main.py

from fastapi import FastAPI, File, UploadFile, HTTPException, Request
from fastapi.responses import HTMLResponse, FileResponse, JSONResponse
from fastapi.staticfiles import StaticFiles
from fastapi.templating import Jinja2Templates
from fastapi.middleware.cors import CORSMiddleware
from gradio_client import Client, handle_file
import os
import tempfile
import base64

app = FastAPI()

# Retrieve Hugging Face token from environment variables
hf_token = os.environ.get('HF_TOKEN')
client = Client("Ashrafb/image-to-sketch-ari", hf_token=hf_token)

# Configure CORS middleware
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],  # Adjust as needed, '*' allows requests from any origin
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

@app.post("/upload/")
async def upload_file(file: UploadFile = File(...)):
    # Save the uploaded file to a temporary location
    with tempfile.NamedTemporaryFile(delete=False) as temp_file:
        temp_file.write(await file.read())
        temp_file_path = temp_file.name

    try:
        # Call the Gradio API
        result = client.predict(
            img=handle_file(temp_file_path),
            api_name="/predict"
        )
        
        # Check if the result is valid
        if result and len(result) == 2:
            # Convert the sketch image to a base64 string
            with open(result[0], "rb") as image_file:
                image_data = base64.b64encode(image_file.read()).decode("utf-8")

            return {
                "sketch_image_base64": f"data:image/png;base64,{image_data}",
                "result_file": result[1]
            }
        else:
            return {"error": "Invalid result from the prediction API."}
    except Exception as e:
        return {"error": str(e)}
    finally:
        # Clean up the temporary file
        if os.path.exists(temp_file_path):
            os.unlink(temp_file_path)