Update app.py

app.py

@@ -1,13 +1,17 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
+from huggingface_hub.errors import HfHubHTTPError
 import os
 import base64
 from PIL import Image
 import io
 from typing import Optional
 
-ACCESS_TOKEN = os.getenv("HF_TOKEN")
-print("Access token loaded.")
+HF_READ_TOKEN = os.getenv("HF_READ_TOKEN")
+if HF_READ_TOKEN:
+    print("Default Hugging Face token available from environment.")
+else:
+    print("No default Hugging Face token configured; relying on user sign-in.")
 
 # Function to encode image to base64
 def encode_image(image_path):
@@ -78,17 +82,19 @@ def respond(
     if oauth_token is not None and getattr(oauth_token, "token", None):
         api_token = oauth_token.token
         print("Using OAuth token from signed-in user for inference.")
-    elif ACCESS_TOKEN:
-        api_token = ACCESS_TOKEN
+    elif HF_READ_TOKEN:
+        api_token = HF_READ_TOKEN
         print("Using server-configured Hugging Face token for inference.")
-    else:
-        raise gr.Error(
-            "No Hugging Face session detected. Please sign in with your Hugging Face account before running the chat."
-        )
 
     # Initialize the Inference Client with default HF inference
-    client = InferenceClient(token=api_token)
-    print("Hugging Face Inference Client initialized with available token.")
+    client_kwargs = {}
+    if api_token is not None:
+        client_kwargs["token"] = api_token
+    else:
+        print("No Hugging Face token available; attempting anonymous inference (may fail for private models).")
+
+    client = InferenceClient(**client_kwargs)
+    print("Hugging Face Inference Client initialized.")
 
     # Convert seed to None if -1 (meaning random)
     if seed == -1:
@@ -215,6 +221,17 @@ def respond(
                         yield response
         
         print()
+    except HfHubHTTPError as e:
+        status = getattr(e.response, "status_code", None)
+        if status in (401, 403):
+            raise gr.Error(
+                "Failed to generate response: {}. Sign in with your Hugging Face account and retry."
+                .format(e)
+            ) from e
+
+        print(f"Error during inference: {e}")
+        response += f"\nError: {str(e)}"
+        yield response
     except Exception as e:
         print(f"Error during inference: {e}")
         response += f"\nError: {str(e)}"
@@ -539,4 +556,4 @@ print("Gradio interface initialized.")
 
 if __name__ == "__main__":
     print("Launching the demo application.")
-    demo.launch(show_api=True)
+    demo.launch(show_api=True, mcp_server=True)