dapla-toolbelt-metadata 0.9.6.dev1759398171__tar.gz → 0.9.7__tar.gz

{dapla_toolbelt_metadata-0.9.6.dev1759398171 → dapla_toolbelt_metadata-0.9.7}/.github/workflows/release.yml

@@ -1,14 +1,11 @@
 name: Release
 
 on:
+  workflow_dispatch:
   push:
     branches:
       - main
       - master
-      - release/0.9.6
-    tags:
-      - 'v*'
-
 
 jobs:
   release:
@@ -58,6 +55,7 @@ jobs:
           uv build
 
       - name: Publish package on PyPI
+        if: steps.check-version.outputs.tag
         uses: pypa/gh-action-pypi-publish@v1.13.0
 
       - name: Publish package on TestPyPI

{dapla_toolbelt_metadata-0.9.6.dev1759398171 → dapla_toolbelt_metadata-0.9.7}/.github/workflows/tests.yml

@@ -141,4 +141,4 @@ jobs:
           SONAR_TOKEN: ${{ secrets.SONAR_TOKEN }}
         # No need to run SonarCloud analysis if dependabot update or token not defined
         if: env.SONAR_TOKEN != '' && (github.actor != 'dependabot[bot]')
-        uses: SonarSource/sonarqube-scan-action@v5.3.1
+        uses: SonarSource/sonarqube-scan-action@v6.0.0

{dapla_toolbelt_metadata-0.9.6.dev1759398171 → dapla_toolbelt_metadata-0.9.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dapla-toolbelt-metadata
-Version: 0.9.6.dev1759398171
+Version: 0.9.7
 Summary: Dapla Toolbelt Metadata
 Project-URL: homepage, https://github.com/statisticsnorway/dapla-toolbelt-metadata
 Project-URL: repository, https://github.com/statisticsnorway/dapla-toolbelt-metadata
@@ -16,15 +16,19 @@ Requires-Dist: beautifulsoup4>=4.12.3
 Requires-Dist: cloudpathlib[gs]>=0.17.0
 Requires-Dist: google-auth>=2.38.0
 Requires-Dist: lxml>=5.3.1
+Requires-Dist: pandas>=2.3.3
 Requires-Dist: pyarrow>=8.0.0
 Requires-Dist: pydantic>=2.5.2
 Requires-Dist: pyjwt>=2.8.0
+Requires-Dist: python-dateutil>=2.9.0.post0
 Requires-Dist: python-dotenv>=1.0.1
+Requires-Dist: pytz>=2025.2
 Requires-Dist: requests>=2.31.0
 Requires-Dist: ruamel-yaml>=0.18.10
 Requires-Dist: ssb-datadoc-model<9.0.0,>=8.0.0
 Requires-Dist: ssb-klass-python>=1.0.1
 Requires-Dist: typing-extensions>=4.12.2
+Requires-Dist: urllib3>=2.5.0
 Description-Content-Type: text/markdown
 
 # Dapla Toolbelt Metadata

{dapla_toolbelt_metadata-0.9.6.dev1759398171 → dapla_toolbelt_metadata-0.9.7}/pyproject.toml

@@ -2,7 +2,7 @@
 name = "dapla-toolbelt-metadata"
 description = "Dapla Toolbelt Metadata"
 license = "MIT"
-version = "0.9.6.dev.1759398171"
+version = "0.9.7"
 classifiers = ["Development Status :: 4 - Beta"]
 readme = "README.md"
 authors = [{ name = "Statistics Norway", email = "metadata@ssb.no" }]
@@ -22,6 +22,10 @@ dependencies = [
     'ruamel-yaml >=0.18.10',
     'google-auth >=2.38.0',
     'lxml >=5.3.1',
+    "pandas>=2.3.3",
+    "urllib3>=2.5.0",
+    "pytz>=2025.2",
+    "python-dateutil>=2.9.0.post0",
 ]
 
 [project.urls]

{dapla_toolbelt_metadata-0.9.6.dev1759398171 → dapla_toolbelt_metadata-0.9.7}/src/dapla_metadata/datasets/core.py

@@ -33,7 +33,12 @@ from dapla_metadata.datasets.utility.constants import (
 from dapla_metadata.datasets.utility.constants import METADATA_DOCUMENT_FILE_SUFFIX
 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.urn import convert_uris_to_urns
+from dapla_metadata.datasets.utility.urn import klass_urn_converter
+from dapla_metadata.datasets.utility.urn import vardef_urn_converter
 from dapla_metadata.datasets.utility.utils import OptionalDatadocMetadataType
+from dapla_metadata.datasets.utility.utils import VariableListType
+from dapla_metadata.datasets.utility.utils import VariableType
 from dapla_metadata.datasets.utility.utils import calculate_percentage
 from dapla_metadata.datasets.utility.utils import derive_assessment_from_state
 from dapla_metadata.datasets.utility.utils import get_timestamp_now
@@ -108,8 +113,8 @@ class Datadoc:
         self.container: all_optional_model.MetadataContainer | None = None
         self.dataset_path: pathlib.Path | CloudPath | None = None
         self.dataset = all_optional_model.Dataset()
-        self.variables: list = []
-        self.variables_lookup: dict[str, all_optional_model.Variable] = {}
+        self.variables: VariableListType = []
+        self.variables_lookup: dict[str, VariableType] = {}
         self.explicitly_defined_metadata_document = False
         self.dataset_consistency_status: list[DatasetConsistencyStatus] = []
         if metadata_document_path:
@@ -204,22 +209,24 @@ class Datadoc:
         else:
             self._set_metadata(existing_metadata or extracted_metadata)
 
-        set_default_values_variables(self.variables)
-        set_default_values_dataset(cast("all_optional_model.Dataset", self.dataset))
-        set_dataset_owner(self.dataset)
-        self._create_variables_lookup()
-
     def _set_metadata(
         self,
-        merged_metadata: OptionalDatadocMetadataType,
+        metadata: OptionalDatadocMetadataType,
     ) -> None:
-        if not merged_metadata or not (
-            merged_metadata.dataset and merged_metadata.variables
-        ):
+        if not metadata or not (metadata.dataset and metadata.variables):
             msg = "Could not read metadata"
             raise ValueError(msg)
-        self.dataset = cast("all_optional_model.Dataset", merged_metadata.dataset)
-        self.variables = merged_metadata.variables
+        self.dataset = cast("all_optional_model.Dataset", metadata.dataset)
+        self.variables = metadata.variables
+
+        set_default_values_variables(self.variables)
+        set_default_values_dataset(cast("all_optional_model.Dataset", self.dataset))
+        set_dataset_owner(self.dataset)
+        convert_uris_to_urns(self.variables, "definition_uri", [vardef_urn_converter])
+        convert_uris_to_urns(
+            self.variables, "classification_uri", [klass_urn_converter]
+        )
+        self._create_variables_lookup()
 
     def _create_variables_lookup(self) -> None:
         self.variables_lookup = {

dapla_toolbelt_metadata-0.9.7/src/dapla_metadata/datasets/utility/urn.py

@@ -0,0 +1,161 @@
+"""Validate, parse and render URNs."""
+
+import logging
+import re
+from collections.abc import Iterable
+from dataclasses import dataclass
+from enum import Enum
+from enum import auto
+
+from pydantic import AnyUrl
+
+from dapla_metadata.datasets.utility.utils import VariableListType
+
+logger = logging.getLogger(__name__)
+
+URN_ERROR_MESSAGE_BASE = "The URL is not in a supported format"
+
+URN_ERROR_MESSAGE_TEMPLATE = (
+    URN_ERROR_MESSAGE_BASE
+    + " for field '{field_name}' of variable '{short_name}'. URL: '{value}'. Please contact Team Metadata if this URL should be supported."
+)
+
+
+VARDEF_URL_TEMPLATE = "https://{subdomain}.{domain}/variable-definitions"
+
+
+class SsbNaisDomains(str, Enum):
+    """The available domains on SSBs Nais instance."""
+
+    TEST_EXTERNAL = "test.ssb.no"
+    TEST_INTERNAL = "intern.test.ssb.no"
+    PROD_EXTERNAL = "ssb.no"
+    PROD_INTERNAL = "intern.ssb.no"
+
+
+class ReferenceUrlTypes(Enum):
+    """The general category of the URL.
+
+    This can be useful to refer to when constructing a URL from a URN for a
+    specific context.
+    """
+
+    API = auto()
+    FRONTEND = auto()
+
+
+@dataclass
+class UrnConverter:
+    """Converts URLs to URNs and vice versa.
+
+    Fields:
+        urn_base: The format for the URN, up to the identifier.
+        id_pattern: A capturing group pattern which matches identifiers for this resource.
+        url_bases: The list of all the different URL representations for a resource. There
+            will typically be a number of URL representations for a particular resource,
+            depending on which system or technology they are accessed through and other
+            technical factors. This list defines which concrete URLs can be considered
+            equivalent to a URN.
+    """
+
+    urn_base: str
+    id_pattern: str
+    url_bases: list[tuple[ReferenceUrlTypes, str]]
+
+    def _extract_id(self, url: str, pattern: re.Pattern[str]) -> str | None:
+        if match := pattern.match(url):
+            return match.group(1)
+        return None
+
+    def _build_pattern(self, url_base: str) -> re.Pattern[str]:
+        return re.compile(f"^{url_base}/{self.id_pattern}")
+
+    def build_urn(self, identifier: str) -> str:
+        """Build a URN for the given identifier."""
+        return f"{self.urn_base}:{identifier}"
+
+    def convert_to_urn(self, url: str | AnyUrl) -> AnyUrl | None:
+        """Convert a URL to a generalized URN for that same resource.
+
+        Args:
+            url (str | AnyUrl): The URL to convert.
+
+        Returns:
+            str | None: The URN or None if it can't be converted.
+        """
+        if str(url).startswith(self.urn_base):
+            # In this case the value is already in the expected format and nothing needs to be done.
+            return AnyUrl(url)
+        patterns = (self._build_pattern(url[-1]) for url in self.url_bases)
+        matches = (self._extract_id(str(url), p) for p in patterns)
+        identifier = next((m for m in matches if m), None)
+        if identifier:
+            return AnyUrl(self.build_urn(identifier))
+
+        return None
+
+
+vardef_urn_converter = UrnConverter(
+    urn_base="urn:ssb:variable-definition:vardef",
+    id_pattern=r"([a-z0-9]{8})",
+    url_bases=[
+        *[
+            (
+                ReferenceUrlTypes.API,
+                VARDEF_URL_TEMPLATE.format(
+                    subdomain="metadata", domain=nais_domain.value
+                ),
+            )
+            for nais_domain in SsbNaisDomains
+        ],
+        *[
+            (
+                ReferenceUrlTypes.FRONTEND,
+                VARDEF_URL_TEMPLATE.format(
+                    subdomain="catalog", domain=nais_domain.value
+                ),
+            )
+            for nais_domain in SsbNaisDomains
+        ],
+    ],
+)
+
+klass_urn_converter = UrnConverter(
+    urn_base="urn:ssb:classification:klass",
+    id_pattern=r"([0-9]{1,5})",
+    url_bases=[
+        (ReferenceUrlTypes.FRONTEND, "https://www.ssb.no/klass/klassifikasjoner"),
+        (ReferenceUrlTypes.FRONTEND, "https://www.ssb.no/en/klass/klassifikasjoner"),
+        (ReferenceUrlTypes.API, "https://data.ssb.no/api/klass/v1/classifications"),
+    ],
+)
+
+
+def convert_uris_to_urns(
+    variables: VariableListType, field_name: str, converters: Iterable[UrnConverter]
+) -> None:
+    """Where URIs are recognized URLs, convert them to URNs.
+
+    Where the value is not a known URL we preserve the value as it is and log an
+    ERROR level message.
+
+    Args:
+        variables (VariableListType): The list of variables.
+        field_name (str): The name of the field which has URLs to convert to URNs
+        converters (Iterable[UrnConverter]): One or more converters which implement
+            conversion of URLs into one specific URN format. These will typically be
+            specific to an individual metadata reference system.
+    """
+    for v in variables:
+        field = getattr(v, field_name, None)
+        if field:
+            if urn := next((c.convert_to_urn(field) for c in converters), None):
+                setattr(v, field_name, urn)
+            else:
+                logger.error(
+                    URN_ERROR_MESSAGE_TEMPLATE.format(
+                        field_name=field_name,
+                        short_name=v.short_name,
+                        value=field,
+                    )
+                )

{dapla_toolbelt_metadata-0.9.6.dev1759398171 → dapla_toolbelt_metadata-0.9.7}/src/dapla_metadata/datasets/utility/utils.py

@@ -121,7 +121,7 @@ def derive_assessment_from_state(state: DataSetState) -> Assessment:
             return Assessment.SENSITIVE
 
 
-def set_default_values_variables(variables: list) -> None:
+def set_default_values_variables(variables: VariableListType) -> None:
     """Set default values on variables.
 
     Args:

dapla_toolbelt_metadata-0.9.7/tests/datasets/test_urn.py

@@ -0,0 +1,162 @@
+import logging
+
+import datadoc_model.all_optional.model as all_optional_model
+import pytest
+from pydantic import AnyUrl
+
+from dapla_metadata.datasets.core import Datadoc
+from dapla_metadata.datasets.utility.urn import URN_ERROR_MESSAGE_BASE
+from dapla_metadata.datasets.utility.urn import SsbNaisDomains
+from dapla_metadata.datasets.utility.urn import klass_urn_converter
+from dapla_metadata.datasets.utility.urn import vardef_urn_converter
+
+EXAMPLE_KLASS_ID = "91"
+EXAMPLE_KLASS_URN = AnyUrl(klass_urn_converter.build_urn(EXAMPLE_KLASS_ID))
+EXAMPLE_VARDEF_ID = "hd8sks89"
+EXAMPLE_VARDEF_URN = AnyUrl(vardef_urn_converter.build_urn(EXAMPLE_VARDEF_ID))
+
+VARIABLE_DEFINITION_URN_TEST_CASES = [
+    *[
+        (
+            f"https://metadata.{domain.value}/variable-definitions/{EXAMPLE_VARDEF_ID}",
+            EXAMPLE_VARDEF_URN,
+            False,
+        )
+        for domain in SsbNaisDomains
+    ],
+    *[
+        (
+            f"https://catalog.{domain.value}/variable-definitions/{EXAMPLE_VARDEF_ID}",
+            EXAMPLE_VARDEF_URN,
+            False,
+        )
+        for domain in SsbNaisDomains
+    ],
+    (
+        EXAMPLE_VARDEF_URN,
+        EXAMPLE_VARDEF_URN,
+        False,
+    ),
+]
+
+
+@pytest.mark.parametrize(
+    ("case", "expected_result", "expect_warning"),
+    [
+        *VARIABLE_DEFINITION_URN_TEST_CASES,
+        ("https://www.vg.no", None, True),
+    ],
+)
+def test_convert_to_vardef_urn(
+    case: str,
+    expected_result: AnyUrl | None,
+    expect_warning: bool,  # noqa: ARG001
+):
+    assert vardef_urn_converter.convert_to_urn(case) == expected_result
+
+
+@pytest.mark.parametrize(
+    ("case", "expected_result", "expect_warning"),
+    [
+        *VARIABLE_DEFINITION_URN_TEST_CASES,
+        ("https://www.vg.no", AnyUrl("https://www.vg.no/"), True),
+        (None, None, False),
+    ],
+)
+def test_convert_to_vardef_urn_end_to_end(
+    case: str | None,
+    expected_result: str | None,
+    expect_warning: bool,
+    caplog: pytest.LogCaptureFixture,
+):
+    caplog.set_level(logging.ERROR)
+    meta = Datadoc()
+    meta._set_metadata(  # noqa: SLF001
+        all_optional_model.DatadocMetadata(
+            dataset=all_optional_model.Dataset(),
+            variables=[
+                all_optional_model.Variable(
+                    short_name="my_variable", definition_uri=case
+                )
+            ],
+        )
+    )
+    if not expected_result:
+        assert meta.variables[0].definition_uri is None
+    else:
+        assert meta.variables[0].definition_uri == expected_result
+    assert (URN_ERROR_MESSAGE_BASE in caplog.text) is expect_warning
+
+
+CLASSIFICATION_URN_TEST_CASES = [
+    (
+        f"https://www.ssb.no/klass/klassifikasjoner/{EXAMPLE_KLASS_ID}",
+        EXAMPLE_KLASS_URN,
+        False,
+    ),
+    (
+        f"https://www.ssb.no/en/klass/klassifikasjoner/{EXAMPLE_KLASS_ID}",
+        EXAMPLE_KLASS_URN,
+        False,
+    ),
+    (
+        f"https://data.ssb.no/api/klass/v1/classifications/{EXAMPLE_KLASS_ID}",
+        EXAMPLE_KLASS_URN,
+        False,
+    ),
+    (
+        f"https://data.ssb.no/api/klass/v1/classifications/{EXAMPLE_KLASS_ID}.json",
+        EXAMPLE_KLASS_URN,
+        False,
+    ),
+    (
+        EXAMPLE_KLASS_URN,
+        EXAMPLE_KLASS_URN,
+        False,
+    ),
+]
+
+
+@pytest.mark.parametrize(
+    ("case", "expected_result", "expect_warning"),
+    CLASSIFICATION_URN_TEST_CASES,
+)
+def test_convert_to_klass_urn(
+    case: str,
+    expected_result: str,
+    expect_warning: bool,  # noqa: ARG001
+):
+    assert klass_urn_converter.convert_to_urn(case) == expected_result
+
+
+@pytest.mark.parametrize(
+    ("case", "expected_result", "expect_warning"),
+    [
+        *CLASSIFICATION_URN_TEST_CASES,
+        ("https://www.vg.no", AnyUrl("https://www.vg.no/"), True),
+        (None, None, False),
+    ],
+)
+def test_convert_to_klass_urn_end_to_end(
+    case: str | None,
+    expected_result: str | None,
+    expect_warning: bool,
+    caplog: pytest.LogCaptureFixture,
+):
+    caplog.set_level(logging.ERROR)
+    meta = Datadoc()
+    meta._set_metadata(  # noqa: SLF001
+        all_optional_model.DatadocMetadata(
+            dataset=all_optional_model.Dataset(),
+            variables=[
+                all_optional_model.Variable(
+                    short_name="my_variable", classification_uri=case
+                )
+            ],
+        )
+    )
+    if not expected_result:
+        assert meta.variables[0].classification_uri is None
+    else:
+        assert meta.variables[0].classification_uri == expected_result
+    assert (URN_ERROR_MESSAGE_BASE in caplog.text) is expect_warning