mew/core/youtube.py

import time
import urllib

from pytube import YouTube

from core import constants, logging_
from core.exceptions import YoutubeStreamNotFound

MAX_ATTEMPTS = 5

logger = logging_.get()


class YoutubeDownloader:
    MAX_ATTEMPTS = 5
    OUTPUT_DIR = constants.APP_ROOT / 'data' / 'youtube'

    @staticmethod
    def get_audio_stream(yt):
        args = [
            {"only_audio": True},
            {"file_extension": 'mp4'}
        ]

        for a in args:
            stream = yt.streams \
                .filter(**a) \
                .first()
            if stream:
                return stream

    @staticmethod
    def get_stream(url, progress_callback=None, complete_callback=None):
        for i in range(1, YoutubeDownloader.MAX_ATTEMPTS + 1):
            logger.info(f"download {url} - attempt {i}/{YoutubeDownloader.MAX_ATTEMPTS}")
            try:
                yt = YouTube(url, progress_callback, complete_callback)

                return YoutubeDownloader.get_audio_stream(yt)

            except urllib.error.HTTPError:
                time.sleep(500)
                continue
        raise YoutubeStreamNotFound('Stream not found')

    @staticmethod
    def download(url, progress_callback=None, complete_callback=None):
        logger.info(f"** Download {url} into {YoutubeDownloader.OUTPUT_DIR}")

        stream = YoutubeDownloader.get_stream(
            url,
            progress_callback,
            complete_callback
        )

        logger.info(f"stream found - start downloading")

        stream.download(
            output_path=YoutubeDownloader.OUTPUT_DIR,
            timeout=5,
            max_retries=3
        )

        output_path = YoutubeDownloader.OUTPUT_DIR / stream.title
        logger.info(f"> file downloaded as {output_path}")
        return output_path


if __name__ == "__main__":

    YoutubeDownloader.download(
        'https://www.youtube.com/watch?v=nOWjX4BpC24XXXXXX',
    )