1
1
mirror of https://github.com/ytdl-org/youtube-dl synced 2024-12-05 05:12:56 +01:00
youtube-dl/youtube_dl/extractor/rutv.py

204 lines
7.5 KiB
Python
Raw Normal View History

# encoding: utf-8
from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..utils import (
ExtractorError,
int_or_none
)
2014-03-16 19:59:40 +01:00
class RUTVIE(InfoExtractor):
IE_DESC = 'RUTV.RU'
_VALID_URL = r'''(?x)
https?://player\.(?:rutv\.ru|vgtrk\.com)/
(?P<path>flash2v/container\.swf\?id=
|iframe/(?P<type>swf|video|live)/id/
|index/iframe/cast_id/)
(?P<id>\d+)'''
_TESTS = [
{
2014-03-16 19:59:40 +01:00
'url': 'http://player.rutv.ru/flash2v/container.swf?id=774471&sid=kultura&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972347/video_id/978186/brand_id/31724',
'info_dict': {
2014-03-16 19:59:40 +01:00
'id': '774471',
'ext': 'mp4',
2014-03-16 19:59:40 +01:00
'title': 'Монологи на все времена',
'description': 'md5:18d8b5e6a41fb1faa53819471852d5d5',
'duration': 2906,
},
'params': {
# m3u8 download
'skip_download': True,
},
},
{
2014-03-16 19:59:40 +01:00
'url': 'https://player.vgtrk.com/flash2v/container.swf?id=774016&sid=russiatv&fbv=true&isPlay=true&ssl=false&i=560&acc_video_id=episode_id/972098/video_id/977760/brand_id/57638',
'info_dict': {
2014-03-16 19:59:40 +01:00
'id': '774016',
'ext': 'mp4',
2014-03-16 19:59:40 +01:00
'title': 'Чужой в семье Сталина',
'description': '',
'duration': 2539,
},
'params': {
# m3u8 download
'skip_download': True,
},
},
{
2014-03-16 19:59:40 +01:00
'url': 'http://player.rutv.ru/iframe/swf/id/766888/sid/hitech/?acc_video_id=4000',
'info_dict': {
'id': '766888',
'ext': 'mp4',
'title': 'Вести.net: интернет-гиганты начали перетягивание программных "одеял"',
'description': 'md5:65ddd47f9830c4f42ed6475f8730c995',
'duration': 279,
},
'params': {
# m3u8 download
'skip_download': True,
},
},
{
2014-03-16 19:59:40 +01:00
'url': 'http://player.rutv.ru/iframe/video/id/771852/start_zoom/true/showZoomBtn/false/sid/russiatv/?acc_video_id=episode_id/970443/video_id/975648/brand_id/5169',
'info_dict': {
2014-03-16 19:59:40 +01:00
'id': '771852',
'ext': 'mp4',
2014-03-16 19:59:40 +01:00
'title': 'Прямой эфир. Жертвы загадочной болезни: смерть от старости в 17 лет',
'description': 'md5:b81c8c55247a4bd996b43ce17395b2d8',
'duration': 3096,
},
'params': {
# m3u8 download
'skip_download': True,
},
},
{
2014-03-16 19:59:40 +01:00
'url': 'http://player.rutv.ru/iframe/live/id/51499/showZoomBtn/false/isPlay/true/sid/sochi2014',
'info_dict': {
'id': '51499',
'ext': 'flv',
'title': 'Сочи-2014. Биатлон. Индивидуальная гонка. Мужчины ',
'description': 'md5:9e0ed5c9d2fa1efbfdfed90c9a6d179c',
},
2015-05-03 06:00:34 +02:00
'skip': 'Translation has finished',
},
{
2015-05-04 17:23:26 +02:00
'url': 'http://player.rutv.ru/iframe/live/id/21/showZoomBtn/false/isPlay/true/',
2015-05-03 06:00:34 +02:00
'info_dict': {
'id': '21',
'ext': 'mp4',
'title': 're:^Россия 24. Прямой эфир [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
'is_live': True,
},
'params': {
2015-05-03 06:00:34 +02:00
# m3u8 download
'skip_download': True,
},
},
]
2014-03-16 19:59:40 +01:00
@classmethod
def _extract_url(cls, webpage):
mobj = re.search(
2015-05-22 21:01:47 +02:00
r'<iframe[^>]+?src=(["\'])(?P<url>https?://player\.(?:rutv\.ru|vgtrk\.com)/(?:iframe/(?:swf|video|live)/id|index/iframe/cast_id)/.+?)\1', webpage)
if mobj:
2014-03-16 19:59:40 +01:00
return mobj.group('url')
mobj = re.search(
2014-09-09 20:09:04 +02:00
r'<meta[^>]+?property=(["\'])og:video\1[^>]+?content=(["\'])(?P<url>https?://player\.(?:rutv\.ru|vgtrk\.com)/flash2v/container\.swf\?id=.+?\2)',
2014-03-16 19:59:40 +01:00
webpage)
if mobj:
2014-03-16 19:59:40 +01:00
return mobj.group('url')
2014-03-16 19:59:40 +01:00
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
video_path = mobj.group('path')
if video_path.startswith('flash2v'):
2014-03-16 19:59:40 +01:00
video_type = 'video'
elif video_path.startswith('iframe'):
video_type = mobj.group('type')
if video_type == 'swf':
video_type = 'video'
elif video_path.startswith('index/iframe/cast_id'):
video_type = 'live'
2015-05-04 17:29:32 +02:00
is_live = video_type == 'live'
json_data = self._download_json(
2015-05-04 17:29:32 +02:00
'http://player.rutv.ru/iframe/%splay/id/%s' % ('live-' if is_live else '', video_id),
video_id, 'Downloading JSON')
if json_data['errors']:
2014-03-16 19:59:40 +01:00
raise ExtractorError('%s said: %s' % (self.IE_NAME, json_data['errors']), expected=True)
playlist = json_data['data']['playlist']
medialist = playlist['medialist']
media = medialist[0]
if media['errors']:
2014-03-16 19:59:40 +01:00
raise ExtractorError('%s said: %s' % (self.IE_NAME, media['errors']), expected=True)
view_count = playlist.get('count_views')
priority_transport = playlist['priority_transport']
thumbnail = media['picture']
2014-03-04 15:40:35 +01:00
width = int_or_none(media['width'])
height = int_or_none(media['height'])
description = media['anons']
title = media['title']
duration = int_or_none(media.get('duration'))
formats = []
for transport, links in media['sources'].items():
for quality, url in links.items():
2015-05-04 17:29:32 +02:00
preference = -1 if priority_transport == transport else -2
if transport == 'rtmp':
mobj = re.search(r'^(?P<url>rtmp://[^/]+/(?P<app>.+))/(?P<playpath>.+)$', url)
if not mobj:
continue
fmt = {
'url': mobj.group('url'),
'play_path': mobj.group('playpath'),
'app': mobj.group('app'),
'page_url': 'http://player.rutv.ru',
'player_url': 'http://player.rutv.ru/flash2v/osmf.swf?i=22',
'rtmp_live': True,
'ext': 'flv',
'vbr': int(quality),
2015-05-04 17:29:32 +02:00
'preference': preference,
}
elif transport == 'm3u8':
2015-05-04 17:29:32 +02:00
formats.extend(self._extract_m3u8_formats(
url, video_id, 'mp4', preference=preference, m3u8_id='hls'))
2015-02-01 18:48:23 +01:00
continue
else:
fmt = {
'url': url
}
fmt.update({
'width': width,
'height': height,
'format_id': '%s-%s' % (transport, quality),
})
formats.append(fmt)
self._sort_formats(formats)
return {
'id': video_id,
2015-05-03 05:56:03 +02:00
'title': self._live_title(title) if is_live else title,
'description': description,
'thumbnail': thumbnail,
'view_count': view_count,
'duration': duration,
'formats': formats,
2015-05-03 05:56:03 +02:00
'is_live': is_live,
2014-11-23 20:41:03 +01:00
}