ds-caselaw-marklogic-api-client 35.1.0__py3-none-any.whl → 36.0.0__py3-none-any.whl

caselawclient/factories.py

@@ -85,7 +85,7 @@ class DocumentFactory:
         document.body = kwargs.pop("body") if "body" in kwargs else DocumentBodyFactory.build()
 
         if identifiers is None:
-            document.identifiers.add(FindCaseLawIdentifier(value="a1b2c3"))
+            document.identifiers.add(FindCaseLawIdentifier(value="tn4t35ts"))
         else:
             for identifier in identifiers:
                 document.identifiers.add(identifier)
@@ -173,7 +173,7 @@ class IdentifierResolutionFactory:
             "documents.compiled_url_slugs.identifier_uuid": resolution_uuid or "24b9a384-8bcf-4f20-996a-5c318f8dc657",
             "documents.compiled_url_slugs.document_uri": document_uri or "/ewca/civ/2003/547.xml",
             "documents.compiled_url_slugs.identifier_slug": identifier_slug or "ewca/civ/2003/54721",
-            "documents.compiled_url_slugs.document_published": "true" if published else "false",
+            "documents.compiled_url_slugs.document_published": published,
             "documents.compiled_url_slugs.identifier_namespace": namespace or "ukncn",
             "documents.compiled_url_slugs.identifier_value": value or "[2003] EWCA 54721 (Civ)",
         }

caselawclient/identifier_resolution.py

@@ -45,7 +45,7 @@ class IdentifierResolution(NamedTuple):
             identifier_uuid=row["documents.compiled_url_slugs.identifier_uuid"],
             document_uri=MarkLogicDocumentURIString(row["documents.compiled_url_slugs.document_uri"]),
             identifier_slug=DocumentIdentifierSlug(row["documents.compiled_url_slugs.identifier_slug"]),
-            document_published=row["documents.compiled_url_slugs.document_published"] == "true",
+            document_published=row["documents.compiled_url_slugs.document_published"],
             identifier_value=DocumentIdentifierValue(row["documents.compiled_url_slugs.identifier_value"]),
             identifier_namespace=identifier_namespace,
             identifier_type=IDENTIFIER_NAMESPACE_MAP[identifier_namespace],

caselawclient/models/identifiers/__init__.py

@@ -6,6 +6,8 @@ from lxml import etree
 
 from caselawclient.types import DocumentIdentifierSlug, DocumentIdentifierValue
 
+from .exceptions import IdentifierValidationException, UUIDMismatchError
+
 IDENTIFIER_PACKABLE_ATTRIBUTES: list[str] = [
     "uuid",
     "value",
@@ -18,14 +20,6 @@ IDENTIFIER_UNPACKABLE_ATTRIBUTES: list[str] = [
 ]
 
 
-class InvalidIdentifierXMLRepresentationException(Exception):
-    pass
-
-
-class UUIDMismatchError(Exception):
-    pass
-
-
 class IdentifierSchema(ABC):
     """
     A base class which describes what an identifier schema should look like.
@@ -89,6 +83,11 @@ class Identifier(ABC):
         return self.value
 
     def __init__(self, value: str, uuid: Optional[str] = None) -> None:
+        if not self.schema.validate_identifier(value=value):
+            raise IdentifierValidationException(
+                f'Identifier value "{value}" is not valid according to the {self.schema.name} schema.'
+            )
+
         self.value = DocumentIdentifierValue(value)
         if uuid:
             self.uuid = uuid

caselawclient/models/identifiers/exceptions.py

@@ -0,0 +1,10 @@
+class InvalidIdentifierXMLRepresentationException(Exception):
+    pass
+
+
+class UUIDMismatchError(Exception):
+    pass
+
+
+class IdentifierValidationException(Exception):
+    pass

caselawclient/models/identifiers/neutral_citation.py

@@ -6,6 +6,7 @@ from ds_caselaw_utils.types import NeutralCitationString
 from caselawclient.types import DocumentIdentifierSlug
 
 from . import Identifier, IdentifierSchema
+from .exceptions import IdentifierValidationException
 
 VALID_NCN_PATTERN = re.compile(r"(^\[([0-9]{4})\] ([a-zA-Z]+)(?: ([a-zA-Z]+))? ([0-9]+)(?: \(([a-zA-Z]+)\))?$)")
 """
@@ -23,6 +24,18 @@ TODO: When these capture groups are being used in anger (eg to build URL slugs)
 """
 
 
+class NCNValidationException(IdentifierValidationException):
+    pass
+
+
+class NCNDoesNotMatchExpectedPatternException(NCNValidationException):
+    pass
+
+
+class NCNCannotConvertToValidURLSlugException(NCNValidationException):
+    pass
+
+
 class NeutralCitationNumberSchema(IdentifierSchema):
     """
     Identifier schema describing a Neutral Citation Number.
@@ -37,7 +50,16 @@ class NeutralCitationNumberSchema(IdentifierSchema):
 
     @classmethod
     def validate_identifier(cls, value: str) -> bool:
-        return bool(VALID_NCN_PATTERN.match(value))
+        # Quick check to see if the NCN matches the expected pattern
+        if not bool(VALID_NCN_PATTERN.match(value)):
+            raise NCNDoesNotMatchExpectedPatternException(f"NCN '{value}' is not in the expected format")
+
+        # Can we convert this to a URL? neutral_url returns False if not.
+        # This functionally tests to see if the court exists, since only valid patterns (where we know how to match the court code) will convert
+        if not neutral_url(NeutralCitationString(value)):
+            raise NCNCannotConvertToValidURLSlugException(f"NCN '{value}' cannot be converted to an NCN-based URL slug")
+
+        return True
 
     @classmethod
     def compile_identifier_url_slug(cls, value: str) -> DocumentIdentifierSlug:
@@ -45,7 +67,7 @@ class NeutralCitationNumberSchema(IdentifierSchema):
             NeutralCitationString(value)
         )  # TODO: At some point this should move out of utils and into this class.
         if not ncn_based_uri_string:
-            raise Exception(f"Unable to convert NCN {value} into NCN-based URL slug")
+            raise NCNCannotConvertToValidURLSlugException(f"NCN '{value}' cannot be converted to an NCN-based URL slug")
         return DocumentIdentifierSlug(ncn_based_uri_string)
 
 

caselawclient/models/identifiers/unpacker.py

@@ -3,7 +3,8 @@ from warnings import warn
 
 from lxml import etree
 
-from . import IDENTIFIER_UNPACKABLE_ATTRIBUTES, Identifier, Identifiers, InvalidIdentifierXMLRepresentationException
+from . import IDENTIFIER_UNPACKABLE_ATTRIBUTES, Identifier, Identifiers
+from .exceptions import InvalidIdentifierXMLRepresentationException
 from .fclid import FindCaseLawIdentifier
 from .neutral_citation import NeutralCitationNumber
 from .press_summary_ncn import PressSummaryRelatedNCNIdentifier

caselawclient/models/utilities/aws.py

@@ -50,6 +50,7 @@ def create_aws_client(service: Literal["s3", "sns"]) -> Any:
     aws = boto3.session.Session()
     return aws.client(
         service,
+        region_name=env("PRIVATE_ASSET_BUCKET_REGION", default=None),
         config=botocore.client.Config(signature_version="s3v4"),
     )
 

{ds_caselaw_marklogic_api_client-35.1.0.dist-info → ds_caselaw_marklogic_api_client-36.0.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ds-caselaw-marklogic-api-client
-Version: 35.1.0
+Version: 36.0.0
 Summary: An API client for interacting with the underlying data in Find Caselaw.
 Home-page: https://github.com/nationalarchives/ds-caselaw-custom-api-client
 Keywords: national archives,caselaw
@@ -10,7 +10,7 @@ Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Requires-Dist: boto3 (>=1.26.112,<2.0.0)
-Requires-Dist: certifi (>=2025.1.31,<2025.2.0)
+Requires-Dist: certifi (>=2025.4.26,<2025.5.0)
 Requires-Dist: charset-normalizer (>=3.0.0,<4.0.0)
 Requires-Dist: django-environ (>=0.12.0)
 Requires-Dist: ds-caselaw-utils (>=2.0.0,<3.0.0)

{ds_caselaw_marklogic_api_client-35.1.0.dist-info → ds_caselaw_marklogic_api_client-36.0.0.dist-info}/RECORD

@@ -4,8 +4,8 @@ caselawclient/client_helpers/__init__.py,sha256=eucyUXwUqI72TPw-C5zLcHlMu4GtFY50
 caselawclient/client_helpers/search_helpers.py,sha256=R99HyRLeYHgsw2L3DOidEqlKLLvs6Tga5rKTuWQViig,1525
 caselawclient/content_hash.py,sha256=0cPC4OoABq0SC2wYFX9-24DodNigeOqksDxgxQH_hUA,2221
 caselawclient/errors.py,sha256=JC16fEGq_MRJX-_KFzfINCV2Cqx8o6OWOt3C16rQd84,3142
-caselawclient/factories.py,sha256=BzxG7rBprmszxS1U1jsyBKaMSSm22RFmFK1nQSOZJlw,6812
-caselawclient/identifier_resolution.py,sha256=BKj7k8ca6b8ZpeoOk2uHYqprOOp9aN5qEB7PWvfJmSw,2439
+caselawclient/factories.py,sha256=tbF3iV3WpHm-z_2aRCV4Dg6ZLBkQuc_egdcd_oWCIlI,6791
+caselawclient/identifier_resolution.py,sha256=B5I1sD7o7YjzsXMECjbKjgiGLDda5bGhejsJ-lYpTIg,2429
 caselawclient/models/__init__.py,sha256=kd23EUpvaC7aLHdgk8farqKAQEx3lf7RvNT2jEatvlg,68
 caselawclient/models/documents/__init__.py,sha256=7U5oN2b8JZV9sFSYglBg2K7fKKQgDjTVU5FT22URx7U,19043
 caselawclient/models/documents/body.py,sha256=mhPOV1cOF3RJr69UzNPlo1KrzePaj7KDPYi1exP06L0,5880
@@ -13,17 +13,18 @@ caselawclient/models/documents/exceptions.py,sha256=Mz1P8uNqf5w6uLnRwJt6xK7efsVq
 caselawclient/models/documents/statuses.py,sha256=Cp4dTQmJOtsU41EJcxy5dV1841pGD2PNWH0VrkDEv4Q,579
 caselawclient/models/documents/transforms/html.xsl,sha256=oSSO-IBX4qLiSWexQYmWJfGNevF09aCBx4D1NYqXxpo,38322
 caselawclient/models/documents/xml.py,sha256=HlmPb63lLMnySSOLP4iexcAyQiLByKBZtTd25f8sY8M,1268
-caselawclient/models/identifiers/__init__.py,sha256=sbS638jek1qtOSyGWVfIgaja0tnpK5iB5FXiYJSbJiU,6666
+caselawclient/models/identifiers/__init__.py,sha256=_iAsx_e-Rxq3ED3TQQ5Vg1ZNnQn6AZQK8QCvHSFhdGg,6848
+caselawclient/models/identifiers/exceptions.py,sha256=ckVsjPzLuTXkbd7KZRJXoxcltQCXPGL2rMyYwE5orgg,177
 caselawclient/models/identifiers/fclid.py,sha256=Gq0G0eLdUH3j7VybluqJGVcHV3Y-hpjty1OdR91dI5Q,1453
-caselawclient/models/identifiers/neutral_citation.py,sha256=O3zLw8lnhh5Mb0COjwhhDfTyvcSt48Td5pfT6yaCCJk,1947
+caselawclient/models/identifiers/neutral_citation.py,sha256=cgB1PO44ra8dKsYGq9_g53be8T1mQatq9Hh_DN5DVg0,2868
 caselawclient/models/identifiers/press_summary_ncn.py,sha256=CKhNTnO6pfriS4Sg-gt0Pf0fRhH53Uf0flULPncJWM4,762
-caselawclient/models/identifiers/unpacker.py,sha256=eApUVdIKg3ABNiG38-OfYwKeaLGOl1Ebe5b9IusuLZM,2276
+caselawclient/models/identifiers/unpacker.py,sha256=fKfiuTKaN-71e40PV8BC9PueKjwgfX9Xw3IVdy8mkqg,2299
 caselawclient/models/judgments.py,sha256=r40irgdEID-NeSNLm3OUdUBznMpRSwjD2SJrGlBgP8o,2208
 caselawclient/models/neutral_citation_mixin.py,sha256=jAac3PPuWyPdj9N-n-U_JfwkbgbSIXaqFVQahfu95do,2086
 caselawclient/models/parser_logs.py,sha256=30kF4w0GcowiMIFtymUkl7ZARanNh_PjDpJZezn-cA8,315
 caselawclient/models/press_summaries.py,sha256=PIq9RceZ7n7Z079tESfxhQbfxCmtTc2V2OeFtcn594s,2144
 caselawclient/models/utilities/__init__.py,sha256=u3yIhbTjFQ1JJyAm5wsMEBswWl4t6Z7UMORF5FqC2xQ,1257
-caselawclient/models/utilities/aws.py,sha256=AMddhGL1E9kG_7yv-18Gn_0ITrvMdgbdA_bB1xiRSWE,8491
+caselawclient/models/utilities/aws.py,sha256=NTF2W2aNgbO72e5WklXZC2U2_GPbVeynjTS1Nqu6DcE,8561
 caselawclient/models/utilities/dates.py,sha256=WwORxVjUHM1ZFcBF6Qtwo3Cj0sATsnSECkUZ6ls1N1Q,492
 caselawclient/models/utilities/move.py,sha256=MXdUqkSiyqRb8YKs_66B6ICWn8EWM6DiJV95fuJO1Us,3610
 caselawclient/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -81,7 +82,7 @@ caselawclient/xquery/validate_document.xqy,sha256=PgaDcnqCRJPIVqfmWsNlXmCLNKd21q
 caselawclient/xquery/xslt.xqy,sha256=w57wNijH3dkwHkpKeAxqjlghVflQwo8cq6jS_sm-erM,199
 caselawclient/xquery/xslt_transform.xqy,sha256=cccaFiGkCcvSfDv007UriZ3I4ak2nTLP1trRZdbOoS8,2462
 caselawclient/xquery_type_dicts.py,sha256=21V6PJ0a8lPZuwzQCWeHNo914hdCbCJtOvnqdHcTpKk,6383
-ds_caselaw_marklogic_api_client-35.1.0.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
-ds_caselaw_marklogic_api_client-35.1.0.dist-info/METADATA,sha256=85B3kbZeAcYmhX5vhjDvHxSz3YnsgWcX3uV7zLQr76s,4206
-ds_caselaw_marklogic_api_client-35.1.0.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-ds_caselaw_marklogic_api_client-35.1.0.dist-info/RECORD,,
+ds_caselaw_marklogic_api_client-36.0.0.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
+ds_caselaw_marklogic_api_client-36.0.0.dist-info/METADATA,sha256=ZIEcK86tOf0ruMjnRFKcq4N65_x8HKUBiF7PNFHj-YM,4206
+ds_caselaw_marklogic_api_client-36.0.0.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+ds_caselaw_marklogic_api_client-36.0.0.dist-info/RECORD,,