#!/usr/bin/env python3
from diffusers import AutoPipelineForInpainting, AutoPipelineForImage2Image
from diffusers.utils import load_image
import torch
from pathlib import Path
import os
from huggingface_hub import HfApi

torch.backends.cuda.matmul.allow_tf32 = True
torch_device = "cuda" if torch.cuda.is_available() else "cpu"
api = HfApi()

pipe = AutoPipelineForInpainting.from_pretrained("runwayml/stable-diffusion-inpainting", torch_dtype=torch.float16, variant="fp16", use_safetensors=True) 

pipe = pipe.to(torch_device)
pipe.enable_xformers_memory_efficient_attention()

img_url = "https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo.png"
mask_url = "https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo_mask.png"

image = load_image(img_url)
mask_image = load_image(mask_url)

prompt = "dslr photography of an empty bench, high quality"
generator = torch.Generator(device="cuda").manual_seed(0)

image = pipe(
    prompt=prompt,
    image=image,
    mask_image=mask_image,
    guidance_scale=8.0,
    num_inference_steps=20,
    generator=generator,
).images[0]


image = image.resize((1024, 1024))

pipe = AutoPipelineForInpainting.from_pretrained("stabilityai/stable-diffusion-xl-refiner-1.0", torch_dtype=torch.float16, variant="fp16", use_safetensors=True) 
pipe.to("cuda")

pipe.enable_xformers_memory_efficient_attention()

image = pipe(
    prompt=prompt,
    image=image,
    mask_image=mask_image,
    guidance_scale=8.0,
    num_inference_steps=100,
    strength=0.2,
    generator=generator,
).images[0]

pipe = AutoPipelineForImage2Image.from_pipe(pipe)
pipe.enable_xformers_memory_efficient_attention()

image = pipe(
    prompt=prompt,
    image=image,
    guidance_scale=8.0,
    num_inference_steps=100,
    strength=0.2,
    generator=generator,
).images[0]

file_name = f"aaa"
path = os.path.join(Path.home(), "images", "ediffi_sdxl", f"{file_name}.png")
image.save(path)

api.upload_file(
    path_or_fileobj=path,
    path_in_repo=path.split("/")[-1],
    repo_id="patrickvonplaten/images",
    repo_type="dataset",
)
print(f"https://huggingface.co/datasets/patrickvonplaten/images/blob/main/{file_name}.png")