import os
import json
import argparse
import copy

import numpy as np
import matplotlib.pyplot as plt
import torch
import tqdm
import librosa
import librosa.display
import soundfile as sf
import gradio as gr
import pytube as pt

from pytube.exceptions import VideoUnavailable

from inference.style_transfer import *


yt_video_dir = "./yt_dir"
os.makedirs(yt_video_dir, exist_ok=True)


def get_audio_from_yt_video(yt_link: str):
    try:
        yt = pt.YouTube(yt_link)
        t = yt.streams.filter(only_audio=True)
        filename = os.path.join(yt_video_dir, binascii.hexlify(os.urandom(8)).decode() + ".wav")
        t[0].download(filename=filename)
    except VideoUnavailable as e:
        warnings.warn(f"Video Not Found at {yt_link} ({e})")
        filename = None

    return filename, filename


def inference(file_uploaded_in, file_uploaded_ref):

    output_wav_path = None

    # Perform music mixing style transfer
    args = set_up()
    
    inference_style_transfer = Mixing_Style_Transfer_Inference(args)
    output_wav_path = inference_style_transfer.inference(None, None)
    
    return output_wav_path, output_wav_path


with gr.Blocks() as demo:
    gr.HTML(
        """
            <div style="text-align: center; max-width: 700px; margin: 0 auto;">
              <div
                style="
                  display: inline-flex;
                  align-items: center;
                  gap: 0.8rem;
                  font-size: 1.75rem;
                "
              >
                <h1 style="font-weight: 900; margin-bottom: 7px;">
                  Music Mixing Style Transfer
                </h1>
              </div>
              <p style="margin-bottom: 10px; font-size: 94%">
                Hugging Face interactive demo of the paper "Music Mixing Style Transfer: A Contrastive Learning Approach to Disentangle Audio Effects" (ICASSP 2023).<br>
                a
            </div>
        """
    )
    with gr.Group():
        with gr.Column():
            with gr.Blocks():
                with gr.Tab("Input Music"):
                    file_uploaded_in = gr.Audio(label="Input track (mix) to be mixing style transferred", type="filepath")
                with gr.Tab("YouTube url"):
                    with gr.Row():
                        yt_link_in = gr.Textbox(
                            label="Enter YouTube Link of the Video", autofocus=True, lines=3
                        )
                        yt_btn = gr.Button("Download Audio from YouTube Link", size="lg")
                    yt_audio_path = gr.Audio(
                        label="Input Audio Extracted from the YouTube Video", interactive=False
                    )
                    yt_btn.click(
                        get_audio_from_yt_video,
                        inputs=[yt_link_in],
                        outputs=[yt_audio_path, file_uploaded_in],
                    )
            with gr.Blocks():
                with gr.Tab("Reference Music"):
                    file_uploaded_ref = gr.Audio(label="Reference track (mix) to copy mixing style", type="filepath")
                with gr.Tab("YouTube url"):
                    with gr.Row():
                        yt_link_ref = gr.Textbox(
                            label="Enter YouTube Link of the Video", autofocus=True, lines=3
                        )
                        yt_btn = gr.Button("Download Audio from YouTube Link", size="lg")
                    yt_audio_path = gr.Audio(
                        label="Reference Audio Extracted from the YouTube Video", interactive=False
                    )
                    yt_btn.click(
                        get_audio_from_yt_video,
                        inputs=[yt_link_ref],
                        outputs=[yt_audio_path, file_uploaded_ref],
                    )
            with gr.Column():
                inference_btn = gr.Button("Perform Style Transfer")
                
    with gr.Group():
        gr.HTML(
            """
            <div> <h3> <center> Mixing Style Transferred Output. </h3> </div>
            """
        )
        with gr.Row().style(mobile_collapse=False, equal_height=True):
            output_mix_l = gr.Audio(label="Listen to output mix")
            output_mix_p = gr.Audio(label="Download output mix")
        with gr.Row():
            output_mix = gr.File(label="Download style transferred music track")
            inference_btn.click(
                inference,
                inputs=[file_uploaded_in, file_uploaded_ref],
                outputs=[output_mix_l, output_mix_p],
            )


    
if __name__ == "__main__":
    demo.launch(debug=True)