Update app.py

app.py

@@ -195,118 +195,4 @@ def website_to_pdf(all_pages, progress_callback):
             progress_callback(f"Processing pages... {progress:.0%}")
             logger.info(f"Generated PDF chunk {i//chunk_size + 1}/{total_chunks}")
 
-        logger.info("Merging PDF chunks...")
-        output_pdf = os.path.join(temp_dir, "final.pdf")
-        merger = PdfMerger()
-        for temp_file in temp_files:
-            merger.append(temp_file)
-        merger.write(output_pdf)
-        merger.close()
-
-        logger.info("PDF generation complete. Reading final PDF...")
-        with open(output_pdf, 'rb') as f:
-            return f.read()
-
-async def process_url(url, depth, progress_callback):
-    try:
-        all_pages = await crawl_pages(url, depth)
-        if not all_pages:
-            return "No pages were successfully crawled. Please check the URL and try again."
-        
-        logger.info("Crawling complete. Starting PDF generation...")
-        # Use ThreadPoolExecutor to run PDF generation in a separate thread
-        loop = asyncio.get_event_loop()
-        pdf_content = await loop.run_in_executor(executor, website_to_pdf, all_pages, progress_callback)
-        logger.info("PDF generation complete.")
-        return pdf_content
-    except Exception as e:
-        logger.error(f"Error in process_url: {str(e)}")
-        return f"An error occurred: {str(e)}"
-
-# App layout
-app.layout = dbc.Container([
-    dcc.Store(id='pdf-store'),
-    dcc.Store(id='progress-store'),
-    dbc.Card(
-        dbc.CardBody([
-            html.H1("Website to PDF Converter", className="text-center mb-4"),
-            html.P("Enter docs URL and crawl depth to convert documentation pages into a PDF. Be responsible for sites you have permission to do this", className="text-center mb-4"),
-            dbc.Input(id="url-input", type="text", placeholder="Enter website URL (e.g., https://www.gradio.app/docs)", className="mb-3"),
-            dcc.Slider(id="depth-slider", min=1, max=10, step=1, value=3, marks={i: str(i) for i in range(1, 11)}, className="mb-3"),
-            dbc.Button("Convert to PDF", id="submit-button", color="primary", className="mb-3 w-100"),
-            dbc.Button("Download PDF", id="download-button", color="secondary", className="mb-3 w-100", disabled=True),
-            html.Div([
-                dbc.Spinner(html.Div(id="progress-message"), color="primary", type="grow", size="lg"),
-            ], className="text-center mb-3"),
-            dcc.Interval(id="progress-interval", interval=1000, n_intervals=0, disabled=True),
-            dcc.Download(id="download-pdf")
-        ]),
-        className="mt-4"
-    )
-], fluid=True)
-
-def update_output(n_clicks, n_intervals, progress_data, url, depth):
-    ctx = dash.callback_context
-    if not ctx.triggered:
-        raise PreventUpdate
-
-    triggered_id = ctx.triggered[0]['prop_id'].split('.')[0]
-
-    if triggered_id == "submit-button":
-        if not url:
-            return True, "secondary", True, None, "Please enter a URL"
-
-        # Start the background task
-        task_id = str(uuid.uuid4())
-        executor.submit(background_task, url, depth, task_id)
-        
-        return True, "secondary", False, None, "Processing... Please wait."
-
-    elif triggered_id == "progress-interval" or triggered_id == "progress-store":
-        if progress_data is None:
-            return True, "secondary", False, None, "Processing... Please wait."
-        
-        if isinstance(progress_data, str):
-            if progress_data.startswith("Error"):
-                return True, "secondary", True, None, progress_data
-            else:
-                return True, "secondary", False, None, progress_data
-        
-        if isinstance(progress_data, bytes):
-            encoded = base64.b64encode(progress_data).decode()
-            return False, "primary", True, encoded, "PDF ready for download!"
-
-    return True, "secondary", False, None, ""
-
-@app.callback(
-    Output("download-pdf", "data"),
-    Input("download-button", "n_clicks"),
-    State("pdf-store", "data"),
-    prevent_initial_call=True
-)
-def download_pdf(n_clicks, pdf_data):
-    if pdf_data is None:
-        raise PreventUpdate
-    
-    decoded = base64.b64decode(pdf_data)
-    return dcc.send_bytes(decoded, f"website_content_{int(time.time())}.pdf")
-
-def background_task(url, depth, task_id):
-    def progress_callback(message):
-        # Update progress in the progress-store
-        app.layout.children[1].data = message
-
-    try:
-        logger.info(f"Starting background task for URL: {url}, depth: {depth}")
-        pdf_content = asyncio.run(process_url(url, depth, progress_callback))
-        logger.info("Background task completed successfully")
-        # Store the PDF content directly in the progress-store
-        app.layout.children[1].data = pdf_content
-    except Exception as e:
-        logger.error(f"Error in background task: {str(e)}")
-        app.layout.children[1].data = f"Error: {str(e)}"
-
-if __name__ == '__main__':
-    print("Starting the Dash application...")
-    app.run(debug=True, host='0.0.0.0', port=7860)
-    print("Dash application has finished running.")
+        logger.info("Mer