Merge branch 'main' into feat/doughnut-http-endpoint-1

.pre-commit-config.yaml

@@ -3,7 +3,7 @@ repos:
     rev: v5.0.0
     hooks:
     -   id: trailing-whitespace
-        exclude: '^(docs)'
+        exclude: '^(docs|data)'
     -   id: end-of-file-fixer
     -   id: check-added-large-files
         args: [-- maxkb=1500]

client/src/nv_ingest_client/client/client.py

@@ -418,7 +418,7 @@ def fetch_with_retries(job_id: str):
                 try:
                     result, _ = handle_future_result(future, timeout=timeout)
                     # Append a tuple of (result data, job_id). (Using result.get("data") if result is valid.)
-                    results.append((result.get("data") if result else None, job_id))
+                    results.append(result.get("data"))
                     # Run the callback if provided and the result is valid
                     if completion_callback and result:
                         completion_callback(result, job_id)

client/src/nv_ingest_client/util/milvus.py

@@ -370,6 +370,28 @@ def _pull_text(element, enable_text: bool, enable_charts: bool, enable_tables: b
     return text
 
 
+def _insert_location_into_content_metadata(element, enable_charts: bool, enable_tables: bool, enable_images: bool):
+    location = max_dimensions = None
+    if element["document_type"] == "structured":
+        location = element["metadata"]["table_metadata"]["table_location"]
+        max_dimensions = element["metadata"]["table_metadata"]["table_location_max_dimensions"]
+        if element["metadata"]["content_metadata"]["subtype"] == "chart" and not enable_charts:
+            location = max_dimensions = None
+        elif element["metadata"]["content_metadata"]["subtype"] == "table" and not enable_tables:
+            location = max_dimensions = None
+    elif element["document_type"] == "image" and enable_images:
+        location = element["metadata"]["image_metadata"]["image_location"]
+        max_dimensions = element["metadata"]["image_metadata"]["image_location_max_dimensions"]
+    if (not location) and (element["document_type"] != "text"):
+        source_name = element["metadata"]["source_metadata"]["source_name"]
+        pg_num = element["metadata"]["content_metadata"]["page_number"]
+        doc_type = element["document_type"]
+        logger.error(f"failed to find location for entity: {source_name} page: {pg_num} type: {doc_type}")
+        location = max_dimensions = None
+    element["metadata"]["content_metadata"]["location"] = location
+    element["metadata"]["content_metadata"]["max_dimensions"] = max_dimensions
+
+
 def write_records_minio(
     records,
     writer: RemoteBulkWriter,
@@ -415,6 +437,7 @@ def write_records_minio(
     for result in records:
         for element in result:
             text = _pull_text(element, enable_text, enable_charts, enable_tables, enable_images)
+            _insert_location_into_content_metadata(element, enable_charts, enable_tables, enable_images)
             if text:
                 if sparse_model is not None:
                     writer.append_row(record_func(text, element, sparse_model.encode_documents([text])))