Prepare vectorizes

app/routers/documents.py

@@ -1,8 +1,6 @@
-# app/routers/documents.py
 import uuid
 import mimetypes
-import base64
-from fastapi import APIRouter, UploadFile, File, HTTPException
+from fastapi import APIRouter, UploadFile, File, HTTPException, BackgroundTasks
 from typing import Dict, Any, List
 from fastapi.concurrency import run_in_threadpool
 from PIL import Image
@@ -10,6 +8,7 @@ from io import BytesIO
 
 from .. import agents
 from ..core.pdf_processor import convert_pdf_to_images, image_to_base64_str
+from ..core.ocr import extract_text_from_images
 
 # Create an APIRouter instance
 router = APIRouter(
@@ -46,8 +45,12 @@ async def multimodal_process_pipeline(doc_id: str, image: Image.Image, page_num:
     db_results[final_result["doc_id"]] = final_result
     return final_result
 
-@router.post("/process", summary="upload and process a document")
-async def upload_and_process_document(file: UploadFile = File(...)):
+
+@router.post("/process", summary="Upload and Process Document")
+async def upload_and_process_document(
+        file: UploadFile = File(...),
+        background_tasks: BackgroundTasks = BackgroundTasks()
+):
     if not file.filename:
         raise HTTPException(status_code=400, detail="No file provided.")
 
@@ -57,7 +60,7 @@ async def upload_and_process_document(file: UploadFile = File(...)):
 
     try:
         file_type = mimetypes.guess_type(file.filename)[0]
-        print(f"File type: {file_type}")
+        print(f"Detected file type: {file_type}")
 
         images: List[Image.Image] = []
         if file_type == 'application/pdf':
@@ -84,18 +87,28 @@ async def upload_and_process_document(file: UploadFile = File(...)):
             elif category == "INVOICE":
                 extraction_result = await agents.agent_extract_invoice_info(images_base64, language)
         else:
-            print(f"The document is classified as '{category}'，skipping extraction。")
+            print(f"Document classified as '{category}'，skipping extraction。")
 
-        # 3. Return a unified result
         final_result = {
             "doc_id": doc_id,
+            "message": "Document processing initiated. Vectorization is running in the background.",
             "page_count": len(images),
             "category": category,
             "language": language,
             "extraction_data": extraction_result.dict() if extraction_result else None,
-            "status": "Processed"
+            "status": "Processing"
         }
-        db_results[doc_id] = final_result
+
+        full_text = await run_in_threadpool(extract_text_from_images, images)
+        background_tasks.add_task(
+            agents.agent_vectorize_and_store,
+            doc_id,
+            full_text,
+            category,
+            language
+        )
+        print("--- [Main] Vectorization job added to background tasks.")
+
         return final_result
 
     except Exception as e: