import gradio as gr
import numpy as np
import random
import torch
from diffusers import DiffusionPipeline, AutoencoderTiny, AutoencoderKL
from live_preview_helpers import flux_pipe_call_that_returns_an_iterable_of_images

dtype = torch.float32  # CPU-friendly
device = "cpu"

# Load models on CPU
taef1 = AutoencoderTiny.from_pretrained("madebyollin/taef1", torch_dtype=dtype).to(device)
good_vae = AutoencoderKL.from_pretrained("black-forest-labs/FLUX.1-dev", subfolder="vae", torch_dtype=dtype).to(device)
pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=dtype, vae=taef1).to(device)

MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 1024  # reduce max size for CPU

# Bind the custom flux method
pipe.flux_pipe_call_that_returns_an_iterable_of_images = flux_pipe_call_that_returns_an_iterable_of_images.__get__(pipe)

def infer(prompt, seed=42, randomize_seed=False, width=512, height=512, guidance_scale=3.5, num_inference_steps=15, progress=gr.Progress(track_tqdm=True)):
    if randomize_seed:
        seed = random.randint(0, MAX_SEED)
    generator = torch.Generator(device=device).manual_seed(seed)
    
    for img in pipe.flux_pipe_call_that_returns_an_iterable_of_images(
            prompt=prompt,
            guidance_scale=guidance_scale,
            num_inference_steps=num_inference_steps,
            width=width,
            height=height,
            generator=generator,
            output_type="pil",
            good_vae=good_vae,
        ):
            yield img, seed

examples = [
    "a tiny astronaut hatching from an egg on the moon",
    "a cat holding a sign that says hello world",
    "an anime illustration of a wiener schnitzel",
]

css="""
#col-container {
    margin: 0 auto;
    max-width: 520px;
}
"""

with gr.Blocks(css=css) as demo:
    
    with gr.Column(elem_id="col-container"):
        gr.Markdown("""
# FLUX.1 [dev]
12B param rectified flow transformer guidance-distilled from [FLUX.1 [pro]](https://blackforestlabs.ai/)  
[[non-commercial license](https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md)]
        """)
        
        with gr.Row():
            prompt = gr.Textbox(
                label="Prompt",
                placeholder="Enter your prompt",
                lines=1
            )
            run_button = gr.Button("Run")
        
        result = gr.Image(label="Result")

        with gr.Accordion("Advanced Settings", open=False):
            seed = gr.Slider(0, MAX_SEED, step=1, label="Seed", value=0)
            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
            width = gr.Slider(256, MAX_IMAGE_SIZE, step=32, label="Width", value=512)
            height = gr.Slider(256, MAX_IMAGE_SIZE, step=32, label="Height", value=512)
            guidance_scale = gr.Slider(1.0, 15.0, step=0.1, label="Guidance Scale", value=3.5)
            num_inference_steps = gr.Slider(1, 30, step=1, label="Inference Steps", value=15)
        
        gr.Examples(
            examples=examples,
            fn=infer,
            inputs=[prompt],
            outputs=[result, seed],
            cache_examples=False
        )

        run_button.click(
            fn=infer,
            inputs=[prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps],
            outputs=[result, seed]
        )
        prompt.submit(
            fn=infer,
            inputs=[prompt, seed, randomize_seed, width, height, guidance_scale, num_inference_steps],
            outputs=[result, seed]
        )

demo.launch()