feat: Add file size check and warnings in handle_upload function to improve user experience with large datasets

app.py

@@ -181,12 +181,34 @@ def extract_output_text(response) -> str:
 def handle_upload(file_path, request: gr.Request):
     """
     1) Take uploaded file path (string)
-    2) Copy to /tmp for a stable path
-    3) Build a public Gradio file URL that the MCP server can fetch via HTTP
+    2) Check file size and show warnings for large datasets
+    3) Copy to /tmp for a stable path
+    4) Build a public Gradio file URL that the MCP server can fetch via HTTP
     """
     if not file_path:
         return None
 
+    # Check file size and add warning if > 1.5MB
+    url_params = ""
+    try:
+        file_size = os.path.getsize(file_path)
+        file_size_mb = file_size / (1024 * 1024)  # Convert to MB
+
+        if file_size_mb > 1.5:
+            # Show Gradio warning with 5-second auto-dismiss
+            gr.Warning(
+                f"Large dataset detected! Your file is {file_size_mb:.1f}MB.",
+                duration=5,
+            )
+            gr.Warning(
+                "For optimal performance, training will use only the first 10,000 rows.",
+                duration=10,
+            )
+
+    except Exception:
+        # If we can't check file size, continue without warnings
+        pass
+
     local_path = file_path
     stable_path = os.path.join("/tmp", os.path.basename(local_path))
     try:
@@ -197,7 +219,7 @@ def handle_upload(file_path, request: gr.Request):
         pass
 
     base_url = str(request.base_url).rstrip("/")
-    public_url = f"{base_url}/gradio_api/file={local_path}"
+    public_url = f"{base_url}/gradio_api/file={local_path}{url_params}"
     return public_url
 
 
@@ -440,4 +462,5 @@ if __name__ == "__main__":
         allowed_paths=["/tmp"],
         ssr_mode=False,
         show_error=True,
+        max_file_size="10mb",
     )