youtool - Easily access YouTube Data API v3 in batches

Python library (and future command-line interface) to crawl YouTube Data API v3 in batch operations and other related tasks. Easier to use than alternatives - you don't need to spend time learning the YouTube API and its caveats. With this library you can get:

• Channel ID from channel URL (scraping) or username (API)
• Channel information (title, subscribers etc.)
• List of playlists for a channel
• List of videos for a playlist
• Video search (many parameters)
• Video information (title, description, likes, comments etc.)
• Comments
• Livechat, including superchat (scraping using chat-downloader)
• Automatic transcription (scraping using yt-dlp)

The library will automatically:

• Try as many keys as you provide
• Use batch of 50 items in supported API endpoints
• Paginate when needed

Installing

pip install youtool

You may also want some extras:

pip install youtool[livechat]
pip install youtool[transcription]

Using as a library

Just follow the tutorial/examples below and check the help() for YouTube methods.

from pprint import pprint
from pathlib import Path

from youtool import YouTube

api_keys = ["key1", "key2", ...]  # Create one in Google Cloud Console
yt = YouTube(api_keys, disable_ipv6=True)  # Will try all keys

channel_id_1 = yt.channel_id_from_url("https://youtube.com/c/PythonicCafe/")
print(f"Pythonic Café's channel ID (got from URL): {channel_id_1}")
channel_id_2 = yt.channel_id_from_username("turicas")
print(f"Turicas' channel ID (got from username): {channel_id_2}")

print('Playlists found on Turicas' channel (the "uploads" playlist is not here):')
# WARNING: this method won't return the main channel playlist ("uploads").
# If you need it, get channel info using `channels_infos` and the `playlist_id` key (or use the hack in the next
# section), so you can pass it to `playlist_videos`.
for playlist in yt.channel_playlists(channel_id_2):
    # `playlist` is a `dict`
    print(f"Playlist: {playlist}")
    for video in yt.playlist_videos(playlist["id"]):
        # `video` is a `dict`, but this endpoint doesn't provide full video information (use `videos_infos` to get them)
        print(f"  Video: {video}")
    print("-" * 80)

# Hack: replace `UC` with `UU` on channel ID to get main playlist ID ("uploads"):
assert channel_id_1[:2] == "UC"
print("Last 3 uploads for Pythonic Café:")
for index, video in enumerate(yt.playlist_videos("UU" + channel_id_1[2:])):
    # `video` is a `dict`, but this endpoint doesn't provide full video information (use `videos_infos` to get them)
    print(f"  Video: {video}")
    if index == 2:  # First 3 results only
        break
print("-" * 80)

print("5 videos found on search:")
# `video_search` has many other parameters also!
# WARNING: each request made by this method will consume 100 units of your quota (out of 10k daily!)
for index, video in enumerate(yt.video_search(term="Álvaro Justen")):  # Will paginate automatically
    # `video` is a `dict`, but this endpoint doesn't provide full video information (use `videos_infos` to get them)
    print(f"  Video: {video}")
    if index == 4:  # First 5 results only
        break
print("-" * 80)

# The method below can be used to get information in batches (50 videos per request) - you can pass a list of video IDs
# (more than 50) and it'll get data in batches from the API.
last_video = list(yt.videos_infos([video["id"]]))[0]
print("Complete information for last video:")
pprint(last_video)
print("-" * 80)

print("Channel information (2 channels in one request):")
for channel in yt.channels_infos([channel_id_1, channel_id_2]):
    # `channel` is a `dict`
    print(channel)
print("-" * 80)

video_id = "b1FjmUzgFB0"
print(f"Comments for video {video_id}:")
for comment in yt.video_comments(video_id):
    # `comment` is a `dict`
    print(comment)
print("-" * 80)

live_video_id = "yyzIPQsa98A"
print(f"Live chat for video {live_video_id}:")
for chat_message in yt.video_livechat(live_video_id):
    # `chat_message` is a `dict`
    print(chat_message)  # It has the superchat information (`money_currency` and `money_amount` keys)
print("-" * 80)

download_path = Path("transcriptions")
if not download_path.exists():
    download_path.mkdir(parents=True)
print(f"Downloading Portuguese (pt) transcriptions for videos {video_id} and {live_video_id} - saving at {download_path.absolute()}")
yt.videos_transcriptions([video_id, live_video_id], language_code="pt", path=download_path)
for vid in [video_id, live_video_id]:
    result = list(download_path.glob(f"{vid}*vtt"))
    if not result:
        print(f"  Transcription for video {vid} could not be downloaded.")
    else:
        filename = result[0]
        print(f"  Downloaded: {filename} ({filename.stat().st_size / 1024:.1f} KiB)")
print("-" * 80)

print("Categories in Brazilian YouTube:")
for category in yt.categories(region_code="BR"):
    # `category` is a `dict`
    print(category)
print("-" * 80)

print("Current most popular videos in Brazil:")
for video in yt.most_popular(region_code="BR"):  # Will paginate automatically
    # `video` is a `dict`, but this endpoint doesn't provide full video information (use `videos_infos` to get them)
    print(f"{video['id']} {video['title']}")
print("-" * 80)

Tests

To run all tests, execute:

make test

Future improvments

Pull requests are welcome! :)

• Command-line interface with the following subcommands:
  • channel-id: get channel IDs from a list of URLs (or CSV filename with URLs inside), generate CSV output (just the IDs)
  • channel-info: get channel info from a list of IDs (or CSV filename with IDs inside), generate CSV output (same schema for channel dicts)
  • video-info: get video info from a list of IDs or URLs (or CSV filename with URLs/IDs inside), generate CSV output (same schema for video dicts)
  • video-search: get video info from a list of IDs or URLs (or CSV filename with URLs/IDs inside), generate CSV output (simplified video dict schema or option to get full video info after)
  • video-comments: get comments from a video ID, generate CSV output (same schema for comment dicts)
  • video-livechat: get comments from a video ID, generate CSV output (same schema for chat_message dicts)
  • video-transcriptions: download video transcriptions based on language code, path and list of video IDs or URLs (or CSV filename with URLs/IDs inside), download files to destination and report results
• Replace dicts with dataclasses
• Create a website with docs/reference
• Deal with quotas (wait some time before using a key, for example)

License

GNU Lesser General Public License (LGPL) version3.

This project was developed in a partnership between Pythonic Café and Novelo Data.