[issue-547] rdf parser: use licenseID from extracted licensing info URIRef if no licenseId node is provided

Signed-off-by: Meret Behrens <[email protected]>

Author: meretp

src/spdx/parser/rdf/extracted_licensing_info_parser.py

@@ -10,9 +10,18 @@
 from spdx.rdfschema.namespace import SPDX_NAMESPACE
 
 
-def parse_extracted_licensing_info(extracted_licensing_info_node: URIRef, graph: Graph) -> ExtractedLicensingInfo:
+def parse_extracted_licensing_info(
+    extracted_licensing_info_node: URIRef, graph: Graph, doc_namespace: str
+) -> ExtractedLicensingInfo:
     logger = Logger()
     license_id = parse_literal(logger, graph, extracted_licensing_info_node, SPDX_NAMESPACE.licenseId)
+    if not license_id:
+        license_id = (
+            extracted_licensing_info_node.fragment
+            if extracted_licensing_info_node.startswith(f"{doc_namespace}#")
+            else extracted_licensing_info_node.toPython()
+        )
+
     extracted_text = parse_literal(logger, graph, extracted_licensing_info_node, SPDX_NAMESPACE.extractedText)
     comment = parse_literal(logger, graph, extracted_licensing_info_node, RDFS.comment)
     license_name = parse_literal_or_no_assertion_or_none(

src/spdx/parser/rdf/rdf_parser.py

@@ -65,7 +65,9 @@ def translate_graph_to_document(graph: Graph) -> Document:
     extracted_licensing_infos = []
     for _, _, extracted_licensing_info_node in graph.triples((None, SPDX_NAMESPACE.hasExtractedLicensingInfo, None)):
         try:
-            extracted_licensing_infos.append(parse_extracted_licensing_info(extracted_licensing_info_node, graph))
+            extracted_licensing_infos.append(
+                parse_extracted_licensing_info(extracted_licensing_info_node, graph, creation_info.document_namespace)
+            )
         except SPDXParsingError as err:
             logger.extend(err.get_messages())
     parsed_fields["extracted_licensing_info"] = extracted_licensing_infos