yt-dlc/youtube_dl/extractor/tvc.py

110 lines
3.8 KiB
Python
Raw Normal View History

2015-06-07 16:25:30 +02:00
# coding: utf-8
from __future__ import unicode_literals
import re
2015-06-07 16:25:30 +02:00
from .common import InfoExtractor
from ..utils import (
2015-06-12 12:13:36 +02:00
clean_html,
2015-06-07 16:25:30 +02:00
int_or_none,
)
2015-06-12 12:24:13 +02:00
class TVCIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?tvc\.ru/video/iframe/id/(?P<id>\d+)'
2015-06-12 12:13:36 +02:00
_TEST = {
'url': 'http://www.tvc.ru/video/iframe/id/74622/isPlay/false/id_stat/channel/?acc_video_id=/channel/brand/id/17/show/episodes/episode_id/39702',
'md5': 'bbc5ff531d1e90e856f60fc4b3afd708',
'info_dict': {
'id': '74622',
'ext': 'mp4',
'title': 'События. "События". Эфир от 22.05.2015 14:30',
'thumbnail': r're:^https?://.*\.jpg$',
2015-06-12 12:13:36 +02:00
'duration': 1122,
2015-06-07 16:25:30 +02:00
},
2015-06-12 12:13:36 +02:00
}
2015-06-07 16:25:30 +02:00
@classmethod
def _extract_url(cls, webpage):
mobj = re.search(
2015-06-12 15:31:52 +02:00
r'<iframe[^>]+?src=(["\'])(?P<url>(?:http:)?//(?:www\.)?tvc\.ru/video/iframe/id/[^"]+)\1', webpage)
if mobj:
return mobj.group('url')
2015-06-07 16:25:30 +02:00
def _real_extract(self, url):
2015-06-12 12:13:36 +02:00
video_id = self._match_id(url)
2015-06-07 16:25:30 +02:00
2015-06-12 12:13:36 +02:00
video = self._download_json(
'http://www.tvc.ru/video/json/id/%s' % video_id, video_id)
2015-06-07 16:25:30 +02:00
formats = []
2015-06-12 12:13:36 +02:00
for info in video.get('path', {}).get('quality', []):
video_url = info.get('url')
if not video_url:
continue
2015-06-07 16:25:30 +02:00
format_id = self._search_regex(
2015-06-12 12:13:36 +02:00
r'cdnvideo/([^/]+?)(?:-[^/]+?)?/', video_url,
'format id', default=None)
2015-06-07 16:25:30 +02:00
formats.append({
2015-06-12 12:13:36 +02:00
'url': video_url,
'format_id': format_id,
2015-06-07 16:25:30 +02:00
'width': int_or_none(info.get('width')),
'height': int_or_none(info.get('height')),
'tbr': int_or_none(info.get('bitrate')),
})
self._sort_formats(formats)
return {
'id': video_id,
2015-06-12 12:13:36 +02:00
'title': video['title'],
'thumbnail': video.get('picture'),
'duration': int_or_none(video.get('duration')),
2015-06-07 16:25:30 +02:00
'formats': formats,
}
2015-06-12 12:13:36 +02:00
2015-06-12 12:24:13 +02:00
class TVCArticleIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?tvc\.ru/(?!video/iframe/id/)(?P<id>[^?#]+)'
2015-06-12 12:13:36 +02:00
_TESTS = [{
'url': 'http://www.tvc.ru/channel/brand/id/29/show/episodes/episode_id/39702/',
'info_dict': {
'id': '74622',
'ext': 'mp4',
'title': 'События. "События". Эфир от 22.05.2015 14:30',
'description': 'md5:ad7aa7db22903f983e687b8a3e98c6dd',
'thumbnail': r're:^https?://.*\.jpg$',
2015-06-12 12:13:36 +02:00
'duration': 1122,
},
}, {
'url': 'http://www.tvc.ru/news/show/id/69944',
'info_dict': {
'id': '75399',
'ext': 'mp4',
'title': 'Эксперты: в столице встал вопрос о максимально безопасных остановках',
'description': 'md5:f2098f71e21f309e89f69b525fd9846e',
'thumbnail': r're:^https?://.*\.jpg$',
2015-06-12 12:13:36 +02:00
'duration': 278,
},
}, {
'url': 'http://www.tvc.ru/channel/brand/id/47/show/episodes#',
'info_dict': {
'id': '2185',
'ext': 'mp4',
'title': 'Ещё не поздно. Эфир от 03.08.2013',
'description': 'md5:51fae9f3f8cfe67abce014e428e5b027',
'thumbnail': r're:^https?://.*\.jpg$',
2015-06-12 12:13:36 +02:00
'duration': 3316,
},
}]
def _real_extract(self, url):
webpage = self._download_webpage(url, self._match_id(url))
return {
'_type': 'url_transparent',
2015-06-12 12:25:26 +02:00
'ie_key': 'TVC',
2015-06-12 12:13:36 +02:00
'url': self._og_search_video_url(webpage),
'title': clean_html(self._og_search_title(webpage)),
'description': clean_html(self._og_search_description(webpage)),
'thumbnail': self._og_search_thumbnail(webpage),
}