fix(harvester): conflict if thesis publication date changed

site/cds_rdm/inspire_harvester/update/config.py

@@ -18,7 +18,10 @@
     IdentifiersFieldUpdate,
     RelatedIdentifiersUpdate,
 )
-from cds_rdm.inspire_harvester.update.fields.metadata import PublicationDateUpdate
+from cds_rdm.inspire_harvester.update.fields.metadata import (
+    PublicationDateUpdate,
+    ThesisPublicationDateUpdate,
+)
 
 UPDATE_STRATEGY_CONFIG = {
     # fields not included in the strategy raise error on update attempt
@@ -29,7 +32,7 @@
     "metadata.contributors": CreatibutorsFieldUpdate(strict=False),
     "metadata.identifiers": IdentifiersFieldUpdate(),
     "metadata.related_identifiers": RelatedIdentifiersUpdate(),
-    "metadata.publication_date": OverwriteFieldUpdate(),
+    "metadata.publication_date": ThesisPublicationDateUpdate(),
     "metadata.subjects": ListOfDictAppendUniqueUpdate(key_field="subject"),
     "metadata.languages": ListOfDictAppendUniqueUpdate(key_field="id"),
     "metadata.description": OverwriteFieldUpdate(),

site/cds_rdm/inspire_harvester/update/fields/metadata.py

@@ -11,11 +11,42 @@
 
 import dateparser
 
+from cds_rdm.inspire_harvester.transform.resource_types import ResourceType
 from cds_rdm.inspire_harvester.update.engine import UpdateConflict, UpdateResult
 from cds_rdm.inspire_harvester.update.field import FieldUpdateBase
+from cds_rdm.inspire_harvester.update.fields.base import OverwriteFieldUpdate
 from cds_rdm.inspire_harvester.utils import get_path, set_path
 
 
+class ThesisPublicationDateUpdate(OverwriteFieldUpdate):
+    """Conflict if a thesis publication_date changed; otherwise overwrite."""
+
+    def update(self, current, incoming, path, ctx):
+        """Return a conflict when a thesis date differs; otherwise overwrite."""
+        current_date = get_path(current, path)
+        incoming_date = get_path(incoming, path)
+        resource_type = get_path(current, "metadata.resource_type.id")
+
+        if resource_type != ResourceType.THESIS:
+            return super().update(current, incoming, path, ctx)
+
+        if current_date and incoming_date and current_date != incoming_date:
+            return UpdateResult(
+                updated=current,
+                conflicts=[
+                    UpdateConflict(
+                        path=path,
+                        kind="date_mismatch",
+                        message="Incoming thesis publication_date differs",
+                        current=current_date,
+                        incoming=incoming_date,
+                    )
+                ],
+            )
+
+        return super().update(current, incoming, path, ctx)
+
+
 class PublicationDateUpdate(FieldUpdateBase):
     """
     Update publication_date only if incoming value is more accurate.

site/tests/inspire_harvester/test_writer.py

@@ -452,7 +452,7 @@ def test_writer_1_existing_found_files_not_changed_metadata_changed(
 def test_writer_updates_publication_date_from_inspire_without_cds_doi(
     running_app, location, transformed_record_1_file, scientific_community
 ):
-    """Test INSPIRE publication-date mismatches update non-CDS DOI records."""
+    """Test a different incoming thesis date is a conflict, even without a CDS DOI."""
     writer = InspireWriter()
     transformed_record = deepcopy(transformed_record_1_file)
 
@@ -469,8 +469,8 @@ def test_writer_updates_publication_date_from_inspire_without_cds_doi(
     RDMRecord.index.refresh()
 
     updated = current_rdm_records_service.read(system_identity, created["id"])
-    assert updated["metadata"]["publication_date"] == "2014"
-    assert not update_entry.errors
+    assert updated["metadata"]["publication_date"] == "2020"
+    assert any("[date_mismatch]" in error for error in update_entry.errors)
 
     _cleanup_record(created["id"])