4 from .common
import InfoExtractor
7 compat_urllib_parse_urlencode
,
21 class SohuIE(InfoExtractor
):
22 _VALID_URL
= r
'https?://(?P<mytv>my\.)?tv\.sohu\.com/.+?/(?(mytv)|n)(?P<id>\d+)\.shtml.*?'
24 # Sohu videos give different MD5 sums on Travis CI and my machine
26 'note': 'This video is available only in Mainland China',
27 'url': 'http://tv.sohu.com/20130724/n382479172.shtml#super',
31 'title': 'MV:Far East Movement《The Illest》',
33 'skip': 'On available in China',
35 'url': 'http://tv.sohu.com/20150305/n409385080.shtml',
39 'title': '《2015湖南卫视羊年元宵晚会》唐嫣《花好月圆》',
41 'skip': 'no longer available',
43 'url': 'http://my.tv.sohu.com/us/232799889/78693464.shtml',
47 'title': '【爱范品】第31期:MWC见不到的奇葩手机',
50 'timestamp': 1425519600,
51 'upload_date': '20150305',
52 'thumbnail': 'http://e3f49eaa46b57.cdn.sohucs.com//group1/M10/83/FA/MTAuMTAuODguODA=/6_14cbccdde5eg104SysCutcloud_78693464_7_0b.jpg',
53 'tags': ['爱范儿', '爱范品', 'MWC', '手机'],
56 'note': 'Multipart video',
57 'url': 'http://my.tv.sohu.com/pl/8384802/78910339.shtml',
60 'title': '【神探苍实战秘籍】第13期 战争之影 赫卡里姆',
63 'timestamp': 1426269360,
64 'upload_date': '20150313',
65 'thumbnail': 'http://e3f49eaa46b57.cdn.sohucs.com//group1/M11/89/57/MTAuMTAuODguODA=/6_14cea022a1dg102SysCutcloud_78910339_8_0b.jpg',
66 'tags': ['小苍MM', '英雄联盟', '实战秘籍'],
70 'id': '78910339_part1',
73 'title': '【神探苍实战秘籍】第13期 战争之影 赫卡里姆',
77 'id': '78910339_part2',
80 'title': '【神探苍实战秘籍】第13期 战争之影 赫卡里姆',
84 'id': '78910339_part3',
87 'title': '【神探苍实战秘籍】第13期 战争之影 赫卡里姆',
91 'note': 'Video with title containing dash',
92 'url': 'http://my.tv.sohu.com/us/249884221/78932792.shtml',
96 'title': 'youtube-dl testing video',
98 'timestamp': 1426348620,
99 'upload_date': '20150314',
100 'thumbnail': 'http://e3f49eaa46b57.cdn.sohucs.com//group1/M02/8A/00/MTAuMTAuODguNzk=/6_14cee1be192g102SysCutcloud_78932792_7_7b.jpg',
104 'skip_download': True
108 def _real_extract(self
, url
):
110 def _fetch_data(vid_id
, mytv
=False):
112 base_data_url
= 'http://my.tv.sohu.com/play/videonew.do?vid='
114 base_data_url
= 'http://hot.vrs.sohu.com/vrs_flash.action?vid='
116 return self
._download
_json
(
117 base_data_url
+ vid_id
, video_id
,
118 'Downloading JSON data for %s' % vid_id
,
119 headers
=self
.geo_verification_headers())
121 mobj
= self
._match
_valid
_url
(url
)
122 video_id
= mobj
.group('id')
123 mytv
= mobj
.group('mytv') is not None
125 webpage
= self
._download
_webpage
(url
, video_id
)
127 title
= re
.sub(r
'( - 高清正版在线观看)? - 搜狐视频$', '', self
._og
_search
_title
(webpage
))
129 vid
= self
._html
_search
_regex
(
130 r
'var vid ?= ?["\'](\d
+)["\']',
131 webpage, 'video path')
132 vid_data = _fetch_data(vid, mytv)
133 if vid_data['play'] != 1:
134 if vid_data.get('status') == 12:
135 raise ExtractorError(
136 '%s said: There\'s something wrong in the video.' % self.IE_NAME,
139 self.raise_geo_restricted(
140 '%s said: The video is only licensed to users in Mainland China.' % self.IE_NAME)
143 for format_id in ('nor', 'high', 'super', 'ori', 'h2644k', 'h2654k'):
144 vid_id = vid_data['data'].get('%sVid' % format_id)
147 vid_id = compat_str(vid_id)
148 formats_json[format_id] = vid_data if vid == vid_id else _fetch_data(vid_id, mytv)
150 part_count = vid_data['data']['totalBlocks']
153 for i in range(part_count):
155 for format_id, format_data in formats_json.items():
156 allot = format_data['allot']
158 data = format_data['data']
159 clip_url = traverse_obj(data, (('clipsURL', 'mp4PlayUrl'), i, {url_or_none}), get_all=False)
161 raise ExtractorError(f'Unable to extract url for clip {i}')
164 video_url = 'newflv.sohu.ccgslb.net'
168 while 'newflv.sohu.ccgslb.net' in video_url:
177 if cdnId is not None:
178 params['idc'] = cdnId
180 download_note = 'Downloading %s video URL part %d of %d' % (
181 format_id, i + 1, part_count)
184 download_note += ' (retry #%d)' % retries
185 part_info = self._parse_json(self._download_webpage(
186 'http://%s/?%s' % (allot, compat_urllib_parse_urlencode(params)),
187 video_id, download_note), video_id)
189 video_url = part_info['url']
190 cdnId = part_info.get('nid')
194 raise ExtractorError('Failed to get video URL')
198 'format_id': format_id,
199 'filesize': int_or_none(
200 try_get(data, lambda x: x['clipsBytes'][i])),
201 'width': int_or_none(data.get('width')),
202 'height': int_or_none(data.get('height')),
203 'fps': int_or_none(data.get('fps')),
207 'id': '%s_part%d' % (video_id, i + 1),
209 'duration': vid_data['data']['clipsDuration'][i],
213 if len(playlist) == 1:
215 info['id'] = video_id
218 '_type': 'multi_video',
222 'duration': traverse_obj(vid_data, ('data', 'totalDuration', {float_or_none})),
226 publish_time = unified_timestamp(self._search_regex(
227 r'publishTime:\s*["\'](\d
+-\d
+-\d
+ \d
+:\d
+)["\']', webpage, 'publish time', fatal=False))
229 publish_time = traverse_obj(vid_data, ('tv_application_time', {unified_timestamp}))
232 'timestamp': publish_time - 8 * 3600 if publish_time else None,
233 **traverse_obj(vid_data, {
234 'alt_title': ('data', 'subName', {str}),
235 'uploader': ('wm_data', 'wm_username', {str}),
236 'thumbnail': ('data', 'coverImg', {url_or_none}),
237 'tags': ('data', 'tag', {str.split}),
243 class SohuVIE(InfoExtractor):
244 _VALID_URL = r'https?://tv\.sohu\.com/v/(?P<id>[\w=-]+)\.html(?:$|[#?])'
247 'note': 'Multipart video',
248 'url': 'https://tv.sohu.com/v/MjAyMzA2MTQvbjYwMTMxNTE5Mi5zaHRtbA==.html',
251 'title': '《淬火丹心》第1集',
252 'alt_title': '“点天灯”发生事故',
253 'duration': 2701.692,
254 'timestamp': 1686758040,
255 'upload_date': '20230614',
256 'thumbnail': 'http://photocdn.tv.sohu.com/img/20230614/vrsa_hor_1686738763256_454010551.jpg',
258 'playlist_mincount': 9,
259 'skip': 'Only available in China',
261 'url': 'https://tv.sohu.com/v/dXMvMjMyNzk5ODg5Lzc4NjkzNDY0LnNodG1s.html',
265 'title': '【爱范品】第31期:MWC见不到的奇葩手机',
268 'timestamp': 1425519600,
269 'upload_date': '20150305',
270 'thumbnail': 'http://e3f49eaa46b57.cdn.sohucs.com//group1/M10/83/FA/MTAuMTAuODguODA=/6_14cbccdde5eg104SysCutcloud_78693464_7_0b.jpg',
271 'tags': ['爱范儿', '爱范品', 'MWC', '手机'],
274 'note': 'Multipart video',
275 'url': 'https://tv.sohu.com/v/dXMvMjQyNTYyMTYzLzc4OTEwMzM5LnNodG1s.html?src=pl',
278 'title': '【神探苍实战秘籍】第13期 战争之影 赫卡里姆',
279 'uploader': '小苍cany',
281 'timestamp': 1426269360,
282 'upload_date': '20150313',
283 'thumbnail': 'http://e3f49eaa46b57.cdn.sohucs.com//group1/M11/89/57/MTAuMTAuODguODA=/6_14cea022a1dg102SysCutcloud_78910339_8_0b.jpg',
284 'tags': ['小苍MM', '英雄联盟', '实战秘籍'],
286 'playlist_mincount': 3,
289 def _real_extract(self, url):
290 encoded_id = self._match_id(url)
291 path = base64.urlsafe_b64decode(encoded_id).decode()
292 subdomain = 'tv' if re.match(r'\d+/n\d+\.shtml', path) else 'my.tv'
293 return self.url_result(urljoin(f'http://{subdomain}.sohu.com/', path), SohuIE)