ds-caselaw-marklogic-api-client 27.2.0__py3-none-any.whl → 27.4.0__py3-none-any.whl

caselawclient/Client.py

@@ -62,6 +62,8 @@ except importlib.metadata.PackageNotFoundError:
     VERSION = "0"
 DEFAULT_USER_AGENT = f"ds-caselaw-marklogic-api-client/{VERSION}"
 
+DEBUG: bool = bool(os.getenv("DEBUG", default=False))
+
 
 class NoResponse(Exception):
     """A requests HTTPError has no response. We expect this will never happen."""
@@ -217,10 +219,10 @@ 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 = {
@@ -728,6 +730,10 @@ class MarklogicApiClient:
             "vars": vars,
         }
         path = "LATEST/eval"
+
+        if DEBUG:
+            print(f"Sending {vars} to {xquery_path}")
+
         response = self.session.request(
             "POST",
             url=self._path_to_request_url(path),
@@ -1080,6 +1086,7 @@ 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 = {
@@ -1087,6 +1094,7 @@ 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(
@@ -1115,11 +1123,13 @@ 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/xquery/get_pending_enrichment_for_version.xqy

@@ -1,35 +1,46 @@
 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
   FROM (
     SELECT
+      propertysummary.published,
       process_data.uri,
       enrich_version_string, enrich_major_version, enrich_minor_version,
       parser_major_version, parser_minor_version,
       DATEDIFF('minute', last_sent_to_enrichment, CURRENT_TIMESTAMP) AS minutes_since_enrichment_request
     FROM documents.process_data
     JOIN documents.process_property_data ON process_data.uri = process_property_data.uri
+    JOIN documents.propertysummary ON process_data.uri = propertysummary.uri
   )
-  WHERE (
-    (enrich_version_string IS NULL) OR
-    (enrich_major_version <= @target_enrichment_major_version AND enrich_minor_version < @target_enrichment_minor_version)
-  ) AND (
-    (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",
+  WHERE 
+    (minutes_since_enrichment_request > 43200 OR minutes_since_enrichment_request IS NULL) AND
+    (propertysummary.published = 'true') AND
+    ( enrich_version_string IS NULL 
+      OR
+      (
+        (enrich_major_version <= @target_enrichment_major_version AND enrich_minor_version < @target_enrichment_minor_version)
+           AND 
+        (parser_major_version = @target_parser_major_version AND parser_minor_version = @target_parser_minor_version)
+      )
+    )
+  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

@@ -88,6 +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
@@ -96,6 +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-27.2.0.dist-info → ds_caselaw_marklogic_api_client-27.4.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ds-caselaw-marklogic-api-client
-Version: 27.2.0
+Version: 27.4.0
 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-27.2.0.dist-info → ds_caselaw_marklogic_api_client-27.4.0.dist-info}/RECORD

@@ -1,4 +1,4 @@
-caselawclient/Client.py,sha256=WKDQaN4b3J3m2pzYdvK4t1MmbeVS9pbtTfqcEH1K-RU,40464
+caselawclient/Client.py,sha256=05vFmk39Gy23i-Mo04pZk_3EmsTe_2feIiwd7fXjDpM,40797
 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
@@ -41,8 +41,8 @@ caselawclient/xquery/get_judgment.xqy,sha256=8V-sEFKmtpf2LIZD9QKVRfpblEsmDpP4BA6
 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=8J5Pi-jMXJd_BgtpK4g6C9uR99HP57JpFv5WkoPfNuo,2016
+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=papUzI0k_8l987IpBXwACpnlL4GU1UX1cLt83HpAgds,5622
-ds_caselaw_marklogic_api_client-27.2.0.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
-ds_caselaw_marklogic_api_client-27.2.0.dist-info/METADATA,sha256=Aoc0502BWIboWQBs72vD4FHlRAqBZ-V3jVlRn51zO64,4232
-ds_caselaw_marklogic_api_client-27.2.0.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-ds_caselaw_marklogic_api_client-27.2.0.dist-info/RECORD,,
+caselawclient/xquery_type_dicts.py,sha256=AUEmycChASKl8sbqVQWe6WA0s-lIfEqh9mA-GdnUCQ8,5692
+ds_caselaw_marklogic_api_client-27.4.0.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
+ds_caselaw_marklogic_api_client-27.4.0.dist-info/METADATA,sha256=9acabO_PzOjK3LScvlra8WgiyVnNRb7ItdcjIBs1zsg,4232
+ds_caselaw_marklogic_api_client-27.4.0.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+ds_caselaw_marklogic_api_client-27.4.0.dist-info/RECORD,,