ds-caselaw-marklogic-api-client 28.1.0__tar.gz → 28.2.0__tar.gz

{ds_caselaw_marklogic_api_client-28.1.0 → ds_caselaw_marklogic_api_client-28.2.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ds-caselaw-marklogic-api-client
-Version: 28.1.0
+Version: 28.2.0
 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
@@ -11,7 +11,7 @@ Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Requires-Dist: boto3 (>=1.26.112,<2.0.0)
-Requires-Dist: certifi (>=2024.8.30,<2024.9.0)
+Requires-Dist: certifi (>=2024.12.14,<2024.13.0)
 Requires-Dist: charset-normalizer (>=3.0.0,<4.0.0)
 Requires-Dist: django-environ (>=0.11.0,<0.12.0)
 Requires-Dist: ds-caselaw-utils (>=2.0.0,<3.0.0)
@@ -25,6 +25,7 @@ Requires-Dist: pytz (>=2024.1,<2025.0)
 Requires-Dist: requests (>=2.28.2,<3.0.0)
 Requires-Dist: requests-toolbelt (>=0.10.1,<1.1.0)
 Requires-Dist: saxonche (>=12.5.0,<13.0.0)
+Requires-Dist: sqids (>=0.5.0,<0.6.0)
 Requires-Dist: typing-extensions (>=4.7.1,<5.0.0)
 Description-Content-Type: text/markdown
 

{ds_caselaw_marklogic_api_client-28.1.0 → ds_caselaw_marklogic_api_client-28.2.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ds-caselaw-marklogic-api-client"
-version = "28.1.0"
+version = "28.2.0"
 description = "An API client for interacting with the underlying data in Find Caselaw."
 authors = ["The National Archives"]
 homepage = "https://github.com/nationalarchives/ds-caselaw-custom-api-client"
@@ -12,7 +12,7 @@ packages = [
 
 [tool.poetry.dependencies]
 python = "^3.9"
-certifi = ">=2024.8.30,<2024.9.0"
+certifi = ">=2024.12.14,<2024.13.0"
 charset-normalizer = "^3.0.0"
 django-environ = "^0.11.0"
 idna = "^3.4"
@@ -28,6 +28,7 @@ mypy-boto3-sns = "^1.26.69"
 pytz = "^2024.1"
 python-dateutil = "^2.9.0-post.0"
 saxonche = "^12.5.0"
+sqids = "^0.5.0"
 
 [tool.poetry.group.dev.dependencies]
 coverage = "^7.2.3"

{ds_caselaw_marklogic_api_client-28.1.0 → ds_caselaw_marklogic_api_client-28.2.0}/src/caselawclient/Client.py

@@ -1217,3 +1217,7 @@ class MarklogicApiClient:
             ),
         )
         return IdentifierResolutions.from_marklogic_output(raw_results)
+
+    def get_next_document_sequence_number(self) -> int:
+        """Increment the MarkLogic sequence number by one and return the value."""
+        return int(self._eval_and_decode({}, "get_next_document_sequence_number.xqy"))

{ds_caselaw_marklogic_api_client-28.1.0 → ds_caselaw_marklogic_api_client-28.2.0}/src/caselawclient/models/documents/__init__.py

@@ -15,6 +15,7 @@ from caselawclient.errors import (
     NotSupportedOnVersion,
     OnlySupportedOnVersion,
 )
+from caselawclient.models.identifiers.fclid import FindCaseLawIdentifier, FindCaseLawIdentifierSchema
 from caselawclient.models.identifiers.unpacker import unpack_all_identifiers_from_etree
 from caselawclient.models.utilities import VersionsDict, extract_version, render_versions
 from caselawclient.models.utilities.aws import (
@@ -432,6 +433,12 @@ class Document:
         if not self.is_publishable:
             raise CannotPublishUnpublishableDocument
 
+        ## If it doesn't already have one, get a new FCLID for this document and assign
+        if len(self.identifiers.of_type(FindCaseLawIdentifier)) < 1:
+            document_fclid = FindCaseLawIdentifierSchema.mint(self.api_client)
+            self.identifiers.add(document_fclid)
+            self.save_identifiers()
+
         publish_documents(uri_for_s3(self.uri))
         self.api_client.set_published(self.uri, True)
         announce_document_event(

{ds_caselaw_marklogic_api_client-28.1.0 → ds_caselaw_marklogic_api_client-28.2.0}/src/caselawclient/models/identifiers/__init__.py

@@ -127,6 +127,11 @@ class Identifiers(dict[str, Identifier]):
         else:
             super().__delitem__(key)
 
+    def of_type(self, identifier_type: type[Identifier]) -> list[Identifier]:
+        """Return a list of all identifiers of a given type."""
+        uuids = self.keys()
+        return [self[uuid] for uuid in list(uuids) if isinstance(self[uuid], identifier_type)]
+
     def delete_type(self, deleted_identifier_type: type[Identifier]) -> None:
         "For when we want an identifier to be the only valid identifier of that type, delete the others first"
         uuids = self.keys()

ds_caselaw_marklogic_api_client-28.2.0/src/caselawclient/models/identifiers/fclid.py

@@ -0,0 +1,48 @@
+import re
+from typing import TYPE_CHECKING
+
+from sqids import Sqids
+
+from . import Identifier, IdentifierSchema
+
+if TYPE_CHECKING:
+    from caselawclient.Client import MarklogicApiClient
+
+
+VALID_FCLID_PATTERN = re.compile(r"^[bcdfghjkmnpqrstvwxyz23456789]{4,}$")
+
+FCLID_MINIMUM_LENGTH = 8
+FCLID_ALPHABET = "bcdfghjkmnpqrstvwxyz23456789"
+
+sqids = Sqids(
+    min_length=FCLID_MINIMUM_LENGTH,
+    alphabet=FCLID_ALPHABET,
+)
+
+
+class FindCaseLawIdentifierSchema(IdentifierSchema):
+    """
+    Identifier schema describing a Find Case Law Identifier.
+    """
+
+    name = "Find Case Law Identifier"
+    namespace = "fclid"
+
+    @classmethod
+    def validate_identifier(cls, value: str) -> bool:
+        return bool(VALID_FCLID_PATTERN.match(value))
+
+    @classmethod
+    def compile_identifier_url_slug(cls, value: str) -> str:
+        return "tna." + value
+
+    @classmethod
+    def mint(cls, api_client: "MarklogicApiClient") -> "FindCaseLawIdentifier":
+        """Generate a totally new Find Case Law identifier."""
+        next_sequence_number = api_client.get_next_document_sequence_number()
+        new_identifier = sqids.encode([next_sequence_number])
+        return FindCaseLawIdentifier(value=new_identifier)
+
+
+class FindCaseLawIdentifier(Identifier):
+    schema = FindCaseLawIdentifierSchema

{ds_caselaw_marklogic_api_client-28.1.0 → ds_caselaw_marklogic_api_client-28.2.0}/src/caselawclient/models/identifiers/unpacker.py

@@ -3,9 +3,11 @@ from typing import Optional
 from lxml import etree
 
 from . import IDENTIFIER_UNPACKABLE_ATTRIBUTES, Identifier, Identifiers, InvalidIdentifierXMLRepresentationException
+from .fclid import FindCaseLawIdentifier
 from .neutral_citation import NeutralCitationNumber
 
 IDENTIFIER_NAMESPACE_MAP: dict[str, type[Identifier]] = {
+    "fclid": FindCaseLawIdentifier,
     "ukncn": NeutralCitationNumber,
 }
 

ds_caselaw_marklogic_api_client-28.2.0/src/caselawclient/xquery/get_next_document_sequence_number.xqy

@@ -0,0 +1,14 @@
+xquery version "1.0-ml";
+declare option xdmp:transaction-mode "update";
+
+let $_ := xdmp:set-transaction-mode("update")
+let $state_doc := fn:doc("state.xml")
+let $counter_node := $state_doc/state/document_counter
+
+let $current_counter := $counter_node/text()
+let $new_counter := fn:sum(($current_counter, 1))
+
+let $_ := xdmp:node-replace($counter_node, <document_counter>{$new_counter}</document_counter>)
+let $_ := xdmp:commit()
+
+return $new_counter