yt-dlp/yt_dlp/extractor/steam.py

130 lines
4.9 KiB
Python
Raw Normal View History

2013-06-23 16:02:56 -04:00
import re
from .common import InfoExtractor
from ..utils import (
2017-10-11 13:49:48 -04:00
extract_attributes,
2013-06-23 16:02:56 -04:00
ExtractorError,
2017-10-11 13:49:48 -04:00
get_element_by_class,
2013-06-23 16:02:56 -04:00
)
class SteamIE(InfoExtractor):
_VALID_URL = r"""(?x)
https?://(?:store\.steampowered|steamcommunity)\.com/
(?:agecheck/)?
(?P<urltype>video|app)/ #If the page is only for videos or for a game
(?P<gameID>\d+)/?
(?P<videoID>\d*)(?P<extra>\??) # For urltype == video we sometimes get the videoID
|
https?://(?:www\.)?steamcommunity\.com/sharedfiles/filedetails/\?id=(?P<fileID>[0-9]+)
"""
2013-06-23 16:02:56 -04:00
_VIDEO_PAGE_TEMPLATE = 'http://store.steampowered.com/video/%s/'
_AGECHECK_TEMPLATE = 'http://store.steampowered.com/agecheck/video/%s/?snr=1_agecheck_agecheck__age-gate&ageDay=1&ageMonth=January&ageYear=1970'
_TESTS = [{
2016-02-14 04:37:17 -05:00
'url': 'http://store.steampowered.com/video/105600/',
'playlist': [
2013-06-27 14:20:00 -04:00
{
'md5': '695242613303ffa2a4c44c9374ddc067',
2016-02-14 04:37:17 -05:00
'info_dict': {
'id': '256785003',
2017-10-11 13:49:48 -04:00
'ext': 'mp4',
'title': 'Terraria video 256785003',
'thumbnail': r're:^https://cdn\.[^\.]+\.steamstatic\.com',
'n_entries': 2,
2013-06-27 14:20:00 -04:00
}
},
{
'md5': '6a294ee0c4b1f47f5bb76a65e31e3592',
2016-02-14 04:37:17 -05:00
'info_dict': {
'id': '2040428',
2017-10-11 13:49:48 -04:00
'ext': 'mp4',
'title': 'Terraria video 2040428',
2014-04-19 13:51:04 -04:00
'playlist_index': 2,
'thumbnail': r're:^https://cdn\.[^\.]+\.steamstatic\.com',
'n_entries': 2,
2013-06-27 14:20:00 -04:00
}
}
2014-04-19 13:55:53 -04:00
],
2017-10-11 13:49:48 -04:00
'info_dict': {
'id': '105600',
'title': 'Terraria',
},
2014-04-19 13:55:53 -04:00
'params': {
'playlistend': 2,
}
}, {
'url': 'https://store.steampowered.com/app/271590/Grand_Theft_Auto_V/',
'info_dict': {
'id': '256757115',
'title': 'Grand Theft Auto V video 256757115',
'ext': 'mp4',
'thumbnail': r're:^https://cdn\.[^\.]+\.steamstatic\.com',
'n_entries': 20,
},
}]
2013-06-27 14:20:00 -04:00
2013-06-23 16:02:56 -04:00
def _real_extract(self, url):
m = self._match_valid_url(url)
fileID = m.group('fileID')
if fileID:
video_url = url
playlist_id = fileID
else:
gameID = m.group('gameID')
playlist_id = gameID
video_url = self._VIDEO_PAGE_TEMPLATE % playlist_id
self._set_cookie('steampowered.com', 'wants_mature_content', '1')
self._set_cookie('steampowered.com', 'birthtime', '944006401')
self._set_cookie('steampowered.com', 'lastagecheckage', '1-0-2000')
webpage = self._download_webpage(video_url, playlist_id)
2013-06-23 16:02:56 -04:00
if re.search('<div[^>]+>Please enter your birth date to continue:</div>', webpage) is not None:
video_url = self._AGECHECK_TEMPLATE % playlist_id
2013-06-23 16:02:56 -04:00
self.report_age_confirmation()
webpage = self._download_webpage(video_url, playlist_id)
2017-10-11 13:49:48 -04:00
videos = re.findall(r'(<div[^>]+id=[\'"]highlight_movie_(\d+)[\'"][^>]+>)', webpage)
2017-10-11 13:49:48 -04:00
entries = []
playlist_title = get_element_by_class('apphub_AppName', webpage)
for movie, movie_id in videos:
if not movie:
continue
movie = extract_attributes(movie)
if not movie_id:
continue
entry = {
'id': movie_id,
'title': f'{playlist_title} video {movie_id}',
}
formats = []
if movie:
entry['thumbnail'] = movie.get('data-poster')
for quality in ('', '-hd'):
for ext in ('webm', 'mp4'):
video_url = movie.get('data-%s%s-source' % (ext, quality))
if video_url:
formats.append({
'format_id': ext + quality,
'url': video_url,
})
self._sort_formats(formats)
entry['formats'] = formats
entries.append(entry)
embedded_videos = re.findall(r'(<iframe[^>]+>)', webpage)
for evideos in embedded_videos:
evideos = extract_attributes(evideos).get('src')
video_id = self._search_regex(r'youtube\.com/embed/([0-9A-Za-z_-]{11})', evideos, 'youtube_video_id', default=None)
if video_id:
2017-10-11 13:49:48 -04:00
entries.append({
'_type': 'url_transparent',
'id': video_id,
'url': video_id,
2017-10-11 13:49:48 -04:00
'ie_key': 'Youtube',
})
if not entries:
raise ExtractorError('Could not find any videos')
2013-06-23 16:02:56 -04:00
2017-10-11 13:49:48 -04:00
return self.playlist_result(entries, playlist_id, playlist_title)