ds-caselaw-marklogic-api-client 30.0.0__tar.gz → 31.0.1__tar.gz

{ds_caselaw_marklogic_api_client-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ds-caselaw-marklogic-api-client
-Version: 30.0.0
+Version: 31.0.1
 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 (>=2024.12.14,<2024.13.0)
+Requires-Dist: certifi (>=2025.1.31,<2025.2.0)
 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)
@@ -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-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ds-caselaw-marklogic-api-client"
-version = "30.0.0"
+version = "31.0.1"
 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 = ">=2024.12.14,<2024.13.0"
+certifi = ">=2025.1.31,<2025.2.0"
 charset-normalizer = "^3.0.0"
 django-environ = ">=0.12.0"
 idna = "^3.4"
@@ -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-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/src/caselawclient/Client.py

@@ -1222,7 +1222,7 @@ class MarklogicApiClient:
         warnings.warn(
             "resolve_from_identifier deprecated, use resolve_from_identifier_slug instead", DeprecationWarning
         )
-        return self.resolve_from_identifier(identifier_uri, published_only)
+        return self.resolve_from_identifier_slug(identifier_uri, published_only)
 
     def resolve_from_identifier_value(
         self, identifier_value: str, published_only: bool = True

{ds_caselaw_marklogic_api_client-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/src/caselawclient/factories.py

@@ -1,10 +1,12 @@
 import datetime
+import json
 from typing import Any, Optional
 from unittest.mock import Mock
 
 from typing_extensions import TypeAlias
 
 from caselawclient.Client import MarklogicApiClient
+from caselawclient.identifier_resolution import IdentifierResolution, IdentifierResolutions
 from caselawclient.models.documents import Document
 from caselawclient.models.documents.body import DocumentBody
 from caselawclient.models.judgments import Judgment
@@ -12,7 +14,16 @@ from caselawclient.models.press_summaries import PressSummary
 from caselawclient.responses.search_result import SearchResult, SearchResultMetadata
 from caselawclient.types import DocumentURIString
 
-DEFAULT_DOCUMENT_BODY_XML = "<akomantoso>This is some XML of a judgment.</akomantoso>"
+DEFAULT_DOCUMENT_BODY_XML = """<akomaNtoso xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0" xmlns:uk="https://caselaw.nationalarchives.gov.uk/akn">
+            <judgment name="decision">
+                <meta/><header/>
+                <judgmentBody>
+                <decision>
+                <p>This is a document.</p>
+                </decision>
+                </judgmentBody>
+            </judgment>
+            </akomaNtoso>"""
 
 
 class DocumentBodyFactory:
@@ -56,7 +67,6 @@ class DocumentFactory:
     def build(
         cls,
         uri: DocumentURIString = DocumentURIString("test/2023/123"),
-        html: str = "<p>This is a judgment.</p>",
         api_client: Optional[MarklogicApiClient] = None,
         **kwargs: Any,
     ) -> target_class:
@@ -66,7 +76,6 @@ class DocumentFactory:
             api_client.get_property_as_node.return_value = None
 
         document = cls.target_class(uri, api_client=api_client)
-        document.content_as_html = Mock(return_value=html)  # type: ignore[method-assign]
         document.body = kwargs.pop("body") if "body" in kwargs else DocumentBodyFactory.build()
 
         for param_name, default_value in cls.PARAMS_MAP.items():
@@ -133,3 +142,33 @@ class SearchResultFactory(SimpleFactory):
         "metadata": SearchResultMetadataFactory.build(),
         "is_failure": False,
     }
+
+
+class IdentifierResolutionFactory:
+    @classmethod
+    def build(
+        self,
+        resolution_uuid: Optional[str] = None,
+        document_uri: Optional[str] = None,
+        identifier_slug: Optional[str] = None,
+        published: Optional[bool] = True,
+        namespace: Optional[str] = None,
+        value: Optional[str] = None,
+    ) -> IdentifierResolution:
+        raw_resolution = {
+            "documents.compiled_url_slugs.identifier_uuid": resolution_uuid or "24b9a384-8bcf-4f20-996a-5c318f8dc657",
+            "documents.compiled_url_slugs.document_uri": document_uri or "/ewca/civ/2003/547.xml",
+            "documents.compiled_url_slugs.identifier_slug": identifier_slug or "ewca/civ/2003/54721",
+            "documents.compiled_url_slugs.document_published": "true" if published else "false",
+            "documents.compiled_url_slugs.identifier_namespace": namespace or "ukncn",
+            "documents.compiled_url_slugs.identifier_value": value or "[2003] EWCA 54721 (Civ)",
+        }
+        return IdentifierResolution.from_marklogic_output(json.dumps(raw_resolution))
+
+
+class IdentifierResolutionsFactory:
+    @classmethod
+    def build(self, resolutions: Optional[list[IdentifierResolution]] = None) -> IdentifierResolutions:
+        if resolutions is None:
+            resolutions = [IdentifierResolutionFactory.build()]
+        return IdentifierResolutions(resolutions)

{ds_caselaw_marklogic_api_client-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/src/caselawclient/models/documents/__init__.py

@@ -6,15 +6,14 @@ from typing import TYPE_CHECKING, Any, Optional
 from ds_caselaw_utils import courts
 from ds_caselaw_utils.courts import CourtNotFoundException
 from ds_caselaw_utils.types import NeutralCitationString
-from lxml import html as html_parser
 from requests_toolbelt.multipart import decoder
 
 from caselawclient.errors import (
     DocumentNotFoundError,
-    GatewayTimeoutError,
     NotSupportedOnVersion,
     OnlySupportedOnVersion,
 )
+from caselawclient.identifier_resolution import IdentifierResolutions
 from caselawclient.models.identifiers import Identifier
 from caselawclient.models.identifiers.fclid import FindCaseLawIdentifier, FindCaseLawIdentifierSchema
 from caselawclient.models.identifiers.unpacker import unpack_all_identifiers_from_etree
@@ -252,39 +251,6 @@ class Document:
         "Is this document a potentially historic version of a document, or is it the main document itself?"
         return extract_version(self.uri) != 0
 
-    def content_as_html(
-        self,
-        version_uri: Optional[DocumentURIString] = None,
-        query: Optional[str] = None,
-    ) -> str:
-        try:
-            results = self.api_client.eval_xslt(
-                self.uri,
-                version_uri,
-                show_unpublished=True,
-                query=query,
-            )
-            multipart_data = decoder.MultipartDecoder.from_response(results)
-            return str(multipart_data.parts[0].text)
-        except GatewayTimeoutError as e:
-            if query is not None:
-                warnings.warn(
-                    (
-                        "Gateway timeout when getting content with query"
-                        "highlighting for document %s, version %s, and query"
-                        '"%s", falling back to unhighlighted content...'
-                    )
-                    % (self.uri, version_uri, query),
-                    GatewayTimeoutGettingHTMLWithQuery,
-                )
-                return self.content_as_html(version_uri)
-            raise e
-
-    def number_of_mentions(self, query: str) -> int:
-        html = self.content_as_html(query=query)
-        tree = html_parser.fromstring(html.encode("utf-8"))
-        return len(tree.findall(".//mark"))
-
     @cached_property
     def is_failure(self) -> bool:
         """
@@ -529,3 +495,21 @@ class Document:
             return getattr(self.body, name)
         except Exception:
             raise AttributeError(f"Neither 'Document' nor 'DocumentBody' objects have an attribute '{name}'")
+
+    def linked_document_resolutions(self, namespaces: list[str], only_published: bool = True) -> IdentifierResolutions:
+        """Get documents which share the same neutral citation as this document."""
+        if not hasattr(self, "neutral_citation") or not self.neutral_citation:
+            return IdentifierResolutions([])
+
+        resolutions = self.api_client.resolve_from_identifier_value(self.neutral_citation)
+        if only_published:
+            resolutions = resolutions.published()
+
+        # only documents which aren't this one and have a right namespace
+        return IdentifierResolutions(
+            [
+                resolution
+                for resolution in resolutions
+                if resolution.document_uri != self.uri.as_marklogic() and resolution.identifier_namespace in namespaces
+            ]
+        )

{ds_caselaw_marklogic_api_client-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/src/caselawclient/models/documents/body.py

@@ -3,7 +3,6 @@ import os
 import warnings
 from functools import cache, cached_property
 from typing import Optional
-from xml.etree.ElementTree import Element
 
 import pytz
 from ds_caselaw_utils.types import CourtCode
@@ -128,13 +127,14 @@ class DocumentBody:
     def has_content(self) -> bool:
         """If we do not have a word document, the XML will not contain
         the contents of the judgment, but will contain a preamble."""
-
-        def stripped_tag_text(tag: Element) -> str:
-            return "".join(tag.itertext()).strip()
-
-        header = self._xml.xml_as_tree.xpath("//akn:header", namespaces=DEFAULT_NAMESPACES)[0]
-        content = self._xml.xml_as_tree.xpath("//akn:judgmentBody", namespaces=DEFAULT_NAMESPACES)[0]
-        return not (stripped_tag_text(header) == "" and stripped_tag_text(content) == "")
+        trailing_tags = self._xml.xml_as_tree.xpath("//*[preceding::akn:meta]", namespaces=DEFAULT_NAMESPACES)
+        for tag in trailing_tags:
+            if tag.tail and tag.tail.strip():
+                return True
+            if tag.text and tag.text.strip():
+                return True
+
+        return False
 
     @cache
     def content_as_html(self, image_base_url: Optional[str] = None) -> Optional[str]:

{ds_caselaw_marklogic_api_client-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/src/caselawclient/models/judgments.py

@@ -5,6 +5,7 @@ from typing import TYPE_CHECKING, Any, Optional
 from ds_caselaw_utils.types import NeutralCitationString
 
 from caselawclient.errors import DocumentNotFoundError
+from caselawclient.identifier_resolution import IdentifierResolutions
 from caselawclient.models.neutral_citation_mixin import NeutralCitationMixin
 
 if TYPE_CHECKING:
@@ -51,3 +52,7 @@ class Judgment(NeutralCitationMixin, Document):
             return PressSummary(uri, self.api_client)
         except DocumentNotFoundError:
             return None
+
+    @cached_property
+    def linked_press_summaries(self, only_published: bool = True) -> "IdentifierResolutions":
+        return self.linked_document_resolutions(["uksummaryofncn"], only_published)

{ds_caselaw_marklogic_api_client-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/src/caselawclient/models/press_summaries.py

@@ -7,6 +7,7 @@ from typing import TYPE_CHECKING, Any, Optional
 from ds_caselaw_utils.types import NeutralCitationString
 
 from caselawclient.errors import DocumentNotFoundError
+from caselawclient.identifier_resolution import IdentifierResolutions
 from caselawclient.models.neutral_citation_mixin import NeutralCitationMixin
 from caselawclient.types import DocumentURIString
 
@@ -51,3 +52,6 @@ class PressSummary(NeutralCitationMixin, Document):
             return Judgment(uri, self.api_client)
         except DocumentNotFoundError:
             return None
+
+    def linked_judgments(self, only_published: bool = True) -> "IdentifierResolutions":
+        return self.linked_document_resolutions(["ukncn"], only_published)

{ds_caselaw_marklogic_api_client-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/src/caselawclient/types.py

@@ -2,6 +2,10 @@ class InvalidDocumentURIException(Exception):
     """The document URI is not valid."""
 
 
+class MarkLogicDocumentURIString(str):
+    pass
+
+
 class DocumentURIString(str):
     """
     This class checks that the string is actually a valid Document URI on creation. It does _not_ manipulate the string.
@@ -22,3 +26,6 @@ class DocumentURIString(str):
 
         # If everything is good, return as usual
         return str.__new__(cls, content)
+
+    def as_marklogic(self) -> MarkLogicDocumentURIString:
+        return MarkLogicDocumentURIString(f"/{self}.xml")

{ds_caselaw_marklogic_api_client-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/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-30.0.0 → ds_caselaw_marklogic_api_client-31.0.1}/src/caselawclient/xquery_type_dicts.py

@@ -8,8 +8,8 @@ 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 MarkLogicDocumentURIString as MarkLogicDocumentURIString
 
-MarkLogicDocumentURIString = NewType("MarkLogicDocumentURIString", str)
 MarkLogicDocumentVersionURIString = NewType("MarkLogicDocumentVersionURIString", MarkLogicDocumentURIString)
 
 MarkLogicPrivilegeURIString = NewType("MarkLogicPrivilegeURIString", str)