Update app.py

app.py

@@ -1,18 +1,45 @@
 import os
+import copy
 import base64
 import requests
+import tempfile
 import secrets
 import gradio as gr
 from huggingface_hub import upload_file, InferenceClient
 from PIL import Image
-import json
 
-# ... (Config and Helper Functions remain the same as previous snippets) ...
-# Ensure HF_VLM_MODEL is set back to the Qwen model you want to use:
-HF_VLM_MODEL = "Qwen/Qwen2.5-VL-7B-Instruct" 
+# --- Config ---
+HF_TOKEN = os.environ.get("HF_CV_ROBOT_TOKEN")
+HF_DATASET_REPO = "OppaAI/Robot_MCP"
+# Model specifically for VLM (image-to-text) tasks on Hugging Face
+HF_VLM_MODEL = "llava-hf/llava-interleave-qwen-0.5b-hf" # A suitable VLM model
+
+if not HF_TOKEN:
+    raise ValueError("HF_TOKEN environment variable not set.")
 
 # Initialize the Hugging Face Inference Client
-hf_client = InferenceClient(token=HF_CV_ROBOT_TOKEN)
+hf_client = InferenceClient(token=HF_TOKEN)
+
+# --- Helper Functions ---
+def save_and_upload_image(image_b64):
+    """Save image to /tmp and upload to HF dataset."""
+    image_bytes = base64.b64decode(image_b64)
+    # Use a unique filename to prevent conflicts in /tmp
+    local_tmp_path = f"/tmp/uploaded_image_{secrets.token_hex(8)}.jpg" 
+    with open(local_tmp_path, "wb") as f:
+        f.write(image_bytes)
+
+    path_in_repo = f"images/uploaded_image_{secrets.token_hex(8)}.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"
+    )
+
+    hf_image_url = f"https://huggingface.co/datasets/{HF_DATASET_REPO}/resolve/main/{path_in_repo}"
+    return local_tmp_path, hf_image_url, path_in_repo, len(image_bytes)
 
 # --- Main MCP function ---
 def process_and_describe(payload: dict):