fix(harvester): put CERN- report numbers in identifiers

site/cds_rdm/inspire_harvester/load/matcher.py

@@ -110,10 +110,10 @@ class ReportNumberMatchFilter(FilterCandidate):
     def query(self):
         """Build the CDS report number query."""
         return [
-            dsl.Q("term", **{"metadata.related_identifiers.scheme": "cdsrn"}),
+            dsl.Q("term", **{"metadata.identifiers.scheme": "cdsrn"}),
             dsl.Q(
                 "term",
-                **{"metadata.related_identifiers.identifier": self.value},
+                **{"metadata.identifiers.identifier": self.value},
             ),
         ]
 
@@ -163,11 +163,12 @@ def _retrieve_identifier(self, identifiers, scheme) -> Optional[str]:
     def _build_filter_priority(self, entry, inspire_id, cdsrdm_id):
         """Build the priority-ordered record match candidates."""
         doi = entry.get("pids", {}).get("doi", {}).get("identifier")
+        identifiers = entry["metadata"].get("identifiers", [])
         related_identifiers = entry["metadata"].get("related_identifiers", [])
 
         cds_id = self._retrieve_identifier(related_identifiers, "cds")
         arxiv_id = self._retrieve_identifier(related_identifiers, "arxiv")
-        report_number = self._retrieve_identifier(related_identifiers, "cdsrn")
+        report_number = self._retrieve_identifier(identifiers, "cdsrn")
         return [
             ParentMatchFilter(value=cdsrdm_id),
             CDSIdentifierMatchFilter(value=cds_id),

site/cds_rdm/inspire_harvester/transform/mappers/identifiers.py

@@ -104,6 +104,13 @@ def map_value(self, src_record, ctx, logger):
                     "Unexpected schema in external_system_identifiers. "
                     f"| details: schema={schema}, value={value}"
                 )
+
+        # CDS report numbers that start with CERN- belong on the record itself.
+        for rn in src_metadata.get("report_numbers", []):
+            value = rn.get("value")
+            if value and value.startswith("CERN-"):
+                identifiers.append({"identifier": value, "scheme": "cdsrn"})
+
         unique_ids = [dict(t) for t in {tuple(sorted(d.items())) for d in identifiers}]
         return unique_ids
 
@@ -211,10 +218,13 @@ def map_value(self, src_record, ctx, logger):
 
             report_numbers = src_metadata.get("report_numbers", [])
             for rn in report_numbers:
+                value = rn.get("value")
+                if not value or value.startswith("CERN-"):
+                    continue
                 identifiers.append(
                     {
                         "scheme": "cdsrn",
-                        "identifier": f"{rn['value']}",
+                        "identifier": value,
                         "relation_type": {"id": "isvariantformof"},
                         "resource_type": {"id": ctx.resource_type.value},
                     }