Set dc.identifier.oclc for digitized theses item submissions

dsc/reports/base.py

@@ -165,7 +165,7 @@ class CreateReport(Report):
 
     @property
     def subject(self) -> str:
-        return f"DSC Create Batch Results - {self.workflow_name}, batch='{self.batch_id}'"
+        return f"[{CONFIG.workspace}] DSC Create Batch Results - {self.workflow_name}, batch='{self.batch_id}'"  # noqa: E501
 
     @property
     def summary_template(self) -> Template:
@@ -196,7 +196,7 @@ def create_errors_csv(self) -> StringIO | None:
 class SubmitReport(Report):
     @property
     def subject(self) -> str:
-        return f"DSC Submit Results - {self.workflow_name}, batch='{self.batch_id}'"
+        return f"[{CONFIG.workspace}] DSC Submit Results - {self.workflow_name}, batch='{self.batch_id}'"  # noqa: E501
 
     @property
     def summary_template(self) -> Template:

dsc/workflows/digitized_theses/workflow.py

@@ -512,9 +512,10 @@ def submit_items(self, collection_handle: str | None = None) -> list:
             try:
                 # get item metadata
                 item_metadata = self._get_transformed_metadata(
+                    item_identifier=item_submission.item_identifier,
                     source_metadata_file=manifest[item_submission.item_identifier][
                         "metadata_file"
-                    ]
+                    ],
                 )
 
                 # prepare submission assets
@@ -620,7 +621,9 @@ def _load_batch_manifest(self) -> dict:
 
         return manifest
 
-    def _get_transformed_metadata(self, source_metadata_file: str) -> dict:
+    def _get_transformed_metadata(
+        self, item_identifier: str, source_metadata_file: str
+    ) -> dict:
         """Get transformed metadata for an item submission.
 
         This method expects a filepath to an Alma MARC XML file.
@@ -638,6 +641,9 @@ def _get_transformed_metadata(self, source_metadata_file: str) -> dict:
 
         transformed_metadata = self.metadata_transformer.transform(source_metadata)
 
+        # set dc.identifier.oclc to item identifier
+        transformed_metadata["dc.identifier.oclc"] = item_identifier
+
         # if replacement thesis, include additional dc.description.provenance entry
         if "replacement-theses" in source_metadata_file:
             replacement_message = f"The thesis import has been updated on {self.run_date.strftime('%Y-%m-%dT%H:%M:%SZ')}"  # noqa: E501

pyproject.toml

@@ -17,6 +17,7 @@ dependencies = [
     "lxml>=6.0.2",
     "pandas>=2.3.3",
     "pynamodb>=6.1.0",
+    "requests>=2.33.0",
     "sentry-sdk>=2.50.0",
     "smart_open",
 ]

tests/test_report.py

@@ -16,7 +16,9 @@ def test_report_init_success():
     assert create_report.workflow_name == "test"
     assert create_report.batch_id == "batch-aaa"
     assert create_report.report_date == "20250101T090000Z"
-    assert create_report.subject == ("DSC Create Batch Results - test, batch='batch-aaa'")
+    assert create_report.subject == (
+        "[test] DSC Create Batch Results - test, batch='batch-aaa'"
+    )
 
 
 def test_report_get_item_submissions(mock_item_submission_db_with_records):

tests/workflows/digitized_theses/test_workflow.py

@@ -486,9 +486,11 @@ def test_workflow_load_batch_manifest(mock_s3_digitized_theses_dsc):
 def test_workflow_get_transformed_metadata(mock_s3_digitized_theses_dsc):
     workflow = DigitizedTheses(batch_id="batch-aaa")
     item_metadata = workflow._get_transformed_metadata(
-        source_metadata_file="tests/fixtures/digitized-theses/batch-aaa/replacement-theses/05588126/05588126.xml"
+        item_identifier="05588126",
+        source_metadata_file="tests/fixtures/digitized-theses/batch-aaa/replacement-theses/05588126/05588126.xml",
     )
 
+    assert item_metadata["dc.identifier.oclc"] == "05588126"
     assert item_metadata["dc.title"] == [
         "Global solvability of invariant differential operators."
     ]