app.py

import gradio as gr
import json
import base64
import os
import requests
from huggingface_hub import upload_file

HF_TOKEN = os.environ.get("HF_CV_ROBOT_TOKEN")
HF_DATASET_REPO = "OppaAI/Robot_MCP"  # Replace with your dataset repo
MODEL = "Qwen/Qwen2.5-VL-7B-Instruct"

def process_and_describe(payload: dict):
    if not HF_TOKEN:
        return {"error": "HF_TOKEN secret not found in Space settings."}

    try:
        robot_id = payload.get("robot_id", "unknown")
        image_b64 = payload["image_b64"]
        image_bytes = base64.b64decode(image_b64)

        # 1️⃣ Save temporarily
        local_tmp_path = "/tmp/uploaded_image.jpg"
        with open(local_tmp_path, "wb") as f:
            f.write(image_bytes)

        # 2️⃣ Upload to HF dataset repo
        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"
        )
        os.remove(local_tmp_path)

        # 3️⃣ Construct public URL
        image_url = f"https://huggingface.co/datasets/{HF_DATASET_REPO}/resolve/main/{path_in_repo}"

        # 4️⃣ Call VLM
        data = {
            "model": MODEL,
            "messages": [
                {
                    "role": "user",
                    "content": [
                        {"type": "text", "text": "Describe this image in detail."},
                        {"type": "image_url", "image_url": image_url}
                    ]
                }
            ]
        }

        resp = requests.post(
            "https://router.huggingface.co/v1/chat/completions",
            headers={"Authorization": f"Bearer {HF_TOKEN}"},
            json=data,
            timeout=60
        )

        if resp.status_code != 200:
            vlm_text = f"HF VLM error: {resp.status_code}, {resp.text}"
        else:
            try:
                vlm_text = resp.json()["choices"][0]["message"]["content"][0]["text"]
            except Exception as e:
                vlm_text = f"Failed to parse VLM response: {e}, Response={resp.text}"

        return {
            "saved_to_hf_hub": True,
            "repo_id": HF_DATASET_REPO,
            "path_in_repo": path_in_repo,
            "image_url": image_url,
            "file_size_bytes": len(image_bytes),
            "robot_id": robot_id,
            "vlm_description": vlm_text
        }

    except Exception as e:
        return {"error": f"Failed to upload/describe image: {str(e)}"}

demo = gr.Interface(
    fn=process_and_describe,
    inputs=gr.JSON(label="Input Payload (Dict format with 'image_b64')"),
    outputs=gr.JSON(label="Reply to Jetson"),
    api_name="predict"
)

if __name__ == "__main__":
    demo.launch(mcp_server=True)