Made task_executor async to speedup parsing (#5530)

### What problem does this PR solve?

Made task_executor async to speedup parsing

### Type of change

- [x] Performance Improvement

api/apps/conversation_app.py

@@ -17,6 +17,7 @@ import json
 import re
 import traceback
 from copy import deepcopy
+import trio
 from api.db.db_models import APIToken
 
 from api.db.services.conversation_service import ConversationService, structure_answer
@@ -386,7 +387,8 @@ def mindmap():
                                            rank_feature=label_question(question, [kb])
                                            )
     mindmap = MindMapExtractor(chat_mdl)
-    mind_map = mindmap([c["content_with_weight"] for c in ranks["chunks"]]).output
+    mind_map = trio.run(mindmap, [c["content_with_weight"] for c in ranks["chunks"]])
+    mind_map = mind_map.output
     if "error" in mind_map:
         return server_error_response(Exception(mind_map["error"]))
     return get_json_result(data=mind_map)

api/db/services/document_service.py

@@ -22,6 +22,7 @@ from concurrent.futures import ThreadPoolExecutor
 from copy import deepcopy
 from datetime import datetime
 from io import BytesIO
+import trio
 
 from peewee import fn
 
@@ -597,8 +598,8 @@ def doc_upload_and_parse(conversation_id, file_objs, user_id):
         if parser_ids[doc_id] != ParserType.PICTURE.value:
             mindmap = MindMapExtractor(llm_bdl)
             try:
-                mind_map = json.dumps(mindmap([c["content_with_weight"] for c in docs if c["doc_id"] == doc_id]).output,
-                                      ensure_ascii=False, indent=2)
+                mind_map = trio.run(mindmap, [c["content_with_weight"] for c in docs if c["doc_id"] == doc_id])
+                mind_map = json.dumps(mind_map.output, ensure_ascii=False, indent=2)
                 if len(mind_map) < 32:
                     raise Exception("Few content: " + mind_map)
                 cks.append({

api/utils/file_utils.py

@@ -17,6 +17,8 @@ import base64
 import json
 import os
 import re
+import sys
+import threading
 from io import BytesIO
 
 import pdfplumber
@@ -30,6 +32,10 @@ from api.constants import IMG_BASE64_PREFIX
 PROJECT_BASE = os.getenv("RAG_PROJECT_BASE") or os.getenv("RAG_DEPLOY_BASE")
 RAG_BASE = os.getenv("RAG_BASE")
 
+LOCK_KEY_pdfplumber = "global_shared_lock_pdfplumber"
+if LOCK_KEY_pdfplumber not in sys.modules:
+    sys.modules[LOCK_KEY_pdfplumber] = threading.Lock()
+
 
 def get_project_base_directory(*args):
     global PROJECT_BASE
@@ -175,19 +181,20 @@ def thumbnail_img(filename, blob):
     """
     filename = filename.lower()
     if re.match(r".*\.pdf$", filename):
-        pdf = pdfplumber.open(BytesIO(blob))
-        buffered = BytesIO()
-        resolution = 32
-        img = None
-        for _ in range(10):
-            # https://github.com/jsvine/pdfplumber?tab=readme-ov-file#creating-a-pageimage-with-to_image
-            pdf.pages[0].to_image(resolution=resolution).annotated.save(buffered, format="png")
-            img = buffered.getvalue()
-            if len(img) >= 64000 and resolution >= 2:
-                resolution = resolution / 2
-                buffered = BytesIO()
-            else:
-                break
+        with sys.modules[LOCK_KEY_pdfplumber]:
+            pdf = pdfplumber.open(BytesIO(blob))
+            buffered = BytesIO()
+            resolution = 32
+            img = None
+            for _ in range(10):
+                # https://github.com/jsvine/pdfplumber?tab=readme-ov-file#creating-a-pageimage-with-to_image
+                pdf.pages[0].to_image(resolution=resolution).annotated.save(buffered, format="png")
+                img = buffered.getvalue()
+                if len(img) >= 64000 and resolution >= 2:
+                    resolution = resolution / 2
+                    buffered = BytesIO()
+                else:
+                    break
         pdf.close()
         return img
 

api/utils/log_utils.py

@@ -18,6 +18,8 @@ import os.path
 import logging
 from logging.handlers import RotatingFileHandler
 
+initialized_root_logger = False
+
 def get_project_base_directory():
     PROJECT_BASE = os.path.abspath(
         os.path.join(
@@ -29,10 +31,13 @@ def get_project_base_directory():
     return PROJECT_BASE
 
 def initRootLogger(logfile_basename: str, log_format: str = "%(asctime)-15s %(levelname)-8s %(process)d %(message)s"):
-    logger = logging.getLogger()
-    if logger.hasHandlers():
+    global initialized_root_logger
+    if initialized_root_logger:
         return
+    initialized_root_logger = True
 
+    logger = logging.getLogger()
+    logger.handlers.clear()
     log_path = os.path.abspath(os.path.join(get_project_base_directory(), "logs", f"{logfile_basename}.log"))
 
     os.makedirs(os.path.dirname(log_path), exist_ok=True)