main.py

#!/usr/bin/python3
# -*- coding: utf-8 -*-
import argparse
import logging
from pathlib import Path
import platform
import re

from project_settings import project_path, log_directory
import log

log.setup(log_directory=log_directory)

import gradio as gr

from toolbox.os.command import Command

main_logger = logging.getLogger("main")


def get_args():
    parser = argparse.ArgumentParser()

    parser.add_argument(
        "--example_wav_dir",
        default=(project_path / "data/examples").as_posix(),
        type=str
    )
    args = parser.parse_args()
    return args


def process_uploaded_file(filename: str) -> str:
    filename = Path(filename).as_posix()

    main_logger.info("asr recognize: {}".format(filename))

    cmd = "build/asr_id --filename {}".format(
        filename
    )
    asr_result = Command.popen(cmd)

    pattern = "text: (.*)textSize: (.*)wordSize: (.*)timeCost: (.+)"
    match = re.search(pattern, asr_result, flags=re.IGNORECASE | re.DOTALL)

    if match is None:
        raise AssertionError("run asr recognize failed: \n{}".format(asr_result))

    text = match.group(1)

    return text


def shell(cmd: str):
    return Command.popen(cmd)


def main():
    args = get_args()

    title = "## 针对电话场景的印尼语ASR."

    # examples
    example_wav_dir = Path(args.example_wav_dir)

    examples = list()
    for filename in example_wav_dir.glob("*.wav"):
        examples.append(
            [
                filename.as_posix()
            ]
        )

    # blocks
    with gr.Blocks() as blocks:
        gr.Markdown(value=title)

        with gr.Tabs():
            with gr.TabItem("Upload from disk"):
                uploaded_file = gr.Audio(
                    sources=["upload"],
                    type="filepath",
                    label="Upload from disk",
                )
                upload_button = gr.Button("Submit for recognition")
                uploaded_output = gr.Textbox(label="Recognized speech from uploaded file")

            gr.Examples(
                examples=examples,
                inputs=[
                    uploaded_file,
                ],
                outputs=[
                    uploaded_output
                ],
                fn=process_uploaded_file
            )

            upload_button.click(
                process_uploaded_file,
                inputs=[
                    uploaded_file,
                ],
                outputs=[
                    uploaded_output
                ],
            )
            with gr.TabItem("shell"):
                shell_text = gr.Textbox(label="cmd")
                shell_button = gr.Button("run")
                shell_output = gr.Textbox(label="output")

            shell_button.click(
                shell,
                inputs=[
                    shell_text,
                ],
                outputs=[
                    shell_output
                ],
            )

    blocks.queue().launch(
        share=False if platform.system() == "Windows" else False,
        server_name="127.0.0.1" if platform.system() == "Windows" else "0.0.0.0",
        server_port=7860
    )

    return


if __name__ == "__main__":
    main()