dsp-tools 17.0.0.post21__tar.gz → 17.0.0.post25__tar.gz

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: dsp-tools
-Version: 17.0.0.post21
+Version: 17.0.0.post25
 Summary: DSP-TOOLS is a Python package with a command line interface that helps you interact with a DaSCH service platform (DSP) server.
 Author: DaSCH - Swiss National Data and Service Center for the Humanities
 Author-email: DaSCH - Swiss National Data and Service Center for the Humanities <info@dasch.swiss>

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/pyproject.toml

@@ -2,7 +2,7 @@
 
 [project]
 name = "dsp-tools"
-version = "17.0.0.post21"
+version = "17.0.0.post25"
 description = "DSP-TOOLS is a Python package with a command line interface that helps you interact with a DaSCH service platform (DSP) server."
 authors = [
     { name = "DaSCH - Swiss National Data and Service Center for the Humanities", email = "info@dasch.swiss" },

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/src/dsp_tools/commands/create/parsing/parsing_utils.py

@@ -4,6 +4,7 @@ import regex
 
 from dsp_tools.commands.create.constants import KNORA_API_STR
 from dsp_tools.commands.create.constants import UNIVERSAL_PREFIXES
+from dsp_tools.utils.data_formats.iri_util import make_dsp_ontology_prefix
 from dsp_tools.utils.data_formats.uri_util import is_uri
 
 
@@ -27,7 +28,7 @@ def create_prefix_lookup(project_json: dict[str, Any], api_url: str) -> dict[str
     shortcode = project_json["project"]["shortcode"]
     for onto in project_json["project"]["ontologies"]:
         onto_name = onto["name"]
-        defined_prefixes[onto_name] = _make_dsp_ontology_prefix(api_url, shortcode, onto_name)
+        defined_prefixes[onto_name] = make_dsp_ontology_prefix(api_url, shortcode, onto_name)
     return defined_prefixes
 
 
@@ -37,7 +38,3 @@ def _correct_external_prefix(prefixes: dict[str, str]) -> dict[str, str]:
             continue
         prefixes[prfx] = f"{namespace}/"
     return prefixes
-
-
-def _make_dsp_ontology_prefix(api_url: str, shortcode: str, onto_name: str) -> str:
-    return f"{api_url}/ontology/{shortcode}/{onto_name}/v2#"

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/src/dsp_tools/commands/validate_data/models/api_responses.py

@@ -16,6 +16,7 @@ class SHACLValidationReport:
 class ProjectDataFromApi:
     all_lists: list[OneList]
     enabled_licenses: EnabledLicenseIris
+    resource_iris_in_db: list[InfoForResourceInDB]
 
 
 @dataclass
@@ -50,3 +51,9 @@ class SHACLListInfo:
 @dataclass
 class EnabledLicenseIris:
     enabled_licenses: list[str]
+
+
+@dataclass
+class InfoForResourceInDB:
+    res_iri: str
+    res_type: str

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/src/dsp_tools/commands/validate_data/models/validation.py

@@ -23,6 +23,7 @@ class RDFGraphs:
     cardinality_shapes: Graph
     content_shapes: Graph
     knora_api: Graph
+    resources_in_db_graph: Graph
 
 
 @dataclass

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/src/dsp_tools/commands/validate_data/prepare_data/make_data_graph.py

@@ -48,7 +48,10 @@ def _make_one_value(val: RdfLikeValue, res_iri: URIRef) -> Graph:
         return g
     if val.knora_type == KnoraValueType.LINK_VALUE:
         link_val = val.user_facing_value if val.user_facing_value else ""
-        triple_object: Literal | URIRef = DATA[link_val]
+        if link_val.startswith("http://rdfh.ch/"):
+            triple_object: Literal | URIRef = URIRef(link_val)
+        else:
+            triple_object = DATA[link_val]
     else:
         triple_object = _make_one_rdflib_object(val.user_facing_value, VALUE_INFO_TRIPLE_OBJECT_TYPE[val.knora_type])
     g.add((val_iri, prop_type_info.knora_prop, triple_object))

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/src/dsp_tools/commands/validate_data/prepare_data/prepare_data.py

@@ -2,13 +2,18 @@ import importlib.resources
 from pathlib import Path
 
 from loguru import logger
+from rdflib import RDF
 from rdflib import Graph
+from rdflib import URIRef
 
 from dsp_tools.clients.authentication_client import AuthenticationClient
 from dsp_tools.clients.legal_info_client_live import LegalInfoClientLive
+from dsp_tools.clients.metadata_client import MetadataRetrieval
+from dsp_tools.clients.metadata_client_live import MetadataClientLive
 from dsp_tools.commands.validate_data.api_clients import ListClient
 from dsp_tools.commands.validate_data.api_clients import OntologyClient
 from dsp_tools.commands.validate_data.models.api_responses import EnabledLicenseIris
+from dsp_tools.commands.validate_data.models.api_responses import InfoForResourceInDB
 from dsp_tools.commands.validate_data.models.api_responses import ListLookup
 from dsp_tools.commands.validate_data.models.api_responses import OneList
 from dsp_tools.commands.validate_data.models.api_responses import ProjectDataFromApi
@@ -42,13 +47,13 @@ def prepare_data_for_validation_from_parsed_resource(
     permission_ids: list[str],
     auth: AuthenticationClient,
     shortcode: str,
-) -> tuple[RDFGraphs, set[str]]:
+) -> tuple[RDFGraphs, set[str], MetadataRetrieval]:
     used_iris = {x.res_type for x in parsed_resources}
-    proj_info = _get_project_specific_information_from_api(auth, shortcode)
+    proj_info, metadata_retrieval_success = _get_project_specific_information_from_api(auth, shortcode)
     list_lookup = _make_list_lookup(proj_info.all_lists)
     data_rdf = _make_data_graph_from_parsed_resources(parsed_resources, authorship_lookup, list_lookup)
     rdf_graphs = _create_graphs(data_rdf, shortcode, auth, proj_info, permission_ids)
-    return rdf_graphs, used_iris
+    return rdf_graphs, used_iris, metadata_retrieval_success
 
 
 def _make_list_lookup(project_lists: list[OneList]) -> ListLookup:
@@ -60,11 +65,23 @@ def _make_list_lookup(project_lists: list[OneList]) -> ListLookup:
     return ListLookup(lookup)
 
 
-def _get_project_specific_information_from_api(auth: AuthenticationClient, shortcode: str) -> ProjectDataFromApi:
+def _get_project_specific_information_from_api(
+    auth: AuthenticationClient, shortcode: str
+) -> tuple[ProjectDataFromApi, MetadataRetrieval]:
     list_client = ListClient(auth.server, shortcode)
     all_lists = list_client.get_lists()
     enabled_licenses = _get_license_iris(shortcode, auth)
-    return ProjectDataFromApi(all_lists, enabled_licenses)
+    retrieval_status, formatted_metadata = _get_metadata_info(auth, shortcode)
+    return ProjectDataFromApi(all_lists, enabled_licenses, formatted_metadata), retrieval_status
+
+
+def _get_metadata_info(
+    auth: AuthenticationClient, shortcode: str
+) -> tuple[MetadataRetrieval, list[InfoForResourceInDB]]:
+    metadata_client = MetadataClientLive(auth.server, auth)
+    retrieval_status, metadata = metadata_client.get_resource_metadata(shortcode)
+    formatted_metadata = [InfoForResourceInDB(x["resourceIri"], x["resourceClassIri"]) for x in metadata]
+    return retrieval_status, formatted_metadata
 
 
 def _make_data_graph_from_parsed_resources(
@@ -99,6 +116,8 @@ def _create_graphs(
     api_card_shapes.parse(str(api_card_path))
     content_shapes = shapes.content + api_shapes
     card_shapes = shapes.cardinality + api_card_shapes
+    resources_in_db = _make_resource_in_db_graph(proj_info.resource_iris_in_db)
+    resources_in_db = _bind_prefixes_to_graph(resources_in_db, onto_iris)
     data_rdf = _bind_prefixes_to_graph(data_rdf, onto_iris)
     ontologies = _bind_prefixes_to_graph(ontologies, onto_iris)
     card_shapes = _bind_prefixes_to_graph(card_shapes, onto_iris)
@@ -110,6 +129,7 @@ def _create_graphs(
         cardinality_shapes=card_shapes,
         content_shapes=content_shapes,
         knora_api=knora_api,
+        resources_in_db_graph=resources_in_db,
     )
 
 
@@ -143,3 +163,10 @@ def _get_license_iris(shortcode: str, auth: AuthenticationClient) -> EnabledLice
     license_info = legal_client.get_licenses_of_a_project()
     iris = [x["id"] for x in license_info]
     return EnabledLicenseIris(iris)
+
+
+def _make_resource_in_db_graph(resources_in_db: list[InfoForResourceInDB]) -> Graph:
+    g = Graph()
+    for r in resources_in_db:
+        g.add((URIRef(r.res_iri), RDF.type, URIRef(r.res_type)))
+    return g

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/src/dsp_tools/commands/validate_data/validate_data.py

@@ -104,7 +104,7 @@ def validate_parsed_resources(
     config: ValidateDataConfig,
     auth: AuthenticationClient,
 ) -> bool:
-    rdf_graphs, used_iris = prepare_data_for_validation_from_parsed_resource(
+    rdf_graphs, used_iris, _ = prepare_data_for_validation_from_parsed_resource(
         parsed_resources=parsed_resources,
         authorship_lookup=authorship_lookup,
         permission_ids=permission_ids,

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/src/dsp_tools/commands/validate_data/validation/get_validation_report.py

@@ -82,10 +82,11 @@ def _create_and_write_graphs(rdf_graphs: RDFGraphs, tmp_path: Path) -> None:
     card_shape_str = rdf_graphs.cardinality_shapes.serialize(format="ttl")
     content_shape_str = rdf_graphs.content_shapes.serialize(format="ttl")
     knora_api_str = rdf_graphs.knora_api.serialize(format="ttl")
+    res_in_db_str = rdf_graphs.resources_in_db_graph.serialize(format="ttl")
     turtle_paths_and_graphs = [
         (tmp_path / CARDINALITY_DATA_TTL, data_str),
         (tmp_path / CARDINALITY_SHACL_TTL, card_shape_str + ontos_str + knora_api_str),
-        (tmp_path / CONTENT_DATA_TTL, data_str + ontos_str + knora_api_str),
+        (tmp_path / CONTENT_DATA_TTL, data_str + ontos_str + knora_api_str + res_in_db_str),
         (tmp_path / CONTENT_SHACL_TTL, content_shape_str + ontos_str + knora_api_str),
     ]
     for f_path, content in turtle_paths_and_graphs:

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/src/dsp_tools/utils/data_formats/iri_util.py

@@ -19,3 +19,12 @@ def from_dsp_iri_to_prefixed_iri(iri: str) -> str:
     if not (found := regex.search(dsp_iri_re, iri)):
         return iri
     return f"{found.group(1)}:{found.group(2)}"
+
+
+def make_dsp_ontology_prefix(api_url: str, shortcode: str, onto_name: str) -> str:
+    api_fixed = convert_api_url_for_correct_iri_namespace_construction(api_url)
+    return f"{api_fixed}/ontology/{shortcode}/{onto_name}/v2#"
+
+
+def convert_api_url_for_correct_iri_namespace_construction(api_url: str) -> str:
+    return regex.sub(r"^https", "http", api_url)

{dsp_tools-17.0.0.post21 → dsp_tools-17.0.0.post25}/src/dsp_tools/utils/xml_parsing/get_parsed_resources.py

@@ -5,6 +5,7 @@ from lxml import etree
 
 from dsp_tools.commands.validate_data.mappers import XML_TAG_TO_VALUE_TYPE_MAPPER
 from dsp_tools.error.exceptions import InputError
+from dsp_tools.utils.data_formats.iri_util import convert_api_url_for_correct_iri_namespace_construction
 from dsp_tools.utils.rdflib_constants import KNORA_API_STR
 from dsp_tools.utils.xml_parsing.models.parsed_resource import KnoraValueType
 from dsp_tools.utils.xml_parsing.models.parsed_resource import ParsedFileValue
@@ -15,7 +16,7 @@ from dsp_tools.utils.xml_parsing.models.parsed_resource import ParsedValue
 
 
 def get_parsed_resources(root: etree._Element, api_url: str) -> list[ParsedResource]:
-    api_url = _convert_api_url_for_correct_iri_namespace_construction(api_url)
+    api_url = convert_api_url_for_correct_iri_namespace_construction(api_url)
     iri_lookup = _create_from_local_name_to_absolute_iri_lookup(root, api_url)
     all_res: list[ParsedResource] = []
     for res in root.iterdescendants(tag="resource"):
@@ -34,10 +35,6 @@ def get_parsed_resources(root: etree._Element, api_url: str) -> list[ParsedResou
     return all_res
 
 
-def _convert_api_url_for_correct_iri_namespace_construction(api_url: str) -> str:
-    return regex.sub(r"^https", "http", api_url)
-
-
 def _create_from_local_name_to_absolute_iri_lookup(root: etree._Element, api_url: str) -> dict[str, str]:
     shortcode = root.attrib["shortcode"]
     default_ontology = root.attrib["default-ontology"]