ds-caselaw-marklogic-api-client 27.1.0__py3-none-any.whl → 27.3.0__py3-none-any.whl

caselawclient/Client.py

@@ -217,10 +217,10 @@ class MarklogicApiClient:
     def get_document_by_uri(
         self,
         uri: DocumentURIString,
-        query: Optional[str] = None,
+        search_query: Optional[str] = None,
     ) -> Document:
         document_type_class = self.get_document_type_from_uri(uri)
-        return document_type_class(uri, self)
+        return document_type_class(uri, self, search_query=search_query)
 
     def get_document_type_from_uri(self, uri: DocumentURIString) -> Type[Document]:
         vars: query_dicts.DocumentCollectionsDict = {
@@ -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(
@@ -1076,6 +1080,7 @@ class MarklogicApiClient:
         self,
         target_enrichment_version: tuple[int, int],
         target_parser_version: tuple[int, int],
+        maximum_records: int = 1000,
     ) -> list[list[Any]]:
         """Retrieve documents which are not yet enriched with a given version."""
         vars: query_dicts.GetPendingEnrichmentForVersionDict = {
@@ -1083,6 +1088,7 @@ class MarklogicApiClient:
             "target_enrichment_minor_version": target_enrichment_version[1],
             "target_parser_major_version": target_parser_version[0],
             "target_parser_minor_version": target_parser_version[1],
+            "maximum_records": maximum_records,
         }
         results: list[list[Any]] = json.loads(
             get_single_string_from_marklogic_response(
@@ -1111,11 +1117,13 @@ class MarklogicApiClient:
     def get_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 = {
             "target_major_version": target_version[0],
             "target_minor_version": target_version[1],
+            "maximum_records": maximum_records,
         }
         results: list[list[Any]] = json.loads(
             get_single_string_from_marklogic_response(

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. """
 

caselawclient/models/documents/transforms/html.xsl

@@ -973,6 +973,14 @@
 	</xsl:element>
 </xsl:template>
 
+<!-- search query numbering -->
+
+<xsl:template match="uk:mark">
+	<xsl:element name="{ local-name(.) }">
+		<xsl:copy-of select="@*"/>
+		<xsl:apply-templates />
+	</xsl:element>
+</xsl:template>
 
 <!-- text -->
 

caselawclient/xquery/get_judgment.xqy

@@ -1,8 +1,38 @@
 xquery version "1.0-ml";
 
+declare namespace xdmp = "http://marklogic.com/xdmp";
+declare namespace cts = "http://marklogic.com/cts";
+declare namespace uk = "https://caselaw.nationalarchives.gov.uk/akn";
+import module namespace helper = "https://caselaw.nationalarchives.gov.uk/helper" at "/judgments/search/helper.xqy";
+
 declare variable $show_unpublished as xs:boolean? external;
 declare variable $uri as xs:string external;
 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 := (
+  <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
+                  xmlns:uk="https://caselaw.nationalarchives.gov.uk/akn"
+                  version="2.0">
+    <xsl:output method="xml" />
+    <xsl:template match="@*|node()">
+      <xsl:copy>
+        <xsl:apply-templates select="@*|node()"/>
+      </xsl:copy>
+    </xsl:template>
+    <xsl:template match="uk:mark">
+      <xsl:copy>
+          <xsl:copy-of select="@*" />
+          <xsl:attribute name="id">
+              <xsl:text>mark_</xsl:text>
+              <xsl:value-of select="count(preceding::uk:mark)"/>
+          </xsl:attribute>
+          <xsl:apply-templates />
+      </xsl:copy>
+    </xsl:template>
+  </xsl:stylesheet>
+)
 
 let $judgment := fn:document($uri)
 let $version := if ($version_uri) then fn:document($version_uri) else ()
@@ -11,11 +41,26 @@ let $is_published := $judgment_published_property/text()
 
 let $document_to_return := if ($version_uri) then $version else $judgment
 
-let $return_value := if ($show_unpublished) then
+
+let $raw_xml := if ($show_unpublished) then
         $document_to_return
     else if (xs:boolean($is_published)) then
         $document_to_return
     else
         ()
 
-return $return_value
+(: If a search query string is present, highlight instances :)
+let $transformed := if($search_query) then
+      xdmp:xslt-eval(
+        $number_marks_xslt,
+        cts:highlight(
+          $raw_xml,
+          helper:make-q-query($search_query),
+          <uk:mark>{$cts:text}</uk:mark>
+        )
+      )
+    else
+      $raw_xml
+
+
+return $transformed

caselawclient/xquery/get_pending_enrichment_for_version.xqy

@@ -1,9 +1,11 @@
 xquery version "1.0-ml";
 
+declare namespace xdmp="http://marklogic.com/xdmp"; 
 declare variable $target_enrichment_major_version as xs:int external;
 declare variable $target_enrichment_minor_version as xs:int external;
 declare variable $target_parser_major_version as xs:int external;
 declare variable $target_parser_minor_version as xs:int external;
+declare variable $maximum_records as xs:int? external := 1000;
 
 xdmp:to-json(xdmp:sql(
   "SELECT process_data.uri, enrich_version_string, minutes_since_enrichment_request
@@ -23,13 +25,16 @@ xdmp:to-json(xdmp:sql(
     (parser_major_version = @target_parser_major_version AND parser_minor_version = @target_parser_minor_version)
   )
   AND (minutes_since_enrichment_request > 43200 OR minutes_since_enrichment_request IS NULL)
-  ORDER BY enrich_major_version ASC NULLS FIRST, enrich_minor_version ASC",
+  ORDER BY enrich_major_version ASC NULLS FIRST, enrich_minor_version ASC
+  LIMIT @maximum_records",
   "array",
   map:new((
     map:entry("target_enrichment_major_version", $target_enrichment_major_version),
     map:entry("target_enrichment_minor_version", $target_enrichment_minor_version),
     map:entry("target_parser_major_version", $target_parser_major_version),
-    map:entry("target_parser_minor_version", $target_parser_minor_version)
+    map:entry("target_parser_minor_version", $target_parser_minor_version),
+    map:entry("maximum_records", $maximum_records)
+    
   ))
 ))
 

caselawclient/xquery/get_pending_parse_for_version.xqy

@@ -2,6 +2,7 @@ xquery version "1.0-ml";
 
 declare variable $target_major_version as xs:int external;
 declare variable $target_minor_version as xs:int external;
+declare variable $maximum_records as xs:int? external := 1000;
 
 xdmp:to-json(xdmp:sql(
   "SELECT process_data.uri, parser_version_string, minutes_since_parse_request
@@ -18,11 +19,13 @@ xdmp:to-json(xdmp:sql(
     (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
+  LIMIT @maximum_records",
   "array",
   map:new((
     map:entry("target_major_version", $target_major_version),
-    map:entry("target_minor_version", $target_minor_version)
+    map:entry("target_minor_version", $target_minor_version),
+    map:entry("maximum_records", $maximum_records)
+    
   ))
-))
-
+))

caselawclient/xquery_type_dicts.py

@@ -64,6 +64,7 @@ class GetComponentsForDocumentDict(MarkLogicAPIDict):
 
 # get_judgment.xqy
 class GetJudgmentDict(MarkLogicAPIDict):
+    search_query: Optional[str]
     show_unpublished: Optional[bool]
     uri: MarkLogicDocumentURIString
     version_uri: Optional[MarkLogicDocumentVersionURIString]
@@ -87,6 +88,7 @@ class GetLastModifiedDict(MarkLogicAPIDict):
 
 # get_pending_enrichment_for_version.xqy
 class GetPendingEnrichmentForVersionDict(MarkLogicAPIDict):
+    maximum_records: Optional[int]
     target_enrichment_major_version: int
     target_enrichment_minor_version: int
     target_parser_major_version: int
@@ -95,6 +97,7 @@ class GetPendingEnrichmentForVersionDict(MarkLogicAPIDict):
 
 # get_pending_parse_for_version.xqy
 class GetPendingParseForVersionDict(MarkLogicAPIDict):
+    maximum_records: Optional[int]
     target_major_version: int
     target_minor_version: int
 

{ds_caselaw_marklogic_api_client-27.1.0.dist-info → ds_caselaw_marklogic_api_client-27.3.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ds-caselaw-marklogic-api-client
-Version: 27.1.0
+Version: 27.3.0
 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

{ds_caselaw_marklogic_api_client-27.1.0.dist-info → ds_caselaw_marklogic_api_client-27.3.0.dist-info}/RECORD

@@ -1,4 +1,4 @@
-caselawclient/Client.py,sha256=CP7wx_Z_RKK6Wg4Ib3JypEXbncL1jNsLmo66MyXxcF0,40295
+caselawclient/Client.py,sha256=oMWfL0FwKdWyMtk_0qf6uxWJvsHLSps8vxDsAC9GxeI,40668
 caselawclient/__init__.py,sha256=DY-caubLDQWWingSdsBWgovDNXh8KcnkI6kwz08eIFk,612
 caselawclient/client_helpers/__init__.py,sha256=fyDNKCdrTb2N0Ks23YDhmvlXKfLTHnYQCXhnZb-QQbg,3832
 caselawclient/client_helpers/search_helpers.py,sha256=R99HyRLeYHgsw2L3DOidEqlKLLvs6Tga5rKTuWQViig,1525
@@ -6,11 +6,11 @@ caselawclient/content_hash.py,sha256=0cPC4OoABq0SC2wYFX9-24DodNigeOqksDxgxQH_hUA
 caselawclient/errors.py,sha256=tV0vs3wYSd331BzmfuRiZV6GAdsd91rtN65ymRaSx3s,3164
 caselawclient/factories.py,sha256=wTGGP8VyUME2gwbnuYE2uC5PdCdJ2PXF9gEET3U9CZg,4356
 caselawclient/models/__init__.py,sha256=kd23EUpvaC7aLHdgk8farqKAQEx3lf7RvNT2jEatvlg,68
-caselawclient/models/documents/__init__.py,sha256=LnAURUtxoY4UdDUk-dFgnyJ-Gn9eB0VfdC4gjNoYacM,16974
+caselawclient/models/documents/__init__.py,sha256=uNjmmNKAthBmB6RSNlmJPir3E2MGrYkcP17iEY-xtkk,17100
 caselawclient/models/documents/body.py,sha256=mtdjmG1WU2qSpyRLS8-PWcSoXpDa2Qz6xlcTbxZgxvA,5603
 caselawclient/models/documents/exceptions.py,sha256=Mz1P8uNqf5w6uLnRwJt6xK7efsVqtd5VA-WXUUH7QLk,285
 caselawclient/models/documents/statuses.py,sha256=Cp4dTQmJOtsU41EJcxy5dV1841pGD2PNWH0VrkDEv4Q,579
-caselawclient/models/documents/transforms/html.xsl,sha256=BTo3dsjJywmw28rnPB-AwRXZk8hY8Bi7KqEkg8tsOuQ,38131
+caselawclient/models/documents/transforms/html.xsl,sha256=oSSO-IBX4qLiSWexQYmWJfGNevF09aCBx4D1NYqXxpo,38322
 caselawclient/models/documents/xml.py,sha256=afEsgcnTThqW_gKYq-VGtFr4ovOoT2J7h2gXX7F8BbE,1267
 caselawclient/models/judgments.py,sha256=SuCNtOD4LElp37df4dvhaD0umTowioWH0sZNmBgFsoE,1739
 caselawclient/models/neutral_citation_mixin.py,sha256=5ktKCPIDidVRwxVTzx5e242O1BxOdP--1dnatZyTbYI,1773
@@ -37,12 +37,12 @@ caselawclient/xquery/get_combined_stats_table.xqy,sha256=cclNqSzIB6sX6A_hgVOFZon
 caselawclient/xquery/get_components_for_document.xqy,sha256=qh_F56XEDyVKMIaN4J1fskcjYK3Mc6JR8ARGwS2b0lw,768
 caselawclient/xquery/get_highest_enrichment_version.xqy,sha256=a0dwVmEZuIMyRjIlvenSmbOaaN0WvgaCZvMtVWoLulQ,247
 caselawclient/xquery/get_highest_parser_version.xqy,sha256=LW3iSg4eWArbfBaCVWWOpr4MoUcDBz514nV48ElOsAM,247
-caselawclient/xquery/get_judgment.xqy,sha256=xg66A1xslgwds670gu_9DFcOmxYGoo8sA2EDoKbz0Co,715
+caselawclient/xquery/get_judgment.xqy,sha256=8V-sEFKmtpf2LIZD9QKVRfpblEsmDpP4BA6QztSCyHA,2216
 caselawclient/xquery/get_judgment_checkout_status.xqy,sha256=mdY9UXLyzQdB7byEERPqentlr0YDLbXRVqH0h4UuZTQ,193
 caselawclient/xquery/get_judgment_version.xqy,sha256=wF9k9-CBrqo8VbxxyTrD-AGzR3-3jMm25tRVCjxPLrU,292
 caselawclient/xquery/get_last_modified.xqy,sha256=8fCm_7o_kkytCEmEeSTLWzLP7iOjuPV01IfHDgf6HaQ,172
-caselawclient/xquery/get_pending_enrichment_for_version.xqy,sha256=jYc2OPnAIU4kfRsyhCam9RtrxOB2KhKX40CALFc2Rwo,1623
-caselawclient/xquery/get_pending_parse_for_version.xqy,sha256=1HRQ3Mwljgp1QbVTiesJrV5HuuMCyKPDMzBOzrT_oQM,1054
+caselawclient/xquery/get_pending_enrichment_for_version.xqy,sha256=S5r5h1iyn_wDQXLEh2MzK8qQ1Jiq-1lRsKgDa8L16yY,1821
+caselawclient/xquery/get_pending_parse_for_version.xqy,sha256=9cjVZtHeBBjm-a7RMsn1PVJt_Ug78YFlmp5CN8VJ1jQ,1197
 caselawclient/xquery/get_properties_for_search_results.xqy,sha256=Tlv3EKwVV_q-JyQyhjWVHIleicPDpucxP4ScuQjpgSw,625
 caselawclient/xquery/get_property.xqy,sha256=RHlOTrK0aH-S7s_ykYzGmUeKOJxXlI4vE5sKRt556NY,209
 caselawclient/xquery/get_version_annotation.xqy,sha256=pFDMGA9SxI59iUPaoAeUsq23kUp4Op7cUg-PFNFqxcI,262
@@ -65,8 +65,8 @@ caselawclient/xquery/validate_all_documents.xqy,sha256=z_0YEXmRcZ-FaJM0ouKiTjdI4
 caselawclient/xquery/validate_document.xqy,sha256=PgaDcnqCRJPIVqfmWsNlXmCLNKd21qkJrvY1RtNP7eA,140
 caselawclient/xquery/xslt.xqy,sha256=w57wNijH3dkwHkpKeAxqjlghVflQwo8cq6jS_sm-erM,199
 caselawclient/xquery/xslt_transform.xqy,sha256=smyFFxqmtkuOzBd2l7uw6K2oAsYctudrP8omdv_XNAM,2463
-caselawclient/xquery_type_dicts.py,sha256=YOrXbEYJU84S-YwergCI12OL5Wrn_wpqMcqWpsQrKek,5590
-ds_caselaw_marklogic_api_client-27.1.0.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
-ds_caselaw_marklogic_api_client-27.1.0.dist-info/METADATA,sha256=gYAgOwD-FrETnEArRn9NSyZ2Yl2EvBOFPse-fWZNI7w,4232
-ds_caselaw_marklogic_api_client-27.1.0.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-ds_caselaw_marklogic_api_client-27.1.0.dist-info/RECORD,,
+caselawclient/xquery_type_dicts.py,sha256=AUEmycChASKl8sbqVQWe6WA0s-lIfEqh9mA-GdnUCQ8,5692
+ds_caselaw_marklogic_api_client-27.3.0.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
+ds_caselaw_marklogic_api_client-27.3.0.dist-info/METADATA,sha256=GltSiGOxA1e5fTsNbAdFhCjVwxQe7XAWS__-eZ165AM,4232
+ds_caselaw_marklogic_api_client-27.3.0.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+ds_caselaw_marklogic_api_client-27.3.0.dist-info/RECORD,,