dapla-toolbelt-metadata 0.9.7__py3-none-any.whl → 0.9.8__py3-none-any.whl

dapla_metadata/_shared/config.py

@@ -13,9 +13,6 @@ from dotenv import load_dotenv
 from dapla_metadata._shared.enums import DaplaEnvironment
 from dapla_metadata._shared.enums import DaplaRegion
 from dapla_metadata._shared.enums import DaplaService
-from dapla_metadata.datasets.utility.constants import (
-    DATADOC_STATISTICAL_SUBJECT_SOURCE_URL,
-)
 
 logger = logging.getLogger(__name__)
 
@@ -28,6 +25,12 @@ DAPLA_SERVICE = "DAPLA_SERVICE"
 DAPLA_GROUP_CONTEXT = "DAPLA_GROUP_CONTEXT"
 OIDC_TOKEN = "OIDC_TOKEN"  # noqa: S105
 
+
+DATADOC_STATISTICAL_SUBJECT_SOURCE_URL_DEFAULT = (
+    "https://www.ssb.no/xp/_/service/mimir/subjectStructurStatistics"
+)
+
+
 env_loaded = False
 
 
@@ -73,7 +76,7 @@ def get_statistical_subject_source_url() -> str | None:
     """Get the URL to the statistical subject source."""
     return (
         get_config_item("DATADOC_STATISTICAL_SUBJECT_SOURCE_URL")
-        or DATADOC_STATISTICAL_SUBJECT_SOURCE_URL
+        or DATADOC_STATISTICAL_SUBJECT_SOURCE_URL_DEFAULT
     )
 
 

dapla_metadata/datasets/utility/constants.py

@@ -94,10 +94,6 @@ DATASET_FIELDS_FROM_EXISTING_METADATA = [
 
 METADATA_DOCUMENT_FILE_SUFFIX = "__DOC.json"
 
-DATADOC_STATISTICAL_SUBJECT_SOURCE_URL = (
-    "https://www.ssb.no/xp/_/service/mimir/subjectStructurStatistics"
-)
-
 PAPIS_STABLE_IDENTIFIER_TYPE = "FREG_SNR"
 PAPIS_ENCRYPTION_KEY_REFERENCE = "papis-common-key-1"
 DAEAD_ENCRYPTION_KEY_REFERENCE = "ssb-common-key-1"

dapla_metadata/datasets/utility/urn.py

@@ -6,9 +6,12 @@ from collections.abc import Iterable
 from dataclasses import dataclass
 from enum import Enum
 from enum import auto
+from typing import Literal
 
 from pydantic import AnyUrl
 
+from dapla_metadata._shared.config import get_dapla_environment
+from dapla_metadata._shared.enums import DaplaEnvironment
 from dapla_metadata.datasets.utility.utils import VariableListType
 
 logger = logging.getLogger(__name__)
@@ -44,6 +47,9 @@ class ReferenceUrlTypes(Enum):
     FRONTEND = auto()
 
 
+UrlVisibility = Literal["public", "internal"]
+
+
 @dataclass
 class UrnConverter:
     """Converts URLs to URNs and vice versa.
@@ -70,11 +76,81 @@ class UrnConverter:
     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:
+    def get_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:
+    def get_url(
+        self,
+        identifier: str,
+        url_type: ReferenceUrlTypes,
+        visibility: Literal["public", "internal"] = "public",
+    ) -> str | None:
+        """Build concrete URL to reference a resource.
+
+        There are typically multiple URLs used to refer to one resource, this method attempts to support known variations.
+
+        Args:
+            identifier (str): The identifier of the resource the URL refers to.
+            url_type (ReferenceUrlTypes): The representation type of the URL
+            visibility (UrlVisibility, optional): Whether the URL should be that which is publicly available or not. Defaults to "public".
+
+        Returns:
+            str | None: The concrete URL. None if we cannot satisfy the supplied requirements.
+        """
+        candidates = [base[-1] for base in self.url_bases if base[0] == url_type]
+
+        def matches_visibility(url: str, visibility: UrlVisibility):
+            return (".intern." in url) is (visibility == "internal")
+
+        def matches_environment(url: str):
+            current_environment = get_dapla_environment()
+            if current_environment == DaplaEnvironment.TEST:
+                return ".test." in url
+            return ".test." not in url
+
+        if url := next(
+            (
+                url
+                for url in candidates
+                if matches_visibility(url, visibility) and matches_environment(url)
+            ),
+            None,
+        ):
+            return url + "/" + identifier
+        return None
+
+    def get_id(self, urn_or_url: str | AnyUrl) -> str | None:
+        """Get an identifier from a URN or URL.
+
+        Args:
+            urn_or_url (str | AnyUrl): The URN or URL refering to a particular resource
+
+        Returns:
+            str | None: The identifier for the resource, or None if it cannot be extracted.
+        """
+        if str(urn_or_url).startswith(self.urn_base):
+            return str(urn_or_url).removeprefix(self.urn_base + ":")
+        return self._extract_id_from_url(urn_or_url)
+
+    def is_id(self, value: str) -> bool:
+        """Check if the value is an identifier for this URN type.
+
+        Args:
+            value (str): The value to check.
+        """
+        if not isinstance(value, str):
+            # Mypy thinks it's impossible to reach this branch, but there are no guarantees in Python.
+            return False  # type: ignore [unreachable]
+        pattern = re.compile(f"^{self.id_pattern}$")
+        return bool(pattern.match(value))
+
+    def _extract_id_from_url(self, url: str | AnyUrl) -> str | None:
+        patterns = (self._build_pattern(url[-1]) for url in self.url_bases)
+        matches = (self._extract_id(str(url), p) for p in patterns)
+        return next((m for m in matches if m), None)
+
+    def convert_url_to_urn(self, url: str | AnyUrl) -> AnyUrl | None:
         """Convert a URL to a generalized URN for that same resource.
 
         Args:
@@ -86,11 +162,8 @@ class UrnConverter:
         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))
+        if identifier := self._extract_id_from_url(url):
+            return AnyUrl(self.get_urn(identifier))
 
         return None
 
@@ -149,7 +222,7 @@ def convert_uris_to_urns(
     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):
+            if urn := next((c.convert_url_to_urn(field) for c in converters), None):
                 setattr(v, field_name, urn)
             else:
                 logger.error(

{dapla_toolbelt_metadata-0.9.7.dist-info → dapla_toolbelt_metadata-0.9.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dapla-toolbelt-metadata
-Version: 0.9.7
+Version: 0.9.8
 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-0.9.7.dist-info → dapla_toolbelt_metadata-0.9.8.dist-info}/RECORD

@@ -1,6 +1,6 @@
 dapla_metadata/__init__.py,sha256=37yh9XWYQoLIVIS_fDdwNN8OXzbYY-6kMYwvjQrLMJQ,428
 dapla_metadata/_shared/__init__.py,sha256=qUFgnVhBVlPRQP0ePmY76c8FvWRrJ-9c5GvzibwERnQ,103
-dapla_metadata/_shared/config.py,sha256=QqXcmP66AfXF8wi6FMsa7et7kH2k4EJPOF4IELKuQig,3213
+dapla_metadata/_shared/config.py,sha256=xRAXDULS85lAi5XyAirh1acDbf7_0xPhElssXvQQk3Y,3245
 dapla_metadata/_shared/enums.py,sha256=WHkH1d8xw41gOly6au_izZB1_-6XTcKu5rhBWUImjp8,509
 dapla_metadata/_shared/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dapla_metadata/dapla/__init__.py,sha256=tkapF-YwmruPPrKvN3pEoCZqb7xvJx_ogBM8XyGMuJI,130
@@ -21,9 +21,9 @@ dapla_metadata/datasets/compatibility/model_backwards_compatibility.py,sha256=W5
 dapla_metadata/datasets/external_sources/__init__.py,sha256=qvIdXwqyEmXNUCB94ZtZXRzifdW4hiXASFFPtC70f6E,83
 dapla_metadata/datasets/external_sources/external_sources.py,sha256=9eIcOIUbaodNX1w9Tj2wl4U4wUmr5kF1R0i01fKUzGs,2974
 dapla_metadata/datasets/utility/__init__.py,sha256=pp6tUcgUbo8iq9OPtFKQrTbLuI3uY7NHptwWSTpasOU,33
-dapla_metadata/datasets/utility/constants.py,sha256=YKsn6GfNIkwLoBp0yq209o0TbsEhsA_jGaZLVR984JU,2933
+dapla_metadata/datasets/utility/constants.py,sha256=4ixDvz5nErQwXa3BEtaGZb2AFUUUldJtNZV46SKwUBc,2817
 dapla_metadata/datasets/utility/enums.py,sha256=i6dcxWya5k4LjLdGGIM_H37rRndizug3peaAgoE5UdM,652
-dapla_metadata/datasets/utility/urn.py,sha256=Y_4wYwWWaFDffIN3uXjCodi-uUUQ7zkX1qEFSwGlVqs,5317
+dapla_metadata/datasets/utility/urn.py,sha256=1NtM9OkTUe4_T5Iy8cR-ofwuR0gUlrtA7oQqEmgWKOc,8141
 dapla_metadata/datasets/utility/utils.py,sha256=q76UJI8W4j2aHSq1jz_AfYnJmLfygEflgUrQpqQEPnY,20157
 dapla_metadata/standards/__init__.py,sha256=n8jnMrudLuScSdfQ4UMJorc-Ptg3Y1-ilT8zAaQnM70,179
 dapla_metadata/standards/name_validator.py,sha256=6-DQE_EKVd6UjL--EXpFcZDQtusVbSFaWaUY-CfOV2c,9184
@@ -91,7 +91,7 @@ dapla_metadata/variable_definitions/_utils/constants.py,sha256=zr5FNVCEz6TM9PVEr
 dapla_metadata/variable_definitions/_utils/files.py,sha256=JbPgPNQ7iA38juMqGEdcg5OjZZUwCb6NQtPL0AEspD0,10933
 dapla_metadata/variable_definitions/_utils/template_files.py,sha256=7fcc7yEHOl5JUZ698kqj4IiikXPHBi3SrAVOk4wqQtw,3308
 dapla_metadata/variable_definitions/_utils/variable_definition_files.py,sha256=sGhcSpckR9NtYGNh2oVkiCd5SI3bbJEBhc1PA2uShs0,4701
-dapla_toolbelt_metadata-0.9.7.dist-info/METADATA,sha256=yb6SMXPh6nMfUPfOMyicbrcConcCoWgEtcgsUsl2jJ0,4854
-dapla_toolbelt_metadata-0.9.7.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-dapla_toolbelt_metadata-0.9.7.dist-info/licenses/LICENSE,sha256=np3IfD5m0ZUofn_kVzDZqliozuiO6wrktw3LRPjyEiI,1073
-dapla_toolbelt_metadata-0.9.7.dist-info/RECORD,,
+dapla_toolbelt_metadata-0.9.8.dist-info/METADATA,sha256=jBwnT3Vs5w-JIXTjzcMDmzhDr8XPlkkdU24zxTNaz4w,4854
+dapla_toolbelt_metadata-0.9.8.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+dapla_toolbelt_metadata-0.9.8.dist-info/licenses/LICENSE,sha256=np3IfD5m0ZUofn_kVzDZqliozuiO6wrktw3LRPjyEiI,1073
+dapla_toolbelt_metadata-0.9.8.dist-info/RECORD,,