Update app.py

app.py

@@ -131,13 +131,14 @@ def convert_pdf_to_images_gradio(pdf_file):
         raise Exception(f"Error converting PDF: {str(e)}")
 
 
-def process_pdf_document(pdf_file, model, progress=gr.Progress()):
+def process_pdf_document(pdf_file, model, progress_callback=None):
     """Process uploaded PDF file page by page"""
     if pdf_file is None:
         return "No PDF file uploaded", ""
     
     try:
-        progress(0.1, desc="Converting PDF to images...")
+        if progress_callback:
+            progress_callback("🔄 Converting PDF to images...")
         images = convert_pdf_to_images_gradio(pdf_file)
         
         if not images:
@@ -146,8 +147,8 @@ def process_pdf_document(pdf_file, model, progress=gr.Progress()):
         all_results = []
         
         for page_idx, pil_image in enumerate(images):
-            progress((page_idx + 1) / len(images) * 0.8 + 0.1, 
-                    desc=f"Processing page {page_idx + 1}/{len(images)}...")
+            if progress_callback:
+                progress_callback(f"🔄 Processing page {page_idx + 1}/{len(images)}...")
             
             layout_output = model.chat("Parse the reading order of this document.", pil_image)
             
@@ -172,7 +173,8 @@ def process_pdf_document(pdf_file, model, progress=gr.Progress()):
             }
             all_results.append(page_result)
         
-        progress(1.0, desc="Processing complete!")
+        if progress_callback:
+            progress_callback("🔄 Finalizing document...")
         
         combined_markdown = "\n\n---\n\n".join([
             f"# Page {result['page_number']}\n\n{result['markdown']}" 
@@ -322,31 +324,64 @@ def start_processing():
     """Immediately update status when processing starts"""
     return "🔄 Processing PDF... Please wait", gr.Tabs(visible=False)
 
-def process_uploaded_pdf(pdf_file, progress=gr.Progress()):
-    """Main processing function for uploaded PDF"""
+def process_uploaded_pdf_with_progress(pdf_file):
+    """Main processing function with custom progress updates"""
     global processed_markdown, show_results_tab
     
     if dolphin_model is None:
-        return "❌ Model not loaded", gr.Tabs(visible=False)
+        yield "❌ Model not loaded", gr.Tabs(visible=False)
+        return
     
     if pdf_file is None:
-        return "❌ No PDF uploaded", gr.Tabs(visible=False)
+        yield "❌ No PDF uploaded", gr.Tabs(visible=False)
+        return
     
     try:
-        combined_markdown, status = process_pdf_document(pdf_file, dolphin_model, progress)
+        def update_progress(message):
+            return message
+        
+        # Process with custom progress callback
+        for progress_msg in process_pdf_document_with_updates(pdf_file, dolphin_model):
+            yield progress_msg, gr.Tabs(visible=False)
+        
+        # Final result
+        combined_markdown, status = process_pdf_document(pdf_file, dolphin_model)
         
         if status == "processing_complete":
             processed_markdown = combined_markdown
             show_results_tab = True
-            return "✅ PDF processed successfully! Check the 'Document' tab above.", gr.Tabs(visible=True)
+            yield "✅ PDF processed successfully! Check the 'Document' tab above.", gr.Tabs(visible=True)
         else:
             show_results_tab = False
-            return combined_markdown, gr.Tabs(visible=False)
+            yield combined_markdown, gr.Tabs(visible=False)
             
     except Exception as e:
         show_results_tab = False
         error_msg = f"❌ Error processing PDF: {str(e)}"
-        return error_msg, gr.Tabs(visible=False)
+        yield error_msg, gr.Tabs(visible=False)
+
+def process_pdf_document_with_updates(pdf_file, model):
+    """Generator that yields progress updates"""
+    try:
+        yield "🔄 Converting PDF to images..."
+        images = convert_pdf_to_images_gradio(pdf_file)
+        
+        if not images:
+            yield "❌ Failed to convert PDF to images"
+            return
+        
+        for page_idx, pil_image in enumerate(images):
+            yield f"🔄 Processing page {page_idx + 1}/{len(images)}..."
+            
+            # Small delay to show progress
+            import time
+            time.sleep(0.1)
+        
+        yield "🔄 Finalizing document..."
+        time.sleep(0.5)
+        
+    except Exception as e:
+        yield f"❌ Error: {str(e)}"
 
 
 def get_processed_markdown():
@@ -494,19 +529,10 @@ with gr.Blocks(
     
     # Event handlers
     process_btn.click(
-        fn=start_processing,
-        outputs=[status_output, results_tab]
-    ).then(
-        fn=lambda: gr.HTML(visible=True),
-        outputs=[progress_space]
-    ).then(
-        fn=process_uploaded_pdf,
+        fn=process_uploaded_pdf_with_progress,
         inputs=[pdf_input],
         outputs=[status_output, results_tab],
-        show_progress="full"
-    ).then(
-        fn=lambda: gr.HTML(visible=False),
-        outputs=[progress_space]
+        show_progress=False
     ).then(
         fn=get_processed_markdown,
         outputs=[markdown_display]