in referenced_by, this is a no-op.
+
+        :param _ReferenceKey old_key: The old key to be renamed.
+        :param _ReferenceKey new_key: The new key to rename to.
+        :raises InternalError: If the new key already exists.
+        """
+        if new_key in self.referenced_by:
+            dbt.exceptions.raise_cache_inconsistent(
+                'in rename of "{}" -> "{}", new name is in the cache already'
+                .format(old_key, new_key)
+            )
+
+        if old_key not in self.referenced_by:
+            return
+        value = self.referenced_by.pop(old_key)
+        self.referenced_by[new_key] = value
+
+    def dump_graph_entry(self):
+        """Return a key/value pair representing this key and its referents.
+
+        return List[str]: The dot-separated form of all referent keys.
+        """
+        return [dot_separated(r) for r in self.referenced_by]
+
+
+def lazy_log(msg, func):
+    if logger.disabled:
+        return
+    logger.debug(msg.format(func()))
+
+
+class RelationsCache:
+    """A cache of the relations known to dbt. Keeps track of relationships
+    declared between tables and handles renames/drops as a real database would.
+
+    :attr Dict[_ReferenceKey, _CachedRelation] relations: The known relations.
+    :attr threading.RLock lock: The lock around relations, held during updates.
+        The adapters also hold this lock while filling the cache.
+    :attr Set[str] schemas: The set of known/cached schemas, all lowercased.
+    """
+    def __init__(self) -> None:
+        self.relations: Dict[_ReferenceKey, _CachedRelation] = {}
+        self.lock = threading.RLock()
+        self.schemas: Set[Tuple[Optional[str], Optional[str]]] = set()
+
+    def add_schema(
+        self, database: Optional[str], schema: Optional[str],
+    ) -> None:
+        """Add a schema to the set of known schemas (case-insensitive)
+
+        :param database: The database name to add.
+        :param schema: The schema name to add.
+        """
+        self.schemas.add((lowercase(database), lowercase(schema)))
+
+    def drop_schema(
+        self, database: Optional[str], schema: Optional[str],
+    ) -> None:
+        """Drop the given schema and remove it from the set of known schemas.
+
+        Then remove all its contents (and their dependents, etc) as well.
+        """
+        key = (lowercase(database), lowercase(schema))
+        if key not in self.schemas:
+            return
+
+        # avoid iterating over self.relations while removing things by
+        # collecting the list first.
+
+        with self.lock:
+            to_remove = self._list_relations_in_schema(database, schema)
+            self._remove_all(to_remove)
+            # handle a drop_schema race by using discard() over remove()
+            self.schemas.discard(key)
+
+    def update_schemas(self, schemas: Iterable[Tuple[Optional[str], str]]):
+        """Add multiple schemas to the set of known schemas (case-insensitive)
+
+        :param schemas: An iterable of the schema names to add.
+        """
+        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
+        schemas.
+
+        :param schema_id: The db name and schema name to look up.
+        """
+        db, schema = schema_id
+        return (lowercase(db), schema.lower()) in self.schemas
+
+    def dump_graph(self):
+        """Dump a key-only representation of the schema to a dictionary. Every
+        known relation is a key with a value of a list of keys it is referenced
+        by.
+        """
+        # we have to hold the lock for the entire dump, if other threads modify
+        # self.relations or any cache entry's referenced_by during iteration
+        # it's a runtime error!
+        with self.lock:
+            return {
+                dot_separated(k): v.dump_graph_entry()
+                for k, v in self.relations.items()
+            }
+
+    def _setdefault(self, relation: _CachedRelation):
+        """Add a relation to the cache, or return it if it already exists.
+
+        :param _CachedRelation relation: The relation to set or get.
+        :return _CachedRelation: The relation stored under the given relation's
+            key
+        """
+        self.add_schema(relation.database, relation.schema)
+        key = relation.key()
+        return self.relations.setdefault(key, relation)
+
+    def _add_link(self, referenced_key, dependent_key):
+        """Add a link between two relations to the database. Both the old and
+        new entries must alraedy exist in the database.
+
+        :param _ReferenceKey referenced_key: The key identifying the referenced
+            model (the one that if dropped will drop the dependent model).
+        :param _ReferenceKey dependent_key: The key identifying the dependent
+            model.
+        :raises InternalError: If either entry does not exist.
+        """
+        referenced = self.relations.get(referenced_key)
+        if referenced is None:
+            return
+        if referenced is None:
+            dbt.exceptions.raise_cache_inconsistent(
+                'in add_link, referenced link key {} not in cache!'
+                .format(referenced_key)
+            )
+
+        dependent = self.relations.get(dependent_key)
+        if dependent is None:
+            dbt.exceptions.raise_cache_inconsistent(
+                'in add_link, dependent link key {} not in cache!'
+                .format(dependent_key)
+            )
+
+        assert dependent is not None  # we just raised!
+
+        referenced.add_reference(dependent)
+
+    def add_link(self, referenced, dependent):
+        """Add a link between two relations to the database. If either relation
+        does not exist, it will be added as an "external" relation.
+
+        The dependent model refers _to_ the referenced model. So, given
+        arguments of (jake_test, bar, jake_test, foo):
+        both values are in the schema jake_test and foo is a view that refers
+        to bar, so "drop bar cascade" will drop foo and all of foo's
+        dependents.
+
+        :param BaseRelation referenced: The referenced model.
+        :param BaseRelation dependent: The dependent model.
+        :raises InternalError: If either entry does not exist.
+        """
+        ref_key = _make_key(referenced)
+        if (ref_key.database, ref_key.schema) not in self:
+            # if we have not cached the referenced schema at all, we must be
+            # referring to a table outside our control. There's no need to make
+            # a link - we will never drop the referenced relation during a run.
+            logger.debug(
+                '{dep!s} references {ref!s} but {ref.database}.{ref.schema} '
+                'is not in the cache, skipping assumed external relation'
+                .format(dep=dependent, ref=ref_key)
+            )
+            return
+        if ref_key not in self.relations:
+            # Insert a dummy "external" relation.
+            referenced = referenced.replace(
+                type=referenced.External
+            )
+            self.add(referenced)
+
+        dep_key = _make_key(dependent)
+        if dep_key not in self.relations:
+            # Insert a dummy "external" relation.
+            dependent = dependent.replace(
+                type=referenced.External
+            )
+            self.add(dependent)
+        logger.debug(
+            'adding link, {!s} references {!s}'.format(dep_key, ref_key)
+        )
+        with self.lock:
+            self._add_link(ref_key, dep_key)
+
+    def add(self, relation):
+        """Add the relation inner to the cache, under the schema schema and
+        identifier identifier
+
+        :param BaseRelation relation: The underlying relation.
+        """
+        cached = _CachedRelation(relation)
+        logger.debug('Adding relation: {!s}'.format(cached))
+
+        lazy_log('before adding: {!s}', self.dump_graph)
+
+        with self.lock:
+            self._setdefault(cached)
+
+        lazy_log('after adding: {!s}', self.dump_graph)
+
+    def _remove_refs(self, keys):
+        """Removes all references to all entries in keys. This does not
+        cascade!
+
+        :param Iterable[_ReferenceKey] keys: The keys to remove.
+        """
+        # remove direct refs
+        for key in keys:
+            del self.relations[key]
+        # then remove all entries from each child
+        for cached in self.relations.values():
+            cached.release_references(keys)
+
+    def _drop_cascade_relation(self, dropped):
+        """Drop the given relation and cascade it appropriately to all
+        dependent relations.
+
+        :param _CachedRelation dropped: An existing _CachedRelation to drop.
+        """
+        if dropped not in self.relations:
+            logger.debug('dropped a nonexistent relationship: {!s}'
+                         .format(dropped))
+            return
+        consequences = self.relations[dropped].collect_consequences()
+        logger.debug(
+            'drop {} is cascading to {}'.format(dropped, consequences)
+        )
+        self._remove_refs(consequences)
+
+    def drop(self, relation):
+        """Drop the named relation and cascade it appropriately to all
+        dependent relations.
+
+        Because dbt proactively does many `drop relation if exist ... cascade`
+        that are noops, nonexistent relation drops cause a debug log and no
+        other actions.
+
+        :param str schema: The schema of the relation to drop.
+        :param str identifier: The identifier of the relation to drop.
+        """
+        dropped = _make_key(relation)
+        logger.debug('Dropping relation: {!s}'.format(dropped))
+        with self.lock:
+            self._drop_cascade_relation(dropped)
+
+    def _rename_relation(self, old_key, new_relation):
+        """Rename a relation named old_key to new_key, updating references.
+        Return whether or not there was a key to rename.
+
+        :param _ReferenceKey old_key: The existing key, to rename from.
+        :param _CachedRelation new_key: The new relation, to rename to.
+        """
+        # On the database level, a rename updates all values that were
+        # previously referenced by old_name to be referenced by new_name.
+        # basically, the name changes but some underlying ID moves. Kind of
+        # like an object reference!
+        relation = self.relations.pop(old_key)
+        new_key = new_relation.key()
+
+        # relaton has to rename its innards, so it needs the _CachedRelation.
+        relation.rename(new_relation)
+        # update all the relations that refer to it
+        for cached in self.relations.values():
+            if cached.is_referenced_by(old_key):
+                logger.debug(
+                    'updated reference from {0} -> {2} to {1} -> {2}'
+                    .format(old_key, new_key, cached.key())
+                )
+                cached.rename_key(old_key, new_key)
+
+        self.relations[new_key] = relation
+        # also fixup the schemas!
+        self.add_schema(new_key.database, new_key.schema)
+
+        return True
+
+    def _check_rename_constraints(self, old_key, new_key):
+        """Check the rename constraints, and return whether or not the rename
+        can proceed.
+
+        If the new key is already present, that is an error.
+        If the old key is absent, we debug log and return False, assuming it's
+        a temp table being renamed.
+
+        :param _ReferenceKey old_key: The existing key, to rename from.
+        :param _ReferenceKey new_key: The new key, to rename to.
+        :return bool: If the old relation exists for renaming.
+        :raises InternalError: If the new key is already present.
+        """
+        if new_key in self.relations:
+            dbt.exceptions.raise_cache_inconsistent(
+                'in rename, new key {} already in cache: {}'
+                .format(new_key, list(self.relations.keys()))
+            )
+
+        if old_key not in self.relations:
+            logger.debug(
+                'old key {} not found in self.relations, assuming temporary'
+                .format(old_key)
+            )
+            return False
+        return True
+
+    def rename(self, old, new):
+        """Rename the old schema/identifier to the new schema/identifier and
+        update references.
+
+        If the new schema/identifier is already present, that is an error.
+        If the schema/identifier key is absent, we only debug log and return,
+        assuming it's a temp table being renamed.
+
+        :param BaseRelation old: The existing relation name information.
+        :param BaseRelation new: The new relation name information.
+        :raises InternalError: If the new key is already present.
+        """
+        old_key = _make_key(old)
+        new_key = _make_key(new)
+        logger.debug('Renaming relation {!s} to {!s}'.format(
+            old_key, new_key
+        ))
+
+        lazy_log('before rename: {!s}', self.dump_graph)
+
+        with self.lock:
+            if self._check_rename_constraints(old_key, new_key):
+                self._rename_relation(old_key, _CachedRelation(new))
+            else:
+                self._setdefault(_CachedRelation(new))
+
+        lazy_log('after rename: {!s}', self.dump_graph)
+
+    def get_relations(
+        self, database: Optional[str], schema: Optional[str]
+    ) -> List[Any]:
+        """Case-insensitively yield all relations matching the given schema.
+
+        :param str schema: The case-insensitive schema name to list from.
+        :return List[BaseRelation]: The list of relations with the given
+            schema
+        """
+        database = lowercase(database)
+        schema = lowercase(schema)
+        with self.lock:
+            results = [
+                r.inner for r in self.relations.values()
+                if (lowercase(r.schema) == schema and
+                    lowercase(r.database) == database)
+            ]
+
+        if None in results:
+            dbt.exceptions.raise_cache_inconsistent(
+                'in get_relations, a None relation was found in the cache!'
+            )
+        return results
+
+    def clear(self):
+        """Clear the cache"""
+        with self.lock:
+            self.relations.clear()
+            self.schemas.clear()
+
+    def _list_relations_in_schema(
+        self, database: Optional[str], schema: Optional[str]
+    ) -> List[_CachedRelation]:
+        """Get the relations in a schema. Callers should hold the lock."""
+        key = (lowercase(database), lowercase(schema))
+
+        to_remove: List[_CachedRelation] = []
+        for cachekey, relation in self.relations.items():
+            if (cachekey.database, cachekey.schema) == key:
+                to_remove.append(relation)
+        return to_remove
+
+    def _remove_all(self, to_remove: List[_CachedRelation]):
+        """Remove all the listed relations. Ignore relations that have been
+        cascaded out.
+        """
+        for relation in to_remove:
+            # it may have been cascaded out already
+            drop_key = _make_key(relation)
+            if drop_key in self.relations:
+                self.drop(drop_key)

core/dbt/adapters/factory.py

@@ -0,0 +1,227 @@
+import threading
+from pathlib import Path
+from importlib import import_module
+from typing import Type, Dict, Any, List, Optional, Set
+
+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.protocol import (
+    AdapterProtocol,
+    AdapterConfig,
+    RelationProtocol,
+)
+from dbt.adapters.base.plugin import AdapterPlugin
+
+
+Adapter = AdapterProtocol
+
+
+class AdapterContainer:
+    def __init__(self):
+        self.lock = threading.Lock()
+        self.adapters: Dict[str, 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_plugin_by_name(self, name: str) -> AdapterPlugin:
+        with self.lock:
+            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_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
+        # and adapter_type entries with the same value, as they're all
+        # singletons
+        try:
+            # mypy doesn't think modules have any attributes.
+            mod: Any = import_module('.' + name, 'dbt.adapters')
+        except ModuleNotFoundError as exc:
+            # if we failed to import the target module in particular, inform
+            # the user about it via a runtime error
+            if exc.name == 'dbt.adapters.' + name:
+                raise RuntimeException(f'Could not find adapter type {name}!')
+            logger.info(f'Error importing adapter: {exc}')
+            # otherwise, the error had to have come from some underlying
+            # library. Log the stack trace.
+            logger.debug('', exc_info=True)
+            raise
+        plugin: AdapterPlugin = mod.Plugin
+        plugin_type = plugin.adapter.type()
+
+        if plugin_type != name:
+            raise RuntimeException(
+                f'Expected to find adapter with type named {name}, got '
+                f'adapter with type {plugin_type}'
+            )
+
+        with self.lock:
+            # things do hold the lock to iterate over it so we need it to add
+            self.plugins[name] = plugin
+
+        self.packages[plugin.project_name] = Path(plugin.include_path)
+
+        for dep in plugin.dependencies:
+            self.load_plugin(dep)
+
+        return plugin.credentials
+
+    def register_adapter(self, config: AdapterRequiredConfig) -> None:
+        adapter_name = config.credentials.type
+        adapter_type = self.get_adapter_class_by_name(adapter_name)
+
+        with self.lock:
+            if adapter_name in self.adapters:
+                # this shouldn't really happen...
+                return
+
+            adapter: Adapter = adapter_type(config)  # type: ignore
+            self.adapters[adapter_name] = adapter
+
+    def lookup_adapter(self, adapter_name: str) -> Adapter:
+        return self.adapters[adapter_name]
+
+    def reset_adapters(self):
+        """Clear the adapters. This is useful for tests, which change configs.
+        """
+        with self.lock:
+            for adapter in self.adapters.values():
+                adapter.cleanup_connections()
+            self.adapters.clear()
+
+    def cleanup_connections(self):
+        """Only clean up the adapter connections list without resetting the
+        actual adapters.
+        """
+        with self.lock:
+            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())
+
+        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:
+    FACTORY.register_adapter(config)
+
+
+def get_adapter(config: AdapterRequiredConfig):
+    return FACTORY.lookup_adapter(config.credentials.type)
+
+
+def reset_adapters():
+    """Clear the adapters. This is useful for tests, which change configs.
+    """
+    FACTORY.reset_adapters()
+
+
+def cleanup_connections():
+    """Only clean up the adapter connections list without resetting the actual
+    adapters.
+    """
+    FACTORY.cleanup_connections()
+
+
+def get_adapter_class_by_name(name: str) -> Type[AdapterProtocol]:
+    return FACTORY.get_adapter_class_by_name(name)
+
+
+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/__init__.py

@@ -0,0 +1,3 @@
+# these are all just exports, #noqa them so flake8 will be happy
+from dbt.adapters.sql.connections import SQLConnectionManager  # noqa
+from dbt.adapters.sql.impl import SQLAdapter  # noqa

core/dbt/adapters/sql/connections.py

@@ -0,0 +1,168 @@
+import abc
+import time
+from typing import List, Optional, Tuple, Any, Iterable, Dict, Union
+
+import agate
+
+import dbt.clients.agate_helper
+import dbt.exceptions
+from dbt.adapters.base import BaseConnectionManager
+from dbt.contracts.connection import (
+    Connection, ConnectionState, AdapterResponse
+)
+from dbt.logger import GLOBAL_LOGGER as logger
+
+
+class SQLConnectionManager(BaseConnectionManager):
+    """The default connection manager with some common SQL methods implemented.
+
+    Methods to implement:
+        - exception_handler
+        - cancel
+        - get_response
+        - open
+    """
+    @abc.abstractmethod
+    def cancel(self, connection: Connection):
+        """Cancel the given connection."""
+        raise dbt.exceptions.NotImplementedException(
+            '`cancel` is not implemented for this adapter!'
+        )
+
+    def cancel_open(self) -> List[str]:
+        names = []
+        this_connection = self.get_if_exists()
+        with self.lock:
+            for connection in self.thread_connections.values():
+                if connection is this_connection:
+                    continue
+
+                # if the connection failed, the handle will be None so we have
+                # nothing to cancel.
+                if (
+                    connection.handle is not None and
+                    connection.state == ConnectionState.OPEN
+                ):
+                    self.cancel(connection)
+                if connection.name is not None:
+                    names.append(connection.name)
+        return names
+
+    def add_query(
+        self,
+        sql: str,
+        auto_begin: bool = True,
+        bindings: Optional[Any] = None,
+        abridge_sql_log: bool = False
+    ) -> Tuple[Connection, Any]:
+        connection = self.get_thread_connection()
+        if auto_begin and connection.transaction_open is False:
+            self.begin()
+
+        logger.debug('Using {} connection "{}".'
+                     .format(self.TYPE, connection.name))
+
+        with self.exception_handler(sql):
+            if abridge_sql_log:
+                log_sql = '{}...'.format(sql[:512])
+            else:
+                log_sql = sql
+
+            logger.debug(
+                'On {connection_name}: {sql}',
+                connection_name=connection.name,
+                sql=log_sql,
+            )
+            pre = time.time()
+
+            cursor = connection.handle.cursor()
+            cursor.execute(sql, bindings)
+            logger.debug(
+                "SQL status: {status} in {elapsed:0.2f} seconds",
+                status=self.get_response(cursor),
+                elapsed=(time.time() - pre)
+            )
+
+            return connection, cursor
+
+    @abc.abstractclassmethod
+    def get_response(cls, cursor: Any) -> Union[AdapterResponse, str]:
+        """Get the status of the cursor."""
+        raise dbt.exceptions.NotImplementedException(
+            '`get_response` is not implemented for this adapter!'
+        )
+
+    @classmethod
+    def process_results(
+        cls,
+        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
+    def get_result_from_cursor(cls, cursor: Any) -> agate.Table:
+        data: List[Any] = []
+        column_names: List[str] = []
+
+        if cursor.description is not None:
+            column_names = [col[0] for col in cursor.description]
+            rows = cursor.fetchall()
+            data = cls.process_results(column_names, rows)
+
+        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[Union[AdapterResponse, str], agate.Table]:
+        sql = self._add_query_comment(sql)
+        _, cursor = self.add_query(sql, auto_begin)
+        response = self.get_response(cursor)
+        if fetch:
+            table = self.get_result_from_cursor(cursor)
+        else:
+            table = dbt.clients.agate_helper.empty_table()
+        return response, table
+
+    def add_begin_query(self):
+        return self.add_query('BEGIN', auto_begin=False)
+
+    def add_commit_query(self):
+        return self.add_query('COMMIT', auto_begin=False)
+
+    def begin(self):
+        connection = self.get_thread_connection()
+        if connection.transaction_open is True:
+            raise dbt.exceptions.InternalException(
+                'Tried to begin a new transaction on connection "{}", but '
+                'it already had one open!'.format(connection.name))
+
+        self.add_begin_query()
+
+        connection.transaction_open = True
+        return connection
+
+    def commit(self):
+        connection = self.get_thread_connection()
+        if connection.transaction_open is False:
+            raise dbt.exceptions.InternalException(
+                'Tried to commit transaction on connection "{}", but '
+                'it does not have one open!'.format(connection.name))
+
+        logger.debug('On {}: COMMIT'.format(connection.name))
+        self.add_commit_query()
+
+        connection.transaction_open = False
+
+        return connection

core/dbt/adapters/sql/impl.py

@@ -0,0 +1,250 @@
+import agate
+from typing import Any, Optional, Tuple, Type, List
+
+import dbt.clients.agate_helper
+from dbt.contracts.connection import Connection
+import dbt.exceptions
+from dbt.adapters.base import BaseAdapter, available
+from dbt.adapters.sql import SQLConnectionManager
+from dbt.logger import GLOBAL_LOGGER as logger
+
+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'
+LIST_SCHEMAS_MACRO_NAME = 'list_schemas'
+CHECK_SCHEMA_EXISTS_MACRO_NAME = 'check_schema_exists'
+CREATE_SCHEMA_MACRO_NAME = 'create_schema'
+DROP_SCHEMA_MACRO_NAME = 'drop_schema'
+RENAME_RELATION_MACRO_NAME = 'rename_relation'
+TRUNCATE_RELATION_MACRO_NAME = 'truncate_relation'
+DROP_RELATION_MACRO_NAME = 'drop_relation'
+ALTER_COLUMN_TYPE_MACRO_NAME = 'alter_column_type'
+
+
+class SQLAdapter(BaseAdapter):
+    """The default adapter with the common agate conversions and some SQL
+    methods implemented. This adapter has a different much shorter list of
+    methods to implement, but some more macros that must be implemented.
+
+    To implement a macro, implement "${adapter_type}__${macro_name}". in the
+    adapter's internal project.
+
+    Methods to implement:
+        - date_function
+
+    Macros to implement:
+        - get_catalog
+        - list_relations_without_caching
+        - get_columns_in_relation
+    """
+
+    ConnectionManager: Type[SQLConnectionManager]
+    connections: SQLConnectionManager
+
+    @available.parse(lambda *a, **k: (None, None))
+    def add_query(
+        self,
+        sql: str,
+        auto_begin: bool = True,
+        bindings: Optional[Any] = None,
+        abridge_sql_log: bool = False,
+    ) -> Tuple[Connection, Any]:
+        """Add a query to the current transaction. A thin wrapper around
+        ConnectionManager.add_query.
+
+        :param sql: The SQL query to add
+        :param auto_begin: If set and there is no transaction in progress,
+            begin a new one.
+        :param bindings: An optional list of bindings for the query.
+        :param abridge_sql_log: If set, limit the raw sql logged to 512
+            characters
+        """
+        return self.connections.add_query(sql, auto_begin, bindings,
+                                          abridge_sql_log)
+
+    @classmethod
+    def convert_text_type(cls, agate_table: agate.Table, col_idx: int) -> str:
+        return "text"
+
+    @classmethod
+    def convert_number_type(
+        cls, agate_table: agate.Table, col_idx: int
+    ) -> str:
+        decimals = agate_table.aggregate(agate.MaxPrecision(col_idx))
+        return "float8" if decimals else "integer"
+
+    @classmethod
+    def convert_boolean_type(
+            cls, agate_table: agate.Table, col_idx: int
+    ) -> str:
+        return "boolean"
+
+    @classmethod
+    def convert_datetime_type(
+            cls, agate_table: agate.Table, col_idx: int
+    ) -> str:
+        return "timestamp without time zone"
+
+    @classmethod
+    def convert_date_type(cls, agate_table: agate.Table, col_idx: int) -> str:
+        return "date"
+
+    @classmethod
+    def convert_time_type(cls, agate_table: agate.Table, col_idx: int) -> str:
+        return "time"
+
+    @classmethod
+    def is_cancelable(cls) -> bool:
+        return True
+
+    def expand_column_types(self, goal, current):
+        reference_columns = {
+            c.name: c for c in
+            self.get_columns_in_relation(goal)
+        }
+
+        target_columns = {
+            c.name: c for c
+            in self.get_columns_in_relation(current)
+        }
+
+        for column_name, reference_column in reference_columns.items():
+            target_column = target_columns.get(column_name)
+
+            if target_column is not None and \
+               target_column.can_expand_to(reference_column):
+                col_string_size = reference_column.string_size()
+                new_type = self.Column.string_type(col_string_size)
+                logger.debug("Changing col type from {} to {} in table {}",
+                             target_column.data_type, new_type, current)
+
+                self.alter_column_type(current, column_name, new_type)
+
+    def alter_column_type(
+            self, relation, column_name, new_column_type
+    ) -> None:
+        """
+        1. Create a new column (w/ temp name and correct type)
+        2. Copy data over to it
+        3. Drop the existing column (cascade!)
+        4. Rename the new column to existing column
+        """
+        kwargs = {
+            'relation': relation,
+            'column_name': column_name,
+            'new_column_type': new_column_type,
+        }
+        self.execute_macro(
+            ALTER_COLUMN_TYPE_MACRO_NAME,
+            kwargs=kwargs
+        )
+
+    def drop_relation(self, relation):
+        if relation.type is None:
+            dbt.exceptions.raise_compiler_error(
+                'Tried to drop relation {}, but its type is null.'
+                .format(relation))
+
+        self.cache_dropped(relation)
+        self.execute_macro(
+            DROP_RELATION_MACRO_NAME,
+            kwargs={'relation': relation}
+        )
+
+    def truncate_relation(self, relation):
+        self.execute_macro(
+            TRUNCATE_RELATION_MACRO_NAME,
+            kwargs={'relation': relation}
+        )
+
+    def rename_relation(self, from_relation, to_relation):
+        self.cache_renamed(from_relation, to_relation)
+
+        kwargs = {'from_relation': from_relation, 'to_relation': to_relation}
+        self.execute_macro(
+            RENAME_RELATION_MACRO_NAME,
+            kwargs=kwargs
+        )
+
+    def get_columns_in_relation(self, relation):
+        return self.execute_macro(
+            GET_COLUMNS_IN_RELATION_MACRO_NAME,
+            kwargs={'relation': relation}
+        )
+
+    def create_schema(self, relation: BaseRelation) -> None:
+        relation = relation.without_identifier()
+        logger.debug('Creating schema "{}"', relation)
+        kwargs = {
+            '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, relation: BaseRelation) -> None:
+        relation = relation.without_identifier()
+        logger.debug('Dropping schema "{}".', relation)
+        kwargs = {
+            'relation': relation,
+        }
+        self.execute_macro(DROP_SCHEMA_MACRO_NAME, kwargs=kwargs)
+        # we can update the cache here
+        self.cache.drop_schema(relation.database, relation.schema)
+
+    def list_relations_without_caching(
+        self, schema_relation: BaseRelation,
+    ) -> List[BaseRelation]:
+        kwargs = {'schema_relation': schema_relation}
+        results = self.execute_macro(
+            LIST_RELATIONS_MACRO_NAME,
+            kwargs=kwargs
+        )
+
+        relations = []
+        quote_policy = {
+            'database': True,
+            'schema': True,
+            'identifier': True
+        }
+        for _database, name, _schema, _type in results:
+            try:
+                _type = self.Relation.get_relation_type(_type)
+            except ValueError:
+                _type = self.Relation.External
+            relations.append(self.Relation.create(
+                database=_database,
+                schema=_schema,
+                identifier=name,
+                quote_policy=quote_policy,
+                type=_type
+            ))
+        return relations
+
+    def quote(self, identifier):
+        return '"{}"'.format(identifier)
+
+    def list_schemas(self, database: str) -> List[str]:
+        results = self.execute_macro(
+            LIST_SCHEMAS_MACRO_NAME,
+            kwargs={'database': database}
+        )
+
+        return [row[0] for row in results]
+
+    def check_schema_exists(self, database: str, schema: str) -> bool:
+        information_schema = self.Relation.create(
+            database=database,
+            schema=schema,
+            identifier='INFORMATION_SCHEMA',
+            quote_policy=self.config.quoting
+        ).information_schema()
+
+        kwargs = {'information_schema': information_schema, 'schema': schema}
+        results = self.execute_macro(
+            CHECK_SCHEMA_EXISTS_MACRO_NAME,
+            kwargs=kwargs
+        )
+        return results[0][0] > 0

core/dbt/clients/_jinja_blocks.py

@@ -0,0 +1,393 @@
+import re
+from collections import namedtuple
+
+import dbt.exceptions
+
+
+def regex(pat):
+    return re.compile(pat, re.DOTALL | re.MULTILINE)
+
+
+class BlockData:
+    """raw plaintext data from the top level of the file."""
+    def __init__(self, contents):
+        self.block_type_name = '__dbt__data'
+        self.contents = contents
+        self.full_block = contents
+
+
+class BlockTag:
+    def __init__(self, block_type_name, block_name, contents=None,
+                 full_block=None, **kw):
+        self.block_type_name = block_type_name
+        self.block_name = block_name
+        self.contents = contents
+        self.full_block = full_block
+
+    def __str__(self):
+        return 'BlockTag({!r}, {!r})'.format(self.block_type_name,
+                                             self.block_name)
+
+    def __repr__(self):
+        return str(self)
+
+    @property
+    def end_block_type_name(self):
+        return 'end{}'.format(self.block_type_name)
+
+    def end_pat(self):
+        # we don't want to use string formatting here because jinja uses most
+        # of the string formatting operators in its syntax...
+        pattern = ''.join((
+            r'(?P<endblock>((?:\s*\{\%\-|\{\%)\s*',
+            self.end_block_type_name,
+            r'\s*(?:\-\%\}\s*|\%\})))',
+        ))
+        return regex(pattern)
+
+
+Tag = namedtuple('Tag', 'block_type_name block_name start end')
+
+
+_NAME_PATTERN = r'[A-Za-z_][A-Za-z_0-9]*'
+
+COMMENT_START_PATTERN = regex(r'(?:(?P<comment_start>(\s*\{\#)))')
+COMMENT_END_PATTERN = regex(r'(.*?)(\s*\#\})')
+RAW_START_PATTERN = regex(
+    r'(?:\s*\{\%\-|\{\%)\s*(?P<raw_start>(raw))\s*(?:\-\%\}\s*|\%\})'
+)
+EXPR_START_PATTERN = regex(r'(?P<expr_start>(\{\{\s*))')
+EXPR_END_PATTERN = regex(r'(?P<expr_end>(\s*\}\}))')
+
+BLOCK_START_PATTERN = regex(''.join((
+    r'(?:\s*\{\%\-|\{\%)\s*',
+    r'(?P<block_type_name>({}))'.format(_NAME_PATTERN),
+    # some blocks have a 'block name'.
+    r'(?:\s+(?P<block_name>({})))?'.format(_NAME_PATTERN),
+)))
+
+
+RAW_BLOCK_PATTERN = regex(''.join((
+    r'(?:\s*\{\%\-|\{\%)\s*raw\s*(?:\-\%\}\s*|\%\})',
+    r'(?:.*?)',
+    r'(?:\s*\{\%\-|\{\%)\s*endraw\s*(?:\-\%\}\s*|\%\})',
+)))
+
+TAG_CLOSE_PATTERN = regex(r'(?:(?P<tag_close>(\-\%\}\s*|\%\})))')
+
+# stolen from jinja's lexer. Note that we've consumed all prefix whitespace by
+# the time we want to use this.
+STRING_PATTERN = regex(
+    r"(?P<string>('([^'\\]*(?:\\.[^'\\]*)*)'|"
+    r'"([^"\\]*(?:\\.[^"\\]*)*)"))'
+)
+
+QUOTE_START_PATTERN = regex(r'''(?P<quote>(['"]))''')
+
+
+class TagIterator:
+    def __init__(self, data):
+        self.data = data
+        self.blocks = []
+        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
+
+    def rewind(self, amount=1):
+        self.pos -= amount
+
+    def _search(self, pattern):
+        return pattern.search(self.data, self.pos)
+
+    def _match(self, pattern):
+        return pattern.match(self.data, self.pos)
+
+    def _first_match(self, *patterns, **kwargs):
+        matches = []
+        for pattern in patterns:
+            # default to 'search', but sometimes we want to 'match'.
+            if kwargs.get('method', 'search') == 'search':
+                match = self._search(pattern)
+            else:
+                match = self._match(pattern)
+            if match:
+                matches.append(match)
+        if not matches:
+            return None
+        # if there are multiple matches, pick the least greedy match
+        # TODO: do I need to account for m.start(), or is this ok?
+        return min(matches, key=lambda m: m.end())
+
+    def _expect_match(self, expected_name, *patterns, **kwargs):
+        match = self._first_match(*patterns, **kwargs)
+        if match is None:
+            msg = 'unexpected EOF, expected {}, got "{}"'.format(
+                expected_name, self.data[self.pos:]
+            )
+            dbt.exceptions.raise_compiler_error(msg)
+        return match
+
+    def handle_expr(self, match):
+        """Handle an expression. At this point we're at a string like:
+            {{ 1 + 2 }}
+            ^ right here
+
+        And the match contains "{{ "
+
+        We expect to find a `}}`, but we might find one in a string before
+        that. Imagine the case of `{{ 2 * "}}" }}`...
+
+        You're not allowed to have blocks or comments inside an expr so it is
+        pretty straightforward, I hope: only strings can get in the way.
+        """
+        self.advance(match.end())
+        while True:
+            match = self._expect_match('}}',
+                                       EXPR_END_PATTERN,
+                                       QUOTE_START_PATTERN)
+            if match.groupdict().get('expr_end') is not None:
+                break
+            else:
+                # it's a quote. we haven't advanced for this match yet, so
+                # just slurp up the whole string, no need to rewind.
+                match = self._expect_match('string', STRING_PATTERN)
+                self.advance(match.end())
+
+        self.advance(match.end())
+
+    def handle_comment(self, match):
+        self.advance(match.end())
+        match = self._expect_match('#}', COMMENT_END_PATTERN)
+        self.advance(match.end())
+
+    def _expect_block_close(self):
+        """Search for the tag close marker.
+        To the right of the type name, there are a few possiblities:
+           - a name (handled by the regex's 'block_name')
+           - any number of: `=`, `(`, `)`, strings, etc (arguments)
+           - nothing
+
+        followed eventually by a %}
+
+        So the only characters we actually have to worry about in this context
+        are quote and `%}` - nothing else can hide the %} and be valid jinja.
+        """
+        while True:
+            end_match = self._expect_match(
+                'tag close ("%}")',
+                QUOTE_START_PATTERN,
+                TAG_CLOSE_PATTERN
+            )
+            self.advance(end_match.end())
+            if end_match.groupdict().get('tag_close') is not None:
+                return
+            # must be a string. Rewind to its start and advance past it.
+            self.rewind()
+            string_match = self._expect_match('string', STRING_PATTERN)
+            self.advance(string_match.end())
+
+    def handle_raw(self):
+        # raw blocks are super special, they are a single complete regex
+        match = self._expect_match('{% raw %}...{% endraw %}',
+                                   RAW_BLOCK_PATTERN)
+        self.advance(match.end())
+        return match.end()
+
+    def handle_tag(self, match):
+        """The tag could be one of a few things:
+
+            {% mytag %}
+            {% mytag x = y %}
+            {% mytag x = "y" %}
+            {% mytag x.y() %}
+            {% mytag foo("a", "b", c="d") %}
+
+        But the key here is that it's always going to be `{% mytag`!
+        """
+        groups = match.groupdict()
+        # always a value
+        block_type_name = groups['block_type_name']
+        # might be None
+        block_name = groups.get('block_name')
+        start_pos = self.pos
+        if block_type_name == 'raw':
+            match = self._expect_match('{% raw %}...{% endraw %}',
+                                       RAW_BLOCK_PATTERN)
+            self.advance(match.end())
+        else:
+            self.advance(match.end())
+            self._expect_block_close()
+        return Tag(
+            block_type_name=block_type_name,
+            block_name=block_name,
+            start=start_pos,
+            end=self.pos
+        )
+
+    def find_tags(self):
+        while True:
+            match = self._first_match(
+                BLOCK_START_PATTERN,
+                COMMENT_START_PATTERN,
+                EXPR_START_PATTERN
+            )
+            if match is None:
+                break
+
+            self.advance(match.start())
+            # start = self.pos
+
+            groups = match.groupdict()
+            comment_start = groups.get('comment_start')
+            expr_start = groups.get('expr_start')
+            block_type_name = groups.get('block_type_name')
+
+            if comment_start is not None:
+                self.handle_comment(match)
+            elif expr_start is not None:
+                self.handle_expr(match)
+            elif block_type_name is not None:
+                yield self.handle_tag(match)
+            else:
+                raise dbt.exceptions.InternalException(
+                    'Invalid regex match in next_block, expected block start, '
+                    'expr start, or comment start'
+                )
+
+    def __iter__(self):
+        return self.find_tags()
+
+
+duplicate_tags = (
+    'Got nested tags: {outer.block_type_name} (started at {outer.start}) did '
+    'not have a matching {{% end{outer.block_type_name} %}} before a '
+    'subsequent {inner.block_type_name} was found (started at {inner.start})'
+)
+
+
+_CONTROL_FLOW_TAGS = {
+    'if': 'endif',
+    'for': 'endfor',
+}
+
+_CONTROL_FLOW_END_TAGS = {
+    v: k
+    for k, v in _CONTROL_FLOW_TAGS.items()
+}
+
+
+class BlockIterator:
+    def __init__(self, data):
+        self.tag_parser = TagIterator(data)
+        self.current = None
+        self.stack = []
+        self.last_position = 0
+
+    @property
+    def current_end(self):
+        if self.current is None:
+            return 0
+        else:
+            return self.current.end
+
+    @property
+    def data(self):
+        return self.tag_parser.data
+
+    def is_current_end(self, tag):
+        return (
+            tag.block_type_name.startswith('end') and
+            self.current is not None and
+            tag.block_type_name[3:] == self.current.block_type_name
+        )
+
+    def find_blocks(self, allowed_blocks=None, collect_raw_data=True):
+        """Find all top-level blocks in the data."""
+        if allowed_blocks is None:
+            allowed_blocks = {'snapshot', 'macro', 'materialization', 'docs'}
+
+        for tag in self.tag_parser.find_tags():
+            if tag.block_type_name in _CONTROL_FLOW_TAGS:
+                self.stack.append(tag.block_type_name)
+            elif tag.block_type_name in _CONTROL_FLOW_END_TAGS:
+                found = None
+                if self.stack:
+                    found = self.stack.pop()
+                else:
+                    expected = _CONTROL_FLOW_END_TAGS[tag.block_type_name]
+                    dbt.exceptions.raise_compiler_error((
+                        'Got an unexpected control flow end tag, got {} but '
+                        'never saw a preceeding {} (@ {})'
+                    ).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,
+                        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(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)
+                    )
+                if collect_raw_data:
+                    raw_data = self.data[self.last_position:tag.start]
+                    self.last_position = tag.start
+                    if raw_data:
+                        yield BlockData(raw_data)
+                self.current = tag
+
+            elif self.is_current_end(tag):
+                self.last_position = tag.end
+                assert self.current is not None
+                yield BlockTag(
+                    block_type_name=self.current.block_type_name,
+                    block_name=self.current.block_name,
+                    contents=self.data[self.current.end:tag.start],
+                    full_block=self.data[self.current.start:tag.end]
+                )
+                self.current = None
+
+        if self.current:
+            linecount = self.data[:self.current.end].count('\n') + 1
+            dbt.exceptions.raise_compiler_error((
+                'Reached EOF without finding a close tag for '
+                '{} (searched from line {})'
+            ).format(self.current.block_type_name, linecount))
+
+        if collect_raw_data:
+            raw_data = self.data[self.last_position:]
+            if raw_data:
+                yield BlockData(raw_data)
+
+    def lex_for_blocks(self, allowed_blocks=None, collect_raw_data=True):
+        return list(self.find_blocks(allowed_blocks=allowed_blocks,
+                                     collect_raw_data=collect_raw_data))

core/dbt/clients/agate_helper.py

@@ -0,0 +1,228 @@
+from codecs import BOM_UTF8
+
+import agate
+import datetime
+import isodate
+import json
+import dbt.utils
+from typing import Iterable, List, Dict, Union, Optional, Any
+
+from dbt.exceptions import RuntimeException
+
+
+BOM = BOM_UTF8.decode('utf-8')  # '\ufeff'
+
+
+class ISODateTime(agate.data_types.DateTime):
+    def cast(self, d):
+        # this is agate.data_types.DateTime.cast with the "clever" bits removed
+        # so we only handle ISO8601 stuff
+        if isinstance(d, datetime.datetime) or d is None:
+            return d
+        elif isinstance(d, datetime.date):
+            return datetime.datetime.combine(d, datetime.time(0, 0, 0))
+        elif isinstance(d, str):
+            d = d.strip()
+            if d.lower() in self.null_values:
+                return None
+        try:
+            return isodate.parse_datetime(d)
+        except:  # noqa
+            pass
+
+        raise agate.exceptions.CastError(
+            'Can not parse value "%s" as datetime.' % d
+        )
+
+
+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', ''),
+                              date_format='%Y-%m-%d'),
+        agate.data_types.DateTime(null_values=('null', ''),
+                                  datetime_format='%Y-%m-%d %H:%M:%S'),
+        ISODateTime(null_values=('null', '')),
+        agate.data_types.Boolean(true_values=('true',),
+                                 false_values=('false',),
+                                 null_values=('null', '')),
+        agate.data_types.Text(null_values=string_null_values)
+    ]
+    force = {
+        k: agate.data_types.Text(null_values=string_null_values)
+        for k in text_columns
+    }
+    return agate.TypeTester(force=force, types=types)
+
+
+DEFAULT_TYPE_TESTER = build_type_tester(())
+
+
+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
+    # from `data` is not preserved. We can use `select` to reorder the columns
+    #
+    # If there is no data, create an empty table with the specified columns
+
+    if len(data) == 0:
+        return agate.Table([], column_names=column_names)
+    else:
+        table = agate.Table.from_object(data, column_types=DEFAULT_TYPE_TESTER)
+        return table.select(column_names)
+
+
+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 col_name in column_names:
+            value = _row[col_name]
+            if isinstance(value, (dict, list, tuple)):
+                # 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 table_from_rows(
+        rows=rows,
+        column_names=column_names,
+        text_only_columns=text_only_columns
+    )
+
+
+def empty_table():
+    "Returns an empty Agate table. To be used in place of None"
+
+    return agate.Table(rows=[])
+
+
+def as_matrix(table):
+    "Return an agate table as a matrix of data sans columns"
+
+    return [r.values() for r in table.rows.values()]
+
+
+def from_csv(abspath, text_columns):
+    type_tester = build_type_tester(text_columns=text_columns)
+    with open(abspath, encoding='utf-8') as fp:
+        if fp.read(1) != BOM:
+            fp.seek(0)
+        return agate.Table.from_csv(fp, column_types=type_tester)
+
+
+class _NullMarker:
+    pass
+
+
+NullableAgateType = Union[agate.data_types.DataType, _NullMarker]
+
+
+class ColumnTypeBuilder(Dict[str, NullableAgateType]):
+    def __init__(self):
+        super().__init__()
+
+    def __setitem__(self, key, value):
+        if key not in self:
+            super().__setitem__(key, value)
+            return
+
+        existing_type = self[key]
+        if isinstance(existing_type, _NullMarker):
+            # overwrite
+            super().__setitem__(key, value)
+        elif isinstance(value, _NullMarker):
+            # use the existing value
+            return
+        elif not isinstance(value, type(existing_type)):
+            # actual type mismatch!
+            raise RuntimeException(
+                f'Tables contain columns with the same names ({key}), '
+                f'but different types ({value} vs {existing_type})'
+            )
+
+    def finalize(self) -> Dict[str, agate.data_types.DataType]:
+        result: Dict[str, agate.data_types.DataType] = {}
+        for key, value in self.items():
+            if isinstance(value, _NullMarker):
+                # this is what agate would do.
+                result[key] = agate.data_types.Number()
+            else:
+                result[key] = value
+        return result
+
+
+def _merged_column_types(
+    tables: List[agate.Table]
+) -> Dict[str, agate.data_types.DataType]:
+    # this is a lot like agate.Table.merge, but with handling for all-null
+    # rows being "any type".
+    new_columns: ColumnTypeBuilder = ColumnTypeBuilder()
+    for table in tables:
+        for i in range(len(table.columns)):
+            column_name: str = table.column_names[i]
+            column_type: NullableAgateType = table.column_types[i]
+            # avoid over-sensitive type inference
+            if all(x is None for x in table.columns[column_name]):
+                column_type = _NullMarker()
+            new_columns[column_name] = column_type
+
+    return new_columns.finalize()
+
+
+def merge_tables(tables: List[agate.Table]) -> agate.Table:
+    """This is similar to agate.Table.merge, but it handles rows of all 'null'
+    values more gracefully during merges.
+    """
+    new_columns = _merged_column_types(tables)
+    column_names = tuple(new_columns.keys())
+    column_types = tuple(new_columns.values())
+
+    rows: List[agate.Row] = []
+    for table in tables:
+        if (
+            table.column_names == column_names and
+            table.column_types == column_types
+        ):
+            rows.extend(table.rows)
+        else:
+            for row in table.rows:
+                data = [row.get(name, None) for name in column_names]
+                rows.append(agate.Row(data, column_names))
+    # _is_fork to tell agate that we already made things into `Row`s.
+    return agate.Table(rows, column_names, column_types, _is_fork=True)

core/dbt/clients/git.py

@@ -0,0 +1,142 @@
+import re
+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 _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'))
+
+    return result
+
+
+def list_tags(cwd):
+    out, err = run_cmd(cwd, ['git', 'tag', '--list'], env={'LC_ALL': 'C'})
+    tags = out.decode('utf-8').strip().split("\n")
+    return tags
+
+
+def _checkout(cwd, repo, revision):
+    logger.debug('  Checking out revision {}.'.format(revision))
+
+    fetch_cmd = ["git", "fetch", "origin", "--depth", "1"]
+
+    if _is_commit(revision):
+        run_cmd(cwd, fetch_cmd + [revision])
+    else:
+        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, revision=None):
+    if revision is None:
+        revision = 'HEAD'
+    try:
+        return _checkout(cwd, repo, revision)
+    except dbt.exceptions.CommandResultError as exc:
+        stderr = exc.stderr.decode('utf-8').strip()
+    dbt.exceptions.bad_package_spec(repo, revision, stderr)
+
+
+def get_current_sha(cwd):
+    out, err = run_cmd(cwd, ['git', 'rev-parse', 'HEAD'], env={'LC_ALL': 'C'})
+
+    return out.decode('utf-8')
+
+
+def remove_remote(cwd):
+    return run_cmd(cwd, ['git', 'remote', 'rm', 'origin'], env={'LC_ALL': 'C'})
+
+
+def clone_and_checkout(repo, cwd, dirname=None, remove_git_dir=False,
+                       revision=None, subdirectory=None):
+    exists = None
+    try:
+        _, 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)
+        if not exists:  # something else is wrong, raise it
+            raise
+
+    directory = None
+    start_sha = None
+    if exists:
+        directory = exists.group(1)
+        logger.debug('Updating existing dependency {}.', directory)
+    else:
+        matches = re.match("Cloning into '(.+)'", err.decode('utf-8'))
+        if matches is None:
+            raise dbt.exceptions.RuntimeException(
+                f'Error cloning {repo} - never saw "Cloning into ..." from git'
+            )
+        directory = matches.group(1)
+        logger.debug('Pulling new dependency {}.', directory)
+    full_path = os.path.join(cwd, directory)
+    start_sha = get_current_sha(full_path)
+    checkout(full_path, repo, revision)
+    end_sha = get_current_sha(full_path)
+    if exists:
+        if start_sha == end_sha:
+            logger.debug('  Already at {}, nothing to do.', start_sha[:7])
+        else:
+            logger.debug('  Updated checkout from {} to {}.',
+                         start_sha[:7], end_sha[:7])
+    else:
+        logger.debug('  Checked out at {}.', end_sha[:7])
+    return os.path.join(directory, subdirectory or '')

core/dbt/clients/jinja.py

@@ -0,0 +1,665 @@
+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, 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_test_macro_name, deep_map
+)
+
+from dbt.clients._jinja_blocks import BlockIterator, BlockData, BlockTag
+from dbt.contracts.graph.compiled import CompiledGenericTestNode
+from dbt.contracts.graph.parsed import ParsedGenericTestNode
+from dbt.exceptions import (
+    InternalException, raise_compiler_error, CompilationException,
+    invalid_materialization_argument, MacroReturn, JinjaRenderingException,
+    UndefinedMacroException
+)
+from dbt import flags
+from dbt.logger import GLOBAL_LOGGER as logger  # noqa
+
+
+def _linecache_inject(source, write):
+    if write:
+        # this is the only reliable way to accomplish this. Obviously, it's
+        # really darn noisy and will fill your temporary directory
+        tmp_file = tempfile.NamedTemporaryFile(
+            prefix='dbt-macro-compiled-',
+            suffix='.py',
+            delete=False,
+            mode='w+',
+            encoding='utf-8',
+        )
+        tmp_file.write(source)
+        filename = tmp_file.name
+    else:
+        # `codecs.encode` actually takes a `bytes` as the first argument if
+        # the second argument is 'hex' - mypy does not know this.
+        rnd = codecs.encode(os.urandom(12), 'hex')  # type: ignore
+        filename = rnd.decode('ascii')
+
+    # put ourselves in the cache
+    cache_entry = (
+        len(source),
+        None,
+        [line + '\n' for line in source.splitlines()],
+        filename
+    )
+    # linecache does in fact have an attribute `cache`, thanks
+    linecache.cache[filename] = cache_entry  # type: ignore
+    return filename
+
+
+class MacroFuzzParser(jinja2.parser.Parser):
+    def parse_macro(self):
+        node = jinja2.nodes.Macro(lineno=next(self.stream).lineno)
+
+        # modified to fuzz macros defined in the same file. this way
+        # dbt can understand the stack of macros being called.
+        #  - @cmcarthur
+        node.name = get_dbt_macro_name(
+            self.parse_assign_target(name_only=True).name)
+
+        self.parse_signature(node)
+        node.body = self.parse_statements(('name:endmacro',),
+                                          drop_needle=True)
+        return node
+
+
+class MacroFuzzEnvironment(jinja2.sandbox.SandboxedEnvironment):
+    def _parse(self, source, name, filename):
+        return MacroFuzzParser(self, source, name, filename).parse()
+
+    def _compile(self, source, filename):
+        """Override jinja's compilation to stash the rendered source inside
+        the python linecache for debugging when the appropriate environment
+        variable is set.
+
+        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 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: Dict[str, jinja2.Template] = {}
+
+    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.macro_sql,
+            ctx={},
+            node=node,
+        )
+
+        self.file_cache[key] = template
+        return template
+
+    def clear(self):
+        self.file_cache.clear()
+
+
+template_cache = TemplateCache()
+
+
+class BaseMacroGenerator:
+    def __init__(self, context: Optional[Dict[str, Any]] = None) -> None:
+        self.context: Optional[Dict[str, Any]] = context
+
+    def get_template(self):
+        raise NotImplementedError('get_template not implemented!')
+
+    def get_name(self) -> str:
+        raise NotImplementedError('get_name not implemented!')
+
+    def get_macro(self):
+        name = self.get_name()
+        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)
+        return macro
+
+    @contextmanager
+    def exception_handler(self) -> Iterator[None]:
+        try:
+            yield
+        except (TypeError, jinja2.exceptions.TemplateRuntimeError) as e:
+            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!'
+            )
+        assert self.context is not None
+
+        macro = self.get_macro()
+
+        with self.exception_handler():
+            try:
+                return macro(*args, **kwargs)
+            except MacroReturn as e:
+                return e.value
+
+
+class MacroStack(threading.local):
+    def __init__(self):
+        super().__init__()
+        self.call_stack = []
+
+    @property
+    def depth(self) -> int:
+        return len(self.call_stack)
+
+    def push(self, name):
+        self.call_stack.append(name)
+
+    def pop(self, name):
+        got = self.call_stack.pop()
+        if got != name:
+            raise InternalException(f'popped {got}, expected {name}')
+
+
+class MacroGenerator(BaseMacroGenerator):
+    def __init__(
+        self,
+        macro,
+        context: Optional[Dict[str, Any]] = None,
+        node: Optional[Any] = None,
+        stack: Optional[MacroStack] = None
+    ) -> None:
+        super().__init__(context)
+        self.macro = macro
+        self.node = node
+        self.stack = stack
+
+    def get_template(self):
+        return template_cache.get_node_template(self.macro)
+
+    def get_name(self) -> str:
+        return self.macro.name
+
+    @contextmanager
+    def exception_handler(self) -> Iterator[None]:
+        try:
+            yield
+        except (TypeError, jinja2.exceptions.TemplateRuntimeError) as e:
+            raise_compiler_error(str(e), self.macro)
+        except CompilationException as e:
+            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:
+            unique_id = self.macro.unique_id
+            depth = self.stack.depth
+            # only mark depth=0 as a dependency
+            if depth == 0:
+                self.node.depends_on.add_macro(unique_id)
+            self.stack.push(unique_id)
+            try:
+                yield
+            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)
+
+
+class QueryStringGenerator(BaseMacroGenerator):
+    def __init__(
+        self, template_str: str, context: Dict[str, Any]
+    ) -> None:
+        super().__init__(context)
+        self.template_str: str = template_str
+        env = get_environment()
+        self.template = env.from_string(
+            self.template_str,
+            globals=self.context,
+        )
+
+    def get_name(self) -> str:
+        return 'query_comment_macro'
+
+    def get_template(self):
+        """Don't use the template cache, we don't have a node"""
+        return self.template
+
+    def __call__(self, connection_name: str, node) -> str:
+        return str(self.call_macro(connection_name, node))
+
+
+class MaterializationExtension(jinja2.ext.Extension):
+    tags = ['materialization']
+
+    def parse(self, parser):
+        node = jinja2.nodes.Macro(lineno=next(parser.stream).lineno)
+        materialization_name = \
+            parser.parse_assign_target(name_only=True).name
+
+        adapter_name = 'default'
+        node.args = []
+        node.defaults = []
+
+        while parser.stream.skip_if('comma'):
+            target = parser.parse_assign_target(name_only=True)
+
+            if target.name == 'default':
+                pass
+
+            elif target.name == 'adapter':
+                parser.stream.expect('assign')
+                value = parser.parse_expression()
+                adapter_name = value.value
+
+            else:
+                invalid_materialization_argument(
+                    materialization_name, target.name
+                )
+
+        node.name = get_materialization_macro_name(
+            materialization_name, adapter_name
+        )
+
+        node.body = parser.parse_statements(('name:endmaterialization',),
+                                            drop_needle=True)
+
+        return node
+
+
+class DocumentationExtension(jinja2.ext.Extension):
+    tags = ['docs']
+
+    def parse(self, parser):
+        node = jinja2.nodes.Macro(lineno=next(parser.stream).lineno)
+        docs_name = parser.parse_assign_target(name_only=True).name
+
+        node.args = []
+        node.defaults = []
+        node.name = get_docs_macro_name(docs_name)
+        node.body = parser.parse_statements(('name:enddocs',),
+                                            drop_needle=True)
+        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_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.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 __getitem__(self, name):
+            # Propagate the undefined value if a caller accesses this as if it
+            # were a dictionary
+            return self
+
+        def __getattr__(self, name):
+            if name == 'name' or _is_dunder_name(name):
+                raise AttributeError(
+                    "'{}' object has no attribute '{}'"
+                    .format(type(self).__name__, name)
+                )
+
+            self.name = name
+
+            return self.__class__(hint=self.hint, name=self.name)
+
+        def __call__(self, *args, **kwargs):
+            return self
+
+        def __reduce__(self):
+            raise_compiler_error(f'{self.name} is undefined', node=node)
+
+    return Undefined
+
+
+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_undefined(node)
+
+    args['extensions'].append(MaterializationExtension)
+    args['extensions'].append(DocumentationExtension)
+    args['extensions'].append(TestExtension)
+
+    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
+def catch_jinja(node=None) -> Iterator[None]:
+    try:
+        yield
+    except jinja2.exceptions.TemplateSyntaxError as e:
+        e.translated = False
+        raise CompilationException(str(e), node) from e
+    except jinja2.exceptions.UndefinedError as e:
+        raise UndefinedMacroException(str(e), node) from e
+    except CompilationException as exc:
+        exc.add_node(node)
+        raise
+
+
+def parse(string):
+    with catch_jinja():
+        return get_environment().parse(str(string))
+
+
+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, native=native)
+
+        template_source = str(string)
+        return env.from_string(template_source, globals=ctx)
+
+
+def render_template(template, ctx: Dict[str, Any], node=None) -> str:
+    with catch_jinja(node):
+        return template.render(ctx)
+
+
+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)
+
+
+def undefined_error(msg) -> NoReturn:
+    raise jinja2.exceptions.UndefinedError(msg)
+
+
+def extract_toplevel_blocks(
+    data: str,
+    allowed_blocks: Optional[Set[str]] = None,
+    collect_raw_data: bool = True,
+) -> List[Union[BlockData, BlockTag]]:
+    """Extract the top level blocks with matching block types from a jinja
+    file, with some special handling for block nesting.
+
+    :param data: The data to extract blocks from.
+    :param allowed_blocks: The names of the blocks to extract from the file.
+        They may not be nested within if/for blocks. If None, use the default
+        values.
+    :param collect_raw_data: If set, raw data between matched blocks will also
+        be part of the results, as `BlockData` objects. They have a
+        `block_type_name` field of `'__dbt_data'` and will never have a
+        `block_name`.
+    :return: A list of `BlockTag`s matching the allowed block types and (if
+        `collect_raw_data` is `True`) `BlockData` objects.
+    """
+    return BlockIterator(data).lex_for_blocks(
+        allowed_blocks=allowed_blocks,
+        collect_raw_data=collect_raw_data
+    )
+
+
+GENERIC_TEST_KWARGS_NAME = '_dbt_generic_test_kwargs'
+
+
+def add_rendered_test_kwargs(
+    context: Dict[str, Any],
+    node: Union[ParsedGenericTestNode, CompiledGenericTestNode],
+    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[GENERIC_TEST_KWARGS_NAME] = kwargs

core/dbt/clients/jinja_static.py

@@ -0,0 +1,154 @@
+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
+    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 == '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
+
+    if db_wrapper:
+        macro = db_wrapper.dispatch(
+            func_name,
+            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]
+        else:
+            packages = []
+        for package_name in packages:
+            possible_macro_calls.append(f'{package_name}.{func_name}')
+
+    return possible_macro_calls

core/dbt/clients/registry.py

@@ -0,0 +1,79 @@
+import functools
+import requests
+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
+
+if os.getenv('DBT_PACKAGE_HUB_URL'):
+    DEFAULT_REGISTRY_BASE_URL = os.getenv('DBT_PACKAGE_HUB_URL')
+else:
+    DEFAULT_REGISTRY_BASE_URL = 'https://hub.getdbt.com/'
+
+
+def _get_url(url, registry_base_url=None):
+    if registry_base_url is None:
+        registry_base_url = DEFAULT_REGISTRY_BASE_URL
+
+    return '{}{}'.format(registry_base_url, url)
+
+
+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)
+
+
+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, timeout=30)
+    logger.debug('Response from registry: GET {} {}'.format(url,
+                                                            resp.status_code))
+    resp.raise_for_status()
+    return resp.json()
+
+
+def index(registry_base_url=None):
+    return _get_with_retries('api/v1/index.json', registry_base_url)
+
+
+index_cached = memoized(index)
+
+
+def packages(registry_base_url=None):
+    return _get_with_retries('api/v1/packages.json', registry_base_url)
+
+
+def package(name, registry_base_url=None):
+    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_with_retries('api/v1/{}/{}.json'.format(name, version), registry_base_url)
+
+
+def get_available_versions(name):
+    response = package(name)
+    return list(response['versions'])

core/dbt/clients/system.py

@@ -0,0 +1,561 @@
+import errno
+import functools
+import fnmatch
+import json
+import os
+import os.path
+import re
+import shutil
+import subprocess
+import sys
+import tarfile
+import requests
+import stat
+from typing import (
+    Type, NoReturn, List, Optional, Dict, Any, Tuple, Callable, Union
+)
+
+import dbt.exceptions
+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(
+    root_path: str,
+    relative_paths_to_search: List[str],
+    file_pattern: str,
+) -> List[Dict[str, Any]]:
+    """
+    Given an absolute `root_path`, a list of relative paths to that
+    absolute root path (`relative_paths_to_search`), and a `file_pattern`
+    like '*.sql', returns information about the files. For example:
+
+    > find_matching('/root/path', ['models'], '*.sql')
+
+      [ { 'absolute_path': '/root/path/models/model_one.sql',
+          'relative_path': 'model_one.sql',
+          'searched_path': 'models' },
+        { 'absolute_path': '/root/path/models/subdirectory/model_two.sql',
+          'relative_path': 'subdirectory/model_two.sql',
+          'searched_path': 'models' } ]
+    """
+    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(
+            root_path, relative_path_to_search)
+        walk_results = os.walk(absolute_path_to_search)
+
+        for current_path, subdirectories, local_files in walk_results:
+            for local_file in local_files:
+                absolute_path = os.path.join(current_path, local_file)
+                relative_path = os.path.relpath(
+                    absolute_path, absolute_path_to_search
+                )
+                modification_time = 0.0
+                try:
+                    modification_time = os.path.getmtime(absolute_path)
+                except OSError:
+                    logger.exception(
+                        f"Error retrieving modification time for file {absolute_path}"
+                    )
+                if reobj.match(local_file):
+                    matching.append({
+                        'searched_path': relative_path_to_search,
+                        'absolute_path': absolute_path,
+                        'relative_path': relative_path,
+                        'modification_time': modification_time,
+                    })
+
+    return 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')
+
+    if strip:
+        to_return = to_return.strip()
+
+    return to_return
+
+
+def make_directory(path: str) -> None:
+    """
+    Make a directory and any intermediate directories that don't already
+    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:
+            os.makedirs(path)
+
+        except OSError as e:
+            if e.errno == errno.EEXIST:
+                pass
+            else:
+                raise e
+
+
+def make_file(path: str, contents: str = '', overwrite: bool = False) -> bool:
+    """
+    Make a file at `path` assuming that the directory it resides in already
+    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
+
+    return False
+
+
+def make_symlink(source: str, link_path: str) -> None:
+    """
+    Create a symlink at `link_path` referring to `source`.
+    """
+    if not supports_symlinks():
+        dbt.exceptions.system_error('create a symbolic link')
+
+    os.symlink(source, link_path)
+
+
+def supports_symlinks() -> bool:
+    return getattr(os, "symlink", None) is not None
+
+
+def write_file(path: str, contents: str = '') -> bool:
+    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))
+
+
+def _windows_rmdir_readonly(
+    func: Callable[[str], Any], path: str, exc: Tuple[Any, OSError, Any]
+):
+    exception_val = exc[1]
+    if exception_val.errno == errno.EACCES:
+        os.chmod(path, stat.S_IWUSR)
+        func(path)
+    else:
+        raise
+
+
+def resolve_path_from_base(path_to_resolve: str, base_path: str) -> str:
+    """
+    If path-to_resolve is a relative path, create an absolute path
+    with base_path as the base.
+
+    If path_to_resolve is an absolute path or a user path (~), just
+    resolve it to an absolute path and return.
+    """
+    return os.path.abspath(
+        os.path.join(
+            base_path,
+            os.path.expanduser(path_to_resolve)))
+
+
+def rmdir(path: str) -> None:
+    """
+    Recursively deletes a directory. Includes an error handler to retry with
+    different permissions on Windows. Otherwise, removing directories (eg.
+    cloned via git) can cause rmtree to throw a PermissionError exception
+    """
+    path = convert_path(path)
+    if sys.platform == 'win32':
+        onerror = _windows_rmdir_readonly
+    else:
+        onerror = 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)
+
+
+def open_dir_cmd() -> str:
+    # https://docs.python.org/2/library/sys.html#sys.platform
+    if sys.platform == 'win32':
+        return 'start'
+
+    elif sys.platform == 'darwin':
+        return 'open'
+
+    else:
+        return 'xdg-open'
+
+
+def _handle_posix_cwd_error(
+    exc: OSError, cwd: str, cmd: List[str]
+) -> NoReturn:
+    if exc.errno == errno.ENOENT:
+        message = 'Directory does not exist'
+    elif exc.errno == errno.EACCES:
+        message = 'Current user cannot access directory, check permissions'
+    elif exc.errno == errno.ENOTDIR:
+        message = 'Not a directory'
+    else:
+        message = 'Unknown OSError: {} - cwd'.format(str(exc))
+    raise dbt.exceptions.WorkingDirectoryError(cwd, cmd, message)
+
+
+def _handle_posix_cmd_error(
+    exc: OSError, cwd: str, cmd: List[str]
+) -> NoReturn:
+    if exc.errno == errno.ENOENT:
+        message = "Could not find command, ensure it is in the user's PATH"
+    elif exc.errno == errno.EACCES:
+        message = 'User does not have permissions for this command'
+    else:
+        message = 'Unknown OSError: {} - cmd'.format(str(exc))
+    raise dbt.exceptions.ExecutableError(cwd, cmd, message)
+
+
+def _handle_posix_error(exc: OSError, cwd: str, cmd: List[str]) -> NoReturn:
+    """OSError handling for posix systems.
+
+    Some things that could happen to trigger an OSError:
+        - cwd could not exist
+            - exc.errno == ENOENT
+            - exc.filename == cwd
+        - cwd could have permissions that prevent the current user moving to it
+            - exc.errno == EACCES
+            - exc.filename == cwd
+        - cwd could exist but not be a directory
+            - exc.errno == ENOTDIR
+            - exc.filename == cwd
+        - cmd[0] could not exist
+            - exc.errno == ENOENT
+            - exc.filename == None(?)
+        - cmd[0] could exist but have permissions that prevents the current
+            user from executing it (executable bit not set for the user)
+            - exc.errno == EACCES
+            - exc.filename == None(?)
+    """
+    if getattr(exc, 'filename', None) == cwd:
+        _handle_posix_cwd_error(exc, cwd, cmd)
+    else:
+        _handle_posix_cmd_error(exc, cwd, cmd)
+
+
+def _handle_windows_error(exc: OSError, cwd: str, cmd: List[str]) -> NoReturn:
+    cls: Type[dbt.exceptions.Exception] = dbt.exceptions.CommandError
+    if exc.errno == errno.ENOENT:
+        message = ("Could not find command, ensure it is in the user's PATH "
+                   "and that the user has permissions to run it")
+        cls = dbt.exceptions.ExecutableError
+    elif exc.errno == errno.ENOEXEC:
+        message = ('Command was not executable, ensure it is valid')
+        cls = dbt.exceptions.ExecutableError
+    elif exc.errno == errno.ENOTDIR:
+        message = ('Unable to cd: path does not exist, user does not have'
+                   ' permissions, or not a directory')
+        cls = dbt.exceptions.WorkingDirectoryError
+    else:
+        message = 'Unknown error: {} (errno={}: "{}")'.format(
+            str(exc), exc.errno, errno.errorcode.get(exc.errno, '<Unknown!>')
+        )
+    raise cls(cwd, cmd, message)
+
+
+def _interpret_oserror(exc: OSError, cwd: str, cmd: List[str]) -> NoReturn:
+    """Interpret an OSError exc and raise the appropriate dbt exception.
+
+    """
+    if len(cmd) == 0:
+        raise dbt.exceptions.CommandError(cwd, cmd)
+
+    # all of these functions raise unconditionally
+    if os.name == 'nt':
+        _handle_windows_error(exc, cwd, cmd)
+    else:
+        _handle_posix_error(exc, cwd, cmd)
+
+    # this should not be reachable, raise _something_ at least!
+    raise dbt.exceptions.InternalException(
+        'Unhandled exception in _interpret_oserror: {}'.format(exc)
+    )
+
+
+def run_cmd(
+    cwd: str, cmd: List[str], env: Optional[Dict[str, Any]] = None
+) -> Tuple[bytes, bytes]:
+    logger.debug('Executing "{}"'.format(' '.join(cmd)))
+    if len(cmd) == 0:
+        raise dbt.exceptions.CommandError(cwd, cmd)
+
+    # the env argument replaces the environment entirely, which has exciting
+    # consequences on Windows! Do an update instead.
+    full_env = env
+    if env is not None:
+        full_env = os.environ.copy()
+        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,
+            stdout=subprocess.PIPE,
+            stderr=subprocess.PIPE,
+            env=full_env)
+
+        out, err = proc.communicate()
+    except OSError as exc:
+        _interpret_oserror(exc, cwd, cmd)
+
+    logger.debug('STDOUT: "{!s}"'.format(out))
+    logger.debug('STDERR: "{!s}"'.format(err))
+
+    if proc.returncode != 0:
+        logger.debug('command return code={}'.format(proc.returncode))
+        raise dbt.exceptions.CommandResultError(cwd, cmd, proc.returncode,
+                                                out, err)
+
+    return out, err
+
+
+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:
+        if is_symlink:
+            remove_file(to_path)
+        else:
+            rmdir(to_path)
+
+    shutil.move(from_path, to_path)
+
+
+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)
+        tar_dir_name = os.path.commonprefix(tarball.getnames())
+    if rename_to:
+        downloaded_path = os.path.join(dest_dir, tar_dir_name)
+        desired_path = os.path.join(dest_dir, rename_to)
+        dbt.clients.system.rename(downloaded_path, desired_path, force=True)
+
+
+def chmod_and_retry(func, path, exc_info):
+    """Define an error handler to pass to shutil.rmtree.
+    On Windows, when a file is marked read-only as git likes to do, rmtree will
+    fail. To handle that, on errors try to make the file writable.
+    We want to retry most operations here, but listdir is one that we know will
+    be useless.
+    """
+    if func is os.listdir or os.name != 'nt':
+        raise
+    os.chmod(path, stat.S_IREAD | stat.S_IWRITE)
+    # on error,this will raise.
+    func(path)
+
+
+def _absnorm(path):
+    return os.path.normcase(os.path.abspath(path))
+
+
+def move(src, dst):
+    """A re-implementation of shutil.move that properly removes the source
+    directory on windows when it has read-only files in it and the move is
+    between two drives.
+
+    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)
+
+    if os.path.isdir(dst):
+        if _absnorm(src) == _absnorm(dst):
+            os.rename(src, dst)
+            return
+
+        dst = os.path.join(dst, os.path.basename(src.rstrip('/\\')))
+        if os.path.exists(dst):
+            raise EnvironmentError("Path '{}' already exists".format(dst))
+
+    try:
+        os.rename(src, dst)
+    except OSError:
+        # probably different drives
+        if os.path.isdir(src):
+            if _absnorm(dst + '\\').startswith(_absnorm(src + '\\')):
+                # dst is inside src
+                raise EnvironmentError(
+                    "Cannot move a directory '{}' into itself '{}'"
+                    .format(src, dst)
+                )
+            shutil.copytree(src, dst, symlinks=True)
+            rmtree(src)
+        else:
+            shutil.copy2(src, dst)
+            os.unlink(src)
+
+
+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 +1,18 @@
-import dbt.compat
 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 = """
@@ -17,7 +27,7 @@ Raw Error:
 
 
 def line_no(i, line, width=3):
-    line_number = dbt.compat.to_string(i).ljust(width)
+    line_number = str(i).ljust(width)
     return "{}| {}".format(line_number, line)
 
 
@@ -45,13 +55,17 @@ 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)
         else:
-            error = dbt.compat.to_string(e)
+            error = str(e)
 
         raise dbt.exceptions.ValidationException(error)

core/dbt/compilation.py

@@ -0,0 +1,553 @@
+import os
+from collections import defaultdict
+from typing import List, Dict, Any, Tuple, cast, Optional
+
+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
+from dbt.contracts.graph.manifest import Manifest, UniqueID
+from dbt.contracts.graph.compiled import (
+    COMPILED_TYPES,
+    CompiledGenericTestNode,
+    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 type(model) not in COMPILED_TYPES:
+        raise InternalException(
+            f'Asked to compile {type(model)} node, but it has no compiled form'
+        )
+    return COMPILED_TYPES[type(model)]
+
+
+def print_compile_stats(stats):
+    names = {
+        NodeType.Model: 'model',
+        NodeType.Test: 'test',
+        NodeType.Snapshot: 'snapshot',
+        NodeType.Analysis: 'analysis',
+        NodeType.Macro: 'macro',
+        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([
+        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: ManifestNode):
+    # Disabled models are already excluded from the manifest
+    if node.resource_type == NodeType.Test and not node.config.enabled:
+        return False
+    else:
+        return True
+
+
+def _generate_stats(manifest: Manifest):
+    stats: Dict[NodeType, int] = defaultdict(int)
+    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
+
+
+def _add_prepended_cte(prepended_ctes, new_cte):
+    for cte in prepended_ctes:
+        if cte.id == new_cte.id:
+            cte.sql = new_cte.sql
+            return
+    prepended_ctes.append(new_cte)
+
+
+def _extend_prepended_ctes(prepended_ctes, new_prepended_ctes):
+    for new_cte in new_prepended_ctes:
+        _add_prepended_cte(prepended_ctes, new_cte)
+
+
+def _get_tests_for_node(manifest: Manifest, unique_id: UniqueID) -> List[UniqueID]:
+    """ Get a list of tests that depend on the node with the
+    provided unique id """
+
+    tests = []
+    if unique_id in manifest.child_map:
+        for child_unique_id in manifest.child_map[unique_id]:
+            if child_unique_id.startswith('test.'):
+                tests.append(child_unique_id)
+
+    return tests
+
+
+class Linker:
+    def __init__(self, data=None):
+        if data is None:
+            data = {}
+        self.graph = nx.DiGraph(**data)
+
+    def edges(self):
+        return self.graph.edges()
+
+    def nodes(self):
+        return self.graph.nodes()
+
+    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)
+
+    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)
+
+    def add_node(self, node):
+        self.graph.add_node(node)
+
+    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:
+            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:
+    def __init__(self, config):
+        self.config = config
+
+    def initialize(self):
+        make_directory(self.config.target_path)
+        make_directory(self.config.packages_install_path)
+
+    # 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, CompiledGenericTestNode):
+            # 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(omit_none=True)
+        data.update({
+            'compiled': False,
+            'compiled_sql': None,
+            'extra_ctes_injected': False,
+            'extra_ctes': [],
+        })
+        compiled_node = _compiled_type_for(node).from_dict(data)
+
+        context = self._create_node_context(
+            compiled_node, manifest, extra_context
+        )
+
+        compiled_node.compiled_sql = jinja.get_rendered(
+            node.raw_sql,
+            context,
+            node,
+        )
+
+        compiled_node.relation_name = self._get_relation_name(node)
+
+        compiled_node.compiled = True
+
+        return compiled_node
+
+    def write_graph_file(self, linker: Linker, manifest: Manifest):
+        filename = graph_file_name
+        graph_path = os.path.join(self.config.target_path, filename)
+        if flags.WRITE_JSON:
+            linker.write_graph(graph_path, 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 dependency in manifest.nodes:
+                linker.dependency(
+                    node.unique_id,
+                    (manifest.nodes[dependency].unique_id)
+                )
+            elif dependency in manifest.sources:
+                linker.dependency(
+                    node.unique_id,
+                    (manifest.sources[dependency].unique_id)
+                )
+            else:
+                dependency_not_found(node, dependency)
+
+    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)
+
+        cycle = linker.find_cycles()
+
+        if cycle:
+            raise RuntimeError("Found a cycle: {}".format(cycle))
+
+        manifest.build_parent_and_child_maps()
+
+        self.resolve_graph(linker, manifest)
+
+    def resolve_graph(self, linker: Linker, manifest: Manifest) -> None:
+        """ This method adds additional edges to the DAG. For a given non-test
+        executable node, add an edge from an upstream test to the given node if
+        the set of nodes the test depends on is a proper/strict subset of the
+        upstream nodes for the given node. """
+
+        # Given a graph:
+        # model1 --> model2 --> model3
+        #   |         |
+        #   |        \/
+        #  \/      test 2
+        # test1
+        #
+        # Produce the following graph:
+        # model1 --> model2 --> model3
+        #   |         |         /\ /\
+        #   |        \/         |  |
+        #  \/      test2 -------   |
+        # test1 -------------------
+
+        for node_id in linker.graph:
+            # If node is executable (in manifest.nodes) and does _not_
+            # represent a test, continue.
+            if (
+                node_id in manifest.nodes and
+                manifest.nodes[node_id].resource_type != NodeType.Test
+            ):
+                # Get *everything* upstream of the node
+                all_upstream_nodes = nx.traversal.bfs_tree(
+                    linker.graph, node_id, reverse=True
+                )
+                # Get the set of upstream nodes not including the current node.
+                upstream_nodes = set([
+                    n for n in all_upstream_nodes if n != node_id
+                ])
+
+                # Get all tests that depend on any upstream nodes.
+                upstream_tests = []
+                for upstream_node in upstream_nodes:
+                    upstream_tests += _get_tests_for_node(
+                        manifest,
+                        upstream_node
+                    )
+
+                for upstream_test in upstream_tests:
+                    # Get the set of all nodes that the test depends on
+                    # including the upstream_node itself. This is necessary
+                    # because tests can depend on multiple nodes (ex:
+                    # relationship tests). Test nodes do not distinguish
+                    # between what node the test is "testing" and what
+                    # node(s) it depends on.
+                    test_depends_on = set(
+                        manifest.nodes[upstream_test].depends_on_nodes
+                    )
+
+                    # If the set of nodes that an upstream test depends on
+                    # is a proper (or strict) subset of all upstream nodes of
+                    # the current node, add an edge from the upstream test
+                    # to the current node. Must be a proper/strict subset to
+                    # avoid adding a circular dependency to the graph.
+                    if (test_depends_on < upstream_nodes):
+                        linker.graph.add_edge(
+                            upstream_test,
+                            node_id
+                        )
+
+    def compile(self, manifest: Manifest, write=True) -> Graph:
+        self.initialize()
+        linker = Linker()
+
+        self.link_graph(linker, manifest)
+
+        stats = _generate_stats(manifest)
+
+        if write:
+            self.write_graph_file(linker, manifest)
+        print_compile_stats(stats)
+
+        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}"')
+
+        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)
+
+        node, _ = self._recursively_prepend_ctes(
+            node, manifest, extra_context
+        )
+        if write:
+            self._write_node(node)
+        return node

core/dbt/config/__init__.py

@@ -0,0 +1,4 @@
+# all these are just exports, they need "noqa" so flake8 will not complain.
+from .profile import Profile, read_user_config  # noqa
+from .project import Project, IsFQNResource  # noqa
+from .runtime import RuntimeConfig, UnsetProfileConfig  # noqa

core/dbt/config/profile.py

@@ -0,0 +1,446 @@
+from dataclasses import dataclass
+from typing import Any, Dict, Optional, Tuple
+import os
+
+from dbt.dataclass_schema import ValidationError
+
+from dbt import flags
+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
+from dbt.exceptions import RuntimeException
+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 ProfileRenderer
+
+DEFAULT_THREADS = 1
+
+DEFAULT_PROFILES_DIR = os.path.join(os.path.expanduser('~'), '.dbt')
+
+INVALID_PROFILE_MESSAGE = """
+dbt encountered an error while trying to read your profiles.yml file.
+
+{error_string}
+"""
+
+
+NO_SUPPLIED_PROFILE_ERROR = """\
+dbt cannot run because no profile was specified for this dbt project.
+To specify a profile for this project, add a line like the this to
+your dbt_project.yml file:
+
+profile: [profile name]
+
+Here, [profile name] should be replaced with a profile name
+defined in your profiles.yml file. You can find profiles.yml here:
+
+{profiles_file}/profiles.yml
+""".format(profiles_file=DEFAULT_PROFILES_DIR)
+
+
+def read_profile(profiles_dir: str) -> Dict[str, Any]:
+    path = os.path.join(profiles_dir, 'profiles.yml')
+
+    contents = None
+    if os.path.isfile(path):
+        try:
+            contents = load_file_contents(path, strip=False)
+            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
+
+    return {}
+
+
+def read_user_config(directory: str) -> UserConfig:
+    try:
+        profile = read_profile(directory)
+        if profile:
+            user_config = coerce_dict_str(profile.get('config', {}))
+            if user_config is not None:
+                UserConfig.validate(user_config)
+                return UserConfig.from_dict(user_config)
+    except (RuntimeException, ValidationError):
+        pass
+    return UserConfig()
+
+
+# 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
+    user_config: UserConfig
+    threads: int
+    credentials: Credentials
+
+    def __init__(
+        self,
+        profile_name: str,
+        target_name: str,
+        user_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.user_config = user_config
+        self.threads = threads
+        self.credentials = credentials
+
+    def to_profile_info(
+        self, serialize_credentials: bool = False
+    ) -> Dict[str, Any]:
+        """Unlike to_project_config, this dict is not a mirror of any existing
+        on-disk data structure. It's used when creating a new profile from an
+        existing one.
+
+        :param serialize_credentials bool: If True, serialize the credentials.
+            Otherwise, the Credentials object will be copied.
+        :returns dict: The serialized profile.
+        """
+        result = {
+            'profile_name': self.profile_name,
+            'target_name': self.target_name,
+            'user_config': self.user_config,
+            'threads': self.threads,
+            'credentials': self.credentials,
+        }
+        if serialize_credentials:
+            result['user_config'] = self.user_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.user_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__)):
+            return NotImplemented
+        return self.to_profile_info() == other.to_profile_info()
+
+    def validate(self):
+        try:
+            if self.credentials:
+                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
+
+    @staticmethod
+    def _credentials_from_profile(
+        profile: Dict[str, Any], profile_name: str, target_name: str
+    ) -> Credentials:
+        # avoid an import cycle
+        from dbt.adapters.factory import load_plugin
+        # credentials carry their 'type' in their actual type, not their
+        # attributes. We do want this in order to pick our Credentials class.
+        if 'type' not in profile:
+            raise DbtProfileError(
+                'required field "type" not found in profile {} and target {}'
+                .format(profile_name, target_name))
+
+        typename = profile.pop('type')
+        try:
+            cls = load_plugin(typename)
+            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(
+                'Credentials in profile "{}", target "{}" invalid: {}'
+                .format(profile_name, target_name, msg)
+            ) from e
+
+        return credentials
+
+    @staticmethod
+    def pick_profile_name(
+        args_profile_name: Optional[str],
+        project_profile_name: Optional[str] = None,
+    ) -> str:
+        profile_name = project_profile_name
+        if args_profile_name is not None:
+            profile_name = args_profile_name
+        if profile_name is None:
+            raise DbtProjectError(NO_SUPPLIED_PROFILE_ERROR)
+        return profile_name
+
+    @staticmethod
+    def _get_profile_data(
+        profile: Dict[str, Any], profile_name: str, target_name: str
+    ) -> Dict[str, Any]:
+        if 'outputs' not in profile:
+            raise DbtProfileError(
+                "outputs not specified in profile '{}'".format(profile_name)
+            )
+        outputs = profile['outputs']
+
+        if target_name not in outputs:
+            outputs = '\n'.join(' - {}'.format(output)
+                                for output in outputs)
+            msg = ("The profile '{}' does not have a target named '{}'. The "
+                   "valid target names for this profile are:\n{}"
+                   .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
+    def from_credentials(
+        cls,
+        credentials: Credentials,
+        threads: int,
+        profile_name: str,
+        target_name: str,
+        user_config: Optional[Dict[str, Any]] = None
+    ) -> 'Profile':
+        """Create a profile from an existing set of Credentials and the
+        remaining information.
+
+        :param credentials: The credentials dict for this profile.
+        :param threads: The number of threads to use for connections.
+        :param profile_name: The profile name used for this profile.
+        :param target_name: The target name used for this profile.
+        :param user_config: The user-level config block from the
+            raw profiles, if specified.
+        :raises DbtProfileError: If the profile is invalid.
+        :returns: The new Profile object.
+        """
+        if user_config is None:
+            user_config = {}
+        UserConfig.validate(user_config)
+        user_config_obj: UserConfig = UserConfig.from_dict(user_config)
+
+        profile = cls(
+            profile_name=profile_name,
+            target_name=target_name,
+            user_config=user_config_obj,
+            threads=threads,
+            credentials=credentials
+        )
+        profile.validate()
+        return profile
+
+    @classmethod
+    def render_profile(
+        cls,
+        raw_profile: Dict[str, Any],
+        profile_name: str,
+        target_override: Optional[str],
+        renderer: ProfileRenderer,
+    ) -> Tuple[str, Dict[str, Any]]:
+        """This is a containment zone for the hateful way we're rendering
+        profiles.
+        """
+        # rendering profiles is a bit complex. Two constraints cause trouble:
+        # 1) users should be able to use environment/cli variables to specify
+        #    the target in their profile.
+        # 2) Missing environment/cli variables in profiles/targets that don't
+        #    end up getting selected should not cause errors.
+        # so first we'll just render the target name, then we use that rendered
+        # name to extract a profile that we can render.
+        if target_override is not None:
+            target_name = target_override
+        elif 'target' in raw_profile:
+            # render the target if it was parsed from yaml
+            target_name = renderer.render_value(raw_profile['target'])
+        else:
+            target_name = 'default'
+            logger.debug(
+                "target not specified in profile '{}', using '{}'"
+                .format(profile_name, target_name)
+            )
+
+        raw_profile_data = cls._get_profile_data(
+            raw_profile, profile_name, target_name
+        )
+
+        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
+    def from_raw_profile_info(
+        cls,
+        raw_profile: Dict[str, Any],
+        profile_name: str,
+        renderer: ProfileRenderer,
+        user_config: Optional[Dict[str, Any]] = None,
+        target_override: Optional[str] = None,
+        threads_override: Optional[int] = None,
+    ) -> 'Profile':
+        """Create a profile from its raw profile information.
+
+         (this is an intermediate step, mostly useful for unit testing)
+
+        :param raw_profile: The profile data for a single profile, from
+            disk as yaml and its values rendered with jinja.
+        :param profile_name: The profile name used.
+        :param renderer: The config renderer.
+        :param user_config: The global config for the user, if it
+            was present.
+        :param target_override: The target to use, if provided on
+            the command line.
+        :param threads_override: The thread count to use, if
+            provided on the command line.
+        :raises DbtProfileError: If the profile is invalid or missing, or the
+            target could not be found
+        :returns: The new Profile object.
+        """
+        # user_config is not rendered.
+        if user_config is None:
+            user_config = 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
+        )
+
+        # valid connections never include the number of threads, but it's
+        # stored on a per-connection level in the raw configs
+        threads = profile_data.pop('threads', DEFAULT_THREADS)
+        if threads_override is not None:
+            threads = threads_override
+
+        credentials: Credentials = cls._credentials_from_profile(
+            profile_data, profile_name, target_name
+        )
+
+        return cls.from_credentials(
+            credentials=credentials,
+            profile_name=profile_name,
+            target_name=target_name,
+            threads=threads,
+            user_config=user_config
+        )
+
+    @classmethod
+    def from_raw_profiles(
+        cls,
+        raw_profiles: Dict[str, Any],
+        profile_name: str,
+        renderer: ProfileRenderer,
+        target_override: Optional[str] = None,
+        threads_override: Optional[int] = None,
+    ) -> 'Profile':
+        """
+        :param raw_profiles: The profile data, from disk as yaml.
+        :param profile_name: The profile name to use.
+        :param renderer: The config renderer.
+        :param target_override: The target to use, if provided on the command
+            line.
+        :param threads_override: The thread count to use, if provided on the
+            command line.
+        :raises DbtProjectError: If there is no profile name specified in the
+            project or the command line arguments
+        :raises DbtProfileError: If the profile is invalid or missing, or the
+            target could not be found
+        :returns: The new Profile object.
+        """
+        if profile_name not in raw_profiles:
+            raise DbtProjectError(
+                "Could not find profile named '{}'".format(profile_name)
+            )
+
+        # 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_config = raw_profiles.get('config')
+
+        return cls.from_raw_profile_info(
+            raw_profile=raw_profile,
+            profile_name=profile_name,
+            renderer=renderer,
+            user_config=user_config,
+            target_override=target_override,
+            threads_override=threads_override,
+        )
+
+    @classmethod
+    def render_from_args(
+        cls,
+        args: Any,
+        renderer: ProfileRenderer,
+        project_profile_name: Optional[str],
+    ) -> 'Profile':
+        """Given the raw profiles as read from disk and the name of the desired
+        profile if specified, return the profile component of the runtime
+        config.
+
+        :param args argparse.Namespace: The arguments as parsed from the cli.
+        :param project_profile_name Optional[str]: The profile name, if
+            specified in a project.
+        :raises DbtProjectError: If there is no profile name specified in the
+            project or the command line arguments, or if the specified profile
+            is not found
+        :raises DbtProfileError: If the profile is invalid or missing, or the
+            target could not be found.
+        :returns Profile: The new Profile object.
+        """
+        threads_override = getattr(args, 'threads', None)
+        target_override = getattr(args, 'target', None)
+        raw_profiles = read_profile(flags.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,
+            renderer=renderer,
+            target_override=target_override,
+            threads_override=threads_override
+        )

core/dbt/config/project.py

@@ -0,0 +1,698 @@
+from copy import deepcopy
+from dataclasses import dataclass, field
+from itertools import chain
+from typing import (
+    List, Dict, Any, Optional, TypeVar, Union, Mapping,
+)
+from typing_extensions import Protocol, runtime_checkable
+
+import hashlib
+import os
+
+from dbt import deprecations
+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 SemverException
+from dbt.exceptions import validator_error_message
+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.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 dbt.dataclass_schema import ValidationError
+from .renderer import DbtProjectYamlRenderer
+from .selectors import (
+    selector_config_from_data,
+    selector_data_from_root,
+    SelectorConfig,
+)
+
+
+INVALID_VERSION_ERROR = """\
+This version of dbt is not supported with the '{package}' package.
+  Installed version of dbt: {installed}
+  Required version of dbt for '{package}': {version_spec}
+Check the requirements for the '{package}' package, or run dbt again with \
+--no-version-check
+"""
+
+
+IMPOSSIBLE_VERSION_ERROR = """\
+The package version requirement can never be satisfied for the '{package}
+package.
+  Required versions of dbt for '{package}': {version_spec}
+Check the requirements for the '{package}' package, or run dbt again with \
+--no-version-check
+"""
+
+MALFORMED_PACKAGE_ERROR = """\
+The packages.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}
+"""
+
+
+@runtime_checkable
+class IsFQNResource(Protocol):
+    fqn: List[str]
+    resource_type: NodeType
+    package_name: str
+
+
+def _load_yaml(path):
+    contents = load_file_contents(path)
+    return load_yaml_text(contents)
+
+
+def package_data_from_root(project_root):
+    package_filepath = resolve_path_from_base(
+        'packages.yml', project_root
+    )
+
+    if path_exists(package_filepath):
+        packages_dict = _load_yaml(package_filepath)
+    else:
+        packages_dict = None
+    return packages_dict
+
+
+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(
+            MALFORMED_PACKAGE_ERROR.format(error=str(e.message))
+        ) from e
+    return packages
+
+
+def _parse_versions(versions: Union[List[str], str]) -> List[VersionSpecifier]:
+    """Parse multiple versions as read from disk. The versions value may be any
+    one of:
+        - a single version string ('>0.12.1')
+        - a single string specifying multiple comma-separated versions
+            ('>0.11.1,<=0.12.2')
+        - an array of single-version strings (['>0.11.1', '<=0.12.2'])
+
+    Regardless, this will return a list of VersionSpecifiers
+    """
+    if isinstance(versions, str):
+        versions = versions.split(',')
+    return [VersionSpecifier.from_version_string(v) for v in versions]
+
+
+def _all_source_paths(
+    model_paths: List[str],
+    seed_paths: List[str],
+    snapshot_paths: List[str],
+    analysis_paths: List[str],
+    macro_paths: List[str],
+) -> List[str]:
+    return list(chain(model_paths, seed_paths, snapshot_paths, analysis_paths,
+                      macro_paths))
+
+
+T = TypeVar('T')
+
+
+def value_or(value: Optional[T], default: T) -> T:
+    if value is None:
+        return default
+    else:
+        return value
+
+
+def _raw_project_from(project_root: str) -> Dict[str, Any]:
+
+    project_root = os.path.normpath(project_root)
+    project_yaml_filepath = os.path.join(project_root, 'dbt_project.yml')
+
+    # get the project.yml contents
+    if not path_exists(project_yaml_filepath):
+        raise DbtProjectError(
+            'no dbt_project.yml found at expected path {}'
+            .format(project_yaml_filepath)
+        )
+
+    project_dict = _load_yaml(project_yaml_filepath)
+
+    if not isinstance(project_dict, dict):
+        raise DbtProjectError(
+            'dbt_project.yml does not parse to a dictionary'
+        )
+
+    return project_dict
+
+
+def _query_comment_from_cfg(
+        cfg_query_comment: Union[QueryComment, NoValue, str, None]
+) -> QueryComment:
+    if not cfg_query_comment:
+        return QueryComment(comment='')
+
+    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)
+
+    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)
+
+
+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
+
+    try:
+        dbt_version = _parse_versions(dbt_raw_version)
+    except SemverException as e:
+        raise DbtProjectError(str(e)) from e
+
+    if verify_version:
+        # no name is also an error that we want to raise
+        if 'name' not in project_dict:
+            raise DbtProjectError(
+                '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:
+            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 check_config_path(self, project_dict, deprecated_path, exp_path):
+        if deprecated_path in project_dict:
+            if exp_path in project_dict:
+                msg = (
+                    '{deprecated_path} and {exp_path} cannot both be defined. The '
+                    '`{deprecated_path}` config has been deprecated in favor of `{exp_path}`. '
+                    'Please update your `dbt_project.yml` configuration to reflect this '
+                    'change.'
+                )
+                raise DbtProjectError(msg.format(deprecated_path=deprecated_path,
+                                                 exp_path=exp_path))
+            deprecations.warn('project_config_path',
+                              deprecated_path=deprecated_path,
+                              exp_path=exp_path)
+
+    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,
+        )
+
+        self.check_config_path(rendered.project_dict, 'source-paths', 'model-paths')
+        self.check_config_path(rendered.project_dict, 'data-paths', 'seed-paths')
+
+        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
+        version = cfg.version
+        # this is added at project_dict parse time and should always be here
+        # once we see it.
+        if cfg.project_root is None:
+            raise DbtProjectError('cfg must have a project root!')
+        else:
+            project_root = cfg.project_root
+        # this is only optional in the sense that if it's not present, it needs
+        # to have been a cli argument.
+        profile_name = cfg.profile
+        # these are all the defaults
+
+        # `source_paths` is deprecated but still allowed. Copy it into
+        # `model_paths` to simlify logic throughout the rest of the system.
+        model_paths: List[str] = value_or(cfg.model_paths
+                                          if 'model-paths' in rendered.project_dict
+                                          else cfg.source_paths, ['models'])
+        macro_paths: List[str] = value_or(cfg.macro_paths, ['macros'])
+        # `data_paths` is deprecated but still allowed. Copy it into
+        # `seed_paths` to simlify logic throughout the rest of the system.
+        seed_paths: List[str] = value_or(cfg.seed_paths
+                                         if 'seed-paths' in rendered.project_dict
+                                         else cfg.data_paths, ['seeds'])
+        test_paths: List[str] = value_or(cfg.test_paths, ['tests'])
+        analysis_paths: List[str] = value_or(cfg.analysis_paths, ['analyses'])
+        snapshot_paths: List[str] = value_or(cfg.snapshot_paths, ['snapshots'])
+
+        all_source_paths: List[str] = _all_source_paths(
+            model_paths, seed_paths, snapshot_paths, analysis_paths,
+            macro_paths
+        )
+
+        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')
+        packages_install_path: str = value_or(cfg.packages_install_path, 'dbt_packages')
+        # in the default case we'll populate this once we know the adapter type
+        # It would be nice to just pass along a Quoting here, but that would
+        # break many things
+        quoting: Dict[str, Any] = {}
+        if cfg.quoting is not None:
+            quoting = cfg.quoting.to_dict(omit_none=True)
+
+        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, [])
+
+        query_comment = _query_comment_from_cfg(cfg.query_comment)
+
+        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'])
+        project = Project(
+            project_name=name,
+            version=version,
+            project_root=project_root,
+            profile_name=profile_name,
+            model_paths=model_paths,
+            macro_paths=macro_paths,
+            seed_paths=seed_paths,
+            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,
+            log_path=log_path,
+            packages_install_path=packages_install_path,
+            quoting=quoting,
+            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]
+    model_paths: List[str]
+    macro_paths: List[str]
+    seed_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
+    packages_install_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.model_paths, self.seed_paths, self.snapshot_paths,
+            self.analysis_paths, self.macro_paths
+        )
+
+    def __str__(self):
+        cfg = self.to_project_config(with_packages=True)
+        return str(cfg)
+
+    def __eq__(self, other):
+        if not (isinstance(other, self.__class__) and
+                isinstance(self, other.__class__)):
+            return False
+        return self.to_project_config(with_packages=True) == \
+            other.to_project_config(with_packages=True)
+
+    def to_project_config(self, with_packages=False):
+        """Return a dict representation of the config that could be written to
+        disk with `yaml.safe_dump` to get this configuration.
+
+        :param with_packages bool: If True, include the serialized packages
+            file in the root.
+        :returns dict: The serialized profile.
+        """
+        result = deepcopy({
+            'name': self.project_name,
+            'version': self.version,
+            'project-root': self.project_root,
+            'profile': self.profile_name,
+            'model-paths': self.model_paths,
+            'macro-paths': self.macro_paths,
+            'seed-paths': self.seed_paths,
+            '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,
+            'log-path': self.log_path,
+            'quoting': self.quoting,
+            '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(omit_none=True))
+
+        return result
+
+    def validate(self):
+        try:
+            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],
+        selectors_dict: Dict[str, Any],
+        renderer: DbtProjectYamlRenderer,
+        *,
+        verify_version: bool = False
+    ) -> 'Project':
+        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: DbtProjectYamlRenderer,
+        *,
+        verify_version: bool = False,
+    ) -> 'Project':
+        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_selector(self, name: str) -> Union[SelectionSpec, bool]:
+        if name not in self.selectors:
+            raise RuntimeException(
+                f'Could not find selector named {name}, expected one of '
+                f'{list(self.selectors)}'
+            )
+        return self.selectors[name]["definition"]
+
+    def get_default_selector_name(self) -> Union[str, None]:
+        """This function fetch the default selector to use on `dbt run` (if any)
+        :return: either a selector if default is set or None
+        :rtype: Union[SelectionSpec, None]
+        """
+        for selector_name, selector in self.selectors.items():
+            if selector["default"] is True:
+                return selector_name
+
+        return None
+
+    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

@@ -0,0 +1,238 @@
+from typing import Dict, Any, Tuple, Optional, Union, Callable
+
+from dbt.clients.jinja import get_rendered, catch_jinja
+
+from dbt.exceptions import (
+    DbtProjectError, CompilationException, RecursionException
+)
+from dbt.node_types import NodeType
+from dbt.utils import deep_map
+
+
+Keypath = Tuple[Union[str, int], ...]
+
+
+class BaseRenderer:
+    def __init__(self, context: Dict[str, Any]) -> None:
+        self.context = context
+
+    @property
+    def name(self):
+        return 'Rendering'
+
+    def should_render_keypath(self, keypath: Keypath) -> bool:
+        return True
+
+    def render_entry(self, value: Any, keypath: Keypath) -> Any:
+        if not self.should_render_keypath(keypath):
+            return value
+
+        return self.render_value(value, keypath)
+
+    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
+        try:
+            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(
+                f'Cycle detected: {self.name} input has a reference to itself',
+                project=data
+            )
+
+
+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
+
+
+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

@@ -0,0 +1,586 @@
+import itertools
+import os
+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 DbtProjectYamlRenderer, ProfileRenderer
+from .utils import parse_cli_vars
+from dbt import flags
+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, Credentials
+from dbt.contracts.graph.manifest import ManifestMetadata
+from dbt.contracts.relation import ComponentName
+from dbt.logger import GLOBAL_LOGGER as logger
+from dbt.ui import warning_tag
+
+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,
+        dependencies: Optional[Mapping[str, 'RuntimeConfig']] = None,
+    ) -> 'RuntimeConfig':
+        """Instantiate a RuntimeConfig from its components.
+
+        :param profile: A parsed dbt Profile.
+        :param project: A parsed dbt Project.
+        :param args: The parsed command-line arguments.
+        :returns RuntimeConfig: The new configuration.
+        """
+        quoting: Dict[str, Any] = (
+            get_relation_class_by_name(profile.credentials.type)
+            .get_default_quote_policy()
+            .replace_dict(_project_quoting_dict(project, profile))
+        ).to_dict(omit_none=True)
+
+        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,
+            model_paths=project.model_paths,
+            macro_paths=project.macro_paths,
+            seed_paths=project.seed_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,
+            packages_install_path=project.packages_install_path,
+            quoting=quoting,
+            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,
+            user_config=profile.user_config,
+            threads=profile.threads,
+            credentials=profile.credentials,
+            args=args,
+            cli_vars=cli_vars,
+            dependencies=dependencies,
+        )
+
+    def new_project(self, project_root: str) -> 'RuntimeConfig':
+        """Given a new project root, read in its project dictionary, supply the
+        existing project's profile info, and create a new project file.
+
+        :param project_root: A filepath to a dbt project.
+        :raises DbtProfileError: If the profile is invalid.
+        :raises DbtProjectError: If project is missing or invalid.
+        :returns: The new configuration.
+        """
+        # copy profile
+        profile = Profile(**self.to_profile_info())
+        profile.validate()
+
+        # load the new project and its packages. Don't pass cli variables.
+        renderer = DbtProjectYamlRenderer(generate_target_context(profile, {}))
+
+        project = Project.from_project_root(
+            project_root,
+            renderer,
+            verify_version=bool(flags.VERSION_CHECK),
+        )
+
+        cfg = self.from_parts(
+            project=project,
+            profile=profile,
+            args=deepcopy(self.args),
+        )
+        # force our quoting back onto the new project.
+        cfg.quoting = deepcopy(self.quoting)
+        return cfg
+
+    def serialize(self) -> Dict[str, Any]:
+        """Serialize the full configuration to a single dictionary. For any
+        instance that has passed validate() (which happens in __init__), it
+        matches the Configuration contract.
+
+        Note that args are not serialized.
+
+        :returns dict: The serialized configuration.
+        """
+        result = self.to_project_config(with_packages=True)
+        result.update(self.to_profile_info(serialize_credentials=True))
+        result['cli_vars'] = deepcopy(self.cli_vars)
+        return result
+
+    def validate(self):
+        """Validate the configuration against its contract.
+
+        :raises DbtProjectError: If the configuration fails validation.
+        """
+        try:
+            Configuration.validate(self.serialize())
+        except ValidationError as e:
+            raise DbtProjectError(validator_error_message(e)) from e
+
+    @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 = bool(flags.VERSION_CHECK)
+        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':
+        """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)
+
+        return cls.from_parts(
+            project=project,
+            profile=profile,
+            args=args,
+        )
+
+    def get_metadata(self) -> ManifestMetadata:
+        return ManifestMetadata(
+            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.packages_install_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.packages_install_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.user_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,
+            model_paths=project.model_paths,
+            macro_paths=project.macro_paths,
+            seed_paths=project.seed_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,
+            packages_install_path=project.packages_install_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='',
+            user_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 +1,201 @@
+from pathlib import Path
+from typing import Dict, Any, Union
+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, Dict[str, Union[SelectionSpec, bool]]]):
+
+    @classmethod
+    def selectors_from_dict(cls, data: Dict[str, Any]) -> 'SelectorConfig':
+        try:
+            SelectorFile.validate(data)
+            selector_file = SelectorFile.from_dict(data)
+            validate_selector_default(selector_file)
+            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
+
+
+def validate_selector_default(selector_file: SelectorFile) -> None:
+    """Check if a selector.yml file has more than 1 default key set to true"""
+    default_set: bool = False
+    default_selector_name: Union[str, None] = None
+
+    for selector in selector_file.selectors:
+        if selector.default is True and default_set is False:
+            default_set = True
+            default_selector_name = selector.name
+            continue
+        if selector.default is True and default_set is True:
+            raise DbtSelectorsError(
+                "Error when parsing the selector file. "
+                "Found multiple selectors with `default: true`:"
+                f"{default_selector_name} and {selector.name}"
+            )
+
+
+# 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

@@ -0,0 +1,537 @@
+import json
+import os
+from typing import (
+    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.events.functions import fire_event
+from dbt.events.types import MacroEventInfo, MacroEventDebug
+from dbt.version import __version__ as dbt_version
+
+# 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]:
+    context_exports = pytz.__all__  # type: ignore
+
+    return {
+        name: getattr(pytz, name) for name in context_exports
+    }
+
+
+def get_datetime_module_context() -> Dict[str, Any]:
+    context_exports = [
+        'date',
+        'datetime',
+        'time',
+        'timedelta',
+        'tzinfo'
+    ]
+
+    return {
+        name: getattr(datetime, name) for name in context_exports
+    }
+
+
+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(),
+    }
+
+
+class ContextMember:
+    def __init__(self, value, name=None):
+        self.name = name
+        self.inner = value
+
+    def key(self, default):
+        if self.name is None:
+            return default
+        return self.name
+
+
+def contextmember(value):
+    if isinstance(value, str):
+        return lambda v: ContextMember(v, name=value)
+    return ContextMember(value)
+
+
+def contextproperty(value):
+    if isinstance(value, str):
+        return lambda v: ContextMember(property(v), name=value)
+    return ContextMember(property(value))
+
+
+class ContextMeta(type):
+    def __new__(mcls, name, bases, dct):
+        context_members = {}
+        context_attrs = {}
+        new_dct = {}
+
+        for base in bases:
+            context_members.update(getattr(base, '_context_members_', {}))
+            context_attrs.update(getattr(base, '_context_attrs_', {}))
+
+        for key, value in dct.items():
+            if isinstance(value, ContextMember):
+                context_key = value.key(key)
+                context_members[context_key] = value.inner
+                context_attrs[context_key] = key
+                value = value.inner
+            new_dct[key] = value
+        new_dct['_context_members_'] = context_members
+        new_dct['_context_attrs_'] = context_attrs
+        return type.__new__(mcls, name, bases, new_dct)
+
+
+class Var:
+    UndefinedVarError = "Required var '{}' not found in config:\nVars "\
+                        "supplied to {} = {}"
+    _VAR_NOTSET = object()
+
+    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()
+
+    def _generate_merged(self) -> Mapping[str, Any]:
+        return self._cli_vars
+
+    @property
+    def node_name(self):
+        if self._node is not None:
+            return self._node.name
+        else:
+            return '<Configuration>'
+
+    def get_missing_var(self, var_name):
+        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.node_name, pretty_vars
+        )
+        raise_compiler_error(msg, self._node)
+
+    def has_var(self, var_name: str):
+        return var_name in self._merged
+
+    def get_rendered_var(self, 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)
+
+    def __call__(self, var_name, default=_VAR_NOTSET):
+        if self.has_var(var_name):
+            return self.get_rendered_var(var_name)
+        elif default is not self._VAR_NOTSET:
+            return default
+        else:
+            return self.get_missing_var(var_name)
+
+
+class BaseContext(metaclass=ContextMeta):
+    def __init__(self, cli_vars):
+        self._ctx = {}
+        self.cli_vars = cli_vars
+
+    def generate_builtins(self):
+        builtins: Dict[str, Any] = {}
+        for key, value in self._context_members_.items():
+            if hasattr(value, '__get__'):
+                # handle properties, bound methods, etc
+                value = value.__get__(self)
+            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()
+        self._ctx['builtins'] = builtins
+        self._ctx.update(builtins)
+        return self._ctx
+
+    @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:
+        """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:
+            return default
+        else:
+            msg = f"Env var required but not provided: '{var}'"
+            undefined_error(msg)
+
+    if os.environ.get('DBT_MACRO_DEBUGGING'):
+        @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)
+            ipdb.set_trace(frame)
+            return ''
+
+    @contextmember('return')
+    @staticmethod
+    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:
+            return default
+
+    @contextmember
+    @staticmethod
+    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:
+            return default
+
+    @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 safe_load(value)
+        except (AttributeError, ValueError, yaml.YAMLError):
+            return default
+
+    # safe_dump defaults to sort_keys=True, but we act like json.dumps (the
+    # opposite)
+    @contextmember
+    @staticmethod
+    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):
+            return default
+
+    @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:
+            fire_event(MacroEventInfo(msg))
+        else:
+            fire_event(MacroEventDebug(msg))
+        return ''
+
+    @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:
+            return None
+
+    @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:
+            return None
+
+    @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 %}
+
+        This supports all flags defined in flags submodule (core/dbt/flags.py)
+        TODO: Replace with object that provides read-only access to flag values
+        """
+        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

@@ -0,0 +1,100 @@
+from typing import Any, Dict
+
+from dbt.contracts.connection import AdapterRequiredConfig
+from dbt.node_types import NodeType
+from dbt.utils import MultiDict
+
+from dbt.context.base import contextproperty, Var
+from dbt.context.target import TargetContext
+
+
+class ConfiguredContext(TargetContext):
+    config: AdapterRequiredConfig
+
+    def __init__(
+        self, config: AdapterRequiredConfig
+    ) -> None:
+        super().__init__(config, config.cli_vars)
+
+    @contextproperty
+    def project_name(self) -> str:
+        return self.config.project_name
+
+
+class FQNLookup:
+    def __init__(self, package_name: str):
+        self.package_name = package_name
+        self.fqn = [package_name]
+        self.resource_type = NodeType.Model
+
+
+class ConfiguredVar(Var):
+    def __init__(
+        self,
+        context: Dict[str, Any],
+        config: AdapterRequiredConfig,
+        project_name: str,
+    ):
+        super().__init__(context, config.cli_vars)
+        self._config = config
+        self._project_name = project_name
+
+    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
+        )
+
+
+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
+        )
+
+
+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

@@ -0,0 +1,86 @@
+from typing import (
+    Any, Dict, Union
+)
+
+from dbt.exceptions import (
+    doc_invalid_args,
+    doc_target_not_found,
+)
+from dbt.config.runtime import RuntimeConfig
+from dbt.contracts.graph.compiled import CompileResultNode
+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 SchemaYamlContext
+
+
+class DocsRuntimeContext(SchemaYamlContext):
+    def __init__(
+        self,
+        config: RuntimeConfig,
+        node: Union[ParsedMacro, CompileResultNode],
+        manifest: Manifest,
+        current_project: str,
+    ) -> None:
+        super().__init__(config, current_project)
+        self.node = node
+        self.manifest = manifest
+
+    @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
+            doc_name = args[0]
+        elif len(args) == 2:
+            doc_package_name, doc_name = args
+        else:
+            doc_invalid_args(self.node, args)
+
+        # ParsedDocumentation
+        target_doc = self.manifest.resolve_doc(
+            doc_name,
+            doc_package_name,
+            self._project_name,
+            self.node.package_name,
+        )
+        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_runtime_docs(
+    config: RuntimeConfig,
+    target: Any,
+    manifest: Manifest,
+    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

@@ -0,0 +1,84 @@
+from typing import Any, Dict
+
+from dbt.contracts.connection import HasCredentials
+
+from dbt.context.base import (
+    BaseContext, contextproperty
+)
+
+
+class TargetContext(BaseContext):
+    def __init__(self, config: HasCredentials, cli_vars: Dict[str, Any]):
+        super().__init__(cli_vars=cli_vars)
+        self.config = config
+
+    @contextproperty
+    def target(self) -> Dict[str, Any]:
+        """`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(
+    config: HasCredentials, cli_vars: Dict[str, Any]
+) -> Dict[str, Any]:
+    ctx = TargetContext(config, cli_vars)
+    return ctx.to_dict()

core/dbt/contracts/connection.py

@@ -0,0 +1,236 @@
+import abc
+import itertools
+import hashlib
+from dataclasses import dataclass, field
+from typing import (
+    Any, ClassVar, Dict, Tuple, Iterable, Optional, List, Callable,
+)
+from dbt.exceptions import InternalException
+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
+
+
+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'
+    CLOSED = 'closed'
+    FAIL = 'fail'
+
+
+@dataclass(init=False)
+class Connection(ExtensibleDbtClassMixin, Replaceable):
+    type: Identifier
+    name: Optional[str] = None
+    state: ConnectionState = ConnectionState.INIT
+    transaction_open: bool = False
+    _handle: Optional[Any] = None
+    _credentials: Optional[Any] = None
+
+    def __init__(
+        self,
+        type: Identifier,
+        name: Optional[str],
+        credentials: dbtClassMixin,
+        state: ConnectionState = ConnectionState.INIT,
+        transaction_open: bool = False,
+        handle: Optional[Any] = None,
+    ) -> None:
+        self.type = type
+        self.name = name
+        self.state = state
+        self.credentials = credentials
+        self.transaction_open = transaction_open
+        self.handle = handle
+
+    @property
+    def credentials(self):
+        return self._credentials
+
+    @credentials.setter
+    def credentials(self, value):
+        self._credentials = value
+
+    @property
+    def handle(self):
+        if isinstance(self._handle, LazyHandle):
+            try:
+                # this will actually change 'self._handle'.
+                self._handle.resolve(self)
+            except RecursionError as exc:
+                raise InternalException(
+                    "A connection's open() method attempted to read the "
+                    "handle value"
+                ) from exc
+        return self._handle
+
+    @handle.setter
+    def handle(self, value):
+        self._handle = value
+
+
+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)
+
+
+# see https://github.com/python/mypy/issues/4717#issuecomment-373932080
+# and https://github.com/python/mypy/issues/5374
+# for why we have type: ignore. Maybe someday dataclasses + abstract classes
+# will work.
+@dataclass  # type: ignore
+class Credentials(
+    ExtensibleDbtClassMixin,
+    Replaceable,
+    metaclass=abc.ABCMeta
+):
+    database: str
+    schema: str
+    _ALIASES: ClassVar[Dict[str, str]] = field(default={}, init=False)
+
+    @abc.abstractproperty
+    def type(self) -> str:
+        raise NotImplementedError(
+            'type not implemented for base credentials class'
+        )
+
+    @property
+    def unique_field(self) -> str:
+        """Hashed and included in anonymous telemetry to track adapter adoption.
+        Return the field from Credentials that can uniquely identify
+        one team/organization using this adapter
+        """
+        raise NotImplementedError(
+            'unique_field 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)
+        connection_keys = set(self._connection_keys())
+        aliases: List[str] = []
+        if with_aliases:
+            aliases = [
+                k for k, v in self._ALIASES.items() if v in connection_keys
+            ]
+        for key in itertools.chain(self._connection_keys(), aliases):
+            if key in as_dict:
+                yield key, as_dict[key]
+
+    @abc.abstractmethod
+    def _connection_keys(self) -> Tuple[str, ...]:
+        raise NotImplementedError
+
+    @classmethod
+    def __pre_deserialize__(cls, data):
+        data = super().__pre_deserialize__(data)
+        data = cls.translate_aliases(data)
+        return data
+
+    @classmethod
+    def translate_aliases(
+        cls, kwargs: Dict[str, Any], recurse: bool = False
+    ) -> Dict[str, Any]:
+        return translate_aliases(kwargs, cls._ALIASES, recurse)
+
+    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 dct
+            })
+        return dct
+
+
+class UserConfigContract(Protocol):
+    send_anonymous_usage_stats: bool
+    use_colors: Optional[bool] = None
+    partial_parse: Optional[bool] = None
+    printer_width: Optional[int] = None
+
+
+class HasCredentials(Protocol):
+    credentials: Credentials
+    profile_name: str
+    user_config: UserConfigContract
+    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: QueryComment
+    cli_vars: Dict[str, Any]
+    target_path: str

core/dbt/contracts/files.py

@@ -0,0 +1,303 @@
+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'
+    SingularTest = 'singular_test'
+    GenericTest = 'generic_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.SingularTest: 'SingularTestParser',
+    ParseFileType.GenericTest: 'GenericTestParser',
+    ParseFileType.Seed: 'SeedParser',
+    ParseFileType.Documentation: 'DocumentationParser',
+    ParseFileType.Schema: 'SchemaParser',
+    ParseFileType.Hook: 'HookParser',
+}
+
+
+@dataclass
+class FilePath(dbtClassMixin):
+    searched_path: str
+    relative_path: str
+    modification_time: float
+    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'
+
+    @property
+    def modification_time(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
+        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

@@ -0,0 +1,227 @@
+from dbt.contracts.graph.parsed import (
+    HasTestMetadata,
+    ParsedNode,
+    ParsedAnalysisNode,
+    ParsedSingularTestNode,
+    ParsedHookNode,
+    ParsedModelNode,
+    ParsedExposure,
+    ParsedResource,
+    ParsedRPCNode,
+    ParsedGenericTestNode,
+    ParsedSeedNode,
+    ParsedSnapshotNode,
+    ParsedSourceDefinition,
+    SeedConfig,
+    TestConfig,
+    same_seeds,
+)
+from dbt.node_types import NodeType
+from dbt.contracts.util import Replaceable
+
+from dbt.dataclass_schema import dbtClassMixin
+from dataclasses import dataclass, field
+from typing import Optional, List, Union, Dict, Type
+
+
+@dataclass
+class InjectedCTE(dbtClassMixin, Replaceable):
+    id: str
+    sql: str
+
+
+@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, CompiledNodeMixin):
+    compiled_sql: Optional[str] = None
+    extra_ctes_injected: bool = False
+    extra_ctes: List[InjectedCTE] = field(default_factory=list)
+    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
+        do if extra_ctes were an OrderedDict
+        """
+        for cte in self.extra_ctes:
+            if cte.id == cte_id:
+                cte.sql = sql
+                break
+        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):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Analysis]})
+
+
+@dataclass
+class CompiledHookNode(CompiledNode):
+    resource_type: NodeType = field(
+        metadata={'restrict': [NodeType.Operation]}
+    )
+    index: Optional[int] = None
+
+
+@dataclass
+class CompiledModelNode(CompiledNode):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Model]})
+
+
+@dataclass
+class CompiledRPCNode(CompiledNode):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.RPCCall]})
+
+
+@dataclass
+class CompiledSeedNode(CompiledNode):
+    # keep this in sync with ParsedSeedNode!
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Seed]})
+    config: SeedConfig = field(default_factory=SeedConfig)
+
+    @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):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Snapshot]})
+
+
+@dataclass
+class CompiledSingularTestNode(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 CompiledGenericTestNode(CompiledNode, HasTestMetadata):
+    # keep this in sync with ParsedGenericTestNode!
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Test]})
+    column_name: Optional[str] = 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
+        )
+
+
+CompiledTestNode = Union[CompiledSingularTestNode, CompiledGenericTestNode]
+
+
+PARSED_TYPES: Dict[Type[CompiledNode], Type[ParsedResource]] = {
+    CompiledAnalysisNode: ParsedAnalysisNode,
+    CompiledModelNode: ParsedModelNode,
+    CompiledHookNode: ParsedHookNode,
+    CompiledRPCNode: ParsedRPCNode,
+    CompiledSeedNode: ParsedSeedNode,
+    CompiledSnapshotNode: ParsedSnapshotNode,
+    CompiledSingularTestNode: ParsedSingularTestNode,
+    CompiledGenericTestNode: ParsedGenericTestNode,
+}
+
+
+COMPILED_TYPES: Dict[Type[ParsedResource], Type[CompiledNode]] = {
+    ParsedAnalysisNode: CompiledAnalysisNode,
+    ParsedModelNode: CompiledModelNode,
+    ParsedHookNode: CompiledHookNode,
+    ParsedRPCNode: CompiledRPCNode,
+    ParsedSeedNode: CompiledSeedNode,
+    ParsedSnapshotNode: CompiledSnapshotNode,
+    ParsedSingularTestNode: CompiledSingularTestNode,
+    ParsedGenericTestNode: CompiledGenericTestNode,
+}
+
+
+# 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(type(compiled))
+    if cls is None:
+        # how???
+        raise ValueError('invalid resource_type: {}'
+                         .format(compiled.resource_type))
+
+    return cls.from_dict(compiled.to_dict(omit_none=True))
+
+
+NonSourceCompiledNode = Union[
+    CompiledAnalysisNode,
+    CompiledSingularTestNode,
+    CompiledModelNode,
+    CompiledHookNode,
+    CompiledRPCNode,
+    CompiledGenericTestNode,
+    CompiledSeedNode,
+    CompiledSnapshotNode,
+]
+
+NonSourceParsedNode = Union[
+    ParsedAnalysisNode,
+    ParsedSingularTestNode,
+    ParsedHookNode,
+    ParsedModelNode,
+    ParsedRPCNode,
+    ParsedGenericTestNode,
+    ParsedSeedNode,
+    ParsedSnapshotNode,
+]
+
+
+# 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[
+    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

@@ -0,0 +1,792 @@
+import os
+import time
+from dataclasses import dataclass, field
+from mashumaro.types import SerializableType
+from pathlib import Path
+from typing import (
+    Optional,
+    Union,
+    List,
+    Dict,
+    Any,
+    Sequence,
+    Tuple,
+    Iterator,
+    TypeVar,
+)
+
+from dbt.dataclass_schema import (
+    dbtClassMixin, ExtensibleDbtClassMixin
+)
+
+from dbt.clients.system import write_file
+from dbt.contracts.files import FileHash, MAXIMUM_SEED_SIZE_NAME
+from dbt.contracts.graph.unparsed import (
+    UnparsedNode, UnparsedDocumentation, Quoting, Docs,
+    UnparsedBaseNode, FreshnessThreshold, ExternalTable,
+    HasYamlMetadata, MacroArgument, UnparsedSourceDefinition,
+    UnparsedSourceTableDefinition, UnparsedColumn, TestDef,
+    ExposureOwner, ExposureType, MaturityType
+)
+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
+
+
+from .model_config import (
+    NodeConfig,
+    SeedConfig,
+    TestConfig,
+    SourceConfig,
+    EmptySnapshotConfig,
+    SnapshotConfig,
+)
+
+
+@dataclass
+class ColumnInfo(
+    AdditionalPropertiesMixin,
+    ExtensibleDbtClassMixin,
+    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(dbtClassMixin, Replaceable):
+    fqn: List[str]
+
+    def same_fqn(self, other: 'HasFqn') -> bool:
+        return self.fqn == other.fqn
+
+
+@dataclass
+class HasUniqueID(dbtClassMixin, Replaceable):
+    unique_id: str
+
+
+@dataclass
+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
+    def add_macro(self, value: str):
+        if value not in self.macros:
+            self.macros.append(value)
+
+
+@dataclass
+class DependsOn(MacroDependsOn):
+    nodes: List[str] = field(default_factory=list)
+
+    def add_node(self, value: str):
+        if value not in self.nodes:
+            self.nodes.append(value)
+
+
+@dataclass
+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(dbtClassMixin):
+    resource_type: NodeType
+    depends_on: DependsOn
+    config: NodeConfig
+
+    @property
+    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'
+
+    @property
+    def is_ephemeral_model(self):
+        return self.is_refable and self.is_ephemeral
+
+    @property
+    def depends_on_nodes(self):
+        return self.depends_on.nodes
+
+    def patch(self, patch: 'ParsedNodePatch'):
+        """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
+        # 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
+
+    def get_materialization(self):
+        return self.config.materialized
+
+
+@dataclass
+class ParsedNodeMandatory(
+    UnparsedNode,
+    HasUniqueID,
+    HasFqn,
+    HasRelationMetadata,
+    Replaceable
+):
+    alias: str
+    checksum: FileHash
+    config: NodeConfig = field(default_factory=NodeConfig)
+
+    @property
+    def identifier(self):
+        return self.alias
+
+
+@dataclass
+class ParsedNodeDefaults(ParsedNodeMandatory):
+    tags: List[str] = field(default_factory=list)
+    refs: List[List[str]] = field(default_factory=list)
+    sources: List[List[str]] = field(default_factory=list)
+    depends_on: DependsOn = field(default_factory=DependsOn)
+    description: str = field(default='')
+    columns: Dict[str, ColumnInfo] = field(default_factory=dict)
+    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, path
+        )
+
+        write_file(full_path, payload)
+        return full_path
+
+
+T = TypeVar('T', bound='ParsedNode')
+
+
+@dataclass
+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 ParsedGenericTestNode.from_dict(dct)
+            else:
+                return ParsedSingularTestNode.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
+class ParsedAnalysisNode(ParsedNode):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Analysis]})
+
+
+@dataclass
+class ParsedHookNode(ParsedNode):
+    resource_type: NodeType = field(
+        metadata={'restrict': [NodeType.Operation]}
+    )
+    index: Optional[int] = None
+
+
+@dataclass
+class ParsedModelNode(ParsedNode):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Model]})
+
+
+@dataclass
+class ParsedRPCNode(ParsedNode):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.RPCCall]})
+
+
+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)
+
+    @property
+    def empty(self):
+        """ Seeds are never empty"""
+        return False
+
+    def same_body(self: T, other: T) -> bool:
+        return same_seeds(self, other)
+
+
+@dataclass
+class TestMetadata(dbtClassMixin, Replaceable):
+    name: str
+    kwargs: Dict[str, Any] = field(default_factory=dict)
+    namespace: Optional[str] = None
+
+
+@dataclass
+class HasTestMetadata(dbtClassMixin):
+    test_metadata: TestMetadata
+
+
+@dataclass
+class ParsedSingularTestNode(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 ParsedGenericTestNode(ParsedNode, HasTestMetadata):
+    # keep this in sync with CompiledGenericTestNode!
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Test]})
+    column_name: Optional[str] = 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
+        )
+
+
+@dataclass
+class IntermediateSnapshotNode(ParsedNode):
+    # at an intermediate stage in parsing, where we've built something better
+    # than an unparsed node for rendering in parse mode, it's pretty possible
+    # that we won't have critical snapshot-related information that is only
+    # defined in config blocks. To fix that, we have an intermediate type that
+    # 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]})
+    config: EmptySnapshotConfig = field(default_factory=EmptySnapshotConfig)
+
+
+@dataclass
+class ParsedSnapshotNode(ParsedNode):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Snapshot]})
+    config: SnapshotConfig
+
+
+@dataclass
+class ParsedPatch(HasYamlMetadata, Replaceable):
+    name: str
+    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
+# regular parsed node. Note that description and columns must be present, but
+# may be empty.
+@dataclass
+class ParsedNodePatch(ParsedPatch):
+    columns: Dict[str, ColumnInfo]
+
+
+@dataclass
+class ParsedMacroPatch(ParsedPatch):
+    arguments: List[MacroArgument] = field(default_factory=list)
+
+
+@dataclass
+class ParsedMacro(UnparsedBaseNode, HasUniqueID):
+    name: str
+    macro_sql: str
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Macro]})
+    # TODO: can macros even have tags?
+    tags: List[str] = field(default_factory=list)
+    # TODO: is this ever populated?
+    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)
+    created_at: int = field(default_factory=lambda: int(time.time()))
+
+    def patch(self, patch: ParsedMacroPatch):
+        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
+
+    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(UnparsedDocumentation, HasUniqueID):
+    name: str
+    block_contents: str
+
+    @property
+    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,
+
+):
+    name: str
+    source_name: str
+    source_description: str
+    loader: str
+    identifier: str
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Source]})
+    quoting: Quoting = field(default_factory=Quoting)
+    loaded_at_field: Optional[str] = None
+    freshness: Optional[FreshnessThreshold] = None
+    external: Optional[ExternalTable] = None
+    description: str = ''
+    columns: Dict[str, ColumnInfo] = field(default_factory=dict)
+    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):
+        return False
+
+    @property
+    def is_ephemeral(self):
+        return False
+
+    @property
+    def is_ephemeral_model(self):
+        return False
+
+    @property
+    def depends_on_nodes(self):
+        return []
+
+    @property
+    def depends_on(self):
+        return DependsOn(macros=[], nodes=[])
+
+    @property
+    def refs(self):
+        return []
+
+    @property
+    def sources(self):
+        return []
+
+    @property
+    def has_freshness(self):
+        return bool(self.freshness) and self.loaded_at_field is not None
+
+    @property
+    def search_name(self):
+        return f'{self.source_name}.{self.name}'
+
+
+@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,
+    ParsedSingularTestNode,
+    ParsedHookNode,
+    ParsedModelNode,
+    ParsedRPCNode,
+    ParsedGenericTestNode,
+    ParsedSeedNode,
+    ParsedSnapshotNode,
+]
+
+
+ParsedResource = Union[
+    ParsedDocumentation,
+    ParsedMacro,
+    ParsedNode,
+    ParsedExposure,
+    ParsedSourceDefinition,
+]

core/dbt/contracts/graph/unparsed.py

@@ -0,0 +1,442 @@
+from dbt.node_types import NodeType
+from dbt.contracts.util import (
+    AdditionalPropertiesMixin,
+    Mergeable,
+    Replaceable,
+)
+# trigger the PathEncoder
+import dbt.helper_types  # noqa:F401
+from dbt.exceptions import CompilationException
+
+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(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:
+    raw_sql: str
+
+    @property
+    def empty(self):
+        return not self.raw_sql.strip()
+
+
+@dataclass
+class UnparsedMacro(UnparsedBaseNode, HasSQL):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Macro]})
+
+
+@dataclass
+class UnparsedGenericTest(UnparsedBaseNode, HasSQL):
+    resource_type: NodeType = field(metadata={'restrict': [NodeType.Macro]})
+
+
+@dataclass
+class UnparsedNode(UnparsedBaseNode, HasSQL):
+    name: str
+    resource_type: NodeType = field(metadata={'restrict': [
+        NodeType.Model,
+        NodeType.Analysis,
+        NodeType.Test,
+        NodeType.Snapshot,
+        NodeType.Operation,
+        NodeType.Seed,
+        NodeType.RPCCall,
+    ]})
+
+    @property
+    def search_name(self):
+        return self.name
+
+
+@dataclass
+class UnparsedRunHook(UnparsedNode):
+    resource_type: NodeType = field(
+        metadata={'restrict': [NodeType.Operation]}
+    )
+    index: Optional[int] = None
+
+
+@dataclass
+class Docs(dbtClassMixin, Replaceable):
+    show: bool = True
+
+
+@dataclass
+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]
+
+
+@dataclass
+class HasTests(HasDocs):
+    tests: Optional[List[TestDef]] = None
+
+    def __post_init__(self):
+        if self.tests is None:
+            self.tests = []
+
+
+@dataclass
+class UnparsedColumn(HasTests):
+    quote: Optional[bool] = None
+    tags: List[str] = field(default_factory=list)
+
+
+@dataclass
+class HasColumnDocs(dbtClassMixin, Replaceable):
+    columns: Sequence[HasDocs] = field(default_factory=list)
+
+
+@dataclass
+class HasColumnTests(HasColumnDocs):
+    columns: Sequence[UnparsedColumn] = field(default_factory=list)
+
+
+@dataclass
+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 HasConfig():
+    config: Dict[str, Any] = field(default_factory=dict)
+
+
+@dataclass
+class UnparsedAnalysisUpdate(HasConfig, HasColumnDocs, HasDocs, HasYamlMetadata):
+    pass
+
+
+@dataclass
+class UnparsedNodeUpdate(HasConfig, HasColumnTests, HasTests, HasYamlMetadata):
+    quote_columns: Optional[bool] = None
+
+
+@dataclass
+class MacroArgument(dbtClassMixin):
+    name: str
+    type: Optional[str] = None
+    description: str = ''
+
+
+@dataclass
+class UnparsedMacroUpdate(HasConfig, HasDocs, HasYamlMetadata):
+    arguments: List[MacroArgument] = field(default_factory=list)
+
+
+class TimePeriod(StrEnum):
+    minute = 'minute'
+    hour = 'hour'
+    day = 'day'
+
+    def plural(self) -> str:
+        return str(self) + 's'
+
+
+@dataclass
+class Time(dbtClassMixin, Replaceable):
+    count: int
+    period: TimePeriod
+
+    def exceeded(self, actual_age: float) -> bool:
+        kwargs = {self.period.plural(): self.count}
+        difference = timedelta(**kwargs).total_seconds()
+        return actual_age > difference
+
+
+@dataclass
+class FreshnessThreshold(dbtClassMixin, Mergeable):
+    warn_after: Optional[Time] = None
+    error_after: Optional[Time] = None
+    filter: Optional[str] = None
+
+    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):
+            return FreshnessStatus.Warn
+        else:
+            return FreshnessStatus.Pass
+
+    def __bool__(self):
+        return self.warn_after is not None or self.error_after is not None
+
+
+@dataclass
+class AdditionalPropertiesAllowed(
+    AdditionalPropertiesMixin,
+    ExtensibleDbtClassMixin
+):
+    _extra: Dict[str, Any] = field(default_factory=dict)
+
+
+@dataclass
+class ExternalPartition(AdditionalPropertiesAllowed, Replaceable):
+    name: str = ''
+    description: str = ''
+    data_type: str = ''
+    meta: Dict[str, Any] = field(default_factory=dict)
+
+    def __post_init__(self):
+        if self.name == '' or self.data_type == '':
+            raise CompilationException(
+                'External partition columns must have names and data types'
+            )
+
+
+@dataclass
+class ExternalTable(AdditionalPropertiesAllowed, Mergeable):
+    location: Optional[str] = None
+    file_format: Optional[str] = None
+    row_format: Optional[str] = None
+    tbl_properties: Optional[str] = None
+    partitions: Optional[List[ExternalPartition]] = None
+
+    def __bool__(self):
+        return self.location is not None
+
+
+@dataclass
+class Quoting(dbtClassMixin, Mergeable):
+    database: Optional[bool] = None
+    schema: Optional[bool] = None
+    identifier: Optional[bool] = None
+    column: Optional[bool] = None
+
+
+@dataclass
+class UnparsedSourceTableDefinition(HasColumnTests, HasTests):
+    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: 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(dbtClassMixin, Replaceable):
+    name: str
+    description: str = ''
+    meta: Dict[str, Any] = field(default_factory=dict)
+    database: Optional[str] = None
+    schema: Optional[str] = None
+    loader: str = ''
+    quoting: Quoting = field(default_factory=Quoting)
+    freshness: Optional[FreshnessThreshold] = field(
+        default_factory=FreshnessThreshold
+    )
+    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 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
+
+    @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

@@ -0,0 +1,272 @@
+from dbt.contracts.util import Replaceable, Mergeable, list_str
+from dbt.contracts.connection import QueryComment, UserConfigContract
+from dbt.helper_types import NoValue
+from dbt.logger import GLOBAL_LOGGER as logger  # noqa
+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
+from mashumaro.types import SerializableType
+
+PIN_PACKAGE_URL = 'https://docs.getdbt.com/docs/package-management#section-specifying-package-versions'  # noqa
+DEFAULT_SEND_ANONYMOUS_USAGE_STATS = True
+
+
+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'.
+register_pattern(
+    SemverString,
+    r'^(?:0|[1-9]\d*)\.(?:0|[1-9]\d*)(\.(?:0|[1-9]\d*))?$',
+)
+
+
+@dataclass
+class Quoting(dbtClassMixin, Mergeable):
+    schema: Optional[bool] = None
+    database: Optional[bool] = None
+    project: Optional[bool] = None
+    identifier: Optional[bool] = None
+
+
+@dataclass
+class Package(Replaceable, HyphenatedDbtClassMixin):
+    pass
+
+
+@dataclass
+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[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[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(dbtClassMixin, Replaceable):
+    packages: List[PackageSpec]
+
+
+@dataclass
+class ProjectPackageMetadata:
+    name: str
+    packages: List[PackageSpec]
+
+    @classmethod
+    def from_project(cls, project):
+        return cls(name=project.project_name,
+                   packages=project.packages.packages)
+
+
+@dataclass
+class Downloads(ExtensibleDbtClassMixin, Replaceable):
+    tarball: str
+
+
+@dataclass
+class RegistryPackageMetadata(
+    ExtensibleDbtClassMixin,
+    ProjectPackageMetadata,
+):
+    downloads: Downloads
+
+
+# A list of all the reserved words that packages may not have as names.
+BANNED_PROJECT_NAMES = {
+    '_sql_results',
+    'adapter',
+    'api',
+    'column',
+    'config',
+    'context',
+    'database',
+    'env',
+    'env_var',
+    'exceptions',
+    'execute',
+    'flags',
+    'fromjson',
+    'fromyaml',
+    'graph',
+    'invocation_id',
+    'load_agate_table',
+    'load_result',
+    'log',
+    'model',
+    'modules',
+    'post_hooks',
+    'pre_hooks',
+    'ref',
+    'render',
+    'return',
+    'run_started_at',
+    'schema',
+    'source',
+    'sql',
+    'sql_now',
+    'store_result',
+    'store_raw_result',
+    'target',
+    'this',
+    'tojson',
+    'toyaml',
+    'try_or_compiler_error',
+    'var',
+    'write',
+}
+
+
+@dataclass
+class Project(HyphenatedDbtClassMixin, Replaceable):
+    name: Name
+    version: Union[SemverString, float]
+    config_version: int
+    project_root: Optional[str] = None
+    source_paths: Optional[List[str]] = None
+    model_paths: Optional[List[str]] = None
+    macro_paths: Optional[List[str]] = None
+    data_paths: Optional[List[str]] = None
+    seed_paths: Optional[List[str]] = None
+    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
+    profile: Optional[str] = None
+    log_path: Optional[str] = None
+    packages_install_path: Optional[str] = None
+    quoting: Optional[Quoting] = None
+    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[QueryComment, NoValue, str]] = NoValue()
+
+    @classmethod
+    def validate(cls, data):
+        super().validate(data)
+        if data['name'] in BANNED_PROJECT_NAMES:
+            raise ValidationError(
+                f"Invalid project name: {data['name']} is a reserved word"
+            )
+        # 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(ExtensibleDbtClassMixin, Replaceable, UserConfigContract):
+    send_anonymous_usage_stats: bool = DEFAULT_SEND_ANONYMOUS_USAGE_STATS
+    use_colors: Optional[bool] = None
+    partial_parse: Optional[bool] = None
+    printer_width: Optional[int] = None
+    write_json: Optional[bool] = None
+    warn_error: Optional[bool] = None
+    log_format: Optional[bool] = None
+    debug: Optional[bool] = None
+    version_check: Optional[bool] = None
+    fail_fast: Optional[bool] = None
+    use_experimental_parser: Optional[bool] = None
+    static_parser: Optional[bool] = None
+
+
+@dataclass
+class ProfileConfig(HyphenatedDbtClassMixin, Replaceable):
+    profile_name: str = field(metadata={'preserve_underscore': True})
+    target_name: str = field(metadata={'preserve_underscore': True})
+    user_config: UserConfig = field(metadata={'preserve_underscore': True})
+    threads: int
+    # TODO: make this a dynamic union of some kind?
+    credentials: Optional[Dict[str, Any]]
+
+
+@dataclass
+class ConfiguredQuoting(Quoting, Replaceable):
+    identifier: bool = True
+    schema: bool = True
+    database: Optional[bool] = None
+    project: Optional[bool] = None
+
+
+@dataclass
+class Configuration(Project, ProfileConfig):
+    cli_vars: Dict[str, Any] = field(
+        default_factory=dict,
+        metadata={'preserve_underscore': True},
+    )
+    quoting: Optional[ConfiguredQuoting] = None
+
+
+@dataclass
+class ProjectList(dbtClassMixin):
+    projects: Dict[str, Project]

core/dbt/contracts/relation.py

@@ -0,0 +1,126 @@
+from collections.abc import Mapping
+from dataclasses import dataclass
+from typing import (
+    Optional, Dict,
+)
+from typing_extensions import Protocol
+
+from dbt.dataclass_schema import dbtClassMixin, StrEnum
+
+from dbt.contracts.util import Replaceable
+from dbt.exceptions import raise_dataclass_not_dict, 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):
+        raise_dataclass_not_dict(self)
+
+    def __len__(self):
+        raise_dataclass_not_dict(self)
+
+    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

@@ -0,0 +1,491 @@
+from dbt.contracts.graph.manifest import CompileResultNode
+from dbt.contracts.graph.unparsed import (
+    FreshnessThreshold
+)
+from dbt.contracts.graph.parsed import ParsedSourceDefinition
+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 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, Sequence,
+)
+
+from dbt.clients.system import write_json
+
+
+@dataclass
+class TimingInfo(dbtClassMixin):
+    name: str
+    started_at: Optional[datetime] = None
+    completed_at: Optional[datetime] = None
+
+    def begin(self):
+        self.started_at = datetime.utcnow()
+
+    def end(self):
+        self.completed_at = datetime.utcnow()
+
+
+class collect_timing_info:
+    def __init__(self, name: str):
+        self.timing_info = TimingInfo(name=name)
+
+    def __enter__(self):
+        self.timing_info.begin()
+        return self.timing_info
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        self.timing_info.end()
+        with JsonOnly(), TimingProcessor(self.timing_info):
+            logger.debug('finished collecting timing info')
+
+
+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
+        }
+    )
+
+    @property
+    def skipped(self):
+        return self.status == RunStatus.Skipped
+
+
+@dataclass
+class ExecutionResult(dbtClassMixin):
+    results: Sequence[BaseResult]
+    elapsed_time: float
+
+    def __len__(self):
+        return len(self.results)
+
+    def __iter__(self):
+        return iter(self.results)
+
+    def __getitem__(self, idx):
+        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', 3)
+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(NodeResult):
+    node: ParsedSourceDefinition
+    status: FreshnessStatus
+    max_loaded_at: datetime
+    snapshotted_at: datetime
+    age: float
+
+    @property
+    def skipped(self):
+        return False
+
+
+class FreshnessErrorEnum(StrEnum):
+    runtime_error = 'runtime error'
+
+
+@dataclass
+class SourceFreshnessRuntimeError(dbtClassMixin):
+    unique_id: str
+    error: Optional[Union[str, int]]
+    status: FreshnessErrorEnum
+
+
+@dataclass
+class SourceFreshnessOutput(dbtClassMixin):
+    unique_id: str
+    max_loaded_at: datetime
+    snapshotted_at: datetime
+    max_loaded_at_time_ago_in_s: float
+    status: FreshnessStatus
+    criteria: FreshnessThreshold
+    adapter_response: Dict[str, Any]
+    timing: List[TimingInfo]
+    thread_id: str
+    execution_time: float
+
+
+@dataclass
+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,
+        timing=result.timing,
+        thread_id=result.thread_id,
+        execution_time=result.execution_time,
+    )
+
+
+@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', 2)
+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', Optional[str]), ('schema', str), ('name', str)]
+)
+
+
+@dataclass
+class StatsItem(dbtClassMixin):
+    id: str
+    label: str
+    value: Primitive
+    include: bool
+    description: Optional[str] = None
+
+
+StatsDict = Dict[str, StatsItem]
+
+
+@dataclass
+class ColumnMetadata(dbtClassMixin):
+    type: str
+    index: int
+    name: str
+    comment: Optional[str] = None
+
+
+ColumnMap = Dict[str, ColumnMetadata]
+
+
+@dataclass
+class TableMetadata(dbtClassMixin):
+    type: str
+    schema: str
+    name: str
+    database: Optional[str] = None
+    comment: Optional[str] = None
+    owner: Optional[str] = None
+
+
+@dataclass
+class CatalogTable(dbtClassMixin, Replaceable):
+    metadata: TableMetadata
+    columns: ColumnMap
+    stats: StatsDict
+    # the same table with two unique IDs will just be listed two times
+    unique_id: Optional[str] = None
+
+    def key(self) -> CatalogKey:
+        return CatalogKey(
+            lowercase(self.metadata.database),
+            self.metadata.schema.lower(),
+            self.metadata.name.lower(),
+        )
+
+
+@dataclass
+class CatalogMetadata(BaseArtifactMetadata):
+    dbt_schema_version: str = field(
+        default_factory=lambda: str(CatalogArtifact.dbt_schema_version)
+    )
+
+
+@dataclass
+class CatalogResults(dbtClassMixin):
+    nodes: Dict[str, CatalogTable]
+    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

@@ -0,0 +1,819 @@
+import enum
+import os
+import uuid
+from dataclasses import dataclass, field
+from datetime import datetime, timedelta
+from typing import Optional, Union, List, Any, Dict, Type, Sequence
+
+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 (
+    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
+
+
+TaskTags = Optional[Dict[str, Any]]
+TaskID = uuid.UUID
+
+# Inputs
+
+
+@dataclass
+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] = 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
+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
+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):
+    resource_types: Optional[List[str]] = None
+    select: Union[None, str, List[str]] = None
+    threads: Optional[int] = None
+    exclude: Union[None, str, List[str]] = None
+    selector: Optional[str] = None
+    state: Optional[str] = None
+    defer: Optional[bool] = None
+
+
+@dataclass
+class RPCCliParameters(RPCParameters):
+    cli: str
+
+
+@dataclass
+class RPCDepsParameters(RPCParameters):
+    pass
+
+
+@dataclass
+class KillParameters(RPCParameters):
+    task_id: TaskID
+
+
+@dataclass
+class PollParameters(RPCParameters):
+    request_token: TaskID
+    logs: bool = True
+    logs_start: int = 0
+
+
+@dataclass
+class PSParameters(RPCParameters):
+    active: bool = True
+    completed: bool = False
+
+
+@dataclass
+class StatusParameters(RPCParameters):
+    pass
+
+
+@dataclass
+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
+    # many tasks in the table
+    reapsize: int
+    # a positive timedelta indicating how far back we should go
+    auto_reap_age: timedelta
+
+
+@dataclass
+class GCParameters(RPCParameters):
+    """The gc endpoint takes three arguments, any of which may be present:
+
+    - task_ids: An optional list of task ID UUIDs to try to GC
+    - before: If provided, should be a datetime string. All tasks that finished
+        before that datetime will be GCed
+    - settings: If provided, should be a GCSettings object in JSON form. It
+        will be applied to the task manager before GC starts. By default the
+        existing gc settings remain.
+    """
+    task_ids: Optional[List[TaskID]] = None
+    before: Optional[datetime] = None
+    settings: Optional[GCSettings] = None
+
+
+@dataclass
+class RPCRunOperationParameters(RPCParameters):
+    macro: str
+    args: Dict[str, Any] = field(default_factory=dict)
+
+
+@dataclass
+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(VersionedSchema):
+    logs: List[LogMessage]
+
+
+@dataclass
+@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):
+    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 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):
+    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(dbtClassMixin):
+    column_names: List[str]
+    rows: List[Any]
+
+
+@dataclass
+@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
+@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,
+    RemoteDepsResult,
+    RemoteRunOperationResult,
+]
+
+
+# GC types
+
+class GCResultState(StrEnum):
+    Deleted = 'deleted'  # successful GC
+    Missing = 'missing'  # nothing to GC
+    Running = 'running'  # can't GC
+
+
+@dataclass
+@schema_version('remote-gc-result', 1)
+class GCResult(RemoteResult):
+    logs: List[LogMessage] = field(default_factory=list)
+    deleted: List[TaskID] = field(default_factory=list)
+    missing: List[TaskID] = field(default_factory=list)
+    running: List[TaskID] = field(default_factory=list)
+
+    def add_result(self, task_id: TaskID, state: GCResultState):
+        if state == GCResultState.Missing:
+            self.missing.append(task_id)
+        elif state == GCResultState.Running:
+            self.running.append(task_id)
+        elif state == GCResultState.Deleted:
+            self.deleted.append(task_id)
+        else:
+            raise InternalException(
+                f'Got invalid state in add_result: {state}'
+            )
+
+# Task management types
+
+
+class TaskHandlerState(StrEnum):
+    NotStarted = 'not started'
+    Initializing = 'initializing'
+    Running = 'running'
+    Success = 'success'
+    Error = 'error'
+    Killed = 'killed'
+    Failed = 'failed'
+
+    def __lt__(self, other) -> bool:
+        """A logical ordering for TaskHandlerState:
+
+        NotStarted < Initializing < Running < (Success, Error, Killed, Failed)
+        """
+        if not isinstance(other, TaskHandlerState):
+            raise TypeError('cannot compare to non-TaskHandlerState')
+        order = (self.NotStarted, self.Initializing, self.Running)
+        smaller = set()
+        for value in order:
+            smaller.add(value)
+            if self == value:
+                return other not in smaller
+
+        return False
+
+    def __le__(self, other) -> bool:
+        # so that ((Success <= Error) is True)
+        return ((self < other) or
+                (self == other) or
+                (self.finished and other.finished))
+
+    def __gt__(self, other) -> bool:
+        if not isinstance(other, TaskHandlerState):
+            raise TypeError('cannot compare to non-TaskHandlerState')
+        order = (self.NotStarted, self.Initializing, self.Running)
+        smaller = set()
+        for value in order:
+            smaller.add(value)
+            if self == value:
+                return other in smaller
+        return other in smaller
+
+    def __ge__(self, other) -> bool:
+        # so that ((Success <= Error) is True)
+        return ((self > other) or
+                (self == other) or
+                (self.finished and other.finished))
+
+    @property
+    def finished(self) -> bool:
+        return self in (self.Error, self.Success, self.Killed, self.Failed)
+
+
+@dataclass
+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_source: str
+    method: str
+    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]
+
+
+class KillResultStatus(StrEnum):
+    Missing = 'missing'
+    NotStarted = 'not_started'
+    Killed = 'killed'
+    Finished = 'finished'
+
+
+@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):
+    Empty = 0
+    BlocksManifestTasks = 1
+    RequiresConfigReloadBefore = 3
+    RequiresManifestReloadAfter = 5
+    Builtin = 8
+
+
+# Polling types
+
+
+@dataclass
+class PollResult(RemoteResult, TaskTiming):
+    state: TaskHandlerState
+    tags: TaskTags
+    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
+@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: RemoteDepsResult,
+        tags: TaskTags,
+        timing: TaskTiming,
+        logs: List[LogMessage],
+    ) -> 'PollRemoteEmptyCompleteResult':
+        return cls(
+            logs=logs,
+            tags=tags,
+            state=timing.state,
+            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),
+    )
+
+
+@dataclass
+@schema_version('poll-remote-execution-result', 1)
+class PollExecuteCompleteResult(
+    RemoteExecutionResult,
+    PollResult,
+):
+    state: TaskHandlerState = field(
+        metadata=restrict_to(TaskHandlerState.Success,
+                             TaskHandlerState.Failed),
+    )
+
+    @classmethod
+    def from_result(
+        cls: Type['PollExecuteCompleteResult'],
+        base: RemoteExecutionResult,
+        tags: TaskTags,
+        timing: TaskTiming,
+        logs: List[LogMessage],
+    ) -> 'PollExecuteCompleteResult':
+        return cls(
+            results=base.results,
+            elapsed_time=base.elapsed_time,
+            logs=logs,
+            tags=tags,
+            state=timing.state,
+            start=timing.start,
+            end=timing.end,
+            elapsed=timing.elapsed,
+            generated_at=base.generated_at,
+        )
+
+
+@dataclass
+@schema_version('poll-remote-compile-result', 1)
+class PollCompileCompleteResult(
+    RemoteCompileResult,
+    PollResult,
+):
+    state: TaskHandlerState = field(
+        metadata=restrict_to(TaskHandlerState.Success,
+                             TaskHandlerState.Failed),
+    )
+
+    @classmethod
+    def from_result(
+        cls: Type['PollCompileCompleteResult'],
+        base: RemoteCompileResult,
+        tags: TaskTags,
+        timing: TaskTiming,
+        logs: List[LogMessage],
+    ) -> 'PollCompileCompleteResult':
+        return cls(
+            raw_sql=base.raw_sql,
+            compiled_sql=base.compiled_sql,
+            node=base.node,
+            timing=base.timing,
+            logs=logs,
+            tags=tags,
+            state=timing.state,
+            start=timing.start,
+            end=timing.end,
+            elapsed=timing.elapsed,
+            generated_at=base.generated_at
+        )
+
+
+@dataclass
+@schema_version('poll-remote-run-result', 1)
+class PollRunCompleteResult(
+    RemoteRunResult,
+    PollResult,
+):
+    state: TaskHandlerState = field(
+        metadata=restrict_to(TaskHandlerState.Success,
+                             TaskHandlerState.Failed),
+    )
+
+    @classmethod
+    def from_result(
+        cls: Type['PollRunCompleteResult'],
+        base: RemoteRunResult,
+        tags: TaskTags,
+        timing: TaskTiming,
+        logs: List[LogMessage],
+    ) -> 'PollRunCompleteResult':
+        return cls(
+            raw_sql=base.raw_sql,
+            compiled_sql=base.compiled_sql,
+            node=base.node,
+            timing=base.timing,
+            logs=logs,
+            table=base.table,
+            tags=tags,
+            state=timing.state,
+            start=timing.start,
+            end=timing.end,
+            elapsed=timing.elapsed,
+            generated_at=base.generated_at
+        )
+
+
+@dataclass
+@schema_version('poll-remote-run-operation-result', 1)
+class PollRunOperationCompleteResult(
+    RemoteRunOperationResult,
+    PollResult,
+):
+    state: TaskHandlerState = field(
+        metadata=restrict_to(TaskHandlerState.Success,
+                             TaskHandlerState.Failed),
+    )
+
+    @classmethod
+    def from_result(
+        cls: Type['PollRunOperationCompleteResult'],
+        base: RemoteRunOperationResult,
+        tags: TaskTags,
+        timing: TaskTiming,
+        logs: List[LogMessage],
+    ) -> 'PollRunOperationCompleteResult':
+        return cls(
+            success=base.success,
+            results=base.results,
+            generated_at=base.generated_at,
+            elapsed_time=base.elapsed_time,
+            logs=logs,
+            tags=tags,
+            state=timing.state,
+            start=timing.start,
+            end=timing.end,
+            elapsed=timing.elapsed,
+        )
+
+
+@dataclass
+@schema_version('poll-remote-catalog-result', 1)
+class PollCatalogCompleteResult(RemoteCatalogResults, PollResult):
+    state: TaskHandlerState = field(
+        metadata=restrict_to(TaskHandlerState.Success,
+                             TaskHandlerState.Failed),
+    )
+
+    @classmethod
+    def from_result(
+        cls: Type['PollCatalogCompleteResult'],
+        base: RemoteCatalogResults,
+        tags: TaskTags,
+        timing: TaskTiming,
+        logs: List[LogMessage],
+    ) -> 'PollCatalogCompleteResult':
+        return cls(
+            nodes=base.nodes,
+            sources=base.sources,
+            generated_at=base.generated_at,
+            errors=base.errors,
+            _compile_results=base._compile_results,
+            logs=logs,
+            tags=tags,
+            state=timing.state,
+            start=timing.start,
+            end=timing.end,
+            elapsed=timing.elapsed,
+        )
+
+
+@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'
+    Ready = 'ready'
+    Error = 'error'
+
+
+@dataclass
+@schema_version('remote-status-result', 1)
+class LastParse(RemoteResult):
+    state: ManifestStatus = ManifestStatus.Init
+    logs: List[LogMessage] = field(default_factory=list)
+    error: Optional[Dict[str, Any]] = None
+    timestamp: datetime = field(default_factory=datetime.utcnow)
+    pid: int = field(default_factory=os.getpid)

core/dbt/contracts/selection.py

@@ -0,0 +1,23 @@
+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 = ''
+    default: bool = False
+
+
+@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 +1,28 @@
+from pathlib import Path
+from .graph.manifest import WritableManifest
+from .results import RunResultsArtifact
+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
+        self.results: Optional[RunResultsArtifact] = 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
+
+        results_path = self.path / 'run_results.json'
+        if results_path.exists() and results_path.is_file():
+            try:
+                self.results = RunResultsArtifact.read(str(results_path))
+            except IncompatibleSchemaException as exc:
+                exc.add_filename(str(results_path))
+                raise

core/dbt/contracts/util.py

@@ -0,0 +1,218 @@
+import dataclasses
+import os
+from datetime import datetime
+from typing import (
+    List, Tuple, ClassVar, Type, TypeVar, Dict, Any, Optional
+)
+
+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:
+
+    from dataclasses import dataclass, field
+    from typing import Optional, List
+
+    @dataclass
+    class Foo:
+        x: Optional[List[str]] = field(default_factory=list)
+
+
+    Because `list` could be any kind of list, I guess
+    """
+    return []
+
+
+class Replaceable:
+    def replace(self, **kwargs):
+        return dataclasses.replace(self, **kwargs)
+
+
+class Mergeable(Replaceable):
+    def merged(self, *args):
+        """Perform a shallow merge, where the last non-None write wins. This is
+        intended to merge dataclasses that are a collection of optional values.
+        """
+        replacements = {}
+        cls = type(self)
+        for arg in args:
+            for field in dataclasses.fields(cls):
+                value = getattr(arg, field.name)
+                if value is not None:
+                    replacements[field.name] = value
+
+        return self.replace(**replacements)
+
+
+class Writable:
+    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 __post_serialize__(self, dct):
+        dct = super().__post_serialize__(dct)
+        if dct['generated_at'] and dct['generated_at'].endswith('+00:00'):
+            dct['generated_at'] = dct['generated_at'].replace('+00:00', '') + "Z"
+        return dct
+
+
+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

@@ -0,0 +1,120 @@
+from typing import Optional, Set, List, Dict, ClassVar
+
+import dbt.exceptions
+from dbt import ui
+
+import dbt.tracking
+
+
+class DBTDeprecation:
+    _name: ClassVar[Optional[str]] = None
+    _description: ClassVar[Optional[str]] = None
+
+    @property
+    def name(self) -> str:
+        if self._name is not None:
+            return self._name
+        raise NotImplementedError(
+            '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:
+            return self._description
+        raise NotImplementedError(
+            'description not implemented for {}'.format(self)
+        )
+
+    def show(self, *args, **kwargs) -> None:
+        if self.name not in active_deprecations:
+            desc = self.description.format(**kwargs)
+            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 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.
+    '''
+
+
+class PackageInstallPathDeprecation(DBTDeprecation):
+    _name = 'install-packages-path'
+    _description = '''\
+    The default package install path has changed from `dbt_modules` to `dbt_packages`.
+    Please update `clean-targets` in `dbt_project.yml` and check `.gitignore` as well.
+    Or, set `packages-install-path: dbt_modules` if you'd like to keep the current value.
+    '''
+
+
+class ConfigPathDeprecation(DBTDeprecation):
+    _name = 'project_config_path'
+    _description = '''\
+    The `{deprecated_path}` config has been deprecated in favor of `{exp_path}`.
+    Please update your `dbt_project.yml` configuration to reflect this change.
+    '''
+
+
+_adapter_renamed_description = """\
+The adapter function `adapter.{old_name}` is deprecated and will be removed in
+a future release of dbt. Please use `adapter.{new_name}` instead.
+
+Documentation for {new_name} can be found here:
+
+    https://docs.getdbt.com/docs/adapter
+"""
+
+
+def renamed_method(old_name: str, new_name: str):
+
+    class AdapterDeprecationWarning(DBTDeprecation):
+        _name = 'adapter:{}'.format(old_name)
+        _description = _adapter_renamed_description.format(old_name=old_name,
+                                                           new_name=new_name)
+
+    dep = AdapterDeprecationWarning()
+    deprecations_list.append(dep)
+    deprecations[dep.name] = dep
+
+
+def warn(name, *args, **kwargs):
+    if name not in deprecations:
+        # this should (hopefully) never happen
+        raise RuntimeError(
+            "Error showing deprecation warning: {}".format(name)
+        )
+
+    deprecations[name].show(*args, **kwargs)
+
+
+# these are globally available
+# since modules are only imported once, active_deprecations is a singleton
+
+active_deprecations: Set[str] = set()
+
+deprecations_list: List[DBTDeprecation] = [
+    ConfigPathDeprecation(),
+    PackageInstallPathDeprecation(),
+    PackageRedirectDeprecation()
+]
+
+deprecations: Dict[str, DBTDeprecation] = {
+    d.name: d for d in deprecations_list
+}
+
+
+def reset_deprecations():
+    active_deprecations.clear()

core/dbt/deps/base.py

@@ -0,0 +1,115 @@
+import abc
+import os
+import tempfile
+from contextlib import contextmanager
+from typing import List, Optional, Generic, TypeVar
+
+from dbt.clients import system
+from dbt.contracts.project import ProjectPackageMetadata
+from dbt.logger import GLOBAL_LOGGER as logger
+
+DOWNLOADS_PATH = None
+
+
+def get_downloads_path():
+    return DOWNLOADS_PATH
+
+
+@contextmanager
+def downloads_directory():
+    global DOWNLOADS_PATH
+    remove_downloads = False
+    # the user might have set an environment variable. Set it to that, and do
+    # not remove it when finished.
+    if DOWNLOADS_PATH is None:
+        DOWNLOADS_PATH = os.getenv('DBT_DOWNLOADS_DIR')
+        remove_downloads = False
+    # if we are making a per-run temp directory, remove it at the end of
+    # successful runs
+    if DOWNLOADS_PATH is None:
+        DOWNLOADS_PATH = tempfile.mkdtemp(prefix='dbt-downloads-')
+        remove_downloads = True
+
+    system.make_directory(DOWNLOADS_PATH)
+    logger.debug("Set downloads directory='{}'".format(DOWNLOADS_PATH))
+
+    yield DOWNLOADS_PATH
+
+    if remove_downloads:
+        system.rmtree(DOWNLOADS_PATH)
+        DOWNLOADS_PATH = None
+
+
+class BasePackage(metaclass=abc.ABCMeta):
+    @abc.abstractproperty
+    def name(self) -> str:
+        raise NotImplementedError
+
+    def all_names(self) -> List[str]:
+        return [self.name]
+
+    @abc.abstractmethod
+    def source_type(self) -> str:
+        raise NotImplementedError
+
+
+class PinnedPackage(BasePackage):
+    def __init__(self) -> None:
+        self._cached_metadata: Optional[ProjectPackageMetadata] = None
+
+    def __str__(self) -> str:
+        version = self.get_version()
+        if not version:
+            return self.name
+
+        return '{}@{}'.format(self.name, version)
+
+    @abc.abstractmethod
+    def get_version(self) -> Optional[str]:
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def _fetch_metadata(self, project, renderer):
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def install(self, project):
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def nice_version_name(self):
+        raise NotImplementedError
+
+    def fetch_metadata(self, project, renderer):
+        if not self._cached_metadata:
+            self._cached_metadata = self._fetch_metadata(project, renderer)
+        return self._cached_metadata
+
+    def get_project_name(self, project, renderer):
+        metadata = self.fetch_metadata(project, renderer)
+        return metadata.name
+
+    def get_installation_path(self, project, renderer):
+        dest_dirname = self.get_project_name(project, renderer)
+        return os.path.join(project.packages_install_path, dest_dirname)
+
+    def get_subdirectory(self):
+        return None
+
+
+SomePinned = TypeVar('SomePinned', bound=PinnedPackage)
+SomeUnpinned = TypeVar('SomeUnpinned', bound='UnpinnedPackage')
+
+
+class UnpinnedPackage(Generic[SomePinned], BasePackage):
+    @abc.abstractclassmethod
+    def from_contract(cls, contract):
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def incorporate(self: SomeUnpinned, other: SomeUnpinned) -> SomeUnpinned:
+        raise NotImplementedError
+
+    @abc.abstractmethod
+    def resolved(self) -> SomePinned:
+        raise NotImplementedError

core/dbt/deps/git.py

@@ -0,0 +1,171 @@
+import os
+import hashlib
+from typing import List, Optional
+
+from dbt.clients import git, system
+from dbt.config import Project
+from dbt.contracts.project import (
+    ProjectPackageMetadata,
+    GitPackage,
+)
+from dbt.deps.base import PinnedPackage, UnpinnedPackage, get_downloads_path
+from dbt.exceptions import (
+    ExecutableError, warn_or_error, raise_dependency_error
+)
+from dbt.logger import GLOBAL_LOGGER as logger
+from dbt import ui
+
+PIN_PACKAGE_URL = 'https://docs.getdbt.com/docs/package-management#section-specifying-package-versions' # noqa
+
+
+def md5sum(s: str):
+    return hashlib.md5(s.encode('latin-1')).hexdigest()
+
+
+class GitPackageMixin:
+    def __init__(self, git: str) -> None:
+        super().__init__()
+        self.git = git
+
+    @property
+    def name(self):
+        return self.git
+
+    def source_type(self) -> str:
+        return 'git'
+
+
+class GitPinnedPackage(GitPackageMixin, PinnedPackage):
+    def __init__(
+        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):
+        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
+        directory. This function can be called repeatedly. If the project has
+        already been checked out at this version, it will be a no-op. Returns
+        the path to the checked out directory."""
+        try:
+            dir_ = git.clone_and_checkout(
+                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':
+                logger.error(
+                    'Make sure git is installed on your machine. More '
+                    'information: '
+                    'https://docs.getdbt.com/docs/package-management'
+                )
+            raise
+        return os.path.join(get_downloads_path(), dir_)
+
+    def _fetch_metadata(self, project, renderer) -> ProjectPackageMetadata:
+        path = self._checkout()
+
+        if self.unpinned_msg() and self.warn_unpinned:
+            warn_or_error(
+                'The git package "{}" \n\tis {}.\n\tThis can introduce '
+                'breaking changes into your project without warning!\n\nSee {}'
+                .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)
+
+    def install(self, project, renderer):
+        dest_path = self.get_installation_path(project, renderer)
+        if os.path.exists(dest_path):
+            if system.path_is_symlink(dest_path):
+                system.remove_file(dest_path)
+            else:
+                system.rmdir(dest_path)
+
+        system.move(self._checkout(), dest_path)
+
+
+class GitUnpinnedPackage(GitPackageMixin, UnpinnedPackage[GitPinnedPackage]):
+    def __init__(
+        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.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, subdirectory=contract.subdirectory)
+
+    def all_names(self) -> List[str]:
+        if self.git.endswith('.git'):
+            other = self.git[:-4]
+        else:
+            other = self.git + '.git'
+        return [self.git, other]
+
+    def incorporate(
+        self, other: 'GitUnpinnedPackage'
+    ) -> 'GitUnpinnedPackage':
+        warn_unpinned = self.warn_unpinned and other.warn_unpinned
+
+        return GitUnpinnedPackage(
+            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 = {'HEAD'}
+        elif len(requested) > 1:
+            raise_dependency_error(
+                'git dependencies should contain exactly one version. '
+                '{} contains: {}'.format(self.git, requested))
+
+        return GitPinnedPackage(
+            git=self.git, revision=requested.pop(),
+            warn_unpinned=self.warn_unpinned, subdirectory=self.subdirectory
+        )

core/dbt/deps/local.py

@@ -0,0 +1,84 @@
+import shutil
+
+from dbt.clients import system
+from dbt.deps.base import PinnedPackage, UnpinnedPackage
+from dbt.contracts.project import (
+    ProjectPackageMetadata,
+    LocalPackage,
+)
+from dbt.logger import GLOBAL_LOGGER as logger
+
+
+class LocalPackageMixin:
+    def __init__(self, local: str) -> None:
+        super().__init__()
+        self.local = local
+
+    @property
+    def name(self):
+        return self.local
+
+    def source_type(self):
+        return 'local'
+
+
+class LocalPinnedPackage(LocalPackageMixin, PinnedPackage):
+    def __init__(self, local: str) -> None:
+        super().__init__(local)
+
+    def get_version(self):
+        return None
+
+    def nice_version_name(self):
+        return '<local @ {}>'.format(self.local)
+
+    def resolve_path(self, project):
+        return system.resolve_path_from_base(
+            self.local,
+            project.project_root,
+        )
+
+    def _fetch_metadata(self, project, renderer):
+        loaded = project.from_project_root(
+            self.resolve_path(project), renderer
+        )
+        return ProjectPackageMetadata.from_project(loaded)
+
+    def install(self, project, renderer):
+        src_path = self.resolve_path(project)
+        dest_path = self.get_installation_path(project, renderer)
+
+        can_create_symlink = system.supports_symlinks()
+
+        if system.path_exists(dest_path):
+            if not system.path_is_symlink(dest_path):
+                system.rmdir(dest_path)
+            else:
+                system.remove_file(dest_path)
+
+        if can_create_symlink:
+            logger.debug('  Creating symlink to local dependency.')
+            system.make_symlink(src_path, dest_path)
+
+        else:
+            logger.debug('  Symlinks are not available on this '
+                         'OS, copying dependency.')
+            shutil.copytree(src_path, dest_path)
+
+
+class LocalUnpinnedPackage(
+    LocalPackageMixin, UnpinnedPackage[LocalPinnedPackage]
+):
+    @classmethod
+    def from_contract(
+        cls, contract: LocalPackage
+    ) -> 'LocalUnpinnedPackage':
+        return cls(local=contract.local)
+
+    def incorporate(
+        self, other: 'LocalUnpinnedPackage'
+    ) -> 'LocalUnpinnedPackage':
+        return LocalUnpinnedPackage(local=self.local)
+
+    def resolved(self) -> LocalPinnedPackage:
+        return LocalPinnedPackage(local=self.local)

core/dbt/deps/registry.py

@@ -0,0 +1,144 @@
+import os
+from typing import List
+
+from dbt import semver
+from dbt.clients import registry, system
+from dbt.contracts.project import (
+    RegistryPackageMetadata,
+    RegistryPackage,
+)
+from dbt.deps.base import PinnedPackage, UnpinnedPackage, get_downloads_path
+from dbt.exceptions import (
+    package_version_not_found,
+    VersionsNotCompatibleException,
+    DependencyException,
+    package_not_found,
+)
+
+
+class RegistryPackageMixin:
+    def __init__(self, package: str) -> None:
+        super().__init__()
+        self.package = package
+
+    @property
+    def name(self):
+        return self.package
+
+    def source_type(self) -> str:
+        return 'hub'
+
+
+class RegistryPinnedPackage(RegistryPackageMixin, PinnedPackage):
+    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):
+        return self.package
+
+    def source_type(self):
+        return 'hub'
+
+    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)
+
+    def _fetch_metadata(self, project, renderer) -> RegistryPackageMetadata:
+        dct = registry.package_version(self.package, self.version)
+        return RegistryPackageMetadata.from_dict(dct)
+
+    def install(self, project, renderer):
+        metadata = self.fetch_metadata(project, renderer)
+
+        tar_name = '{}.{}.tar.gz'.format(self.package, self.version)
+        tar_path = os.path.realpath(
+            os.path.join(get_downloads_path(), tar_name)
+        )
+        system.make_directory(os.path.dirname(tar_path))
+
+        download_url = metadata.downloads.tarball
+        system.download_with_retries(download_url, tar_path)
+        deps_path = project.packages_install_path
+        package_name = self.get_project_name(project, renderer)
+        system.untar_package(tar_path, deps_path, package_name)
+
+
+class RegistryUnpinnedPackage(
+    RegistryPackageMixin, UnpinnedPackage[RegistryPinnedPackage]
+):
+    def __init__(
+        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()
+        if self.package not in index:
+            package_not_found(self.package)
+
+    @classmethod
+    def from_contract(
+        cls, contract: RegistryPackage
+    ) -> 'RegistryUnpinnedPackage':
+        raw_version = contract.get_versions()
+
+        versions = [
+            semver.VersionSpecifier.from_version_string(v)
+            for v in raw_version
+        ]
+        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,
+        )
+
+    def resolved(self) -> RegistryPinnedPackage:
+        self._check_in_index()
+        try:
+            range_ = semver.reduce_versions(*self.versions)
+        except VersionsNotCompatibleException as e:
+            new_msg = ('Version error for package {}: {}'
+                       .format(self.name, e))
+            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_, installable)
+        if not 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

@@ -0,0 +1,143 @@
+from dataclasses import dataclass, field
+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, 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
+from dbt.deps.registry import RegistryUnpinnedPackage
+
+from dbt.contracts.project import (
+    LocalPackage,
+    GitPackage,
+    RegistryPackage,
+)
+
+PackageContract = Union[LocalPackage, GitPackage, RegistryPackage]
+
+
+@dataclass
+class PackageListing:
+    packages: Dict[str, UnpinnedPackage] = field(default_factory=dict)
+
+    def __len__(self):
+        return len(self.packages)
+
+    def __bool__(self):
+        return bool(self.packages)
+
+    def _pick_key(self, key: BasePackage) -> str:
+        for name in key.all_names():
+            if name in self.packages:
+                return name
+        return key.name
+
+    def __contains__(self, key: BasePackage):
+        for name in key.all_names():
+            if name in self.packages:
+                return True
+
+    def __getitem__(self, key: BasePackage):
+        key_str: str = self._pick_key(key)
+        return self.packages[key_str]
+
+    def __setitem__(self, key: BasePackage, value):
+        key_str: str = self._pick_key(key)
+        self.packages[key_str] = value
+
+    def _mismatched_types(
+        self, old: UnpinnedPackage, new: UnpinnedPackage
+    ) -> NoReturn:
+        raise_dependency_error(
+            f'Cannot incorporate {new} ({new.__class__.__name__}) in {old} '
+            f'({old.__class__.__name__}): mismatched types'
+        )
+
+    def incorporate(self, package: UnpinnedPackage):
+        key: str = self._pick_key(package)
+        if key in self.packages:
+            existing: UnpinnedPackage = self.packages[key]
+            if not isinstance(existing, type(package)):
+                self._mismatched_types(existing, package)
+            self.packages[key] = existing.incorporate(package)
+        else:
+            self.packages[key] = package
+
+    def update_from(self, src: List[PackageContract]) -> None:
+        pkg: UnpinnedPackage
+        for contract in src:
+            if isinstance(contract, LocalPackage):
+                pkg = LocalUnpinnedPackage.from_contract(contract)
+            elif isinstance(contract, GitPackage):
+                pkg = GitUnpinnedPackage.from_contract(contract)
+            elif isinstance(contract, RegistryPackage):
+                pkg = RegistryUnpinnedPackage.from_contract(contract)
+            else:
+                raise InternalException(
+                    'Invalid package type {}'.format(type(contract))
+                )
+            self.incorporate(pkg)
+
+    @classmethod
+    def from_contracts(
+        cls: Type['PackageListing'], src: List[PackageContract]
+    ) -> 'PackageListing':
+        self = cls({})
+        self.update_from(src)
+        return self
+
+    def resolved(self) -> List[PinnedPackage]:
+        return [p.resolved() for p in self.packages.values()]
+
+    def __iter__(self) -> Iterator[UnpinnedPackage]:
+        return iter(self.packages.values())
+
+
+def _check_for_duplicate_project_names(
+    final_deps: List[PinnedPackage],
+    config: Project,
+    renderer: DbtProjectYamlRenderer,
+):
+    seen: Set[str] = set()
+    for package in final_deps:
+        project_name = package.get_project_name(config, renderer)
+        if project_name in seen:
+            raise_dependency_error(
+                f'Found duplicate project "{project_name}". This occurs when '
+                'a dependency has the same project name as some other '
+                'dependency.'
+            )
+        elif project_name == config.project_name:
+            raise_dependency_error(
+                'Found a dependency with the same name as the root project '
+                f'"{project_name}". Package names must be unique in a project.'
+                ' Please rename one of these packages.'
+            )
+        seen.add(project_name)
+
+
+def resolve_packages(
+    packages: List[PackageContract], config: RuntimeConfig
+) -> List[PinnedPackage]:
+    pending = PackageListing.from_contracts(packages)
+    final = PackageListing()
+
+    ctx = generate_target_context(config, config.cli_vars)
+    renderer = DbtProjectYamlRenderer(ctx)
+
+    while pending:
+        next_pending = PackageListing()
+        # resolve the dependency in question
+        for package in pending:
+            final.incorporate(package)
+            target = final[package].resolved().fetch_metadata(config, renderer)
+            next_pending.update_from(target.packages)
+        pending = next_pending
+
+    resolved = final.resolved()
+    _check_for_duplicate_project_names(resolved, config, renderer)
+    return resolved

core/dbt/events/README.md

@@ -0,0 +1,9 @@
+# Events Module
+
+The Events module is the implmentation for structured logging. These events represent both a programatic interface to dbt processes as well as human-readable messaging in one centralized place. The centralization allows for leveraging mypy to enforce interface invariants across all dbt events, and the distinct type layer allows for decoupling events and libraries such as loggers.
+
+# Using the Events Module
+The event module provides types that represent what is happening in dbt in `events.types`. These types are intended to represent an exhaustive list of all things happening within dbt that will need to be logged, streamed, or printed. To fire an event, `events.functions::fire_event` is the entry point to the module from everywhere in dbt.
+
+# Adding a New Event
+In `events.types` add a new class that represents the new event. This may be a simple class with no values, or it may be a dataclass with some values to construct downstream messaging. Only include the data necessary to construct this message within this class. You must extend all destinations (e.g. - if your log message belongs on the cli, extend `CliEventABC`) as well as the loglevel this event belongs to.