Resolve identifier URIs to MarkLogic URIs

script/build_xquery_type_dicts

@@ -54,7 +54,7 @@ def ml_type_to_python_type_declaration(variable_name: str, variable_type: str):
         variable_type = "MarkLogicDocumentVersionURIString"
     elif variable_name == "privilege_uri":
         variable_type = "MarkLogicPrivilegeURIString"
-    elif variable_name == "parent_uri":
+    elif variable_name in ["identifier_uri", "parent_uri"]:
         variable_type = "DocumentURIString"
     elif variable_name == "uri" or variable_name.endswith("_uri"):
         variable_type = "MarkLogicDocumentURIString"

src/caselawclient/Client.py

@@ -20,6 +20,7 @@
 
 from caselawclient import xquery_type_dicts as query_dicts
 from caselawclient.client_helpers import VersionAnnotation
+from caselawclient.identifiers import IdentifierResolutions
 from caselawclient.models.documents import (
     DOCUMENT_COLLECTION_URI_JUDGMENT,
     DOCUMENT_COLLECTION_URI_PRESS_SUMMARY,
@@ -1201,3 +1202,15 @@ def get_recently_parsed(
         )
 
         return results
+
+    def resolve_from_identifier(self, identifier_uri: str) -> IdentifierResolutions:
+        """Given a PUI/EUI url, look up the precomputed slug and return the
+        MarkLogic document URIs which match that slug. Multiple returns should be anticipated"""
+        vars: query_dicts.ResolveFromIdentifierDict = {"identifier_uri": DocumentURIString(identifier_uri)}
+        raw_results: list[str] = get_multipart_strings_from_marklogic_response(
+            self._send_to_eval(
+                vars,
+                "resolve_from_identifier.xqy",
+            ),
+        )
+        return IdentifierResolutions.from_marklogic_output(raw_results)

src/caselawclient/identifiers.py

@@ -0,0 +1,31 @@
+import json
+from typing import NamedTuple
+
+from caselawclient.models.documents import DocumentURIString
+from caselawclient.xquery_type_dicts import MarkLogicDocumentURIString
+
+
+class IdentifierResolution(NamedTuple):
+    identifier_uuid: str
+    document_uri: MarkLogicDocumentURIString
+    identifier_slug: DocumentURIString
+    document_published: bool
+
+    @staticmethod
+    def from_marklogic_output(raw_row: str) -> "IdentifierResolution":
+        row = json.loads(raw_row)
+        return IdentifierResolution(
+            identifier_uuid=row["documents.compiled_url_slugs.identifier_uuid"],
+            document_uri=MarkLogicDocumentURIString(row["documents.compiled_url_slugs.document_uri"]),
+            identifier_slug=DocumentURIString(row["documents.compiled_url_slugs.identifier_slug"]),
+            document_published=row["documents.compiled_url_slugs.document_published"] == "true",
+        )
+
+
+class IdentifierResolutions(list[IdentifierResolution]):
+    @staticmethod
+    def from_marklogic_output(table: list[str]) -> "IdentifierResolutions":
+        return IdentifierResolutions(list(IdentifierResolution.from_marklogic_output(row) for row in table))
+
+    def published(self) -> "IdentifierResolutions":
+        return IdentifierResolutions(list(x for x in self if x.document_published))

src/caselawclient/xquery/resolve_from_identifier.xqy

@@ -1,12 +1,12 @@
 xquery version "1.0-ml";
 
 declare namespace xdmp="http://marklogic.com/xdmp"; 
-declare variable $uri as xs:string external;
+declare variable $identifier_uri as xs:string external;
 
 xdmp:sql(
   "SELECT * from compiled_url_slugs WHERE documents.compiled_url_slugs.identifier_slug = @uri",
-  "array",
+  "map",
   map:new((
-    map:entry("uri", $uri)
+    map:entry("uri", $identifier_uri)
   ))
 )

src/caselawclient/xquery_type_dicts.py

@@ -141,6 +141,11 @@ class ListJudgmentVersionsDict(MarkLogicAPIDict):
     uri: MarkLogicDocumentURIString
 
 
+# resolve_from_identifier.xqy
+class ResolveFromIdentifierDict(MarkLogicAPIDict):
+    identifier_uri: DocumentURIString
+
+
 # set_boolean_property.xqy
 class SetBooleanPropertyDict(MarkLogicAPIDict):
     name: str