yt-dlp/yt_dlp/extractor/tva.py

96 lines
3.7 KiB
Python
Raw Normal View History

import functools
import re
2017-01-26 18:36:28 +01:00
from .common import InfoExtractor
from ..utils import float_or_none, int_or_none, smuggle_url, strip_or_none
from ..utils.traversal import traverse_obj
2017-01-26 18:36:28 +01:00
class TVAIE(InfoExtractor):
2020-01-30 21:49:16 +01:00
_VALID_URL = r'https?://videos?\.tva\.ca/details/_(?P<id>\d+)'
_TESTS = [{
2017-10-11 16:14:51 +02:00
'url': 'https://videos.tva.ca/details/_5596811470001',
2017-01-26 18:36:28 +01:00
'info_dict': {
2017-10-11 16:14:51 +02:00
'id': '5596811470001',
2017-01-26 18:36:28 +01:00
'ext': 'mp4',
2017-10-11 16:14:51 +02:00
'title': 'Un extrait de l\'épisode du dimanche 8 octobre 2017 !',
'uploader_id': '5481942443001',
'upload_date': '20171003',
'timestamp': 1507064617,
2017-01-26 18:36:28 +01:00
},
'params': {
# m3u8 download
'skip_download': True,
2021-01-01 13:26:37 +01:00
},
'skip': 'HTTP Error 404: Not Found',
2020-01-30 21:49:16 +01:00
}, {
'url': 'https://video.tva.ca/details/_5596811470001',
'only_matching': True,
}]
2017-10-11 16:14:51 +02:00
BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/5481942443001/default_default/index.html?videoId=%s'
2017-01-26 18:36:28 +01:00
def _real_extract(self, url):
video_id = self._match_id(url)
return {
'_type': 'url_transparent',
'id': video_id,
2017-10-11 16:14:51 +02:00
'url': smuggle_url(self.BRIGHTCOVE_URL_TEMPLATE % video_id, {'geo_countries': ['CA']}),
'ie_key': 'BrightcoveNew',
2017-01-26 18:36:28 +01:00
}
2021-01-01 13:26:37 +01:00
class QubIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?qub\.ca/(?:[^/]+/)*[0-9a-z-]+-(?P<id>\d+)'
_TESTS = [{
'url': 'https://www.qub.ca/tvaplus/tva/alerte-amber/saison-1/episode-01-1000036619',
'md5': '949490fd0e7aee11d0543777611fbd53',
'info_dict': {
'id': '6084352463001',
'ext': 'mp4',
'title': 'Ép 01. Mon dernier jour',
2021-01-01 13:26:37 +01:00
'uploader_id': '5481942443001',
'upload_date': '20190907',
'timestamp': 1567899756,
'description': 'md5:9c0d7fbb90939420c651fd977df90145',
'thumbnail': r're:https://.+\.jpg',
'episode': 'Ép 01. Mon dernier jour',
'episode_number': 1,
'tags': ['alerte amber', 'alerte amber saison 1', 'surdemande'],
'duration': 2625.963,
'season': 'Season 1',
'season_number': 1,
'series': 'Alerte Amber',
'channel': 'TVA',
2021-01-01 13:26:37 +01:00
},
}, {
'url': 'https://www.qub.ca/tele/video/lcn-ca-vous-regarde-rev-30s-ap369664-1009357943',
'only_matching': True,
}]
# reference_id also works with old account_id(5481942443001)
# BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/5813221784001/default_default/index.html?videoId=ref:%s'
def _real_extract(self, url):
entity_id = self._match_id(url)
webpage = self._download_webpage(url, entity_id)
entity = self._search_nextjs_data(webpage, entity_id)['props']['initialProps']['pageProps']['fallbackData']
2021-01-01 13:26:37 +01:00
video_id = entity['videoId']
episode = strip_or_none(entity.get('name'))
return {
'_type': 'url_transparent',
'url': f'https://videos.tva.ca/details/_{video_id}',
'ie_key': TVAIE.ie_key(),
2021-01-01 13:26:37 +01:00
'id': video_id,
'title': episode,
'episode': episode,
**traverse_obj(entity, {
'description': ('longDescription', {str}),
'duration': ('durationMillis', {functools.partial(float_or_none, scale=1000)}),
'channel': ('knownEntities', 'channel', 'name', {str}),
'series': ('knownEntities', 'videoShow', 'name', {str}),
'season_number': ('slug', {lambda x: re.search(r'/s(?:ai|ea)son-(\d+)/', x)}, 1, {int_or_none}),
'episode_number': ('episodeNumber', {int_or_none}),
}),
2021-01-01 13:26:37 +01:00
}