dapla-toolbelt-metadata 0.9.3__tar.gz → 0.9.4__tar.gz

{dapla_toolbelt_metadata-0.9.3 → dapla_toolbelt_metadata-0.9.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dapla-toolbelt-metadata
-Version: 0.9.3
+Version: 0.9.4
 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-0.9.3 → dapla_toolbelt_metadata-0.9.4}/pyproject.toml

@@ -2,7 +2,7 @@
 name = "dapla-toolbelt-metadata"
 description = "Dapla Toolbelt Metadata"
 license = "MIT"
-version = "0.9.3"
+version = "0.9.4"
 classifiers = ["Development Status :: 4 - Beta"]
 readme = "README.md"
 authors = [{ name = "Statistics Norway", email = "metadata@ssb.no" }]

{dapla_toolbelt_metadata-0.9.3 → dapla_toolbelt_metadata-0.9.4}/src/dapla_metadata/datasets/model_validation.py

@@ -13,20 +13,19 @@ from typing_extensions import Self
 
 from dapla_metadata.datasets.utility.constants import DATE_VALIDATION_MESSAGE
 from dapla_metadata.datasets.utility.constants import NUM_OBLIGATORY_DATASET_FIELDS
-from dapla_metadata.datasets.utility.constants import NUM_OBLIGATORY_VARIABLES_FIELDS
 from dapla_metadata.datasets.utility.constants import OBLIGATORY_METADATA_WARNING
 from dapla_metadata.datasets.utility.utils import get_missing_obligatory_dataset_fields
 from dapla_metadata.datasets.utility.utils import (
     get_missing_obligatory_variables_fields,
 )
+from dapla_metadata.datasets.utility.utils import (
+    get_missing_obligatory_variables_pseudo_fields,
+)
 from dapla_metadata.datasets.utility.utils import get_timestamp_now
 from dapla_metadata.datasets.utility.utils import incorrect_date_order
 from dapla_metadata.datasets.utility.utils import (
     num_obligatory_dataset_fields_completed,
 )
-from dapla_metadata.datasets.utility.utils import (
-    num_obligatory_variables_fields_completed,
-)
 from dapla_metadata.datasets.utility.utils import set_variables_inherit_from_dataset
 
 if TYPE_CHECKING:
@@ -146,21 +145,31 @@ class ValidateDatadocMetadata(model.DatadocMetadata):
             ObligatoryVariableWarning: If not all obligatory variable metadata fields
                 are filled in.
         """
-        if self.variables is not None and num_obligatory_variables_fields_completed(
-            self.variables,
-        ) != (NUM_OBLIGATORY_VARIABLES_FIELDS * len(self.variables)):
-            warnings.warn(
-                f"{OBLIGATORY_METADATA_WARNING} {get_missing_obligatory_variables_fields(self.variables)}",
-                ObligatoryVariableWarning,
-                stacklevel=2,
-            )
-            logger.warning(
-                "Type warning: %s.%s %s",
-                ObligatoryVariableWarning,
-                OBLIGATORY_METADATA_WARNING,
-                get_missing_obligatory_variables_fields(self.variables),
-            )
-
+        if self.variables is not None:
+            missing_fields_dict = {}
+            for d in get_missing_obligatory_variables_fields(self.variables):
+                for var, fields in d.items():
+                    missing_fields_dict[var] = fields.copy()
+
+            for d in get_missing_obligatory_variables_pseudo_fields(self.variables):
+                for var, fields in d.items():
+                    if var in missing_fields_dict:
+                        missing_fields_dict[var].extend(fields)
+                    else:
+                        missing_fields_dict[var] = fields.copy()
+
+            missing_fields = [
+                {var: fields} for var, fields in missing_fields_dict.items()
+            ]
+            if missing_fields:
+                message = f"{OBLIGATORY_METADATA_WARNING} {missing_fields}"
+                warnings.warn(message, ObligatoryVariableWarning, stacklevel=2)
+                logger.warning(
+                    "Type warning: %s.%s %s",
+                    ObligatoryVariableWarning,
+                    OBLIGATORY_METADATA_WARNING,
+                    missing_fields,
+                )
         return self
 
 

{dapla_toolbelt_metadata-0.9.3 → dapla_toolbelt_metadata-0.9.4}/src/dapla_metadata/datasets/utility/constants.py

@@ -47,9 +47,9 @@ OBLIGATORY_VARIABLES_METADATA_IDENTIFIERS = [
     "temporality_type",
 ]
 
-OBLIGATORY_VARIABLES_PESUODONYMIZATION_IDENTIFIERS = [
+OBLIGATORY_VARIABLES_PSEUDONYMIZATION_IDENTIFIERS = [
     "encryption_algorithm",
-    "encryption_key_refrence",
+    "encryption_key_reference",
 ]
 
 

{dapla_toolbelt_metadata-0.9.3 → dapla_toolbelt_metadata-0.9.4}/src/dapla_metadata/datasets/utility/utils.py

@@ -36,6 +36,9 @@ from dapla_metadata.datasets.utility.constants import (
 from dapla_metadata.datasets.utility.constants import (
     OBLIGATORY_VARIABLES_METADATA_IDENTIFIERS_MULTILANGUAGE,
 )
+from dapla_metadata.datasets.utility.constants import (
+    OBLIGATORY_VARIABLES_PSEUDONYMIZATION_IDENTIFIERS,
+)
 from dapla_metadata.datasets.utility.constants import PAPIS_ENCRYPTION_KEY_REFERENCE
 from dapla_metadata.datasets.utility.constants import PAPIS_STABLE_IDENTIFIER_TYPE
 from dapla_metadata.datasets.utility.enums import EncryptionAlgorithm
@@ -367,6 +370,25 @@ def num_obligatory_variable_fields_completed(
     return NUM_OBLIGATORY_VARIABLES_FIELDS - len(missing_metadata)
 
 
+def num_obligatory_pseudo_fields_missing(
+    variables: list[all_optional_model.Variable],
+) -> int:
+    """Counts the number of obligatory pseudonymization fields are missing.
+
+    Args:
+        variables: The variables to count obligatory fields for.
+
+    Returns:
+        The number of obligatory pseudonymization fields that are missing.
+    """
+    return sum(
+        getattr(v.pseudonymization, field, None) is None
+        for v in variables
+        if v.pseudonymization is not None
+        for field in OBLIGATORY_VARIABLES_PSEUDONYMIZATION_IDENTIFIERS
+    )
+
+
 def get_missing_obligatory_dataset_fields(
     dataset: DatasetType,
 ) -> list:
@@ -433,6 +455,40 @@ def get_missing_obligatory_variables_fields(variables: list) -> list[dict]:
     return [item for item in missing_variables_fields if next(iter(item.values()))]
 
 
+def get_missing_obligatory_variables_pseudo_fields(
+    variables: list[all_optional_model.Variable],
+) -> list[dict]:
+    """Identify obligatory variable pseudonymization fields that are missing values for each variable.
+
+    This function checks for obligatory fields that are directly missing
+    (i.e., set to `None`).
+
+    Args:
+        variables: A list of variable objects to check for missing obligatory pseudonymization fields.
+
+    Returns:
+        A list of dictionaries with variable short names as keys and list of missing
+        obligatory variable pseudonymization fields as values. This includes:
+            - Fields that are directly `None` and are listed as obligatory metadata.
+    """
+    return [
+        {
+            v.short_name: [
+                key
+                for key, value in v.pseudonymization.model_dump().items()
+                if _is_missing_metadata(
+                    key,
+                    value,
+                    OBLIGATORY_VARIABLES_PSEUDONYMIZATION_IDENTIFIERS,
+                    [],
+                )
+            ]
+        }
+        for v in variables
+        if v.pseudonymization is not None
+    ]
+
+
 def running_in_notebook() -> bool:
     """Return True if running in Jupyter Notebook."""
     try:

{dapla_toolbelt_metadata-0.9.3 → dapla_toolbelt_metadata-0.9.4}/tests/datasets/test_validators.py

@@ -263,3 +263,19 @@ def test_obligatory_metadata_variables_warning_name(metadata: Datadoc):
     ) as record2:
         metadata.write_metadata_document()
     assert variable_with_name not in str(record2[1].message)
+
+
+def test_obligatory_metadata_variables_warning_pseudonymization(metadata: Datadoc):
+    metadata.variables_lookup["pers_id"].pseudonymization = model.Pseudonymization(
+        pseudonymization_time="2022-10-07T07:35:01Z",
+        stable_identifier_type="",
+        stable_identifier_version="",
+        encryption_algorithm=None,
+        encryption_algorithm_parameters=[],
+    )
+    with warnings.catch_warnings(record=True) as w:
+        warnings.simplefilter("always")
+        metadata.write_metadata_document()
+        if issubclass(w[1].category, ObligatoryVariableWarning):
+            missing_obligatory_dataset = str(w[1].message)
+    assert "encryption_algorithm" in str(missing_obligatory_dataset)