2013-08-02 11:58:46 +02:00
|
|
|
|
# encoding: utf-8
|
2014-11-26 12:53:55 +01:00
|
|
|
|
from __future__ import unicode_literals
|
2013-08-02 11:58:46 +02:00
|
|
|
|
|
2013-08-28 13:59:08 +02:00
|
|
|
|
import re
|
2013-08-02 11:58:46 +02:00
|
|
|
|
|
|
|
|
|
from .common import InfoExtractor
|
2015-03-05 19:48:27 +01:00
|
|
|
|
from ..compat import (
|
|
|
|
|
compat_str,
|
|
|
|
|
compat_urllib_request
|
|
|
|
|
)
|
2015-05-09 07:54:28 +02:00
|
|
|
|
from ..utils import ExtractorError
|
2013-08-02 11:58:46 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class SohuIE(InfoExtractor):
|
2013-09-09 19:56:16 +02:00
|
|
|
|
_VALID_URL = r'https?://(?P<mytv>my\.)?tv\.sohu\.com/.+?/(?(mytv)|n)(?P<id>\d+)\.shtml.*?'
|
2013-08-02 11:58:46 +02:00
|
|
|
|
|
2015-03-05 19:43:05 +01:00
|
|
|
|
_TESTS = [{
|
|
|
|
|
'note': 'This video is available only in Mainland China',
|
2014-11-26 12:53:55 +01:00
|
|
|
|
'url': 'http://tv.sohu.com/20130724/n382479172.shtml#super',
|
2015-03-05 19:43:05 +01:00
|
|
|
|
'md5': '29175c8cadd8b5cc4055001e85d6b372',
|
2014-11-26 12:53:55 +01:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '382479172',
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
'title': 'MV:Far East Movement《The Illest》',
|
2013-08-02 11:58:46 +02:00
|
|
|
|
},
|
2015-05-20 08:08:23 +02:00
|
|
|
|
'skip': 'On available in China',
|
2015-03-05 19:43:05 +01:00
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://tv.sohu.com/20150305/n409385080.shtml',
|
2015-05-09 07:54:28 +02:00
|
|
|
|
'md5': 'ac9a5d322b4bf9ae184d53e4711e4f1a',
|
2015-03-05 19:43:05 +01:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '409385080',
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
'title': '《2015湖南卫视羊年元宵晚会》唐嫣《花好月圆》',
|
|
|
|
|
}
|
|
|
|
|
}, {
|
|
|
|
|
'url': 'http://my.tv.sohu.com/us/232799889/78693464.shtml',
|
2015-05-09 07:54:28 +02:00
|
|
|
|
'md5': '49308ff6dafde5ece51137d04aec311e',
|
2015-03-05 19:43:05 +01:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '78693464',
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
'title': '【爱范品】第31期:MWC见不到的奇葩手机',
|
|
|
|
|
}
|
2015-03-14 17:59:49 +01:00
|
|
|
|
}, {
|
|
|
|
|
'note': 'Multipart video',
|
|
|
|
|
'url': 'http://my.tv.sohu.com/pl/8384802/78910339.shtml',
|
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '78910339',
|
2015-04-19 13:19:44 +02:00
|
|
|
|
'title': '【神探苍实战秘籍】第13期 战争之影 赫卡里姆',
|
2015-03-14 17:59:49 +01:00
|
|
|
|
},
|
|
|
|
|
'playlist': [{
|
2015-05-09 07:54:28 +02:00
|
|
|
|
'md5': '492923eac023ba2f13ff69617c32754a',
|
2015-03-14 17:59:49 +01:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '78910339_part1',
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
'duration': 294,
|
|
|
|
|
'title': '【神探苍实战秘籍】第13期 战争之影 赫卡里姆',
|
|
|
|
|
}
|
|
|
|
|
}, {
|
2015-05-09 07:54:28 +02:00
|
|
|
|
'md5': 'de604848c0e8e9c4a4dde7e1347c0637',
|
2015-03-14 17:59:49 +01:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '78910339_part2',
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
'duration': 300,
|
|
|
|
|
'title': '【神探苍实战秘籍】第13期 战争之影 赫卡里姆',
|
|
|
|
|
}
|
|
|
|
|
}, {
|
2015-05-09 07:54:28 +02:00
|
|
|
|
'md5': '93584716ee0657c0b205b8aa3d27aa13',
|
2015-03-14 17:59:49 +01:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '78910339_part3',
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
'duration': 150,
|
|
|
|
|
'title': '【神探苍实战秘籍】第13期 战争之影 赫卡里姆',
|
|
|
|
|
}
|
|
|
|
|
}]
|
2015-03-14 18:05:01 +01:00
|
|
|
|
}, {
|
2015-03-17 16:39:31 +01:00
|
|
|
|
'note': 'Video with title containing dash',
|
2015-03-14 18:05:01 +01:00
|
|
|
|
'url': 'http://my.tv.sohu.com/us/249884221/78932792.shtml',
|
|
|
|
|
'info_dict': {
|
|
|
|
|
'id': '78932792',
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
'title': 'youtube-dl testing video',
|
|
|
|
|
},
|
|
|
|
|
'params': {
|
|
|
|
|
'skip_download': True
|
|
|
|
|
}
|
2015-03-05 19:43:05 +01:00
|
|
|
|
}]
|
2013-08-02 11:58:46 +02:00
|
|
|
|
|
|
|
|
|
def _real_extract(self, url):
|
2013-08-28 13:59:08 +02:00
|
|
|
|
|
2013-09-09 19:56:16 +02:00
|
|
|
|
def _fetch_data(vid_id, mytv=False):
|
|
|
|
|
if mytv:
|
|
|
|
|
base_data_url = 'http://my.tv.sohu.com/play/videonew.do?vid='
|
|
|
|
|
else:
|
2014-11-26 12:53:55 +01:00
|
|
|
|
base_data_url = 'http://hot.vrs.sohu.com/vrs_flash.action?vid='
|
2014-12-25 17:25:05 +01:00
|
|
|
|
|
2015-03-05 19:43:05 +01:00
|
|
|
|
req = compat_urllib_request.Request(base_data_url + vid_id)
|
|
|
|
|
|
|
|
|
|
cn_verification_proxy = self._downloader.params.get('cn_verification_proxy')
|
|
|
|
|
if cn_verification_proxy:
|
|
|
|
|
req.add_header('Ytdl-request-proxy', cn_verification_proxy)
|
|
|
|
|
|
2015-03-17 16:39:31 +01:00
|
|
|
|
return self._download_json(
|
|
|
|
|
req, video_id,
|
|
|
|
|
'Downloading JSON data for %s' % vid_id)
|
2013-08-28 13:59:08 +02:00
|
|
|
|
|
2013-08-02 11:58:46 +02:00
|
|
|
|
mobj = re.match(self._VALID_URL, url)
|
|
|
|
|
video_id = mobj.group('id')
|
2013-09-09 19:56:16 +02:00
|
|
|
|
mytv = mobj.group('mytv') is not None
|
2013-08-28 13:59:08 +02:00
|
|
|
|
|
2013-08-02 11:58:46 +02:00
|
|
|
|
webpage = self._download_webpage(url, video_id)
|
2015-03-14 18:05:01 +01:00
|
|
|
|
|
2015-04-19 13:19:44 +02:00
|
|
|
|
title = re.sub(r' - 搜狐视频$', '', self._og_search_title(webpage))
|
2013-08-02 11:58:46 +02:00
|
|
|
|
|
2014-12-25 17:25:05 +01:00
|
|
|
|
vid = self._html_search_regex(
|
|
|
|
|
r'var vid ?= ?["\'](\d+)["\']',
|
|
|
|
|
webpage, 'video path')
|
|
|
|
|
vid_data = _fetch_data(vid, mytv)
|
2015-05-09 07:19:54 +02:00
|
|
|
|
if vid_data['play'] != 1:
|
|
|
|
|
if vid_data.get('status') == 12:
|
|
|
|
|
raise ExtractorError(
|
|
|
|
|
'Sohu said: There\'s something wrong in the video.',
|
|
|
|
|
expected=True)
|
|
|
|
|
else:
|
|
|
|
|
raise ExtractorError(
|
|
|
|
|
'Sohu said: The video is only licensed to users in Mainland China.',
|
|
|
|
|
expected=True)
|
2013-08-28 13:59:08 +02:00
|
|
|
|
|
2014-12-25 17:25:05 +01:00
|
|
|
|
formats_json = {}
|
|
|
|
|
for format_id in ('nor', 'high', 'super', 'ori', 'h2644k', 'h2654k'):
|
|
|
|
|
vid_id = vid_data['data'].get('%sVid' % format_id)
|
|
|
|
|
if not vid_id:
|
|
|
|
|
continue
|
|
|
|
|
vid_id = compat_str(vid_id)
|
|
|
|
|
formats_json[format_id] = vid_data if vid == vid_id else _fetch_data(vid_id, mytv)
|
2013-08-28 13:59:08 +02:00
|
|
|
|
|
2014-12-25 17:25:05 +01:00
|
|
|
|
part_count = vid_data['data']['totalBlocks']
|
2013-08-28 13:59:08 +02:00
|
|
|
|
|
|
|
|
|
playlist = []
|
|
|
|
|
for i in range(part_count):
|
2014-12-25 17:25:05 +01:00
|
|
|
|
formats = []
|
|
|
|
|
for format_id, format_data in formats_json.items():
|
|
|
|
|
data = format_data['data']
|
2015-03-05 19:43:05 +01:00
|
|
|
|
|
2015-05-09 07:54:28 +02:00
|
|
|
|
# URLs starts with http://newflv.sohu.ccgslb.net/ is not usable
|
|
|
|
|
# so retry until got a working URL
|
|
|
|
|
video_url = 'newflv.sohu.ccgslb.net'
|
|
|
|
|
retries = 0
|
|
|
|
|
while 'newflv.sohu.ccgslb.net' in video_url and retries < 5:
|
|
|
|
|
download_note = 'Download information from CDN gateway for format ' + format_id
|
|
|
|
|
if retries > 0:
|
|
|
|
|
download_note += ' (retry #%d)' % retries
|
|
|
|
|
retries += 1
|
|
|
|
|
cdn_info = self._download_json(
|
|
|
|
|
'http://data.vod.itc.cn/cdnList?new=' + data['su'][i],
|
|
|
|
|
video_id, download_note)
|
|
|
|
|
video_url = cdn_info['url']
|
2014-12-25 17:25:05 +01:00
|
|
|
|
|
|
|
|
|
formats.append({
|
|
|
|
|
'url': video_url,
|
|
|
|
|
'format_id': format_id,
|
|
|
|
|
'filesize': data['clipsBytes'][i],
|
|
|
|
|
'width': data['width'],
|
|
|
|
|
'height': data['height'],
|
|
|
|
|
'fps': data['fps'],
|
|
|
|
|
})
|
|
|
|
|
self._sort_formats(formats)
|
|
|
|
|
|
|
|
|
|
playlist.append({
|
|
|
|
|
'id': '%s_part%d' % (video_id, i + 1),
|
2013-08-02 11:58:46 +02:00
|
|
|
|
'title': title,
|
2014-12-25 17:25:05 +01:00
|
|
|
|
'duration': vid_data['data']['clipsDuration'][i],
|
|
|
|
|
'formats': formats,
|
|
|
|
|
})
|
2013-08-28 13:59:08 +02:00
|
|
|
|
|
|
|
|
|
if len(playlist) == 1:
|
|
|
|
|
info = playlist[0]
|
2013-08-03 04:29:58 +02:00
|
|
|
|
info['id'] = video_id
|
2013-08-28 13:59:08 +02:00
|
|
|
|
else:
|
|
|
|
|
info = {
|
2015-04-19 13:19:44 +02:00
|
|
|
|
'_type': 'multi_video',
|
2013-08-28 13:59:08 +02:00
|
|
|
|
'entries': playlist,
|
|
|
|
|
'id': video_id,
|
2015-04-19 13:19:44 +02:00
|
|
|
|
'title': title,
|
2013-08-28 13:59:08 +02:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return info
|