mutil thread

api/core/indexing_runner.py

@@ -3,6 +3,7 @@ import datetime
 import json
 import logging
 import re
+import threading
 import time
 import uuid
 from concurrent.futures import ThreadPoolExecutor, as_completed
@@ -496,9 +497,7 @@ class IndexingRunner:
             # parse document to nodes
             documents = splitter.split_documents([text_doc])
 
-            split_documents = []
-
-            def format_document(flask_app: Flask, document_node: Document) -> List[Document]:
+            def format_document(flask_app: Flask, document_node: Document, split_documents: List) -> List[Document]:
                 with flask_app.app_context():
                     print("process:"+document_node.page_content)
                     format_documents = []
@@ -529,12 +528,20 @@ class IndexingRunner:
                             qa_documents.append(qa_document)
                         format_documents.extend(qa_documents)
 
-                    return format_documents
-
-            with ThreadPoolExecutor() as executor:
-                future_to_doc = {executor.submit(format_document, current_app._get_current_object(), doc): doc for doc in documents}
-                for future in concurrent.futures.as_completed(future_to_doc):
-                    split_documents.extend(future.result())
+                    split_documents.append(format_documents)
+            split_documents = []
+            threads = []
+            for doc in documents:
+                document_format_thread = threading.Thread(target=format_document, kwargs={
+                    'flask_app': current_app._get_current_object(), 'document_node': doc, 'split_documents': split_documents})
+                threads.append(document_format_thread)
+                document_format_thread.start()
+            for thread in threads:
+                thread.join()
+            # with ThreadPoolExecutor() as executor:
+            #     future_to_doc = {executor.submit(format_document, current_app._get_current_object(), doc): doc for doc in documents}
+            #     for future in concurrent.futures.as_completed(future_to_doc):
+            #         split_documents.extend(future.result())
 
             all_documents.extend(split_documents)