import torch, os, gc, random
import gradio as gr
from PIL import Image
from diffusers.utils import load_image
from accelerate import Accelerator
from diffusers import StableDiffusionXLPipeline, EulerDiscreteScheduler
accelerator = Accelerator(cpu=True)

pipe = accelerator.prepare(StableDiffusionXLPipeline.from_single_file("https://huggingface.co/lllyasviel/fav_models/resolve/main/fav/juggernautXL_version6Rundiffusion.safetensors", torch_dtype=torch.bfloat16, use_safetensors=True, variant=None, safety_checker=False))
##pipe.scheduler = accelerator.prepare(EulerDiscreteScheduler.from_config(pipe.scheduler.config))
##pipe.unet.to(memory_format=torch.channels_last)
pipe.to("cpu")
apol=[]
def plex(prompt,neg_prompt,stips,nut):
    apol=[]
    if nut == 0:
        nm = random.randint(1, 2147483616)
        while nm % 32 != 0:
            nm = random.randint(1, 2147483616)
    else:
        nm=nut
    generator = torch.Generator(device="cpu").manual_seed(nm)
    image = pipe(prompt=prompt, negative_prompt=neg_prompt, denoising_end=1.0,num_inference_steps=stips, output_type="pil",generator=generator)
    for i, imge in enumerate(image["images"]):
        apol.append(imge)
    return apol

iface = gr.Interface(fn=plex, inputs=[gr.Textbox(label="prompt"),gr.Textbox(label="negative prompt",value="ugly, blurry, poor quality"), gr.Slider(label="num inference steps", minimum=1, step=1, maximum=10, value=6),gr.Slider(label="manual seed (leave 0 for random)", minimum=0,step=32,maximum=2147483616,value=0)], outputs=gr.Gallery(label="out", columns=1),description="Running on cpu, very slow! by JoPmt.")
iface.queue(max_size=1,api_open=False)
iface.launch(max_threads=1)