#1022 - Temporarily skip validator during harvest process.

dags/shared_tasks/mapping_tasks.py

@@ -151,25 +151,36 @@ def validate_collection_task(
         "[3433/vernacular_metadata_v1/mapped_metadata_v1/3.jsonl]"
     ]
     """
-    mapped_page_batches = [json.loads(batch) for batch in mapped_page_batches]
-    mapped_pages = list(chain.from_iterable(mapped_page_batches))
-    mapped_pages = [path for path in mapped_pages if 'children' not in path]
+    send_event_to_sns(context, {
+        "validation": "skipped", 
+        "collection_id": collection_id,
+        "mapped_page_batches": mapped_page_batches
+    })
+    return (
+        f"Skipping validation of {mapped_page_batches} for {collection_id} "
+        "until we can re-implement validator to compare against OpenSearch "
+        "data."
+    )
 
-    num_rows, version_page = create_collection_validation_csv(
-        collection_id, mapped_pages)
+    # mapped_page_batches = [json.loads(batch) for batch in mapped_page_batches]
+    # mapped_pages = list(chain.from_iterable(mapped_page_batches))
+    # mapped_pages = [path for path in mapped_pages if 'children' not in path]
 
-    status = ValidationReportStatus(
-        filepath=version_page,
-        num_validation_errors=num_rows,
-        mapped_version=get_version(collection_id, mapped_pages[0])
-    )
+    # num_rows, version_page = create_collection_validation_csv(
+    #     collection_id, mapped_pages)
+
+    # status = ValidationReportStatus(
+    #     filepath=version_page,
+    #     num_validation_errors=num_rows,
+    #     mapped_version=get_version(collection_id, mapped_pages[0])
+    # )
 
-    print(f"Output {num_rows} rows to {version_page}")
-    print_s3_link(version_page, get_version(collection_id, mapped_pages[0]))
+    # print(f"Output {num_rows} rows to {version_page}")
+    # print_s3_link(version_page, get_version(collection_id, mapped_pages[0]))
 
-    send_event_to_sns(context, asdict(status))
+    # send_event_to_sns(context, asdict(status))
 
-    return version_page
+    # return version_page
 
 
 @task_group(group_id='mapping')