Update app.py

app.py

@@ -46,6 +46,9 @@ ssl_context.verify_mode = ssl.CERT_NONE
 # ThreadPoolExecutor for background tasks
 executor = ThreadPoolExecutor(max_workers=4)
 
+# Global variable to store the generated PDF
+generated_file = None
+
 @contextmanager
 def get_db_connection():
     if not hasattr(thread_local, "connection"):
@@ -226,7 +229,7 @@ async def process_url(url, depth, progress_callback):
 # App layout
 app.layout = dbc.Container([
     dcc.Store(id='pdf-store'),
-    dcc.Store(id='progress-store'),
+    dcc.Store(id='progress-store', data={'progress': 0, 'message': ''}),
     dbc.Card(
         dbc.CardBody([
             html.H1("Website to PDF Converter", className="text-center mb-4"),
@@ -236,21 +239,22 @@ app.layout = dbc.Container([
             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),
+                dbc.Progress(id="progress-bar", value=0, className="mb-3"),
+                html.Div(id="progress-message", className="text-center"),
+            ], className="mb-3"),
+            dcc.Interval(id="progress-interval", interval=1000, n_intervals=0),
             dcc.Download(id="download-pdf")
         ]),
         className="mt-4"
     )
 ], fluid=True)
 
-@callback(
+@app.callback(
     Output("submit-button", "disabled"),
     Output("download-button", "disabled"),
-    Output("progress-interval", "disabled"),
-    Output("pdf-store", "data"),
+    Output("progress-bar", "value"),
     Output("progress-message", "children"),
+    Output("pdf-store", "data"),
     Input("submit-button", "n_clicks"),
     Input("progress-interval", "n_intervals"),
     State("url-input", "value"),
@@ -267,31 +271,26 @@ def update_output(n_clicks, n_intervals, url, depth, progress_data):
 
     if triggered_id == "submit-button":
         if not url:
-            return False, True, True, None, "Please enter a URL"
+            return False, True, 0, "Please enter a URL", None
 
         # Start the background task
         task_id = str(uuid.uuid4())
         executor.submit(background_task, url, depth, task_id)
         
-        return True, True, False, None, "Processing... Please wait."
+        return True, True, 0, "Processing... Please wait.", None
 
     elif triggered_id == "progress-interval":
-        if progress_data is None:
-            return True, True, False, None, "Processing... Please wait."
-        
-        if isinstance(progress_data, str):
-            if progress_data.startswith("Error"):
-                return False, True, True, None, progress_data
-            elif progress_data == "PDF generation complete":
-                return False, False, True, dash.no_update, "PDF ready for download!"
-            else:
-                return True, True, False, None, progress_data
-        
-        if isinstance(progress_data, bytes):
-            encoded = base64.b64encode(progress_data).decode()
-            return False, False, True, encoded, "PDF ready for download!"
+        progress = progress_data['progress']
+        message = progress_data['message']
 
-    return False, True, True, None, ""
+        if isinstance(message, str) and message.startswith("Error"):
+            return False, True, 100, message, None
+        elif progress == 100 and generated_file:
+            return False, False, 100, "PDF ready for download!", generated_file
+        else:
+            return True, True, progress, message, None
+
+    return False, True, 0, "", None
 
 @app.callback(
     Output("download-pdf", "data"),
@@ -303,25 +302,24 @@ 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")
+    return dcc.send_bytes(pdf_data, f"website_content_{int(time.time())}.pdf")
 
 def background_task(url, depth, task_id):
+    global generated_file
+
     def progress_callback(message):
-        # Update progress in the progress-store
-        app.layout.children[1].data = message
+        progress = int(float(message.split()[-1].strip('%')))
+        app.layout.children[1].data = {'progress': progress, 'message': 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
-        # Set a flag to indicate completion
-        app.layout.children[1].data = "PDF generation complete"
+        generated_file = pdf_content
+        app.layout.children[1].data = {'progress': 100, 'message': "PDF generation complete"}
     except Exception as e:
         logger.error(f"Error in background task: {str(e)}")
-        app.layout.children[1].data = f"Error: {str(e)}"
+        app.layout.children[1].data = {'progress': 100, 'message': f"Error: {str(e)}"}
 
 if __name__ == '__main__':
     print("Starting the Dash application...")