import gradio as gr
import torch
import math
from PIL import Image
from diffusers import ShapEImg2ImgPipeline

import inspect
print(inspect.signature(ShapEImg2ImgPipeline.__call__))
# Set device
device = "cuda" if torch.cuda.is_available() else "cpu"

# Load the pipeline
pipe = ShapEImg2ImgPipeline.from_pretrained("openai/shap-e-img2img", torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32)
pipe = pipe.to(device)

# Resize function
def resize(value, img_path):
    img = Image.open(img_path)
    img = img.resize((value, value))
    return img

# Inference function
def infer(source_img, prompt, steps, seed, strength):
    generator = torch.Generator(device).manual_seed(seed)
    if int(steps * strength) < 1:
        steps = math.ceil(1 / max(0.10, strength))
    source_image = resize(512, source_img)
    source_image.save('source.png')
    # Adjust this line based on the correct method signature
    result = pipe(images=source_image, strength=strength, guidance_scale=7.5, num_inference_steps=steps, generator=generator)
    return result.images[0]

# Gradio interface
gr.Interface(
    fn=infer,
    inputs=[
        gr.Image(sources=["upload", "webcam", "clipboard"], type="filepath", label="Raw Image"), 
        gr.Textbox(label='Creative Touch (prompt)'),
        gr.Slider(1, 50, value=25, step=1, label='Number of Iterations'),
        gr.Slider(minimum=0, maximum=987654321987654321, step=1, randomize=True, label='Seed'), 
        gr.Slider(minimum=0.1, maximum=1, step=0.05, value=0.5, label='Strength')
    ],
    outputs='image',
    title="Creative Touch"
).queue(max_size=10).launch()