dapla-toolbelt-metadata 0.9.8__tar.gz → 0.9.9__tar.gz

{dapla_toolbelt_metadata-0.9.8 → dapla_toolbelt_metadata-0.9.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dapla-toolbelt-metadata
-Version: 0.9.8
+Version: 0.9.9
 Summary: Dapla Toolbelt Metadata
 Project-URL: homepage, https://github.com/statisticsnorway/dapla-toolbelt-metadata
 Project-URL: repository, https://github.com/statisticsnorway/dapla-toolbelt-metadata

{dapla_toolbelt_metadata-0.9.8 → dapla_toolbelt_metadata-0.9.9}/pyproject.toml

@@ -2,7 +2,7 @@
 name = "dapla-toolbelt-metadata"
 description = "Dapla Toolbelt Metadata"
 license = "MIT"
-version = "0.9.8"
+version = "0.9.9"
 classifiers = ["Development Status :: 4 - Beta"]
 readme = "README.md"
 authors = [{ name = "Statistics Norway", email = "metadata@ssb.no" }]

{dapla_toolbelt_metadata-0.9.8 → dapla_toolbelt_metadata-0.9.9}/src/dapla_metadata/datasets/_merge.py

@@ -78,7 +78,7 @@ class DatasetConsistencyStatus:
 
 def check_dataset_consistency(
     new_dataset_path: Path | CloudPath,
-    existing_dataset_path: Path,
+    existing_dataset_path: Path | CloudPath,
 ) -> list[DatasetConsistencyStatus]:
     """Run consistency tests.
 

{dapla_toolbelt_metadata-0.9.8 → dapla_toolbelt_metadata-0.9.9}/src/dapla_metadata/datasets/core.py

@@ -6,7 +6,6 @@ import copy
 import json
 import logging
 from concurrent.futures import ThreadPoolExecutor
-from pathlib import Path
 from typing import TYPE_CHECKING
 from typing import cast
 
@@ -175,7 +174,7 @@ class Datadoc:
         ):
             self.dataset_consistency_status = check_dataset_consistency(
                 self.dataset_path,
-                Path(self.metadata_document),
+                self.metadata_document,
             )
             self.dataset_consistency_status.extend(
                 check_variables_consistency(

{dapla_toolbelt_metadata-0.9.8 → dapla_toolbelt_metadata-0.9.9}/src/dapla_metadata/variable_definitions/vardok_id.py

@@ -33,7 +33,11 @@ class VardokId(VardokIdResponse):
         model: VardokIdResponse,
     ) -> "VardokId":
         """Create a VariableDefinition instance from a CompleteResponse."""
-        return VardokId.model_construct(**model.model_dump())
+        self = VardokId.from_dict(model.model_dump())
+        if not self:
+            msg = f"Could not construct a VardokId instance from {model}"
+            raise ValueError(msg)
+        return self
 
     def __str__(self) -> str:
         """Format as indented YAML."""

{dapla_toolbelt_metadata-0.9.8 → dapla_toolbelt_metadata-0.9.9}/src/dapla_metadata/variable_definitions/vardok_vardef_id_pair.py

@@ -32,7 +32,11 @@ class VardokVardefIdPair(VardokVardefIdPairResponse):
         model: VardokVardefIdPairResponse,
     ) -> "VardokVardefIdPair":
         """Create a VardokVardefIdPair instance from a VardokVardefIdPairResponse."""
-        return VardokVardefIdPair.model_construct(**model.model_dump())
+        self = VardokVardefIdPair.from_dict(model.model_dump())
+        if not self:
+            msg = f"Could not construct a VardokVardefIdPair instance from {model}"
+            raise ValueError(msg)
+        return self
 
     def __str__(self) -> str:
         """Format as indented YAML."""

{dapla_toolbelt_metadata-0.9.8 → dapla_toolbelt_metadata-0.9.9}/src/dapla_metadata/variable_definitions/variable_definition.py

@@ -82,7 +82,11 @@ class VariableDefinition(CompleteResponse):
         model: CompleteResponse,
     ) -> "VariableDefinition":
         """Create a VariableDefinition instance from a CompleteResponse."""
-        return VariableDefinition.model_construct(**model.model_dump())
+        self = VariableDefinition.from_dict(model.model_dump())
+        if not self:
+            msg = f"Could not construct a VariableDefinition instance from {model}"
+            raise ValueError(msg)
+        return self
 
     @vardef_exception_handler
     def list_validity_periods(self) -> list["VariableDefinition"]:

{dapla_toolbelt_metadata-0.9.8 → dapla_toolbelt_metadata-0.9.9}/tests/datasets/test_dataset_consistency.py

@@ -2,6 +2,8 @@ import contextlib
 from pathlib import Path
 
 import pytest
+from cloudpathlib.local import LocalGSClient
+from cloudpathlib.local import LocalGSPath
 from datadoc_model.all_optional.model import DatadocMetadata
 from datadoc_model.all_optional.model import DataType
 from datadoc_model.all_optional.model import Variable
@@ -21,6 +23,7 @@ from dapla_metadata.datasets._merge import InconsistentDatasetsWarning
 from dapla_metadata.datasets._merge import check_dataset_consistency
 from dapla_metadata.datasets._merge import check_ready_to_merge
 from dapla_metadata.datasets._merge import check_variables_consistency
+from dapla_metadata.datasets.core import Datadoc
 from tests.datasets.constants import TEST_BUCKET_NAMING_STANDARD_COMPATIBLE_PATH
 from tests.datasets.constants import VARIABLE_DATA_TYPES
 from tests.datasets.constants import VARIABLE_SHORT_NAMES
@@ -227,3 +230,61 @@ def test_check_variables_consistency(
                 assert r.success, f"'{r.message}' failed but should have passed"
     else:
         assert all(r.success for r in results)
+
+
+def test_bucket_check_ok_when_both_params_are_gs_paths(mocker):
+    """Test the functionality of `check_dataset_consistency`.
+
+    Verify the desired behavior when BOTH parameters are gs-paths (no Path() conversion occurs). In
+    this case no `Bucket name` warning should be issued.
+    """
+    _patch_gs_utils(mocker)
+    resource_json = (
+        Path(__file__).parent
+        / "resources"
+        / "existing_metadata_file"
+        / "person_testdata_p2020-12-31_p2020-12-31_v1__DOC.json"
+    )
+    resource_parquet = (
+        Path(__file__).parent
+        / "resources"
+        / "datasets"
+        / "person_testdata_p2021-12-31_p2021-12-31_v1.parquet"
+    )
+    assert resource_parquet.exists(), f"Parquet resource missing: {resource_parquet}"
+    assert resource_json.exists(), f"JSON resource missing: {resource_json}"
+    directory = "ssb-dapla-felles-data-produkt-test/datadoc/brukertest/10/sykefratot/klargjorte_data/"
+    dataset_gs = LocalGSPath(
+        f"gs://{directory}person_testdata_p2021-12-31_p2021-12-31_v1.parquet"
+    )
+    metadata_doc_gs = LocalGSPath(
+        f"gs://{directory}person_testdata_p2021-12-31_p2021-12-31_v1__DOC.json"
+    )
+    metadata_doc_gs.parent.mkdir(parents=True, exist_ok=True)
+    metadata_doc_gs.write_text(resource_json.read_text(), encoding="utf-8")
+    assert metadata_doc_gs.exists(), "Metadata document was not written to LocalGS"
+    dataset_gs.parent.mkdir(parents=True, exist_ok=True)
+    dataset_gs.write_bytes(resource_parquet.read_bytes())
+    assert dataset_gs.exists(), "Parquet file was not written to LocalGS"
+    dd = Datadoc(
+        dataset_path=str(dataset_gs),
+        metadata_document_path=str(metadata_doc_gs),
+        errors_as_warnings=True,
+    )
+    _assert_bucket_ok(dd.dataset_consistency_status, BUCKET_NAME_MESSAGE)
+
+
+def _assert_bucket_ok(results: list, expected_msg: str):
+    bucket_checks = [r for r in results if expected_msg in r.message]
+    assert bucket_checks, f"Expected at least one check for {expected_msg}."
+    assert all(r.success for r in bucket_checks), (
+        f"Expected {expected_msg} to have success == True when both are gs-paths."
+    )
+
+
+def _patch_gs_utils(mocker):
+    mocker.patch(
+        "dapla_metadata.datasets.utility.utils.google.auth.default", autospec=True
+    )
+    mocker.patch("dapla_metadata.datasets.utility.utils.GSClient", LocalGSClient)
+    mocker.patch("dapla_metadata.datasets.utility.utils.GSPath", LocalGSPath)

{dapla_toolbelt_metadata-0.9.8 → dapla_toolbelt_metadata-0.9.9}/tests/variable_definitions/test_variable_definition.py

@@ -253,6 +253,30 @@ def test_create_patch_change_short_name(variable_definition: VariableDefinition)
     mock_create_patch.assert_not_called()
 
 
+def test_from_model(
+    variable_definition: VariableDefinition,
+):
+    """This covers a bug that was reported in.
+
+    It was caused by an incorrect implementation in the from_model
+    method. The only thing we're testing is that no exceptions are
+    raised in this flow.
+    """
+    mock_api_response = MagicMock(name="VariableDefinition")
+    mock_api_response.classification_reference = "702"
+    with patch.object(
+        VariableDefinition,
+        "create_patch",
+        return_value=mock_api_response,
+    ) as mock_create_patch:
+        my_patch = VariableDefinition.from_model(variable_definition)
+        my_patch.create_patch_from_file(
+            file_path=VARIABLE_DEFINITION_EDITING_FILES_DIR
+            / "variable_definition_landbak_wypvb3wd_2025-05-02T10-06-20.yaml",
+        )
+        mock_create_patch.assert_called_once()
+
+
 def test_create_patch_from_file_path_not_set(variable_definition: VariableDefinition):
     my_patch = variable_definition
     my_patch.set_file_path(file_path=None)