Update main.py

main.py

@@ -6,6 +6,7 @@ from PIL import Image
 from io import BytesIO
 from starlette.middleware import Middleware
 from starlette.middleware.cors import CORSMiddleware
+from pdf2image import convert_from_bytes
 
 app = FastAPI()
 
@@ -73,28 +74,12 @@ async def pdf_question_answering(
         # Read the uploaded file as bytes
         contents = await file.read()
 
-        # Initialize an empty list to store image bytes
-        images = []
-
-        # Use PyMuPDF to process the PDF and convert each page to an image
-        pdf_document = fitz.open_memo(contents, "pdf")
-        
-        for page_num in range(pdf_document.page_count):
-            page = pdf_document.load_page(page_num)
-            print(f"Converting page {page_num + 1} to image...")
-            
-            # Convert the page to an image
-            image = Image.frombytes("RGB", page.get_size(), page.get_pixmap().samples)
-            
-            # Convert the image to bytes
-            img_byte_array = BytesIO()
-            image.save(img_byte_array, format='PNG')
-            images.append(img_byte_array.getvalue())
+        # Convert PDF to images
+        images = convert_from_bytes(contents)
 
         # Perform document question answering for each image
         answers_dict = {}
-        for idx, image_bytes in enumerate(images):
-            image = Image.open(BytesIO(image_bytes))
+        for idx, image in enumerate(images):
             for question in questions.split(','):
                 result = nlp_qa(
                     image,