import spaces
import torch

import gradio as gr
import yt_dlp as youtube_dl
from transformers import pipeline
from transformers.pipelines.audio_utils import ffmpeg_read

import tempfile
import os

MODEL_NAME = "openai/whisper-large-v3"
BATCH_SIZE = 8
FILE_LIMIT_MB = 1000
YT_LENGTH_LIMIT_S = 3600  # limit to 1 hour YouTube files

device = 0 if torch.cuda.is_available() else "cpu"

pipe = pipeline(
    task="automatic-speech-recognition",
    model=MODEL_NAME,
    chunk_length_s=30,
    device=device,
)


@spaces.GPU
def transcribe(inputs, task):
    if inputs is None:
        raise gr.Error("No audio file submitted! Please upload or record an audio file before submitting your request.")

    result = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps="word")
    
    text = result["text"]
    timestamps = result["chunks"]  # each chunk contains the word and its timestamps
    
    word_timestamps = []
    for chunk in timestamps:
        for word_info in chunk["words"]:
            word_timestamps.append(f"{word_info['word']} [{word_info['start']}-{word_info['end']}]")
    
    return "\n".join(word_timestamps)


@spaces.GPU
def yt_transcribe(yt_url, task, max_filesize=75.0):
    html_embed_str = _return_yt_html_embed(yt_url)

    with tempfile.TemporaryDirectory() as tmpdirname:
        filepath = os.path.join(tmpdirname, "video.mp4")
        download_yt_audio(yt_url, filepath)
        with open(filepath, "rb") as f:
            inputs = f.read()

    inputs = ffmpeg_read(inputs, pipe.feature_extractor.sampling_rate)
    inputs = {"array": inputs, "sampling_rate": pipe.feature_extractor.sampling_rate}

    result = pipe(inputs, batch_size=BATCH_SIZE, generate_kwargs={"task": task}, return_timestamps="word")
    
    text = result["text"]
    timestamps = result["chunks"]

    word_timestamps = []
    for chunk in timestamps:
        for word_info in chunk["words"]:
            word_timestamps.append(f"{word_info['word']} [{word_info['start']}-{word_info['end']}]")
    
    return html_embed_str, "\n".join(word_timestamps)



demo = gr.Blocks()

mf_transcribe = gr.Interface(
    fn=transcribe,
    inputs=[
        gr.Audio(sources="microphone", type="filepath"),
        gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
    ],
    outputs=["text", "text"],  # Output both text and timestamps
    title="Whisper Large V3: Transcribe Audio",
    description=(
        "Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the"
        f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe audio files"
        " of arbitrary length."
    ),
    allow_flagging="never",
)

file_transcribe = gr.Interface(
    fn=transcribe,
    inputs=[
        gr.Audio(sources="upload", type="filepath", label="Audio file"),
        gr.Radio(["transcribe", "translate"], label="Task", value="transcribe"),
    ],
    outputs=["text", "text"],  # Output both text and timestamps
    title="Whisper Large V3: Transcribe Audio",
    description=(
        "Transcribe long-form microphone or audio inputs with the click of a button! Demo uses the"
        f" checkpoint [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe audio files"
        " of arbitrary length."
    ),
    allow_flagging="never",
)

yt_transcribe = gr.Interface(
    fn=yt_transcribe,
    inputs=[
        gr.Textbox(lines=1, placeholder="Paste the URL to a YouTube video here", label="YouTube URL"),
        gr.Radio(["transcribe", "translate"], label="Task", value="transcribe")
    ],
    outputs=["html", "text", "text"],  # Output both text and timestamps
    title="Whisper Large V3: Transcribe YouTube",
    description=(
        "Transcribe long-form YouTube videos with the click of a button! Demo uses the checkpoint"
        f" [{MODEL_NAME}](https://huggingface.co/{MODEL_NAME}) and 🤗 Transformers to transcribe video files of"
        " arbitrary length."
    ),
    allow_flagging="never",
)

with demo:
    gr.TabbedInterface([mf_transcribe, file_transcribe, yt_transcribe], ["Microphone", "Audio file", "YouTube"])

demo.queue().launch()