ds-caselaw-marklogic-api-client 31.0.0__tar.gz → 31.1.0__tar.gz

{ds_caselaw_marklogic_api_client-31.0.0 → ds_caselaw_marklogic_api_client-31.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ds-caselaw-marklogic-api-client
-Version: 31.0.0
+Version: 31.1.0
 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
@@ -21,7 +21,7 @@ Requires-Dist: memoization (>=0.4.0,<0.5.0)
 Requires-Dist: mypy-boto3-s3 (>=1.26.104,<2.0.0)
 Requires-Dist: mypy-boto3-sns (>=1.26.69,<2.0.0)
 Requires-Dist: python-dateutil (>=2.9.0-post.0,<3.0.0)
-Requires-Dist: pytz (>=2024.1,<2025.0)
+Requires-Dist: pytz (>2024)
 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)

{ds_caselaw_marklogic_api_client-31.0.0 → ds_caselaw_marklogic_api_client-31.1.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ds-caselaw-marklogic-api-client"
-version = "31.0.0"
+version = "31.1.0"
 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"
@@ -25,7 +25,7 @@ boto3 = "^1.26.112"
 typing-extensions = "^4.7.1"
 mypy-boto3-s3 = "^1.26.104"
 mypy-boto3-sns = "^1.26.69"
-pytz = "^2024.1"
+pytz = ">2024"
 python-dateutil = "^2.9.0-post.0"
 saxonche = "^12.5.0"
 sqids = "^0.5.0"

{ds_caselaw_marklogic_api_client-31.0.0 → ds_caselaw_marklogic_api_client-31.1.0}/src/caselawclient/Client.py

@@ -30,7 +30,7 @@ from caselawclient.models.judgments import Judgment
 from caselawclient.models.press_summaries import PressSummary
 from caselawclient.models.utilities import move
 from caselawclient.search_parameters import SearchParameters
-from caselawclient.types import DocumentURIString
+from caselawclient.types import DocumentIdentifierSlug, DocumentIdentifierValue, DocumentURIString
 from caselawclient.xquery_type_dicts import (
     MarkLogicDocumentURIString,
     MarkLogicDocumentVersionURIString,
@@ -1203,11 +1203,13 @@ class MarklogicApiClient:
 
         return results
 
-    def resolve_from_identifier_slug(self, identifier_uri: str, published_only: bool = True) -> IdentifierResolutions:
+    def resolve_from_identifier_slug(
+        self, identifier_slug: DocumentIdentifierSlug, published_only: bool = True
+    ) -> 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.ResolveFromIdentifierSlugDict = {
-            "identifier_uri": DocumentURIString(identifier_uri),
+            "identifier_slug": identifier_slug,
             "published_only": int(published_only),
         }
         raw_results: list[str] = get_multipart_strings_from_marklogic_response(
@@ -1218,14 +1220,8 @@ class MarklogicApiClient:
         )
         return IdentifierResolutions.from_marklogic_output(raw_results)
 
-    def resolve_from_identifier(self, identifier_uri: str, published_only: bool = True) -> IdentifierResolutions:
-        warnings.warn(
-            "resolve_from_identifier deprecated, use resolve_from_identifier_slug instead", DeprecationWarning
-        )
-        return self.resolve_from_identifier(identifier_uri, published_only)
-
     def resolve_from_identifier_value(
-        self, identifier_value: str, published_only: bool = True
+        self, identifier_value: DocumentIdentifierValue, published_only: bool = True
     ) -> 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"""

{ds_caselaw_marklogic_api_client-31.0.0 → ds_caselaw_marklogic_api_client-31.1.0}/src/caselawclient/identifier_resolution.py

@@ -3,7 +3,7 @@ from typing import NamedTuple
 
 from caselawclient.models.identifiers import Identifier
 from caselawclient.models.identifiers.unpacker import IDENTIFIER_NAMESPACE_MAP
-from caselawclient.types import DocumentURIString
+from caselawclient.types import DocumentIdentifierSlug, DocumentIdentifierValue
 from caselawclient.xquery_type_dicts import MarkLogicDocumentURIString
 
 
@@ -31,9 +31,9 @@ class IdentifierResolution(NamedTuple):
 
     identifier_uuid: str
     document_uri: MarkLogicDocumentURIString
-    identifier_slug: DocumentURIString
+    identifier_slug: DocumentIdentifierSlug
     document_published: bool
-    identifier_value: str
+    identifier_value: DocumentIdentifierValue
     identifier_namespace: str
     identifier_type: type[Identifier]
 
@@ -44,9 +44,9 @@ class IdentifierResolution(NamedTuple):
         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"]),
+            identifier_slug=DocumentIdentifierSlug(row["documents.compiled_url_slugs.identifier_slug"]),
             document_published=row["documents.compiled_url_slugs.document_published"] == "true",
-            identifier_value=row["documents.compiled_url_slugs.identifier_value"],
+            identifier_value=DocumentIdentifierValue(row["documents.compiled_url_slugs.identifier_value"]),
             identifier_namespace=identifier_namespace,
             identifier_type=IDENTIFIER_NAMESPACE_MAP[identifier_namespace],
         )

{ds_caselaw_marklogic_api_client-31.0.0 → ds_caselaw_marklogic_api_client-31.1.0}/src/caselawclient/types.py

@@ -2,8 +2,29 @@ class InvalidDocumentURIException(Exception):
     """The document URI is not valid."""
 
 
+class InvalidMarkLogicDocumentURIException(Exception):
+    """The MarkLogic document URI is not valid."""
+
+
 class MarkLogicDocumentURIString(str):
-    pass
+    def __new__(cls, content: str) -> "MarkLogicDocumentURIString":
+        # Check that the URI begins with a slash
+        if content[0] != "/":
+            raise InvalidMarkLogicDocumentURIException(
+                f'"{content}" is not a valid MarkLogic document URI; URIs must begin with a slash.'
+            )
+
+        # Check that the URI ends with ".xml"
+        if not content.endswith(".xml"):
+            raise InvalidMarkLogicDocumentURIException(
+                f'"{content}" is not a valid MarkLogic document URI; URIs must end with ".xml". '
+            )
+
+        # If everything is good, return as usual
+        return str.__new__(cls, content)
+
+    def as_document_uri(self) -> "DocumentURIString":
+        return DocumentURIString(self.lstrip("/").rstrip(".xml"))
 
 
 class DocumentURIString(str):
@@ -29,3 +50,11 @@ class DocumentURIString(str):
 
     def as_marklogic(self) -> MarkLogicDocumentURIString:
         return MarkLogicDocumentURIString(f"/{self}.xml")
+
+
+class DocumentIdentifierSlug(str):
+    pass
+
+
+class DocumentIdentifierValue(str):
+    pass

{ds_caselaw_marklogic_api_client-31.0.0 → ds_caselaw_marklogic_api_client-31.1.0}/src/caselawclient/xquery/resolve_from_identifier_slug.xqy

@@ -1,7 +1,7 @@
 xquery version "1.0-ml";
 
 declare namespace xdmp="http://marklogic.com/xdmp"; 
-declare variable $identifier_uri as xs:string external;
+declare variable $identifier_slug as xs:string external;
 declare variable $published_only as xs:int? external := 1;
 
 let $published_query := if ($published_only) then " AND document_published = 'true'" else ""
@@ -11,7 +11,6 @@ return xdmp:sql(
   $query,
   "map",
   map:new((
-    map:entry("uri", $identifier_uri)
+    map:entry("uri", $identifier_slug)
   ))
 )
-

{ds_caselaw_marklogic_api_client-31.0.0 → ds_caselaw_marklogic_api_client-31.1.0}/src/caselawclient/xquery/xslt_transform.xqy

@@ -28,7 +28,7 @@ let $number_marks_xslt := (
     </xsl:template>
     <xsl:template match="mark">
       <xsl:copy>
-          <xsl:copy-of select="@*" />
+          <xsl:copy-of select="*" />
           <xsl:attribute name="id">
               <xsl:text>mark_</xsl:text>
               <xsl:value-of select="count(preceding::mark)"/>

{ds_caselaw_marklogic_api_client-31.0.0 → ds_caselaw_marklogic_api_client-31.1.0}/src/caselawclient/xquery_type_dicts.py

@@ -7,7 +7,7 @@ checks. They are used to enforce appropriately typed variables being passed in t
 """
 
 from typing import Any, NewType, Optional, TypedDict
-from caselawclient.types import DocumentURIString
+from caselawclient.types import DocumentURIString, DocumentIdentifierSlug, DocumentIdentifierValue
 from caselawclient.types import MarkLogicDocumentURIString as MarkLogicDocumentURIString
 
 MarkLogicDocumentVersionURIString = NewType("MarkLogicDocumentVersionURIString", MarkLogicDocumentURIString)
@@ -143,13 +143,13 @@ class ListJudgmentVersionsDict(MarkLogicAPIDict):
 
 # resolve_from_identifier_slug.xqy
 class ResolveFromIdentifierSlugDict(MarkLogicAPIDict):
-    identifier_uri: DocumentURIString
+    identifier_slug: DocumentIdentifierSlug
     published_only: Optional[int]
 
 
 # resolve_from_identifier_value.xqy
 class ResolveFromIdentifierValueDict(MarkLogicAPIDict):
-    identifier_value: str
+    identifier_value: DocumentIdentifierValue
     published_only: Optional[int]