Update app.py

app.py

@@ -35,24 +35,31 @@ def get_ocr(lang, use_gpu=False):
 # Function to extract images from PDF
 def pdf_to_images(uploaded_file):
     try:
-        doc = fitz.open(stream=uploaded_file.read(),filetype="pdf")
+        # Load PDF from the uploaded file
+        doc = fitz.open(stream=uploaded_file.read(), filetype="pdf")
+        logger.info(f"PDF loaded successfully with {len(doc)} pages.")
+        
         image_parts = []
-
         for page_number in range(len(doc)):
-            page = pdf_document.load_page(page_number)
+            page = doc.load_page(page_number)
             pix = page.get_pixmap()
             image_data = pix.tobytes("png")
             
+            # Log progress for each page
+            logger.info(f"Processed page {page_number + 1}/{len(doc)}.")
+            
             image_parts.append({
                 "mime_type": "image/png",
                 "data": image_data
             })
-    
+
+        logger.info(f"PDF to image conversion completed with {len(image_parts)} images.")
         return image_parts
+    
     except Exception as e:
         logger.error(f"Error processing PDF: {str(e)}")
         raise HTTPException(status_code=500, detail="Error processing PDF file")
-
+        
 @app.post("/ocr")
 async def create_upload_file(
     file: UploadFile = File(...),