ds-caselaw-marklogic-api-client 28.2.0__py3-none-any.whl → 29.0.0a2__py3-none-any.whl

caselawclient/Client.py

@@ -20,7 +20,6 @@ from requests_toolbelt.multipart import decoder
 
 from caselawclient import xquery_type_dicts as query_dicts
 from caselawclient.client_helpers import VersionAnnotation
-from caselawclient.identifier_resolution import IdentifierResolutions
 from caselawclient.models.documents import (
     DOCUMENT_COLLECTION_URI_JUDGMENT,
     DOCUMENT_COLLECTION_URI_PRESS_SUMMARY,
@@ -1202,22 +1201,3 @@ class MarklogicApiClient:
         )
 
         return results
-
-    def resolve_from_identifier(self, identifier_uri: str, published_only: bool = True) -> IdentifierResolutions:
-        """Given a PUI/EUI url, look up the precomputed slug and return the
-        MarkLogic document URIs which match that slug. Multiple returns should be anticipated"""
-        vars: query_dicts.ResolveFromIdentifierDict = {
-            "identifier_uri": DocumentURIString(identifier_uri),
-            "published_only": int(published_only),
-        }
-        raw_results: list[str] = get_multipart_strings_from_marklogic_response(
-            self._send_to_eval(
-                vars,
-                "resolve_from_identifier.xqy",
-            ),
-        )
-        return IdentifierResolutions.from_marklogic_output(raw_results)
-
-    def get_next_document_sequence_number(self) -> int:
-        """Increment the MarkLogic sequence number by one and return the value."""
-        return int(self._eval_and_decode({}, "get_next_document_sequence_number.xqy"))

caselawclient/models/documents/__init__.py

@@ -15,7 +15,6 @@ from caselawclient.errors import (
     NotSupportedOnVersion,
     OnlySupportedOnVersion,
 )
-from caselawclient.models.identifiers.fclid import FindCaseLawIdentifier, FindCaseLawIdentifierSchema
 from caselawclient.models.identifiers.unpacker import unpack_all_identifiers_from_etree
 from caselawclient.models.utilities import VersionsDict, extract_version, render_versions
 from caselawclient.models.utilities.aws import (
@@ -433,12 +432,6 @@ class Document:
         if not self.is_publishable:
             raise CannotPublishUnpublishableDocument
 
-        ## If it doesn't already have one, get a new FCLID for this document and assign
-        if len(self.identifiers.of_type(FindCaseLawIdentifier)) < 1:
-            document_fclid = FindCaseLawIdentifierSchema.mint(self.api_client)
-            self.identifiers.add(document_fclid)
-            self.save_identifiers()
-
         publish_documents(uri_for_s3(self.uri))
         self.api_client.set_published(self.uri, True)
         announce_document_event(
@@ -537,11 +530,6 @@ class Document:
         """
         return self.docx_exists()
 
-    def save_identifiers(self) -> None:
-        """Save the current state of this Document's identifiers to MarkLogic."""
-        self.identifiers.validate()
-        self.api_client.set_property_as_node(self.uri, "identifiers", self.identifiers.as_etree)
-
     def __getattr__(self, name: str) -> Any:
         warnings.warn(f"{name} no longer exists on Document, using Document.body instead", DeprecationWarning)
         try:

caselawclient/models/identifiers/__init__.py

@@ -101,10 +101,6 @@ class Identifier(ABC):
     def url_slug(self) -> str:
         return self.schema.compile_identifier_url_slug(self.value)
 
-    def same_as(self, other: "Identifier") -> bool:
-        "Is this the same as another identifier (in value and schema)?"
-        return self.value == other.value and self.schema == other.schema
-
 
 class Identifiers(dict[str, Identifier]):
     def validate(self) -> None:
@@ -113,13 +109,8 @@ class Identifiers(dict[str, Identifier]):
                 msg = "Key of {identifier} in Identifiers is {uuid} not {identifier.uuid}"
                 raise UUIDMismatchError(msg)
 
-    def contains(self, other_identifier: Identifier) -> bool:
-        "Do the identifier's value and namespace already exist in this group?"
-        return any(other_identifier.same_as(identifier) for identifier in self.values())
-
     def add(self, identifier: Identifier) -> None:
-        if not self.contains(identifier):
-            self[identifier.uuid] = identifier
+        self[identifier.uuid] = identifier
 
     def __delitem__(self, key: Union[Identifier, str]) -> None:
         if isinstance(key, Identifier):
@@ -127,19 +118,6 @@ class Identifiers(dict[str, Identifier]):
         else:
             super().__delitem__(key)
 
-    def of_type(self, identifier_type: type[Identifier]) -> list[Identifier]:
-        """Return a list of all identifiers of a given type."""
-        uuids = self.keys()
-        return [self[uuid] for uuid in list(uuids) if isinstance(self[uuid], identifier_type)]
-
-    def delete_type(self, deleted_identifier_type: type[Identifier]) -> None:
-        "For when we want an identifier to be the only valid identifier of that type, delete the others first"
-        uuids = self.keys()
-        for uuid in list(uuids):
-            # we could use compare to .schema instead, which would have diffferent behaviour for subclasses
-            if isinstance(self[uuid], deleted_identifier_type):
-                del self[uuid]
-
     @property
     def as_etree(self) -> etree._Element:
         """Return an etree representation of all the Document's identifiers."""
@@ -149,3 +127,8 @@ class Identifiers(dict[str, Identifier]):
             identifiers_root.append(identifier.as_xml_tree)
 
         return identifiers_root
+
+    def save(self, document) -> None:  # type: ignore[no-untyped-def, unused-ignore]
+        """Save the current state of this Document's identifiers to MarkLogic."""
+        self.validate()
+        document.api_client.set_property_as_node(document.uri, "identifiers", self.as_etree)

caselawclient/models/identifiers/unpacker.py

@@ -3,11 +3,9 @@ from typing import Optional
 from lxml import etree
 
 from . import IDENTIFIER_UNPACKABLE_ATTRIBUTES, Identifier, Identifiers, InvalidIdentifierXMLRepresentationException
-from .fclid import FindCaseLawIdentifier
 from .neutral_citation import NeutralCitationNumber
 
 IDENTIFIER_NAMESPACE_MAP: dict[str, type[Identifier]] = {
-    "fclid": FindCaseLawIdentifier,
     "ukncn": NeutralCitationNumber,
 }
 

caselawclient/xquery_type_dicts.py

@@ -141,12 +141,6 @@ class ListJudgmentVersionsDict(MarkLogicAPIDict):
     uri: MarkLogicDocumentURIString
 
 
-# resolve_from_identifier.xqy
-class ResolveFromIdentifierDict(MarkLogicAPIDict):
-    identifier_uri: DocumentURIString
-    published_only: Optional[int]
-
-
 # set_boolean_property.xqy
 class SetBooleanPropertyDict(MarkLogicAPIDict):
     name: str

{ds_caselaw_marklogic_api_client-28.2.0.dist-info → ds_caselaw_marklogic_api_client-29.0.0a2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ds-caselaw-marklogic-api-client
-Version: 28.2.0
+Version: 29.0.0a2
 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
@@ -11,7 +11,7 @@ Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Requires-Dist: boto3 (>=1.26.112,<2.0.0)
-Requires-Dist: certifi (>=2024.12.14,<2024.13.0)
+Requires-Dist: certifi (>=2024.8.30,<2024.9.0)
 Requires-Dist: charset-normalizer (>=3.0.0,<4.0.0)
 Requires-Dist: django-environ (>=0.11.0,<0.12.0)
 Requires-Dist: ds-caselaw-utils (>=2.0.0,<3.0.0)
@@ -25,7 +25,6 @@ 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: sqids (>=0.5.0,<0.6.0)
 Requires-Dist: typing-extensions (>=4.7.1,<5.0.0)
 Description-Content-Type: text/markdown
 

{ds_caselaw_marklogic_api_client-28.2.0.dist-info → ds_caselaw_marklogic_api_client-29.0.0a2.dist-info}/RECORD

@@ -1,22 +1,20 @@
-caselawclient/Client.py,sha256=6BaVpKSvMcNhSaDXzTkN1h4iV7KFrY7SgSHNGqGQ_2o,43863
+caselawclient/Client.py,sha256=BD6XvObJHILh8C4NvLJUpkEUJ6KpxHOHKokJOrYWaKY,42818
 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
 caselawclient/content_hash.py,sha256=0cPC4OoABq0SC2wYFX9-24DodNigeOqksDxgxQH_hUA,2221
 caselawclient/errors.py,sha256=tV0vs3wYSd331BzmfuRiZV6GAdsd91rtN65ymRaSx3s,3164
 caselawclient/factories.py,sha256=6-xZMVmvtXA8AnyWJgJTums1EWfM6lPIhrWQu0NopJo,4472
-caselawclient/identifier_resolution.py,sha256=IOqrZcIHoHhNOCAkNveOBcWddBNpkOB8cz1r0zFa8mQ,1829
 caselawclient/models/__init__.py,sha256=kd23EUpvaC7aLHdgk8farqKAQEx3lf7RvNT2jEatvlg,68
-caselawclient/models/documents/__init__.py,sha256=DS2jqj4ShRHxAWIqQeeyCCcXctLor4xUvT3sc903D2E,19186
+caselawclient/models/documents/__init__.py,sha256=ZrdtHZ1M8kTODD2Zky6EdUjj8r0KQKIZ6nWtXmuZKuo,18506
 caselawclient/models/documents/body.py,sha256=mtdjmG1WU2qSpyRLS8-PWcSoXpDa2Qz6xlcTbxZgxvA,5603
 caselawclient/models/documents/exceptions.py,sha256=rw1xId16vBKvBImgFmFUpeFgKqU7VTNtVLIEVBPGKyk,374
 caselawclient/models/documents/statuses.py,sha256=Cp4dTQmJOtsU41EJcxy5dV1841pGD2PNWH0VrkDEv4Q,579
 caselawclient/models/documents/transforms/html.xsl,sha256=oSSO-IBX4qLiSWexQYmWJfGNevF09aCBx4D1NYqXxpo,38322
 caselawclient/models/documents/xml.py,sha256=HlmPb63lLMnySSOLP4iexcAyQiLByKBZtTd25f8sY8M,1268
-caselawclient/models/identifiers/__init__.py,sha256=CxKu5iZqPb6BOjrQ3upkzyjU3YSsJhBZd7VATYxDI-A,5061
-caselawclient/models/identifiers/fclid.py,sha256=pAxZKKlKRSHwJqrEHOSlbuCt3gBOgS3sNv98bBjuNBc,1295
+caselawclient/models/identifiers/__init__.py,sha256=FSrhDXMkjgaYnOlBaTDcz9ouzL9ERxEhlqud0fdVT8A,4151
 caselawclient/models/identifiers/neutral_citation.py,sha256=yddlfumdnkrNpoTIOf8dB1foA7hE41-zmlfa17-Ulug,1790
-caselawclient/models/identifiers/unpacker.py,sha256=01gWWlOd_2PxT1GJzOIXbp0G4iGGIxHaTaJMBOQ1TDs,1834
+caselawclient/models/identifiers/unpacker.py,sha256=V79m4rd0FZx5TRueL1m3MbrSUWO8c0f5NoNxSP3dlFw,1757
 caselawclient/models/judgments.py,sha256=NVOg4ZTU7Jtr33UuswL2TXCaN6_W0fKFPK4EdQ-jUhE,1915
 caselawclient/models/neutral_citation_mixin.py,sha256=5ktKCPIDidVRwxVTzx5e242O1BxOdP--1dnatZyTbYI,1773
 caselawclient/models/press_summaries.py,sha256=06flQ8wSLnNxoQtXO0ckmotFKszYZcub0oPcDzYbVQw,1879
@@ -46,7 +44,6 @@ 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_next_document_sequence_number.xqy,sha256=LkGoaS7jZfaKDuZLi0apP5qHP1rpcM1HbqX3RUwquKY,450
 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
@@ -58,7 +55,6 @@ caselawclient/xquery/get_version_annotation.xqy,sha256=pFDMGA9SxI59iUPaoAeUsq23k
 caselawclient/xquery/get_version_created.xqy,sha256=bRweaXFtwMBNzL16SlOdiOxHkbqNUwpwDHLxpZYVCh0,250
 caselawclient/xquery/insert_document.xqy,sha256=iP2xTaLGa-u6X9KfS1yJ6yPCKQUWQFYdEW1S4YcMY7w,531
 caselawclient/xquery/list_judgment_versions.xqy,sha256=WShga8igeD21hSLfVSvCOiDMPDhNH6KGf1OW6G0SAkY,190
-caselawclient/xquery/resolve_from_identifier.xqy,sha256=Fa-RSw9ZwD__BmT5LLJ0J0HcDstDbedkEccv45M3L4g,484
 caselawclient/xquery/set_boolean_property.xqy,sha256=8Vg3yDWqeDynUJQHw2OF4daDIKTnp8ARol1_OCqY0Dk,355
 caselawclient/xquery/set_metadata_citation.xqy,sha256=ImwijXowvOCiH_br_LepnKsEpys9tg4Cf3uz6MoC5-c,659
 caselawclient/xquery/set_metadata_court.xqy,sha256=xQGR3e4pdJuDPMlzdAdzrBDSeQbEFiLVIm2z_KQI_Ds,996
@@ -76,8 +72,8 @@ 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=kybL-YzwK34Fr6MeWfqVOJHYrs0ZNeDWXDsp8o2Yb1U,6114
-ds_caselaw_marklogic_api_client-28.2.0.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
-ds_caselaw_marklogic_api_client-28.2.0.dist-info/METADATA,sha256=i7v85V8SC6lm8oYboo4aGxSmVVaO4QZwj3WVSWDU5Cs,4272
-ds_caselaw_marklogic_api_client-28.2.0.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-ds_caselaw_marklogic_api_client-28.2.0.dist-info/RECORD,,
+caselawclient/xquery_type_dicts.py,sha256=TN-4jPUsus22yBUM6_e4ZaLvYZk_qIhz2gU3Eqg0aaY,5959
+ds_caselaw_marklogic_api_client-29.0.0a2.dist-info/LICENSE.md,sha256=fGMzyyLuQW-IAXUeDSCrRdsYW536aEWThdbpCjo6ZKg,1108
+ds_caselaw_marklogic_api_client-29.0.0a2.dist-info/METADATA,sha256=Tlh3W1mHe_3Th1A3U_TnBhxa64njIGMH6jot3Y_O6vE,4234
+ds_caselaw_marklogic_api_client-29.0.0a2.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+ds_caselaw_marklogic_api_client-29.0.0a2.dist-info/RECORD,,

caselawclient/identifier_resolution.py

@@ -1,43 +0,0 @@
-import json
-from typing import NamedTuple
-
-from caselawclient.models.documents import DocumentURIString
-from caselawclient.xquery_type_dicts import MarkLogicDocumentURIString
-
-
-class IdentifierResolutions(list["IdentifierResolution"]):
-    """
-    A list of candidate MarkLogic documents which correspond to a Public UI uri
-
-    MarkLogic returns a list of dictionaries; IdentifierResolution handles a single dictionary
-    which corresponds to a single identifier to MarkLogic document mapping.
-
-    see `xquery/resolve_from_identifier.xqy` and `resolve_from_identifier` in `Client.py`
-    """
-
-    @staticmethod
-    def from_marklogic_output(table: list[str]) -> "IdentifierResolutions":
-        return IdentifierResolutions(list(IdentifierResolution.from_marklogic_output(row) for row in table))
-
-    def published(self) -> "IdentifierResolutions":
-        "Filter the list so that only published documents are returned"
-        return IdentifierResolutions(list(x for x in self if x.document_published))
-
-
-class IdentifierResolution(NamedTuple):
-    """A single response from MarkLogic about a single identifier / document mapping"""
-
-    identifier_uuid: str
-    document_uri: MarkLogicDocumentURIString
-    identifier_slug: DocumentURIString
-    document_published: bool
-
-    @staticmethod
-    def from_marklogic_output(raw_row: str) -> "IdentifierResolution":
-        row = json.loads(raw_row)
-        return IdentifierResolution(
-            identifier_uuid=row["documents.compiled_url_slugs.identifier_uuid"],
-            document_uri=MarkLogicDocumentURIString(row["documents.compiled_url_slugs.document_uri"]),
-            identifier_slug=DocumentURIString(row["documents.compiled_url_slugs.identifier_slug"]),
-            document_published=row["documents.compiled_url_slugs.document_published"] == "true",
-        )

caselawclient/models/identifiers/fclid.py

@@ -1,48 +0,0 @@
-import re
-from typing import TYPE_CHECKING
-
-from sqids import Sqids
-
-from . import Identifier, IdentifierSchema
-
-if TYPE_CHECKING:
-    from caselawclient.Client import MarklogicApiClient
-
-
-VALID_FCLID_PATTERN = re.compile(r"^[bcdfghjkmnpqrstvwxyz23456789]{4,}$")
-
-FCLID_MINIMUM_LENGTH = 8
-FCLID_ALPHABET = "bcdfghjkmnpqrstvwxyz23456789"
-
-sqids = Sqids(
-    min_length=FCLID_MINIMUM_LENGTH,
-    alphabet=FCLID_ALPHABET,
-)
-
-
-class FindCaseLawIdentifierSchema(IdentifierSchema):
-    """
-    Identifier schema describing a Find Case Law Identifier.
-    """
-
-    name = "Find Case Law Identifier"
-    namespace = "fclid"
-
-    @classmethod
-    def validate_identifier(cls, value: str) -> bool:
-        return bool(VALID_FCLID_PATTERN.match(value))
-
-    @classmethod
-    def compile_identifier_url_slug(cls, value: str) -> str:
-        return "tna." + value
-
-    @classmethod
-    def mint(cls, api_client: "MarklogicApiClient") -> "FindCaseLawIdentifier":
-        """Generate a totally new Find Case Law identifier."""
-        next_sequence_number = api_client.get_next_document_sequence_number()
-        new_identifier = sqids.encode([next_sequence_number])
-        return FindCaseLawIdentifier(value=new_identifier)
-
-
-class FindCaseLawIdentifier(Identifier):
-    schema = FindCaseLawIdentifierSchema

caselawclient/xquery/get_next_document_sequence_number.xqy

@@ -1,14 +0,0 @@
-xquery version "1.0-ml";
-declare option xdmp:transaction-mode "update";
-
-let $_ := xdmp:set-transaction-mode("update")
-let $state_doc := fn:doc("state.xml")
-let $counter_node := $state_doc/state/document_counter
-
-let $current_counter := $counter_node/text()
-let $new_counter := fn:sum(($current_counter, 1))
-
-let $_ := xdmp:node-replace($counter_node, <document_counter>{$new_counter}</document_counter>)
-let $_ := xdmp:commit()
-
-return $new_counter

caselawclient/xquery/resolve_from_identifier.xqy

@@ -1,17 +0,0 @@
-xquery version "1.0-ml";
-
-declare namespace xdmp="http://marklogic.com/xdmp"; 
-declare variable $identifier_uri as xs:string external;
-declare variable $published_only as xs:int? external := 1;
-
-let $published_query := if ($published_only) then " AND document_published = 'true'" else ""
-let $query := "SELECT * from compiled_url_slugs WHERE (identifier_slug = @uri)" ||  $published_query
-
-return xdmp:sql(
-  $query,
-  "map",
-  map:new((
-    map:entry("uri", $identifier_uri)
-  ))
-)
-