ds-caselaw-marklogic-api-client 27.0.0__tar.gz → 27.1.0__tar.gz

{ds_caselaw_marklogic_api_client-27.0.0 → ds_caselaw_marklogic_api_client-27.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ds-caselaw-marklogic-api-client
-Version: 27.0.0
+Version: 27.1.0
 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
@@ -24,6 +24,7 @@ Requires-Dist: python-dateutil (>=2.9.0-post.0,<3.0.0)
 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: typing-extensions (>=4.7.1,<5.0.0)
 Description-Content-Type: text/markdown
 

{ds_caselaw_marklogic_api_client-27.0.0 → ds_caselaw_marklogic_api_client-27.1.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ds-caselaw-marklogic-api-client"
-version = "27.0.0"
+version = "27.1.0"
 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"
@@ -27,10 +27,12 @@ mypy-boto3-s3 = "^1.26.104"
 mypy-boto3-sns = "^1.26.69"
 pytz = "^2024.1"
 python-dateutil = "^2.9.0-post.0"
+saxonche = "^12.5.0"
 
 [tool.poetry.group.dev.dependencies]
 coverage = "^7.2.3"
 pytest = "^8.0.0"
+beautifulsoup4 = "^4.12.3"
 responses = "^0.25.0"
 python-dotenv = "^1.0.0"
 time-machine = "^2.13.0"

{ds_caselaw_marklogic_api_client-27.0.0 → ds_caselaw_marklogic_api_client-27.1.0}/src/caselawclient/Client.py

@@ -634,12 +634,14 @@ class MarklogicApiClient:
         judgment_uri: DocumentURIString,
         annotation: str = "",
         expires_at_midnight: bool = False,
+        timeout_seconds: int = -1,
     ) -> requests.Response:
+        """If timeout_seconds is -1, the lock never times out"""
         uri = self._format_uri_for_marklogic(judgment_uri)
         vars: query_dicts.CheckoutJudgmentDict = {
             "uri": uri,
             "annotation": annotation,
-            "timeout": -1,
+            "timeout": timeout_seconds,
         }
 
         if expires_at_midnight:

{ds_caselaw_marklogic_api_client-27.0.0 → ds_caselaw_marklogic_api_client-27.1.0}/src/caselawclient/factories.py

@@ -1,6 +1,6 @@
 import datetime
-from typing import Any, Optional, cast
-from unittest.mock import Mock, patch
+from typing import Any, Optional
+from unittest.mock import Mock
 
 from typing_extensions import TypeAlias
 
@@ -63,10 +63,8 @@ class DocumentFactory:
             api_client = Mock(spec=MarklogicApiClient)
             api_client.get_judgment_xml_bytestring.return_value = DEFAULT_DOCUMENT_BODY_XML.encode(encoding="utf-8")
 
-        with patch.object(cls.target_class, "content_as_html") as mock_content_as_html:
-            mock_content_as_html.return_value = html
-            document = cls.target_class(uri, api_client=api_client)
-
+        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():
@@ -77,55 +75,24 @@ class DocumentFactory:
 
 
 class JudgmentFactory(DocumentFactory):
-    target_class = Judgment
-
-    def __init__(self) -> None:
-        self.PARAMS_MAP = self.PARAMS_MAP | {
-            "neutral_citation": "[2023] Test 123",
-            "best_human_identifier": "[2023] Test 123",
-        }
-
-        super().__init__()
-
-    @classmethod
-    def build(
-        cls,
-        uri: str = "test/2023/123",
-        html: str = "<p>This is a judgment.</p>",
-        api_client: Optional[MarklogicApiClient] = None,
-        **kwargs: Any,
-    ) -> Judgment:
-        return cast(Judgment, super().build(uri, html, api_client, **kwargs))
+    target_class: TypeAlias = Judgment
+    PARAMS_MAP = DocumentFactory.PARAMS_MAP | {
+        "neutral_citation": "[2023] Test 123",
+    }
 
 
 class PressSummaryFactory(DocumentFactory):
-    target_class = PressSummary
-
-    def __init__(self) -> None:
-        self.PARAMS_MAP = self.PARAMS_MAP | {
-            "neutral_citation": "[2023] Test 123",
-            "best_human_identifier": "[2023] Test 123",
-        }
-
-        super().__init__()
-
-    @classmethod
-    def build(
-        cls,
-        uri: str = "test/2023/123/press-summary/1",
-        html: str = "<p>This is a judgment.</p>",
-        api_client: Optional[MarklogicApiClient] = None,
-        **kwargs: Any,
-    ) -> PressSummary:
-        return cast(PressSummary, super().build(uri, html, api_client, **kwargs))
+    target_class: TypeAlias = PressSummary
+    PARAMS_MAP = DocumentFactory.PARAMS_MAP | {
+        "neutral_citation": "[2023] Test 123",
+    }
 
 
 class SimpleFactory:
+    target_class: TypeAlias = object
     # "name_of_attribute": "default value"
     PARAMS_MAP: dict[str, Any]
 
-    target_class: TypeAlias = object
-
     @classmethod
     def build(cls, **kwargs: Any) -> target_class:
         mock_object = Mock(spec=cls.target_class, autospec=True)
@@ -147,6 +114,7 @@ class SearchResultMetadataFactory(SimpleFactory):
         "author_email": "fake.email@gov.invalid",
         "consignment_reference": "TDR-2023-ABC",
         "submission_datetime": datetime.datetime(2023, 2, 3, 9, 12, 34),
+        "editor_status": "New",
     }
 
 

{ds_caselaw_marklogic_api_client-27.0.0 → ds_caselaw_marklogic_api_client-27.1.0}/src/caselawclient/models/documents/__init__.py

@@ -495,3 +495,10 @@ class Document:
         Is it sensible to reparse this document?
         """
         return self.docx_exists()
+
+    def __getattr__(self, name: str) -> Any:
+        warnings.warn(f"{name} no longer exists on Document, using Document.body instead", DeprecationWarning)
+        try:
+            return getattr(self.body, name)
+        except Exception:
+            raise AttributeError(f"Neither 'Document' nor 'DocumentBody' objects have an attribute '{name}'")

{ds_caselaw_marklogic_api_client-27.0.0 → ds_caselaw_marklogic_api_client-27.1.0}/src/caselawclient/models/documents/body.py

@@ -1,10 +1,12 @@
 import datetime
+import os
 import warnings
-from functools import cached_property
+from functools import cache, cached_property
 from typing import Optional
 
 import pytz
 from ds_caselaw_utils.types import CourtCode
+from saxonche import PySaxonProcessor
 
 from caselawclient.models.utilities.dates import parse_string_date_as_utc
 
@@ -129,6 +131,23 @@ class DocumentBody:
     def content_as_xml(self) -> str:
         return self._xml.xml_as_string
 
+    @cache
+    def content_as_html(self, image_base_url: Optional[str] = None) -> str:
+        """Convert the XML representation of the Document into HTML for rendering."""
+
+        html_xslt_location = os.path.join(os.path.dirname(os.path.realpath(__file__)), "transforms", "html.xsl")
+
+        with PySaxonProcessor() as proc:
+            xslt_processor = proc.new_xslt30_processor()
+            document = proc.parse_xml(xml_text=self._xml.xml_as_string)
+
+            executable = xslt_processor.compile_stylesheet(stylesheet_file=html_xslt_location)
+
+            if image_base_url:
+                executable.set_parameter("image-base", proc.make_string_value(image_base_url))
+
+            return str(executable.transform_to_string(xdm_node=document))
+
     @cached_property
     def failed_to_parse(self) -> bool:
         """