Update app.py

app.py

@@ -2,64 +2,67 @@ import gradio as gr
 import json
 import base64
 import os
-from huggingface_hub import upload_file
+from huggingface_hub import HfApi
+import gradio as gr
 
-# --- Configuration for Hugging Face Hub Upload ---
-# The HF_TOKEN secret is automatically loaded into the environment by Hugging Face Spaces
-HF_TOKEN = os.environ.get("HF_CV_ROBOT_TOKEN")
+# =====================================================
+# Settings
+# =====================================================
+REPO_ID = "OppaAi/Robot-MCP"   # 你的 HuggingFace Space repo
+REPO_TYPE = "space"
+UPLOAD_FILENAME = "tmp.jpg"
 
-# !! REPLACE THIS with your actual dataset repo ID (e.g., "Wauplin/my-image-data") !!
-HF_DATASET_REPO = "OppaAI/ROBOT_MCP" 
+api = HfApi()
 
-def process_and_upload(payload: dict):
-    if not HF_TOKEN:
-        return {"error": "HF_TOKEN secret not found in Space settings."}
-        
+# =====================================================
+# Function: Save & Upload Base64 Image
+# =====================================================
+def save_and_upload_base64_image(base64_str):
     try:
-        image_b64 = payload["image_b64"]
-        image_bytes = base64.b64decode(image_b64)
+        # Remove base64 header (if exists)
+        if "," in base64_str:
+            base64_str = base64_str.split(",", 1)[1]
 
-        # 1. Save temporarily to the local ephemeral storage (/tmp) first
-        # This file will be immediately deleted after the function finishes or the space restarts
-        local_tmp_path = "/tmp/uploaded_image.jpg"
-        with open(local_tmp_path, "wb") as f:
-            f.write(image_bytes)
+        # Decode and save locally
+        with open(UPLOAD_FILENAME, "wb") as f:
+            f.write(base64.b64decode(base64_str))
 
-        # 2. Upload the temporary file to the persistent Hugging Face Dataset Repo
-        # The path in the repo can be dynamic, e.g., using a timestamp
-        path_in_repo = f"images/uploaded_image_{len(image_bytes)}.jpg" 
-        
-        upload_file(
-            path_or_fileobj=local_tmp_path,
-            path_in_repo=path_in_repo,
-            repo_id=HF_DATASET_REPO,
-            token=HF_TOKEN,
-            repo_type="dataset", 
-        )
+        print("Saved tmp.jpg locally.")
 
-        # 3. Clean up the local temporary file
-        os.remove(local_tmp_path)
+        # Upload to repo so you can see it in Files
+        api.upload_file(
+            path_or_fileobj=UPLOAD_FILENAME,
+            path_in_repo=UPLOAD_FILENAME,   # Upload to root folder
+            repo_id=REPO_ID,
+            repo_type=REPO_TYPE,
+        )
 
-        # 4. Return success message
-        return {
-            "saved_to_hf_hub": True,
-            "repo_id": HF_DATASET_REPO,
-            "path_in_repo": path_in_repo,
-            "file_size_bytes": len(image_bytes)
-        }
+        print("Uploaded tmp.jpg to HuggingFace Repo.")
+        return "Success: tmp.jpg uploaded to HF repo!"
 
     except Exception as e:
-        # Check the HF Space logs for full traceback if an error occurs
-        print(f"Upload failed: {e}") 
-        return {"error": f"Failed to upload to HF Hub: {str(e)}"}
+        print("Error:", str(e))
+        return f"Error: {str(e)}"
+
+
+# =====================================================
+# Gradio UI
+# =====================================================
+def handle_input(base64_image):
+    print("Received base64 image.")
+    return save_and_upload_base64_image(base64_image)
+
+
+with gr.Blocks() as demo:
+    gr.Markdown("## 🚀 Base64 → tmp.jpg → Upload to HF Repo")
+    
+    base64_input = gr.Textbox(label="Input Base64 Image String")
+    output_msg = gr.Textbox(label="Result")
 
+    run_btn = gr.Button("Upload")
+    run_btn.click(handle_input, inputs=base64_input, outputs=output_msg)
 
-demo = gr.Interface(
-    fn=process_and_upload,
-    inputs=gr.JSON(label="Input Payload (Dict format with 'image_b64')"),
-    outputs=gr.JSON(label="Reply"),
-    api_name="predict"
-)
 
+# Run Gradio
 if __name__ == "__main__":
-    demo.launch(mcp_server=True)
+    demo.launch()