ds-caselaw-marklogic-api-client 43.1.0__tar.gz → 44.4.1__tar.gz

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: ds-caselaw-marklogic-api-client
-Version: 43.1.0
+Version: 44.4.1
 Summary: An API client for interacting with the underlying data in Find Caselaw.
 Keywords: national archives,caselaw
 Author: The National Archives
@@ -9,11 +9,11 @@ Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Requires-Dist: boto3 (>=1.26.112,<2.0.0)
-Requires-Dist: certifi (>=2025.10.5,<2025.11.0)
+Requires-Dist: certifi (>=2025.11.12,<2025.12.0)
 Requires-Dist: charset-normalizer (>=3.0.0,<4.0.0)
 Requires-Dist: defusedxml (>=0.7.1,<0.8.0)
 Requires-Dist: django-environ (>=0.12.0)
-Requires-Dist: ds-caselaw-utils (>=2.0.0,<3.0.0)
+Requires-Dist: ds-caselaw-utils (>=2.10.0,<3.0.0)
 Requires-Dist: idna (>=3.4,<4.0)
 Requires-Dist: lxml (>=6.0.0,<7.0.0)
 Requires-Dist: memoization (>=0.4.0,<0.5.0)

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "ds-caselaw-marklogic-api-client"
-version = "43.1.0"
+version = "44.4.1"
 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 = "^3.12.0"
-certifi = ">=2025.10.5,<2025.11.0"
+certifi = ">=2025.11.12,<2025.12.0"
 charset-normalizer = "^3.0.0"
 django-environ = ">=0.12.0"
 idna = "^3.4"
@@ -20,7 +20,7 @@ requests = "^2.28.2"
 requests-toolbelt = ">=0.10.1,<1.1.0"
 memoization = "^0.4.0"
 lxml = "^6.0.0"
-ds-caselaw-utils = "^2.0.0"
+ds-caselaw-utils = "^2.10.0"
 boto3 = "^1.26.112"
 typing-extensions = "^4.7.1"
 mypy-boto3-s3 = "^1.26.104"
@@ -33,14 +33,14 @@ defusedxml = "^0.7.1"
 pydantic = "^2.12.3"
 
 [tool.poetry.group.dev.dependencies]
-coverage = "7.11.0"
-pytest = "8.4.2"
+coverage = "7.13.1"
+pytest = "9.0.2"
 pytest-cov = "7.0.0"
-beautifulsoup4 = "4.14.2"
+beautifulsoup4 = "4.14.3"
 responses = "0.25.8"
 python-dotenv = "1.2.1"
-time-machine = "2.19.0"
-moto = {version = "5.1.16", extras = ["all"]}
+time-machine = "3.2.0"
+moto = {version = "5.1.19", extras = ["all"]}
 
 [tool.poetry.group.docs]
 optional = true

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/Client.py

@@ -32,7 +32,7 @@ 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 DocumentIdentifierSlug, DocumentIdentifierValue, DocumentURIString
+from caselawclient.types import DocumentIdentifierSlug, DocumentIdentifierValue, DocumentLock, DocumentURIString
 from caselawclient.xquery_type_dicts import (
     CheckContentHashUniqueByUriDict,
     MarkLogicDocumentURIString,
@@ -1222,13 +1222,13 @@ class MarklogicApiClient:
 
         return (int(table[1][1]), int(table[1][2]))
 
-    def get_pending_parse_for_version(
+    def get_documents_pending_parse_for_version(
         self,
         target_version: tuple[int, int],
         maximum_records: int = 1000,
     ) -> list[list[Any]]:
-        """Retrieve documents which are not yet parsed with a given version."""
-        vars: query_dicts.GetPendingParseForVersionDict = {
+        """Retrieve a list of documents which are not yet parsed with a given version."""
+        vars: query_dicts.GetPendingParseForVersionDocumentsDict = {
             "target_major_version": target_version[0],
             "target_minor_version": target_version[1],
             "maximum_records": maximum_records,
@@ -1237,13 +1237,33 @@ class MarklogicApiClient:
             get_single_string_from_marklogic_response(
                 self._send_to_eval(
                     vars,
-                    "get_pending_parse_for_version.xqy",
+                    "get_pending_parse_for_version_documents.xqy",
                 ),
             ),
         )
 
         return results
 
+    def get_count_pending_parse_for_version(
+        self,
+        target_version: tuple[int, int],
+    ) -> int:
+        """Get the total number of documents which are not yet parsed with a given version."""
+        vars: query_dicts.GetPendingParseForVersionCountDict = {
+            "target_major_version": target_version[0],
+            "target_minor_version": target_version[1],
+        }
+        results = json.loads(
+            get_single_string_from_marklogic_response(
+                self._send_to_eval(
+                    vars,
+                    "get_pending_parse_for_version_count.xqy",
+                ),
+            ),
+        )
+
+        return int(results[1][0])
+
     def get_recently_parsed(
         self,
     ) -> list[list[Any]]:
@@ -1259,6 +1279,19 @@ class MarklogicApiClient:
 
         return results
 
+    def get_locked_documents(
+        self,
+    ) -> list[DocumentLock]:
+        """Retrieve all currently locked documents."""
+        results = [
+            DocumentLock.from_string(lock)
+            for lock in get_multipart_strings_from_marklogic_response(
+                self._send_to_eval({}, "get_locked_documents.xqy")
+            )
+        ]
+
+        return sorted(results, key=lambda lock: lock.timestamp)
+
     def get_missing_fclid(
         self,
         maximum_records: int = 50,

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/factories.py

@@ -20,7 +20,7 @@ T = TypeVar("T")
 
 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/>
+                <meta/><header><p>Header contains text</p></header>
                 <judgmentBody>
                 <decision>
                 <p>This is a document.</p>

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/models/documents/__init__.py

@@ -26,6 +26,7 @@ from caselawclient.models.utilities import VersionsDict, extract_version, render
 from caselawclient.models.utilities.aws import (
     ParserInstructionsDict,
     announce_document_event,
+    are_unpublished_assets_clean,
     check_docx_exists,
     delete_documents_from_private_bucket,
     generate_docx_url,
@@ -102,6 +103,11 @@ class Document:
             True,
             "There is another document with identical content",
         ),
+        (
+            "has_only_clean_assets",
+            True,
+            "An uncleaned asset exists for this document",
+        ),
     ]
     """
     A list of tuples in the form:
@@ -252,7 +258,7 @@ class Document:
         """
         if self.is_version:
             raise NotSupportedOnVersion(
-                "Cannot get versions of a version for {self.uri}",
+                f"Cannot get versions of a version for {self.uri}",
             )
         docs = []
         for version in self.versions:
@@ -370,6 +376,12 @@ class Document:
         """Check if the content hash of this document is unique compared to all other documents in MarkLogic."""
         return self.api_client.has_unique_content_hash(self.uri)
 
+    @cached_property
+    def has_only_clean_assets(self) -> bool:
+        """False if any non-tar.gz assets associated with this document have not been cleaned."""
+        return True  # TODO: Remove this once we have enabled the asset cleaning pipeline.
+        return are_unpublished_assets_clean(self.uri)
+
     @cached_property
     def version_created_datetime(self) -> datetime.datetime:
         return self.api_client.get_version_created_datetime(self.uri)

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/models/documents/body.py

@@ -11,14 +11,10 @@ from saxonche import PySaxonProcessor
 
 from caselawclient.models.utilities.dates import parse_string_date_as_utc
 from caselawclient.types import DocumentCategory
+from caselawclient.xml_helpers import DEFAULT_NAMESPACES
 
 from .xml import XML
 
-DEFAULT_NAMESPACES = {
-    "uk": "https://caselaw.nationalarchives.gov.uk/akn",
-    "akn": "http://docs.oasis-open.org/legaldocml/ns/akn/3.0",
-}
-
 
 class UnparsableDate(Warning):
     pass
@@ -176,15 +172,12 @@ class DocumentBody:
     @cached_property
     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."""
-        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
+        the contents of the judgment, but will have content in the header if a judgment.
+        All press summaries (which have <doc> not <judgment> tags) are assumed to have content"""
+        return bool(
+            self._xml.xml_as_tree.xpath("//akn:header[normalize-space(string(.))]", namespaces=DEFAULT_NAMESPACES)
+            or self._xml.xml_as_tree.xpath("//akn:doc", namespaces=DEFAULT_NAMESPACES)
+        )
 
     @cached_property
     def has_external_data(self) -> bool:

ds_caselaw_marklogic_api_client-44.4.1/src/caselawclient/models/documents/stub.py

@@ -0,0 +1,50 @@
+from pathlib import Path
+from typing import Literal
+
+from ds_caselaw_utils.courts import courts
+from ds_caselaw_utils.types import CourtCode
+from jinja2 import StrictUndefined, Template
+from typing_extensions import TypedDict
+
+
+class PartyData(TypedDict):
+    role: Literal["Claimant", "Respondent", "Appellant", "Defendant"]
+    name: str
+
+
+class EditorStubData(TypedDict):
+    decision_date: str  # day precision
+    transform_datetime: str  # second precision
+    court_code: str
+    title: str
+    year: str
+    case_numbers: list[str]  # can be none
+    parties: list[PartyData]  # (type (claimant|defendant), name)
+
+
+class RendererStubData(EditorStubData):
+    court_url: str  # should be populated from utils/courts.cs
+    court_full_name: str  # ditto
+
+
+def add_other_stub_fields(editor_data: EditorStubData) -> RendererStubData:
+    court = courts.get_court_by_code(CourtCode(editor_data["court_code"].upper()))
+    return {
+        **editor_data,
+        "court_url": court.identifier_iri,
+        "court_full_name": court.long_name,
+    }
+
+
+def render_stub_xml(editor_data: EditorStubData) -> bytes:
+    render_data = add_other_stub_fields(editor_data)
+    from caselawclient.Client import ROOT_DIR
+
+    judgment_path = Path(ROOT_DIR) / "models" / "documents" / "templates" / "judgment.xml"
+
+    with (judgment_path).open("r") as f:
+        template = f.read()
+
+    rendered = bytes(Template(template, undefined=StrictUndefined).render(render_data).encode("utf-8"))
+
+    return rendered

ds_caselaw_marklogic_api_client-44.4.1/src/caselawclient/models/documents/templates/judgment.xml

@@ -0,0 +1,67 @@
+<?xml version="1.0" encoding="utf-8" ?>
+<akomaNtoso xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0" xmlns:uk="https://caselaw.nationalarchives.gov.uk/akn">
+  <judgment name="decision">
+    <meta>
+      <identification source="#tna">
+        <FRBRWork>
+          <FRBRthis value="" />
+          <FRBRuri value="" />
+          <FRBRdate date="{{decision_date}}" name="decision" />
+          <FRBRauthor href="#{{court_code.lower()}}" />
+          <FRBRcountry value="GB-UKM" />
+          <FRBRname value="{{title}}" />
+        </FRBRWork>
+        <FRBRExpression>
+          <FRBRthis value="" />
+          <FRBRuri value="" />
+          <FRBRdate date="{{decision_date}}" name="decision" />
+          <FRBRauthor href="#{{court_code.lower()}}" />
+          <FRBRlanguage language="eng" />
+        </FRBRExpression>
+        <FRBRManifestation>
+          <FRBRthis value="" />
+          <FRBRuri value="" />
+          <FRBRdate date="{{transform_datetime}}" name="transform" />
+          <FRBRauthor href="#tna" />
+          <FRBRformat value="application/xml" />
+        </FRBRManifestation>
+      </identification>
+      <lifecycle source="#">
+        <eventRef date="{{decision_date}}" refersTo="#decision" source="#" />
+      </lifecycle>
+      <references source="#tna">
+        <TLCOrganization
+                    eId="tna"
+                    href="https://www.nationalarchives.gov.uk/"
+                    showAs="The National Archives"
+                />
+        <TLCOrganization
+                    eId="{{court_code.lower()}}"
+                    href="{{court_url}}"
+                    showAs="{{court_full_name}}"
+                />
+        <TLCEvent eId="decision" href="#" showAs="decision" />
+      </references>
+      <proprietary
+                xmlns:uk="https://caselaw.nationalarchives.gov.uk/akn"
+                source="#"
+            >
+        <uk:court>{{court_code.upper()}}</uk:court>
+        <uk:year>{{year}}</uk:year>
+        {% for case_number in case_numbers %}
+            <uk:caseNumber>{{case_number}}</uk:caseNumber>
+        {% endfor %}
+        {% for party in parties %}
+           <uk:party role="{{party.role}}">{{party.name}}</uk:party>
+        {% endfor %}
+        <uk:sourceFormat>application/pdf</uk:sourceFormat>
+      </proprietary>
+    </meta>
+    <header />
+    <judgmentBody>
+      <decision>
+        <p />
+      </decision>
+    </judgmentBody>
+  </judgment>
+</akomaNtoso>

ds_caselaw_marklogic_api_client-44.4.1/src/caselawclient/models/documents/templates/original.xml

@@ -0,0 +1,68 @@
+<?xml version="1.0" encoding="utf-8" ?>
+<!-- from https://github.com/nationalarchives/tna-judgments-parser/blob/main/test/backlog/expected-output/Money%20Worries%20Ltd%20v%20Office%20of%20Fair%20Trading.xml -->
+<!-- generated by github.com/nationalarchives/tna-judgments-parser/blob/main/backlog/src/Stub.cs -->
+<!-- we assume there are no press-releases for these documents -->
+<akomaNtoso xmlns="http://docs.oasis-open.org/legaldocml/ns/akn/3.0">
+  <judgment name="decision">
+    <meta>
+      <identification source="#tna">
+        <FRBRWork>
+          <FRBRthis value="" />
+          <FRBRuri value="" />
+          <FRBRdate date="2013-01-18" name="decision" />
+          <FRBRauthor href="#ukftt-grc" />
+          <FRBRcountry value="GB-UKM" />
+          <FRBRname value="Money Worries Ltd v Office of Fair Trading" />
+        </FRBRWork>
+        <FRBRExpression>
+          <FRBRthis value="" />
+          <FRBRuri value="" />
+          <FRBRdate date="2013-01-18" name="decision" />
+          <FRBRauthor href="#ukftt-grc" />
+          <FRBRlanguage language="eng" />
+        </FRBRExpression>
+        <FRBRManifestation>
+          <FRBRthis value="" />
+          <FRBRuri value="" />
+          <FRBRdate date="2025-07-29T12:47:42" name="transform" />
+          <FRBRauthor href="#tna" />
+          <FRBRformat value="application/xml" />
+        </FRBRManifestation>
+      </identification>
+      <lifecycle source="#">
+        <eventRef date="2013-01-18" refersTo="#decision" source="#" />
+      </lifecycle>
+      <references source="#tna">
+        <TLCOrganization
+                    eId="tna"
+                    href="https://www.nationalarchives.gov.uk/"
+                    showAs="The National Archives"
+                />
+        <TLCOrganization
+                    eId="ukftt-grc"
+                    href="https://www.gov.uk/courts-tribunals/first-tier-tribunal-general-regulatory-chamber"
+                    showAs="United Kingdom First-tier Tribunal (General Regulatory Chamber)"
+                />
+        <TLCEvent eId="decision" href="#" showAs="decision" />
+      </references>
+      <proprietary
+                xmlns:uk="https://caselaw.nationalarchives.gov.uk/akn"
+                source="#"
+            >
+        <uk:court>UKFTT-GRC</uk:court>
+        <uk:year>2013</uk:year>
+        <uk:caseNumber>CCA/2012/0008</uk:caseNumber>
+        <uk:party role="Claimant">Money Worries Ltd</uk:party>
+        <uk:party role="Respondent">Office of Fair Trading</uk:party>
+        <uk:sourceFormat>application/pdf</uk:sourceFormat>
+        <uk:parser>0.27.1</uk:parser>
+      </proprietary>
+    </meta>
+    <header />
+    <judgmentBody>
+      <decision>
+        <p />
+      </decision>
+    </judgmentBody>
+  </judgment>
+</akomaNtoso>

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/models/documents/transforms/html.xsl

@@ -734,7 +734,7 @@
 </xsl:template>
 
 <xsl:template match="img">
-	<img>
+	<img loading="lazy">
 		<xsl:apply-templates select="@*" />
 		<xsl:apply-templates />
 	</img>

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/models/utilities/aws.py

@@ -208,6 +208,13 @@ def announce_document_event(uri: DocumentURIString, status: str, enrich: bool =
     )
 
 
+def upload_asset_to_private_bucket(body: bytes, s3_key: str) -> None:
+    """Upload an asset to the private bucket."""
+    bucket: str = env("PRIVATE_ASSET_BUCKET")
+    s3client = create_s3_client()
+    s3client.put_object(Body=body, Bucket=bucket, Key=s3_key, Tagging="pdfsource=custom-pdfs")
+
+
 def copy_assets(old_uri: DocumentURIString, new_uri: DocumentURIString) -> None:
     """
     Copy *unpublished* assets from one path to another,

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/types.py

@@ -1,4 +1,7 @@
 from dataclasses import dataclass, field
+from datetime import UTC, datetime
+
+from lxml import etree
 
 
 @dataclass
@@ -106,3 +109,42 @@ def SuccessTuple() -> SuccessFailureMessageTuple:
 def FailureTuple(message: str | list[str]) -> SuccessFailureMessageTuple:
     messages = message if isinstance(message, list) else [message]
     return SuccessFailureMessageTuple(success=False, messages=messages)
+
+
+@dataclass
+class DocumentLock:
+    document_uri: DocumentURIString
+    owner: str
+    timestamp: datetime
+    timeout: int
+
+    NAMESPACES = {
+        "lock": "http://marklogic.com/xdmp/lock",
+    }
+
+    @classmethod
+    def from_string(cls, xml_string: str) -> "DocumentLock":
+        root = etree.fromstring(xml_string.encode("utf-8"))
+
+        return cls(
+            document_uri=MarkLogicDocumentURIString(root.xpath("/lock/document/text()")[0]).as_document_uri(),
+            owner=root.xpath(
+                "/lock/details/lock:lock/lock:active-locks/lock:active-lock/lock:owner/text()",
+                namespaces=cls.NAMESPACES,
+            )[0],
+            timestamp=datetime.fromtimestamp(
+                int(
+                    root.xpath(
+                        "/lock/details/lock:lock/lock:active-locks/lock:active-lock/lock:timestamp/text()",
+                        namespaces=cls.NAMESPACES,
+                    )[0]
+                ),
+                tz=UTC,
+            ),
+            timeout=int(
+                root.xpath(
+                    "/lock/details/lock:lock/lock:active-locks/lock:active-lock/lock:timeout/text()",
+                    namespaces=cls.NAMESPACES,
+                )[0]
+            ),
+        )

ds_caselaw_marklogic_api_client-44.4.1/src/caselawclient/xquery/check_content_hash_unique_by_uri.xqy

@@ -0,0 +1,18 @@
+xquery version "1.0-ml";
+declare namespace akn = "http://docs.oasis-open.org/legaldocml/ns/akn/3.0";
+declare namespace uk = "https://caselaw.nationalarchives.gov.uk/akn";
+declare variable $uri as xs:string external;
+
+let $doc := doc($uri)
+let $hash := $doc//uk:hash/text()
+let $count :=
+  if (not($hash)) then 1
+  else
+    count(cts:uris(
+      (), (),
+      cts:and-query((
+        cts:element-value-query(xs:QName("uk:hash"), $hash),
+        cts:collection-query("http://marklogic.com/collections/dls/latest-version")
+      ))
+    ))
+return ($count = 1)

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/xquery/get_judgment.xqy

@@ -11,7 +11,7 @@ declare variable $version_uri as xs:string? external;
 declare variable $search_query as xs:string? external;
 
 (: Note that `xsl:output method` is changed from `html` to `xml` and we've namespaced the tag :)
-let $number_marks_xslt := (
+let $delete_meta_marks_xslt := (
   <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
                   xmlns:uk="https://caselaw.nationalarchives.gov.uk/akn"
                   xmlns:akn="http://docs.oasis-open.org/legaldocml/ns/akn/3.0"
@@ -25,16 +25,6 @@ let $number_marks_xslt := (
     <xsl:template match="//akn:meta//uk:mark">
           <xsl:apply-templates />
     </xsl:template>
-    <xsl:template match="uk:mark">
-      <xsl:copy>
-          <xsl:copy-of select="@*" />
-          <xsl:attribute name="id">
-              <xsl:text>mark_</xsl:text>
-              <xsl:number count="//uk:mark" level="any" from="//*[ancestor::akn:meta]" />
-          </xsl:attribute>
-          <xsl:apply-templates />
-      </xsl:copy>
-    </xsl:template>
   </xsl:stylesheet>
 )
 
@@ -56,7 +46,7 @@ let $raw_xml := if ($show_unpublished) then
 (: If a search query string is present, highlight instances :)
 let $transformed := if($search_query) then
       xdmp:xslt-eval(
-        $number_marks_xslt,
+        $delete_meta_marks_xslt,
         cts:highlight(
           $raw_xml,
           helper:make-q-query($search_query),

ds_caselaw_marklogic_api_client-44.4.1/src/caselawclient/xquery/get_locked_documents.xqy

@@ -0,0 +1,10 @@
+xquery version "1.0-ml";
+
+for $lock in xdmp:document-locks()
+let $uri := xdmp:node-uri($lock)
+
+return
+    <lock>
+        <document>{$uri}</document>
+        <details>{$lock}</details>
+    </lock>

ds_caselaw_marklogic_api_client-44.4.1/src/caselawclient/xquery/get_pending_parse_for_version_count.xqy

@@ -0,0 +1,27 @@
+xquery version "1.0-ml";
+
+declare variable $target_major_version as xs:int external;
+declare variable $target_minor_version as xs:int external;
+
+xdmp:to-json(xdmp:sql(
+  "SELECT COUNT(*) as count
+  FROM (
+    SELECT 
+      process_data.uri,
+      parser_version_string, parser_major_version, parser_minor_version,
+      DATEDIFF('minute', last_sent_to_parser, CURRENT_TIMESTAMP) AS minutes_since_parse_request
+    FROM documents.process_data
+    JOIN documents.process_property_data ON process_data.uri = process_property_data.uri
+  )
+  WHERE (
+    (parser_version_string IS NULL) OR
+    (parser_major_version < @target_major_version) OR
+    (parser_major_version <= @target_major_version AND parser_minor_version < @target_minor_version)
+  )
+  AND (minutes_since_parse_request > 43200 OR minutes_since_parse_request IS NULL)",
+  "array",
+  map:new((
+    map:entry("target_major_version", $target_major_version),
+    map:entry("target_minor_version", $target_minor_version)
+  ))
+))

ds_caselaw_marklogic_api_client-43.1.0/src/caselawclient/xquery/get_pending_parse_for_version.xqy → ds_caselaw_marklogic_api_client-44.4.1/src/caselawclient/xquery/get_pending_parse_for_version_documents.xqy

@@ -16,10 +16,11 @@ xdmp:to-json(xdmp:sql(
   )
   WHERE (
     (parser_version_string IS NULL) OR
+    (parser_major_version < @target_major_version) OR
     (parser_major_version <= @target_major_version AND parser_minor_version < @target_minor_version)
   )
   AND (minutes_since_parse_request > 43200 OR minutes_since_parse_request IS NULL)
-  ORDER BY parser_major_version ASC NULLS FIRST, parser_minor_version ASC
+  ORDER BY parser_major_version ASC NULLS FIRST, parser_minor_version ASC, minutes_since_parse_request DESC
   LIMIT @maximum_records",
   "array",
   map:new((
@@ -28,4 +29,4 @@ xdmp:to-json(xdmp:sql(
     map:entry("maximum_records", $maximum_records)
     
   ))
-))
+))

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/xquery/set_metadata_jurisdiction.xqy

@@ -34,4 +34,4 @@ cts:search(doc($uri),
 cts:element-query(xs:QName('uk:jurisdiction'),cts:and-query(()))))) then
     if ($content = "") then local:delete($uri) else local:edit($uri, $content)
 else
-    local:add($uri, $content)
+    if ($content != "") then local:add($uri, $content) else ()

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/xquery/xslt_transform.xqy

@@ -17,27 +17,6 @@ let $xsl_path := fn:concat("judgments/xslts/", $xsl_filename)
 
 let $params := map:map()
 
-let $number_marks_xslt := (
-  <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
-                  version="2.0">
-    <xsl:output method="html" />
-    <xsl:template match="@*|node()">
-      <xsl:copy>
-        <xsl:apply-templates select="@*|node()"/>
-      </xsl:copy>
-    </xsl:template>
-    <xsl:template match="mark">
-      <xsl:copy>
-          <xsl:copy-of select="*" />
-          <xsl:attribute name="id">
-              <xsl:text>mark_</xsl:text>
-              <xsl:value-of select="count(preceding::mark)"/>
-          </xsl:attribute>
-          <xsl:apply-templates />
-      </xsl:copy>
-    </xsl:template>
-  </xsl:stylesheet>
-)
 (: change the image-base of the document to match the location of the assets in $image_base
    so that references to images point to the correct places on the internet :)
 let $_put := map:put(
@@ -59,13 +38,10 @@ let $retrieved_value := if (xs:boolean($is_published) or $show_unpublished) then
         ()
 
 let $return_value := if($query) then
-      xdmp:xslt-eval(
-        $number_marks_xslt,
-        cts:highlight(
-          $retrieved_value,
-          helper:make-q-query($query),
-          <mark>{$cts:text}</mark>
-        )
+      cts:highlight(
+        $retrieved_value,
+        helper:make-q-query($query),
+        <mark>{$cts:text}</mark>
       )
     else
       $retrieved_value

{ds_caselaw_marklogic_api_client-43.1.0 → ds_caselaw_marklogic_api_client-44.4.1}/src/caselawclient/xquery_type_dicts.py

@@ -105,8 +105,14 @@ class GetPendingEnrichmentForVersionDict(MarkLogicAPIDict):
     target_parser_minor_version: int
 
 
-# get_pending_parse_for_version.xqy
-class GetPendingParseForVersionDict(MarkLogicAPIDict):
+# get_pending_parse_for_version_count.xqy
+class GetPendingParseForVersionCountDict(MarkLogicAPIDict):
+    target_major_version: int
+    target_minor_version: int
+
+
+# get_pending_parse_for_version_documents.xqy
+class GetPendingParseForVersionDocumentsDict(MarkLogicAPIDict):
     maximum_records: Optional[int]
     target_major_version: int
     target_minor_version: int

ds_caselaw_marklogic_api_client-43.1.0/src/caselawclient/xquery/check_content_hash_unique_by_uri.xqy

@@ -1,15 +0,0 @@
-xquery version "1.0-ml";
-declare namespace akn = "http://docs.oasis-open.org/legaldocml/ns/akn/3.0";
-declare namespace uk = "https://caselaw.nationalarchives.gov.uk/akn";
-declare variable $uri as xs:string external;
-
-let $doc := doc($uri)
-let $hash := $doc//uk:hash/text()
-let $count := count(cts:uris(
-  (), (),
-  cts:and-query((
-    cts:element-value-query(xs:QName("uk:hash"), $hash),
-    cts:collection-query("http://marklogic.com/collections/dls/latest-version")
-  ))
-))
-return $count = 1