import torch
import gradio as gr
from PIL import Image
import qrcode
from pathlib import Path
import requests
import io
import os
import spaces
import random

from diffusers import (
    StableDiffusionXLControlNetPipeline,
    ControlNetModel,
    AutoencoderKL,
    DiffusionPipeline,
    DDIMScheduler,
    DPMSolverMultistepScheduler,
    DEISMultistepScheduler,
    HeunDiscreteScheduler,
    EulerDiscreteScheduler,
)

MAX_SEED = 2**32 - 1

# QR Code generation setup
qrcode_generator = qrcode.QRCode(
    version=1,
    error_correction=qrcode.ERROR_CORRECT_H,
    box_size=16,
    border=4,
)

# SDXL and ControlNet setup
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)

controlnet = ControlNetModel.from_pretrained(
    "AGCobra/1",
    torch_dtype=torch.float16
).to(device)

pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
    "stabilityai/stable-diffusion-xl-base-1.0",
    vae=vae,
    controlnet=controlnet,
    torch_dtype=torch.float16,
    use_safetensors=True,
    variant="fp16",
).to(device)

# Sampler setup
SAMPLER_MAP = {
    "DPM++ Karras SDE": lambda config: DPMSolverMultistepScheduler.from_config(config, use_karras=True, algorithm_type="sde-dpmsolver++"),
    "DPM++ Karras": lambda config: DPMSolverMultistepScheduler.from_config(config, use_karras=True),
    "Heun": lambda config: HeunDiscreteScheduler.from_config(config),
    "Euler": lambda config: EulerDiscreteScheduler.from_config(config),
    "DDIM": lambda config: DDIMScheduler.from_config(config),
    "DEIS": lambda config: DEISMultistepScheduler.from_config(config),
}

def resize_for_condition_image(input_image: Image.Image, resolution: int):
    input_image = input_image.convert("RGB")
    W, H = input_image.size
    k = float(resolution) / min(H, W)
    H *= k
    W *= k
    H = int(round(H / 64.0)) * 64
    W = int(round(W / 64.0)) * 64
    img = input_image.resize((W, H), resample=Image.LANCZOS)
    return img

@spaces.GPU()
def inference(
    qr_code_content: str,
    prompt: str,
    negative_prompt: str,
    guidance_scale: float = 7.5,
    controlnet_conditioning_scale: float = 1.1,
    strength: float = 0.9,
    seed: int = -1,
    sampler: str = "DPM++ Karras SDE",
):
    if prompt is None or prompt == "":
        raise gr.Error("Prompt is required")

    if qr_code_content == "":
        raise gr.Error("QR Code Content is required")

    pipe.scheduler = SAMPLER_MAP[sampler](pipe.scheduler.config)

    if seed == -1:
        seed = random.randint(0, MAX_SEED)
    
    # Use a sub-seed for additional randomness
    subseed = random.randint(0, MAX_SEED)
    generator = torch.Generator(device=device).manual_seed(seed + subseed)

    print("Generating QR Code from content")
    qr = qrcode.QRCode(
        version=1,
        error_correction=qrcode.constants.ERROR_CORRECT_H,
        box_size=16,
        border=4,
    )
    qr.add_data(qr_code_content)
    qr.make(fit=True)

    qrcode_image = qr.make_image(fill_color="black", back_color="white")
    qrcode_image = resize_for_condition_image(qrcode_image, 1024)

    init_image = qrcode_image

    out = pipe(
        prompt=prompt,
        negative_prompt=negative_prompt,
        image=init_image,
        control_image=qrcode_image,
        controlnet_conditioning_scale=float(controlnet_conditioning_scale),
        guidance_scale=float(guidance_scale),
        generator=generator,
        strength=float(strength),
        num_inference_steps=30,
    )
    return out.images[0]

with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column():
            qr_code_content = gr.Textbox(
                label="QR Code Content",
                info="QR Code Content or URL",
                value="",
            )

            prompt = gr.Textbox(
                label="Prompt",
                info="Prompt that guides the generation towards",
            )
            negative_prompt = gr.Textbox(
                label="Negative Prompt",
                value="ugly, disfigured, low quality, blurry",
            )

            with gr.Accordion(
                label="Advanced Parameters",
                open=True,
            ):
                controlnet_conditioning_scale = gr.Slider(
                    minimum=0.0,
                    maximum=2.0,
                    step=0.01,
                    value=1.1,
                    label="Controlnet Conditioning Scale",
                )
                strength = gr.Slider(
                    minimum=0.0, maximum=1.0, step=0.01, value=0.9, label="Strength"
                )
                guidance_scale = gr.Slider(
                    minimum=0.0,
                    maximum=50.0,
                    step=0.25,
                    value=7.5,
                    label="Guidance Scale",
                )
                sampler = gr.Dropdown(choices=list(SAMPLER_MAP.keys()), value="DPM++ Karras SDE", label="Sampler")
                seed = gr.Slider(
                    minimum=-1,
                    maximum=MAX_SEED,
                    step=1,
                    value=-1,
                    label="Seed",
                    randomize=True,
                )
            with gr.Row():
                run_btn = gr.Button("Run")
        with gr.Column():
            result_image = gr.Image(label="Result Image")
    run_btn.click(
        inference,
        inputs=[
            qr_code_content,
            prompt,
            negative_prompt,
            guidance_scale,
            controlnet_conditioning_scale,
            strength,
            seed,
            sampler,
        ],
        outputs=[result_image],
    )

demo.queue(max_size=20).launch(share=bool(os.environ.get("SHARE", False)))