import gradio as gr
import numpy as np
import random
#import spaces
import torch
import devicetorch
from diffusers import DiffusionPipeline
import os


# Quant
from optimum.quanto import freeze, qfloat8, quantize
from diffusers import FlowMatchEulerDiscreteScheduler, AutoencoderKL
from diffusers.models.transformers.transformer_flux import FluxTransformer2DModel
from diffusers.pipelines.flux.pipeline_flux import FluxPipeline
from transformers import CLIPTextModel, CLIPTokenizer,T5EncoderModel, T5TokenizerFast


DEVICE = devicetorch.get(torch)
#if device == "cuda":
#    dtype = torch.bfloat16
#elif device == "mps":
#    dtype = torch.float16
#else:
#    dtype = torch.float32
##dtype = torch.bfloat16
##device = "cuda" if torch.cuda.is_available() else "cpu"
#
##pipe = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=torch.bfloat16, revision="refs/pr/1").to(device)
#pipe = DiffusionPipeline.from_pretrained("cocktailpeanut/xulf-s", torch_dtype=dtype).to(device)

MAX_SEED = np.iinfo(np.int32).max
MAX_IMAGE_SIZE = 2048


def init():
    global pipe

    dtype = torch.bfloat16

    # schnell is the distilled turbo model. For the CFG distilled model, use:
    # bfl_repo = "black-forest-labs/FLUX.1-dev"
    # revision = "refs/pr/3"
    #
    # The undistilled model that uses CFG ("pro") which can use negative prompts
    # was not released.
    bfl_repo = "cocktailpeanut/xulf-s"
    te_repo = "comfyanonymous/flux_text_encoders"


    scheduler = FlowMatchEulerDiscreteScheduler.from_pretrained(bfl_repo, subfolder="scheduler")
    #text_encoder = CLIPTextModel.from_pretrained("openai/clip-vit-large-patch14", torch_dtype=dtype)
    text_encoder = CLIPTextModel.from_pretrained(os.path.join(os.getcwd(), "flux_text_encoders/clip_l.safetensors"), torch_dtype=dtype)
    tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-large-patch14", torch_dtype=dtype)
    #text_encoder_2 = T5EncoderModel.from_pretrained(bfl_repo, subfolder="text_encoder_2", torch_dtype=dtype)
    text_encoder_2 = T5EncoderModel.from_pretrained(os.path.join(os.getcwd(), "flux_text_encoders/t5xxl_fp8_e4m3fn.safetensors"), torch_dtype=dtype)
    tokenizer_2 = T5TokenizerFast.from_pretrained(bfl_repo, subfolder="tokenizer_2", torch_dtype=dtype)
    vae = AutoencoderKL.from_pretrained(bfl_repo, subfolder="vae", torch_dtype=dtype)
    transformer = FluxTransformer2DModel.from_pretrained(bfl_repo, subfolder="transformer", torch_dtype=dtype)

    # Experimental: Try this to load in 4-bit for <16GB cards.
    #
    # from optimum.quanto import qint4
    # quantize(transformer, weights=qint4, exclude=["proj_out", "x_embedder", "norm_out", "context_embedder"])
    # freeze(transformer)
    quantize(transformer, weights=qfloat8)
    freeze(transformer)

    quantize(text_encoder_2, weights=qfloat8)
    freeze(text_encoder_2)

    pipe = FluxPipeline(
        scheduler=scheduler,
        text_encoder=text_encoder,
        tokenizer=tokenizer,
        text_encoder_2=None,
        tokenizer_2=tokenizer_2,
        vae=vae,
        transformer=None,
    )
    pipe.text_encoder_2 = text_encoder_2
    pipe.transformer = transformer
    if DEVICE == "cuda":
        pipe.enable_model_cpu_offload()
    pipe.to(DEVICE)

#    generator = torch.Generator().manual_seed(12345)
#    image = pipe(
#        prompt='nekomusume cat girl, digital painting',
#        width=1024,
#        height=1024,
#        num_inference_steps=4,
#        generator=generator,
#        guidance_scale=3.5,
#    ).images[0]

#@spaces.GPU()
def infer(prompt, seed=42, randomize_seed=False, width=1024, height=1024, num_inference_steps=4, progress=gr.Progress(track_tqdm=True)):
    global pipe
    if randomize_seed:
        seed = random.randint(0, MAX_SEED)
    generator = torch.Generator().manual_seed(seed)
    image = pipe(
            prompt = prompt, 
            width = width,
            height = height,
            num_inference_steps = num_inference_steps, 
            generator = generator,
            guidance_scale=0.0
    ).images[0] 
    return image, 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(f"""# FLUX.1 [schnell]
12B param rectified flow transformer distilled from [FLUX.1 [pro]](https://blackforestlabs.ai/) for 4 step generation
[[blog](https://blackforestlabs.ai/2024/07/31/announcing-black-forest-labs/)] [[model](https://huggingface.co/black-forest-labs/FLUX.1-schnell)]
        """)
        
        with gr.Row():
            
            prompt = gr.Text(
                label="Prompt",
                show_label=False,
                max_lines=1,
                placeholder="Enter your prompt",
                container=False,
            )
            
            run_button = gr.Button("Run", scale=0)
        
        result = gr.Image(label="Result", show_label=False)
        
        with gr.Accordion("Advanced Settings", open=False):
            
            seed = gr.Slider(
                label="Seed",
                minimum=0,
                maximum=MAX_SEED,
                step=1,
                value=0,
            )
            
            randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
            
            with gr.Row():
                
                width = gr.Slider(
                    label="Width",
                    minimum=256,
                    maximum=MAX_IMAGE_SIZE,
                    step=32,
                    value=1024,
                )
                
                height = gr.Slider(
                    label="Height",
                    minimum=256,
                    maximum=MAX_IMAGE_SIZE,
                    step=32,
                    value=1024,
                )
            
            with gr.Row():
                
  
                num_inference_steps = gr.Slider(
                    label="Number of inference steps",
                    minimum=1,
                    maximum=50,
                    step=1,
                    value=4,
                )
        
        gr.Examples(
            examples = examples,
            fn = infer,
            inputs = [prompt],
            outputs = [result, seed],
            cache_examples="lazy"
        )

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

init()
demo.launch()