ds-caselaw-marklogic-api-client 27.0.1__py3-none-any.whl → 27.2.0__py3-none-any.whl

caselawclient/Client.py

@@ -403,6 +403,7 @@ class MarklogicApiClient:
         judgment_uri: DocumentURIString,
         version_uri: Optional[DocumentURIString] = None,
         show_unpublished: bool = False,
+        search_query: Optional[str] = None,
     ) -> bytes:
         marklogic_document_uri = self._format_uri_for_marklogic(judgment_uri)
         marklogic_document_version_uri = (
@@ -418,6 +419,7 @@ class MarklogicApiClient:
             "uri": marklogic_document_uri,
             "version_uri": marklogic_document_version_uri,
             "show_unpublished": show_unpublished,
+            "search_query": search_query,
         }
 
         response = self._eval_as_bytes(vars, "get_judgment.xqy")
@@ -433,11 +435,13 @@ class MarklogicApiClient:
         judgment_uri: DocumentURIString,
         version_uri: Optional[DocumentURIString] = None,
         show_unpublished: bool = False,
+        search_query: Optional[str] = None,
     ) -> str:
         return self.get_judgment_xml_bytestring(
             judgment_uri,
             version_uri,
             show_unpublished,
+            search_query=search_query,
         ).decode(encoding="utf-8")
 
     def set_document_name(
@@ -634,12 +638,14 @@ 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:

caselawclient/factories.py

@@ -1,5 +1,5 @@
 import datetime
-from typing import Any, Optional, cast
+from typing import Any, Optional
 from unittest.mock import Mock
 
 from typing_extensions import TypeAlias
@@ -75,55 +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)
@@ -145,6 +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",
     }
 
 

caselawclient/models/documents/__init__.py

@@ -105,7 +105,7 @@ class Document:
     Individual document classes should extend this list where necessary to validate document type-specific attributes.
     """
 
-    def __init__(self, uri: str, api_client: "MarklogicApiClient"):
+    def __init__(self, uri: str, api_client: "MarklogicApiClient", search_query: Optional[str] = None):
         """
         :param uri: For historical reasons this accepts a pseudo-URI which may include leading or trailing slashes.
 
@@ -117,7 +117,11 @@ class Document:
             raise DocumentNotFoundError(f"Document {self.uri} does not exist")
 
         self.body: DocumentBody = DocumentBody(
-            xml_bytestring=self.api_client.get_judgment_xml_bytestring(self.uri, show_unpublished=True),
+            xml_bytestring=self.api_client.get_judgment_xml_bytestring(
+                self.uri,
+                show_unpublished=True,
+                search_query=search_query,
+            ),
         )
         """ `Document.body` represents the XML of the document itself, without any information such as version tracking or properties. """
 
@@ -495,3 +499,10 @@ 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}'")

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:
         """