Compare commits

...

11 Commits

Author SHA1 Message Date
MichelleArk
54b9aa83db Merge branch 'main' into mjsqu-meta-require-return 2025-12-17 12:40:58 -05:00
Michelle Ark
33e89a9956 Support aliased configs jsonschemas (#12291) 2025-12-17 12:31:52 -05:00
MichelleArk
7c43365e72 Add missing return statement to RuntimeConfigObject.meta_require method 2025-12-17 12:16:26 -05:00
MichelleArk
537e3f0fba rename 2025-12-17 12:00:03 -05:00
MichelleArk
b4ed7699ac ensure coverage when validating configs + unit test 2025-12-17 11:59:24 -05:00
MichelleArk
e2ac1922f0 changelog entry 2025-12-17 10:59:33 -05:00
MichelleArk
46b5e90e0b support extending jsonschemas with aliases 2025-12-17 10:58:27 -05:00
Emily Rockman
046e08ea20 add -v (#12269) 2025-12-17 09:16:38 -05:00
Gerda Shank
2308179389 Update schema file order test (#12061) 2025-12-16 14:30:51 -05:00
Michelle Ark
63ae772dc5 Propagate exceptions for NodeFinished callbacks in dbtRunner (#12286) 2025-12-16 14:28:16 -05:00
Michelle Ark
3f297cb4e3 fix test_config_with_meta_key (#12284) 2025-12-15 16:13:14 -05:00
15 changed files with 170 additions and 44 deletions

View File

@@ -0,0 +1,6 @@
kind: Fixes
body: ':bug: :snowman:Propagate exceptions for NodeFinished callbacks in dbtRunner'
time: 2025-12-16T12:07:27.576087-05:00
custom:
Author: michelleark
Issue: "11612"

View File

@@ -0,0 +1,6 @@
kind: Fixes
body: Adds omitted return statement to RuntimeConfigObject.meta_require method
time: 2025-12-17T00:28:13.015416197Z
custom:
Author: mjsqu
Issue: "12288"

View File

@@ -0,0 +1,6 @@
kind: Fixes
body: Do not raise deprecation warning when encountering dataset or project configs for bigquery
time: 2025-12-17T10:59:18.372968-05:00
custom:
Author: michelleark
Issue: "12285"

View File

@@ -0,0 +1,6 @@
kind: Under the Hood
body: Update schema file order test
time: 2025-09-29T15:11:59.611595-04:00
custom:
Author: gshank
Issue: "11869"

View File

@@ -0,0 +1,6 @@
kind: Under the Hood
body: Bump lower bound for dbt-common to 1.37.2
time: 2025-12-15T15:50:46.857793-05:00
custom:
Author: michelleark
Issue: "12284"

View File

@@ -62,7 +62,7 @@ jobs:
python -m pip --version
python -m pip install hatch
cd core
hatch run setup
hatch -v run setup
- name: Verify dbt installation
run: |
@@ -106,7 +106,7 @@ jobs:
with:
timeout_minutes: 10
max_attempts: 3
command: cd core && hatch run ci:unit-tests
command: cd core && hatch -v run ci:unit-tests
- name: Get current date
if: always()
@@ -230,7 +230,7 @@ jobs:
timeout_minutes: 30
max_attempts: 3
shell: bash
command: cd core && hatch run ci:integration-tests -- --ddtrace --splits ${{ env.PYTHON_INTEGRATION_TEST_WORKERS }} --group ${{ matrix.split-group }}
command: cd core && hatch -v run ci:integration-tests -- --ddtrace --splits ${{ env.PYTHON_INTEGRATION_TEST_WORKERS }} --group ${{ matrix.split-group }}
- name: Get current date
if: always()
@@ -311,7 +311,7 @@ jobs:
timeout_minutes: 30
max_attempts: 3
shell: bash
command: cd core && hatch run ci:integration-tests -- --ddtrace --splits ${{ env.PYTHON_INTEGRATION_TEST_WORKERS }} --group ${{ matrix.split-group }}
command: cd core && hatch -v run ci:integration-tests -- --ddtrace --splits ${{ env.PYTHON_INTEGRATION_TEST_WORKERS }} --group ${{ matrix.split-group }}
- name: Get current date
if: always()
@@ -326,7 +326,7 @@ jobs:
name: logs_${{ matrix.python-version }}_${{ matrix.os }}_${{ matrix.split-group }}_${{ steps.date.outputs.date }}
path: ./logs
- name: Upload Integration Test Coverage to Codecov
- name: Upload Integration Test Coverage
if: ${{ matrix.python-version == '3.11' }}
uses: codecov/codecov-action@5a1091511ad55cbe89839c7260b706298ca349f7 # codecov/codecov-action@v5
with:
@@ -380,4 +380,4 @@ jobs:
- name: Check and verify distributions
run: |
cd core
hatch run build:check-all
hatch -v run build:check-all

View File

@@ -123,7 +123,7 @@ jobs:
with:
timeout_minutes: 30
max_attempts: 3
command: cd core && hatch run ci:integration-tests -- -nauto
command: cd core && hatch -v run ci:integration-tests -- -nauto
env:
PYTEST_ADDOPTS: ${{ format('--splits {0} --group {1}', env.PYTHON_INTEGRATION_TEST_WORKERS, matrix.split-group) }}

View File

@@ -608,6 +608,8 @@ class RuntimeConfigObject(Config):
if validator is not None:
self._validate(validator, to_return)
return to_return
def get(self, name, default=None, validator=None):
to_return = self._lookup(name, default)

View File

@@ -37,6 +37,10 @@ _HIERARCHICAL_CONFIG_KEYS = {
"unit_tests",
}
_ADAPTER_TO_CONFIG_ALIASES = {
"bigquery": ["dataset", "project"],
}
def load_json_from_package(jsonschema_type: str, filename: str) -> Dict[str, Any]:
"""Loads a JSON file from within a package."""
@@ -106,6 +110,16 @@ def _validate_with_schema(
return validator.iter_errors(json)
def _get_allowed_config_key_aliases() -> List[str]:
config_aliases = []
invocation_context = get_invocation_context()
for adapter in invocation_context.adapter_types:
if adapter in _ADAPTER_TO_CONFIG_ALIASES:
config_aliases.extend(_ADAPTER_TO_CONFIG_ALIASES[adapter])
return config_aliases
def _get_allowed_config_fields_from_error_path(
yml_schema: Dict[str, Any], error_path: List[Union[str, int]]
) -> Optional[List[str]]:
@@ -135,6 +149,7 @@ def _get_allowed_config_fields_from_error_path(
][0]["$ref"].split("/")[-1]
allowed_config_fields = list(set(yml_schema["definitions"][config_field_name]["properties"]))
allowed_config_fields.extend(_get_allowed_config_key_aliases())
return allowed_config_fields
@@ -169,7 +184,6 @@ def jsonschema_validate(schema: Dict[str, Any], json: Dict[str, Any], file_path:
continue
if key == "overrides" and key_path.startswith("sources"):
deprecations.warn(
"source-override-deprecation",
source_name=key_path.split(".")[-1],
@@ -205,6 +219,9 @@ def jsonschema_validate(schema: Dict[str, Any], json: Dict[str, Any], file_path:
keys = _additional_properties_violation_keys(sub_error)
key_path = error_path_to_string(error)
for key in keys:
if key in _get_allowed_config_key_aliases():
continue
deprecations.warn(
"custom-key-in-config-deprecation",
key=key,

View File

@@ -249,34 +249,17 @@ class GraphRunnableTask(ConfiguredTask):
thread_exception = e
finally:
if result is not None:
fire_event(
NodeFinished(
node_info=runner.node.node_info,
run_result=result.to_msg_dict(),
try:
fire_event(
NodeFinished(
node_info=runner.node.node_info,
run_result=result.to_msg_dict(),
)
)
)
except Exception as e:
result = self._handle_thread_exception(runner, e)
else:
msg = f"Exception on worker thread. {thread_exception}"
fire_event(
GenericExceptionOnRun(
unique_id=runner.node.unique_id,
exc=str(thread_exception),
node_info=runner.node.node_info,
)
)
result = RunResult(
status=RunStatus.Error, # type: ignore
timing=[],
thread_id="",
execution_time=0.0,
adapter_response={},
message=msg,
failures=None,
batch_results=None,
node=runner.node,
)
result = self._handle_thread_exception(runner, thread_exception)
# `_event_status` dict is only used for logging. Make sure
# it gets deleted when we're done with it
@@ -365,6 +348,32 @@ class GraphRunnableTask(ConfiguredTask):
args = [runner]
self._submit(pool, args, callback)
def _handle_thread_exception(
self,
runner: BaseRunner,
thread_exception: Optional[Union[KeyboardInterrupt, SystemExit, Exception]],
) -> RunResult:
msg = f"Exception on worker thread. {thread_exception}"
fire_event(
GenericExceptionOnRun(
unique_id=runner.node.unique_id,
exc=str(thread_exception),
node_info=runner.node.node_info,
)
)
return RunResult(
status=RunStatus.Error, # type: ignore
timing=[],
thread_id="",
execution_time=0.0,
adapter_response={},
message=msg,
failures=None,
batch_results=None,
node=runner.node,
)
def _handle_result(self, result: RunResult) -> None:
"""Mark the result as completed, insert the `CompileResultNode` into
the manifest, and mark any descendants (potentially with a 'cause' if

View File

@@ -54,7 +54,7 @@ dependencies = [
"dbt-extractor>=0.5.0,<=0.6",
"dbt-semantic-interfaces>=0.9.0,<0.10",
# Minor versions for these are expected to be backwards-compatible
"dbt-common>=1.37.0,<2.0",
"dbt-common>=1.37.2,<2.0",
"dbt-adapters>=1.15.5,<2.0",
"dbt-protos>=1.0.405,<2.0",
"pydantic<3",

View File

@@ -33,7 +33,7 @@ select {{ config.require('meta_key') }} as col_value
meta_model_meta_require_sql = """
-- models/meta_model.sql
select {{ config.require('meta_key') }} as col_value
select {{ config.meta_require('meta_key') }} as col_value
"""
@@ -66,11 +66,11 @@ class TestConfigGetMeta:
self,
project,
):
# This test runs a model with a config.get(key, default)
# This test runs a model with a config.get(key, default) -> default value returned
results = run_dbt(["run"], expect_pass=False)
assert len(results) == 1
assert str(results[0].status) == "error"
assert 'column "my_meta_value" does not exist' in results[0].message
assert 'column "meta_default_value" does not exist' in results[0].message
write_file(meta_model_meta_get_sql, "models", "meta_model.sql")
results = run_dbt(["run"], expect_pass=False)
@@ -95,7 +95,7 @@ class TestConfigGetMetaRequire:
results = run_dbt(["run"], expect_pass=False)
assert len(results) == 1
assert str(results[0].status) == "error"
assert 'column "my_meta_value" does not exist' in results[0].message
assert "does not define a required config parameter 'meta_key'" in results[0].message
write_file(meta_model_meta_require_sql, "models", "meta_model.sql")
results = run_dbt(["run"], expect_pass=False)

View File

@@ -55,6 +55,22 @@ class TestDbtRunner:
dbt.invoke(["debug"])
mock_callback.assert_called()
def test_callback_node_finished_exceptions_are_raised(self, project):
from dbt_common.events.base_types import EventMsg
def callback_with_exception(event: EventMsg):
if event.info.name == "NodeFinished":
raise Exception("This should let continue the execution registering the failure")
dbt = dbtRunner(callbacks=[callback_with_exception])
result = dbt.invoke(["run", "--select", "models"])
assert result is not None
assert (
result.result.results[0].message
== "Exception on worker thread. This should let continue the execution registering the failure"
)
def test_invoke_kwargs(self, project, dbt):
res = dbt.invoke(
["run"],

View File

@@ -220,10 +220,16 @@ models_yml = """
models:
- name: abcd
description: "abcd model"
versions:
- v: 1
- name: efgh
description: "efgh model"
versions:
- v: 1
- name: ijkl
description: "ijkl model"
versions:
- v: 1
"""
append_sources_yml = """
@@ -233,6 +239,8 @@ append_sources_yml = """
append_models_yml = """
- name: mnop
description: "mnop model"
versions:
- v: 1
"""
mnop_sql = """
@@ -245,9 +253,9 @@ class TestSourcesAndSchemaFiles:
def models(self):
return {
"sources.yml": sources_yml,
"abcd.sql": abcd_sql,
"efgh.sql": efgh_sql,
"ijkl.sql": ijkl_sql,
"abcd_v1.sql": abcd_sql,
"efgh_v1.sql": efgh_sql,
"ijkl_v1.sql": ijkl_sql,
"_models.yml": models_yml,
}
@@ -258,7 +266,7 @@ class TestSourcesAndSchemaFiles:
assert len(manifest.nodes) == 3
write_file(models_yml + append_models_yml, project.project_root, "models", "_models.yml")
write_file(mnop_sql, project.project_root, "models", "mnop.sql")
write_file(mnop_sql, project.project_root, "models", "mnop_v1.sql")
write_file(sources_yml + append_sources_yml, project.project_root, "models", "sources.yml")
manifest = run_dbt(["parse"])
@@ -268,3 +276,4 @@ class TestSourcesAndSchemaFiles:
# the patch updates, including description, so description will be ""
for node in manifest.nodes.values():
assert node.description == f"{node.name} model"
assert node.unique_id.endswith(".v1")

View File

@@ -1,9 +1,17 @@
import pytest
from dbt.deprecations import (
CustomKeyInConfigDeprecation,
CustomKeyInObjectDeprecation,
GenericJSONSchemaValidationDeprecation,
active_deprecations,
reset_deprecations,
)
from dbt.jsonschemas.jsonschemas import (
jsonschema_validate,
resources_schema,
validate_model_config,
)
from dbt.jsonschemas.jsonschemas import validate_model_config
from dbt.tests.util import safe_set_invocation_context
from dbt_common.context import get_invocation_context
from dbt_common.events.event_catcher import EventCatcher
@@ -48,3 +56,38 @@ class TestValidateModelConfigNoError:
assert len(ckicd_catcher.caught_events) == 1
assert ckicd_catcher.caught_events[0].data.key == "non_existent_config"
assert len(gjsvd_catcher.caught_events) == 0
class TestValidateJsonSchema:
@pytest.fixture(scope="class")
def model_bigquery_alias_config_contents(self):
return {
"models": [
{
"name": "model_1",
"config": {
"dataset": "dataset_1",
"project": "project_1",
},
}
],
}
def test_validate_json_schema_no_error_aliases(self, model_bigquery_alias_config_contents):
reset_deprecations()
safe_set_invocation_context()
get_invocation_context().uses_adapter("bigquery")
jsonschema_validate(resources_schema(), model_bigquery_alias_config_contents, "test.yml")
assert active_deprecations == {}
def test_validate_json_schema_has_error_aliases(self, model_bigquery_alias_config_contents):
reset_deprecations()
safe_set_invocation_context()
# Set to adapter that doesn't support aliases specified
get_invocation_context().uses_adapter("snowflake")
jsonschema_validate(resources_schema(), model_bigquery_alias_config_contents, "test.yml")
assert active_deprecations == {"custom-key-in-config-deprecation": 2}