yt-dlp/youtube_dl/extractor/rtbf.py

83 lines
2.5 KiB
Python
Raw Normal View History

2014-04-28 20:32:13 +02:00
# coding: utf-8
from __future__ import unicode_literals
from .common import InfoExtractor
2015-05-24 17:09:08 +02:00
from ..utils import (
int_or_none,
unescapeHTML,
)
2014-04-28 20:32:13 +02:00
class RTBFIE(InfoExtractor):
_VALID_URL = r'''(?x)
https?://www\.rtbf\.be/
(?:
video/[^\?]+\?id=|
ouftivi/heros/[^&]+&videoId=
)
(?P<id>\d+)
'''
_TESTS = [
{
'url': 'https://www.rtbf.be/video/detail_les-diables-au-coeur-episode-2?id=1921274',
'md5': '799f334ddf2c0a582ba80c44655be570',
'info_dict': {
'id': '1921274',
'ext': 'mp4',
'title': 'Les Diables au coeur (épisode 2)',
'duration': 3099,
}
},
{
'url': 'http://www.rtbf.be/ouftivi/heros/detail_scooby-doo-mysteres-associes?id=1097&videoId=2057442',
'md5': '25aea17e949e1e0c7c41270d60d25f22',
'info_dict': {
'id': '2057442',
'ext': 'mp4',
'title': 'Scooby-Doo, myst\xe8res associ\xe9s',
'duration': 1279,
}
},
]
2014-04-28 20:32:13 +02:00
_QUALITIES = [
('mobile', 'mobile'),
('web', 'SD'),
('url', 'MD'),
('high', 'HD'),
]
2014-04-28 20:32:13 +02:00
def _real_extract(self, url):
2015-05-24 17:09:08 +02:00
video_id = self._match_id(url)
2014-04-28 20:32:13 +02:00
2015-05-24 17:09:08 +02:00
webpage = self._download_webpage(
'http://www.rtbf.be/video/embed?id=%s' % video_id, video_id)
2015-05-24 17:09:08 +02:00
data = self._parse_json(
unescapeHTML(self._search_regex(
2015-10-11 22:45:13 +02:00
r'data-media="([^"]+)"', webpage, 'data video')),
2015-05-24 17:09:08 +02:00
video_id)
2014-04-28 20:32:13 +02:00
2015-05-24 17:09:08 +02:00
if data.get('provider').lower() == 'youtube':
video_url = data.get('downloadUrl') or data.get('url')
return self.url_result(video_url, 'Youtube')
formats = []
for key, format_id in self._QUALITIES:
format_url = data['sources'].get(key)
if format_url:
formats.append({
'format_id': format_id,
'url': format_url,
})
2014-04-28 20:32:13 +02:00
return {
'id': video_id,
'formats': formats,
'title': data['title'],
'description': data.get('description') or data.get('subtitle'),
2015-05-24 17:09:08 +02:00
'thumbnail': data.get('thumbnail'),
'duration': data.get('duration') or data.get('realDuration'),
2015-05-24 17:09:08 +02:00
'timestamp': int_or_none(data.get('created')),
'view_count': int_or_none(data.get('viewCount')),
2014-04-28 20:32:13 +02:00
}