ds-caselaw-marklogic-api-client 38.0.0__py3-none-any.whl → 39.0.0__py3-none-any.whl

caselawclient/models/documents/__init__.py

@@ -31,7 +31,7 @@ from caselawclient.models.utilities.aws import (
     request_parse,
     unpublish_documents,
 )
-from caselawclient.types import DocumentURIString
+from caselawclient.types import DocumentURIString, SuccessFailureMessageTuple
 
 from .body import DocumentBody
 from .exceptions import CannotEnrichUnenrichableDocument, CannotPublishUnpublishableDocument, DocumentNotSafeForDeletion
@@ -521,9 +521,12 @@ class Document:
         """
         return self.body.has_content
 
+    def validate_identifiers(self) -> SuccessFailureMessageTuple:
+        return self.identifiers.perform_all_validations(document_type=type(self), api_client=self.api_client)
+
     def save_identifiers(self) -> None:
         """Validate the identifiers, and if the validation passes save them to MarkLogic"""
-        validations = self.identifiers.perform_all_validations(document_type=type(self), api_client=self.api_client)
+        validations = self.validate_identifiers()
         if validations.success is True:
             self.api_client.set_property_as_node(self.uri, "identifiers", self.identifiers.as_etree)
         else:
@@ -570,7 +573,7 @@ class Document:
     def xml_with_correct_frbr(self) -> bytes:
         """Dynamically modify FRBR uris to reflect current storage location and FCL id"""
         fcl_identifiers = self.identifiers.of_type(FindCaseLawIdentifier)
-        work_uri = f"https://caselaw.nationalarchives.gov.uk/id/{fcl_identifiers[0].url_slug}"
+        work_uri = f"https://caselaw.nationalarchives.gov.uk/id/doc/{fcl_identifiers[0].value}"
         expression_uri = f"https://caselaw.nationalarchives.gov.uk/{self.uri.lstrip('/')}"
         manifestation_uri = f"https://caselaw.nationalarchives.gov.uk/{self.uri.lstrip('/')}/data.xml"
         return self.body.apply_xslt(

caselawclient/models/documents/xml.py

@@ -58,7 +58,8 @@ class XML:
         """XSLT transform this XML, given a stylesheet"""
         passable_values = {k: etree.XSLT.strparam(v) for k, v in values.items()}
         xslt_transform = etree.XSLT(etree.fromstring(xslt))
-        return etree.tostring(xslt_transform(self.xml_as_tree, profile_run=False, **passable_values))
+        noncanonical_xml = xslt_transform(self.xml_as_tree, profile_run=False, **passable_values)
+        return etree.tostring(noncanonical_xml, method="c14n2")
 
     def apply_xslt(self, xslt_filename: str, **values: str) -> bytes:
         """XSLT transform this XML, given a path to a stylesheet"""

caselawclient/models/identifiers/__init__.py

@@ -149,6 +149,9 @@ class Identifier(ABC):
     @property
     def score(self) -> float:
         """Return the score of this identifier, used to calculate the preferred identifier for a document."""
+        if self.deprecated:
+            return 0
+
         return 1 * self.schema.base_score_multiplier
 
     def same_as(self, other: "Identifier") -> bool:

caselawclient/models/identifiers/collection.py

@@ -4,8 +4,7 @@ from lxml import etree
 
 from caselawclient.types import SuccessFailureMessageTuple
 
-from . import Identifier
-from .exceptions import UUIDMismatchError
+from . import Identifier, IdentifierSchema
 from .fclid import FindCaseLawIdentifier
 from .neutral_citation import NeutralCitationNumber
 from .press_summary_ncn import PressSummaryRelatedNCNIdentifier
@@ -22,16 +21,61 @@ SUPPORTED_IDENTIFIER_TYPES: list[type["Identifier"]] = [
 
 
 class IdentifiersCollection(dict[str, Identifier]):
-    def validate_uuids_match_keys(self) -> None:
+    def validate_uuids_match_keys(self) -> SuccessFailureMessageTuple:
         for uuid, identifier in self.items():
             if uuid != identifier.uuid:
-                msg = "Key of {identifier} in Identifiers is {uuid} not {identifier.uuid}"
-                raise UUIDMismatchError(msg)
+                return SuccessFailureMessageTuple(
+                    False, [f"Key of {identifier} in Identifiers is {uuid} not {identifier.uuid}"]
+                )
 
-    def perform_all_validations(
+        return SuccessFailureMessageTuple(True, [])
+
+    def _list_all_identifiers_by_schema(self) -> dict[type[IdentifierSchema], list[Identifier]]:
+        """Get a list of all identifiers, grouped by their schema."""
+        identifiers_by_schema: dict[type[IdentifierSchema], list[Identifier]] = {}
+
+        for identifier in self.values():
+            identifiers_by_schema.setdefault(identifier.schema, []).append(identifier)
+
+        return identifiers_by_schema
+
+    def check_only_single_non_deprecated_identifier_where_multiples_not_allowed(self) -> SuccessFailureMessageTuple:
+        """Check that only one non-deprecated identifier exists per schema where that schema does not allow multiples."""
+
+        for schema, identifiers in self._list_all_identifiers_by_schema().items():
+            non_deprecated_identifiers = [i for i in identifiers if not i.deprecated]
+            if len(non_deprecated_identifiers) > 1:
+                return SuccessFailureMessageTuple(
+                    False,
+                    [
+                        f"Multiple non-deprecated identifiers found for schema '{schema.name}': {', '.join(i.value for i in non_deprecated_identifiers)}"
+                    ],
+                )
+
+        return SuccessFailureMessageTuple(True, [])
+
+    def _perform_collection_level_validations(self) -> SuccessFailureMessageTuple:
+        """Perform identifier validations which are only possible at the collection level, such as UUID integrity and identifying exclusivity problems."""
+
+        success = True
+        messages: list[str] = []
+
+        collection_validations_to_run: list[SuccessFailureMessageTuple] = [
+            self.validate_uuids_match_keys(),
+            self.check_only_single_non_deprecated_identifier_where_multiples_not_allowed(),
+        ]
+
+        for validation in collection_validations_to_run:
+            if not validation.success:
+                success = False
+                messages += validation.messages
+
+        return SuccessFailureMessageTuple(success, messages)
+
+    def _perform_identifier_level_validations(
         self, document_type: type["Document"], api_client: "MarklogicApiClient"
     ) -> SuccessFailureMessageTuple:
-        self.validate_uuids_match_keys()
+        """Perform identifier validations at the individual identifier level."""
 
         success = True
         messages: list[str] = []
@@ -45,14 +89,38 @@ class IdentifiersCollection(dict[str, Identifier]):
 
         return SuccessFailureMessageTuple(success, messages)
 
+    def perform_all_validations(
+        self, document_type: type["Document"], api_client: "MarklogicApiClient"
+    ) -> SuccessFailureMessageTuple:
+        """Perform all possible identifier validations on this collection, both at the individual and collection level."""
+
+        identifier_level_success, identifier_level_messages = self._perform_identifier_level_validations(
+            document_type=document_type, api_client=api_client
+        )
+        collection_level_success, collection_level_messages = self._perform_collection_level_validations()
+
+        success = all([identifier_level_success, collection_level_success])
+        all_messages = identifier_level_messages + collection_level_messages
+
+        return SuccessFailureMessageTuple(success, all_messages)
+
     def contains(self, other_identifier: Identifier) -> bool:
-        "Do the identifier's value and namespace already exist in this group?"
+        """Does the identifier's value and namespace already exist in this group?"""
         return any(other_identifier.same_as(identifier) for identifier in self.values())
 
     def add(self, identifier: Identifier) -> None:
         if not self.contains(identifier):
             self[identifier.uuid] = identifier
 
+    def valid_new_identifier_types(self, document_type: type["Document"]) -> list[type[Identifier]]:
+        """Return a list of identifier types which can be added to a document of the given type, given identifiers already in this collection."""
+        return [
+            t
+            for t in SUPPORTED_IDENTIFIER_TYPES
+            if t.schema.allow_editing
+            and (not t.schema.document_types or document_type.__name__ in t.schema.document_types)
+        ]
+
     def __delitem__(self, key: Union[Identifier, str]) -> None:
         if isinstance(key, Identifier):
             super().__delitem__(key.uuid)

caselawclient/models/identifiers/exceptions.py

@@ -2,9 +2,5 @@ class InvalidIdentifierXMLRepresentationException(Exception):
     pass
 
 
-class UUIDMismatchError(Exception):
-    pass
-
-
 class IdentifierValidationException(Exception):
     pass

caselawclient/xslt/modify_xml_live.xsl

@@ -1,6 +1,9 @@
 <?xml version="1.0"?>
-<xsl:stylesheet version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform" xmlns:uk='https://caselaw.nationalarchives.gov.uk/akn' xmlns:akn='http://docs.oasis-open.org/legaldocml/ns/akn/3.0'>
-
+<xsl:stylesheet version="1.0"
+    xmlns='http://docs.oasis-open.org/legaldocml/ns/akn/3.0'
+    xmlns:akn='http://docs.oasis-open.org/legaldocml/ns/akn/3.0'
+    xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
+    xmlns:uk='https://caselaw.nationalarchives.gov.uk/akn'>
     <xsl:param name="work_uri" />
     <xsl:param name="expression_uri" />
     <xsl:param name="manifestation_uri" />
@@ -16,51 +19,51 @@
     <!-- <xsl:template match="akn:identification/FRBRWork/FRBRthistext/text()"><xsl:copy-of select="$cat" /></xsl:template> -->
 
     <xsl:template match="akn:identification/akn:FRBRWork/akn:FRBRthis">
-        <akn:FRBRthis>
+        <FRBRthis>
             <xsl:attribute name="value">
                 <xsl:value-of select="$work_uri" />
             </xsl:attribute>
-        </akn:FRBRthis>
+        </FRBRthis>
     </xsl:template>
 
     <xsl:template match="akn:identification/akn:FRBRWork/akn:FRBRuri">
-        <akn:FRBRuri>
+        <FRBRuri>
             <xsl:attribute name="value">
                 <xsl:value-of select="$work_uri" />
             </xsl:attribute>
-        </akn:FRBRuri>
+        </FRBRuri>
     </xsl:template>
 
     <xsl:template match="akn:identification/akn:FRBRExpression/akn:FRBRthis">
-        <akn:FRBRthis>
+        <FRBRthis>
             <xsl:attribute name="value">
                 <xsl:value-of select="$expression_uri" />
             </xsl:attribute>
-        </akn:FRBRthis>
+        </FRBRthis>
     </xsl:template>
 
     <xsl:template match="akn:identification/akn:FRBRExpression/akn:FRBRuri">
-        <akn:FRBRuri>
+        <FRBRuri>
             <xsl:attribute name="value">
                 <xsl:value-of select="$expression_uri" />
             </xsl:attribute>
-        </akn:FRBRuri>
+        </FRBRuri>
     </xsl:template>
 
     <xsl:template match="akn:identification/akn:FRBRManifestation/akn:FRBRthis">
-        <akn:FRBRthis>
+        <FRBRthis>
             <xsl:attribute name="value">
                 <xsl:value-of select="$manifestation_uri" />
             </xsl:attribute>
-        </akn:FRBRthis>
+        </FRBRthis>
     </xsl:template>
 
     <xsl:template match="akn:identification/akn:FRBRManifestation/akn:FRBRuri">
-        <akn:FRBRuri>
+        <FRBRuri>
             <xsl:attribute name="value">
                 <xsl:value-of select="$manifestation_uri" />
             </xsl:attribute>
-        </akn:FRBRuri>
+        </FRBRuri>
     </xsl:template>
 
 

{ds_caselaw_marklogic_api_client-38.0.0.dist-info → ds_caselaw_marklogic_api_client-39.0.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: ds-caselaw-marklogic-api-client
-Version: 38.0.0
+Version: 39.0.0
 Summary: An API client for interacting with the underlying data in Find Caselaw.
 Keywords: national archives,caselaw
 Author: The National Archives
@@ -11,7 +11,7 @@ Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Requires-Dist: boto3 (>=1.26.112,<2.0.0)
-Requires-Dist: certifi (>=2025.6.15,<2025.7.0)
+Requires-Dist: certifi (>=2025.7.9,<2025.8.0)
 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-38.0.0.dist-info → ds_caselaw_marklogic_api_client-39.0.0.dist-info}/RECORD

@@ -7,15 +7,15 @@ caselawclient/errors.py,sha256=JC16fEGq_MRJX-_KFzfINCV2Cqx8o6OWOt3C16rQd84,3142
 caselawclient/factories.py,sha256=eGj9TiZpmF3todW-08Ps7bHNMvByHqwEbgujRhvU_Yc,7382
 caselawclient/identifier_resolution.py,sha256=B5I1sD7o7YjzsXMECjbKjgiGLDda5bGhejsJ-lYpTIg,2429
 caselawclient/models/__init__.py,sha256=kd23EUpvaC7aLHdgk8farqKAQEx3lf7RvNT2jEatvlg,68
-caselawclient/models/documents/__init__.py,sha256=3iEzgjX4ooLUKEuAeOylEUTm5rUQkrRztBmv4Rp1Ub4,21117
+caselawclient/models/documents/__init__.py,sha256=0MYorde8Ku9qUWLa5G_aQUBKwDeTRZ_1ebsV9RxIBuI,21256
 caselawclient/models/documents/body.py,sha256=O1ZTV3KHo-YNi7Syd4oCV1CVSuRF7mcLXojwshyY4jg,6601
 caselawclient/models/documents/exceptions.py,sha256=te7PPQTDHjZ9EYVg5pVaiZfF00lMBFy333PHj8_mkC4,443
 caselawclient/models/documents/statuses.py,sha256=Cp4dTQmJOtsU41EJcxy5dV1841pGD2PNWH0VrkDEv4Q,579
 caselawclient/models/documents/transforms/html.xsl,sha256=XyUQLFcJ7_GwthWQ6ShU0bmzrgpl7xDFU-U8VLgOvEs,38258
-caselawclient/models/documents/xml.py,sha256=UPGL4Bm-RRde3ezkm1s1zcL-XdNqy1ncRcbsZNLtGn4,2135
-caselawclient/models/identifiers/__init__.py,sha256=IOMGUT4he1sZ7imjJupkyataB8OHMlsidoHFZJTWFsU,7581
-caselawclient/models/identifiers/collection.py,sha256=zcdFE831vYBh1dueULdKu3cdVEgUlNwCIwq6XNEC36M,4062
-caselawclient/models/identifiers/exceptions.py,sha256=ckVsjPzLuTXkbd7KZRJXoxcltQCXPGL2rMyYwE5orgg,177
+caselawclient/models/documents/xml.py,sha256=BVra2VL_0JyImM8GC3wdouu1tApy79C-e2dHvQyrXPE,2195
+caselawclient/models/identifiers/__init__.py,sha256=N2xcIOUlnG5w7yCzCAvGnlkN_rm6ceMHwA-CS1ZNoGk,7631
+caselawclient/models/identifiers/collection.py,sha256=kGlziJiLAqoyd6LaaZ5tsgUf2fD6Y-7fv1It9S4-Otw,7448
+caselawclient/models/identifiers/exceptions.py,sha256=6LVjvx-UOwqkrpxU19ydmrphKNw0rcG5GXwjTFyf8Dk,130
 caselawclient/models/identifiers/fclid.py,sha256=hj8z-VhXFrUHKOY6k_ItPvOakIvbhJ5xEbZ04E2j7t8,1521
 caselawclient/models/identifiers/neutral_citation.py,sha256=bYAeXHVm_ls0aDTeYI4uv35iZmJGSKU4-H-iLh2xED0,2912
 caselawclient/models/identifiers/press_summary_ncn.py,sha256=t-x6PsEe2tz1uO1qZKXKK0TugkQYb_49O_xgjd_oiE4,801
@@ -84,9 +84,9 @@ 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=zuyDGTkcN6voOXCm3APXItZ-Ey6tZ2hdZummZWzjl50,6489
-caselawclient/xslt/modify_xml_live.xsl,sha256=F-0FM0OhT9Ua1WVEVeHYwwVkktmTOERykOuGhMrKans,2346
+caselawclient/xslt/modify_xml_live.xsl,sha256=gNjwBun2-UzOeeuf0wNjFtN3jXm1yrwqv_KT8r1slXw,2370
 caselawclient/xslt/sample.xsl,sha256=IG-v77stjwqiw25pguh391K-5DTKiX651WqILDZixm0,825
-ds_caselaw_marklogic_api_client-38.0.0.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
-ds_caselaw_marklogic_api_client-38.0.0.dist-info/METADATA,sha256=uvG9_FZesFrKHYJLzQpYw1JokKUIyrqAe7WPf1oVq2A,4320
-ds_caselaw_marklogic_api_client-38.0.0.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
-ds_caselaw_marklogic_api_client-38.0.0.dist-info/RECORD,,
+ds_caselaw_marklogic_api_client-39.0.0.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
+ds_caselaw_marklogic_api_client-39.0.0.dist-info/METADATA,sha256=hn5g5HHm3R3WIpBTeSKsv0IY6uVxKY6m7AkCo-pr_y4,4319
+ds_caselaw_marklogic_api_client-39.0.0.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
+ds_caselaw_marklogic_api_client-39.0.0.dist-info/RECORD,,