from flask import Flask, request, jsonify
from diffusers import DDPMScheduler, DiffusionPipeline
import torch
import base64
from io import BytesIO
from PIL import Image

app = Flask(__name__)

scheduler = DDPMScheduler.from_pretrained("stabilityai/stable-diffusion-2-1-base", subfolder="scheduler")
pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1-base", scheduler=scheduler)

print("Loading UNet weights...")
ckpt_path = './models/model.bin'
try:
    unet_state_dict = torch.load(ckpt_path)
    pipe.unet.load_state_dict(unet_state_dict)
    if torch.cuda.is_available():
        pipe = pipe.to("cuda")
        pipe.to(torch.float16)
        print("UNet weights loaded successfully and model moved to CUDA.")
    else:
        print("CUDA is not available. Running on CPU (will be slow).")
        pipe = pipe.to("cpu")
        pipe.to(torch.float32)
except FileNotFoundError:
    print(f"Error: Model not found at {ckpt_path}. Please ensure the file exists.")
    exit()
except Exception as e:
    print(f"An error occurred while loading UNet weights: {e}")
    exit()

@app.route('/api/generate', methods=['POST'])
def generate_image():
    prompt = request.json.get('prompt')
    if not prompt:
        return jsonify({"error": "Prompt is required in the request body."}), 400

    print(f"Generating image for prompt: '{prompt}'...")
    try:
        image = pipe(prompt=prompt, num_inference_steps=1, guidance_scale=0, timesteps=[999]).images[0]
        image = image.resize((128, 128), Image.LANCZOS)

        # Convert image to base64
        buffered = BytesIO()
        image.save(buffered, format="JPEG") # You can choose JPEG for smaller size, but PNG is lossless
        img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")

        return jsonify({"image": img_str})

    except Exception as e:
        print(f"Error during image generation: {e}")
        return jsonify({"error": "An error occurred during image generation."}), 500

if __name__ == '__main__':
    # Make sure to run this with 'python app.py'
    # It will be accessible at http://127.0.0.1:5000/
    app.run(host='127.0.0.1', port=5000, debug=True) # debug=True is good for development