Update app.py

app.py

@@ -131,14 +131,13 @@ def convert_pdf_to_images_gradio(pdf_file):
         raise Exception(f"Error converting PDF: {str(e)}")
 
 
-def process_pdf_document(pdf_file, model, progress_callback=None):
+def process_pdf_document(pdf_file, model, progress=gr.Progress()):
     """Process uploaded PDF file page by page"""
     if pdf_file is None:
         return "No PDF file uploaded", ""
     
     try:
-        if progress_callback:
-            progress_callback("🔄 Converting PDF to images...")
+        progress(0.1, desc="Converting PDF to images...")
         images = convert_pdf_to_images_gradio(pdf_file)
         
         if not images:
@@ -147,8 +146,8 @@ def process_pdf_document(pdf_file, model, progress_callback=None):
         all_results = []
         
         for page_idx, pil_image in enumerate(images):
-            if progress_callback:
-                progress_callback(f"🔄 Processing page {page_idx + 1}/{len(images)}...")
+            progress((page_idx + 1) / len(images) * 0.8 + 0.1, 
+                    desc=f"Processing page {page_idx + 1}/{len(images)}...")
             
             layout_output = model.chat("Parse the reading order of this document.", pil_image)
             
@@ -173,8 +172,7 @@ def process_pdf_document(pdf_file, model, progress_callback=None):
             }
             all_results.append(page_result)
         
-        if progress_callback:
-            progress_callback("🔄 Finalizing document...")
+        progress(1.0, desc="Processing complete!")
         
         combined_markdown = "\n\n---\n\n".join([
             f"# Page {result['page_number']}\n\n{result['markdown']}" 
@@ -320,68 +318,31 @@ processed_markdown = ""
 show_results_tab = False
 
 
-def start_processing():
-    """Immediately update status when processing starts"""
-    return "🔄 Processing PDF... Please wait", gr.Tabs(visible=False)
-
-def process_uploaded_pdf_with_progress(pdf_file):
-    """Main processing function with custom progress updates"""
+def process_uploaded_pdf(pdf_file, progress=gr.Progress()):
+    """Main processing function for uploaded PDF"""
     global processed_markdown, show_results_tab
     
     if dolphin_model is None:
-        yield "❌ Model not loaded", gr.Tabs(visible=False)
-        return
+        return "❌ Model not loaded", gr.Tabs(visible=False)
     
     if pdf_file is None:
-        yield "❌ No PDF uploaded", gr.Tabs(visible=False)
-        return
+        return "❌ No PDF uploaded", gr.Tabs(visible=False)
     
     try:
-        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)
+        combined_markdown, status = process_pdf_document(pdf_file, dolphin_model, progress)
         
         if status == "processing_complete":
             processed_markdown = combined_markdown
             show_results_tab = True
-            yield "✅ PDF processed successfully! Check the 'Document' tab above.", gr.Tabs(visible=True)
+            return "✅ PDF processed successfully! Check the 'Document' tab above.", gr.Tabs(visible=True)
         else:
             show_results_tab = False
-            yield combined_markdown, gr.Tabs(visible=False)
+            return combined_markdown, gr.Tabs(visible=False)
             
     except Exception as e:
         show_results_tab = False
         error_msg = f"❌ Error processing PDF: {str(e)}"
-        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)}"
+        return error_msg, gr.Tabs(visible=False)
 
 
 def get_processed_markdown():
@@ -395,7 +356,7 @@ def clear_all():
     global processed_markdown, show_results_tab
     processed_markdown = ""
     show_results_tab = False
-    return None, "Upload a PDF to get started", gr.Tabs(visible=False)
+    return None, "✅ Ready to process your PDF", gr.Tabs(visible=False)
 
 
 # Create Gradio interface
@@ -484,9 +445,9 @@ with gr.Blocks(
                 elem_id="progress-container"
             )
             
-            # Status and progress
+            # Status output (hidden during processing)
             status_output = gr.Markdown(
-                "Upload a PDF to get started",
+                "✅ Ready to process your PDF",
                 elem_classes="status-message"
             )
         
@@ -529,10 +490,10 @@ with gr.Blocks(
     
     # Event handlers
     process_btn.click(
-        fn=process_uploaded_pdf_with_progress,
+        fn=process_uploaded_pdf,
         inputs=[pdf_input],
         outputs=[status_output, results_tab],
-        show_progress=False
+        show_progress=True
     ).then(
         fn=get_processed_markdown,
         outputs=[markdown_display]