ds-caselaw-marklogic-api-client 37.0.2__tar.gz → 37.1.0__tar.gz

{ds_caselaw_marklogic_api_client-37.0.2 → ds_caselaw_marklogic_api_client-37.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ds-caselaw-marklogic-api-client
-Version: 37.0.2
+Version: 37.1.0
 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

{ds_caselaw_marklogic_api_client-37.0.2 → ds_caselaw_marklogic_api_client-37.1.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ds-caselaw-marklogic-api-client"
-version = "37.0.2"
+version = "37.1.0"
 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"

{ds_caselaw_marklogic_api_client-37.0.2 → ds_caselaw_marklogic_api_client-37.1.0}/src/caselawclient/Client.py

@@ -1198,6 +1198,24 @@ class MarklogicApiClient:
 
         return results
 
+    def get_missing_fclid(
+        self,
+        maximum_records: int = 50,
+    ) -> list[str]:
+        """Retrieve the URIs of published documents which do not have an identifier in the `fclid` schema."""
+        vars: query_dicts.GetMissingFclidDict = {
+            "maximum_records": maximum_records,
+        }
+
+        results: list[str] = get_multipart_strings_from_marklogic_response(
+            self._send_to_eval(
+                vars,
+                "get_missing_fclid.xqy",
+            )
+        )
+
+        return results
+
     def resolve_from_identifier_slug(
         self, identifier_slug: DocumentIdentifierSlug, published_only: bool = True
     ) -> IdentifierResolutions:

{ds_caselaw_marklogic_api_client-37.0.2 → ds_caselaw_marklogic_api_client-37.1.0}/src/caselawclient/models/documents/__init__.py

@@ -378,6 +378,16 @@ class Document:
         """
         return self.api_client.validate_document(self.uri)
 
+    def assign_fclid_if_missing(self) -> Optional[FindCaseLawIdentifier]:
+        """If the document does not have an FCLID already, mint a new one and save it."""
+        if len(self.identifiers.of_type(FindCaseLawIdentifier)) == 0:
+            document_fclid = FindCaseLawIdentifierSchema.mint(self.api_client)
+            self.identifiers.add(document_fclid)
+            self.save_identifiers()
+            return document_fclid
+
+        return None
+
     def publish(self) -> None:
         """
         :raises CannotPublishUnpublishableDocument: This document has not passed the checks in `is_publishable`, and as
@@ -386,11 +396,8 @@ 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()
+        ## Make sure the document has an FCLID
+        self.assign_fclid_if_missing()
 
         publish_documents(self.uri)
         self.api_client.set_published(self.uri, True)

ds_caselaw_marklogic_api_client-37.1.0/src/caselawclient/xquery/get_missing_fclid.xqy

@@ -0,0 +1,25 @@
+xquery version "1.0-ml";
+
+declare variable $maximum_records as xs:int? external := 1000;
+
+for $doc in fn:subsequence(
+  cts:search(
+    fn:collection(),
+    cts:and-query(
+      (
+        cts:properties-fragment-query(
+          cts:element-value-query(xs:QName("published"), "true")
+        ),
+        cts:not-query(
+          cts:properties-fragment-query(
+            cts:element-value-query(xs:QName("namespace"), "fclid")
+          )
+        )
+      )
+    )
+  ),
+  1,
+  $maximum_records
+)
+
+return fn:base-uri($doc)

{ds_caselaw_marklogic_api_client-37.0.2 → ds_caselaw_marklogic_api_client-37.1.0}/src/caselawclient/xquery_type_dicts.py

@@ -86,6 +86,11 @@ class GetLastModifiedDict(MarkLogicAPIDict):
     uri: MarkLogicDocumentURIString
 
 
+# get_missing_fclid.xqy
+class GetMissingFclidDict(MarkLogicAPIDict):
+    maximum_records: Optional[int]
+
+
 # get_pending_enrichment_for_version.xqy
 class GetPendingEnrichmentForVersionDict(MarkLogicAPIDict):
     maximum_records: Optional[int]