2 from __future__ import unicode_literals
9 from .common import InfoExtractor
20 class YoukuIE(InfoExtractor):
26 (?:v|player)\.youku\.com/(?:v_show/id_|player\.php/sid/)|
27 video\.tudou\.com/v/)|
29 (?P<id>[A-Za-z0-9]+)(?:\.html|/v\.swf|)
34 'url': 'http://v.youku.com/v_show/id_XMTc1ODE5Njcy.html',
36 'id': 'XMTc1ODE5Njcy',
37 'title': '★Smile﹗♡ Git Fresh -Booty Music舞蹈.',
40 'thumbnail': r're:^https?://.*',
42 'uploader_id': '36017967',
43 'uploader_url': 'http://i.youku.com/u/UMTQ0MDcxODY4',
47 'url': 'http://player.youku.com/player.php/sid/XNDgyMDQ2NTQw/v.swf',
48 'only_matching': True,
50 'url': 'http://v.youku.com/v_show/id_XODgxNjg1Mzk2_ev_1.html',
52 'id': 'XODgxNjg1Mzk2',
56 'thumbnail': r're:^https?://.*',
58 'uploader_id': '62583473',
59 'uploader_url': 'http://i.youku.com/u/UMjUwMzMzODky',
63 'url': 'http://v.youku.com/v_show/id_XMTI1OTczNDM5Mg==.html',
65 'id': 'XMTI1OTczNDM5Mg',
69 'thumbnail': r're:^https?://.*',
70 'uploader': '放剧场-花千骨',
71 'uploader_id': '772849359',
72 'uploader_url': 'http://i.youku.com/u/UMzA5MTM5NzQzNg==',
76 'url': 'http://v.youku.com/v_show/id_XNjA1NzA2Njgw.html',
77 'note': 'Video protected with password',
79 'id': 'XNjA1NzA2Njgw',
81 'title': '邢義田复旦讲座之想象中的胡人—从“左衽孔子”说起',
83 'thumbnail': r're:^https?://.*',
84 'uploader': 'FoxJin1006',
85 'uploader_id': '322014285',
86 'uploader_url': 'http://i.youku.com/u/UMTI4ODA1NzE0MA==',
90 'videopassword': '100600',
93 # /play/get.json contains streams with "channel_type":"tail"
94 'url': 'http://v.youku.com/v_show/id_XOTUxMzg4NDMy.html',
96 'id': 'XOTUxMzg4NDMy',
98 'title': '我的世界☆明月庄主☆车震猎杀☆杀人艺术Minecraft',
100 'thumbnail': r're:^https?://.*',
101 'uploader': '明月庄主moon',
102 'uploader_id': '38465621',
103 'uploader_url': 'http://i.youku.com/u/UMTUzODYyNDg0',
107 'url': 'http://video.tudou.com/v/XMjIyNzAzMTQ4NA==.html?f=46177805',
109 'id': 'XMjIyNzAzMTQ4NA',
111 'title': '卡马乔国足开大脚长传冲吊集锦',
113 'thumbnail': r're:^https?://.*',
114 'uploader': '阿卜杜拉之星',
115 'uploader_id': '2382249',
116 'uploader_url': 'http://i.youku.com/u/UOTUyODk5Ng==',
120 'url': 'http://video.tudou.com/v/XMjE4ODI3OTg2MA==.html',
121 'only_matching': True,
126 return '%d%s' % (int(time.time()), ''.join([
127 random.choice(string.ascii_letters) for i in range(3)]))
129 def get_format_name(self, fm):
144 def _real_extract(self, url):
145 video_id = self._match_id(url)
147 self._set_cookie('youku.com', '__ysuid', self.get_ysuid())
148 self._set_cookie('youku.com', 'xreferrer', 'http://www.youku.com')
150 _, urlh = self._download_webpage_handle(
151 'https://log.mmstat.com/eg.js', video_id, 'Retrieving cna info')
152 # The etag header is '"foobar"'; let's remove the double quotes
153 cna = urlh.headers['etag'][1:-1]
156 basic_data_params = {
158 'ccode': '0402' if 'tudou.com' in url else '0401',
159 'client_ip': '192.168.1.1',
161 'client_ts': time.time() / 1000,
164 video_password = self._downloader.params.get('videopassword')
166 basic_data_params['password'] = video_password
171 headers.update(self.geo_verification_headers())
172 data = self._download_json(
173 'https://ups.youku.com/ups/get.json', video_id,
174 'Downloading JSON metadata',
175 query=basic_data_params, headers=headers)['data']
177 error = data.get('error')
179 error_note = error.get('note')
180 if error_note is not None and '因版权原因无法观看此视频' in error_note:
181 raise ExtractorError(
182 'Youku said: Sorry, this video is available in China only', expected=True)
183 elif error_note and '该视频被设为私密' in error_note:
184 raise ExtractorError(
185 'Youku said: Sorry, this video is private', expected=True)
187 msg = 'Youku server reported error %i' % error.get('code')
188 if error_note is not None:
189 msg += ': ' + error_note
190 raise ExtractorError(msg)
193 video_data = data['video']
194 title = video_data['title']
197 'url': stream['m3u8_url'],
198 'format_id': self.get_format_name(stream.get('stream_type')),
200 'protocol': 'm3u8_native',
201 'filesize': int(stream.get('size')),
202 'width': stream.get('width'),
203 'height': stream.get('height'),
204 } for stream in data['stream'] if stream.get('channel_type') != 'tail']
205 self._sort_formats(formats)
211 'duration': video_data.get('seconds'),
212 'thumbnail': video_data.get('logo'),
213 'uploader': video_data.get('username'),
214 'uploader_id': str_or_none(video_data.get('userid')),
215 'uploader_url': data.get('uploader', {}).get('homepage'),
216 'tags': video_data.get('tags'),
220 class YoukuShowIE(InfoExtractor):
221 _VALID_URL = r'https?://list\.youku\.com/show/id_(?P<id>[0-9a-z]+)\.html'
222 IE_NAME = 'youku:show'
225 'url': 'http://list.youku.com/show/id_zc7c670be07ff11e48b3f.html',
227 'id': 'zc7c670be07ff11e48b3f',
229 'description': 'md5:a1ae6f5618571bbeb5c9821f9c81b558',
231 'playlist_count': 50,
233 # Episode number not starting from 1
234 'url': 'http://list.youku.com/show/id_zefbfbd70efbfbd780bef.html',
236 'id': 'zefbfbd70efbfbd780bef',
238 'description': 'md5:275715156abebe5ccc2a1992e9d56b98',
240 'playlist_count': 24,
243 def _extract_entries(self, playlist_data_url, show_id, idx, query, url):
244 query['callback'] = 'cb'
245 playlist_data = self._download_json(
246 playlist_data_url, show_id, query=query,
247 note='Downloading playlist data page %d' % (idx + 1),
248 transform_source=lambda s: js_to_json(strip_jsonp(s)))['html']
249 drama_list = (get_element_by_class('p-drama-grid', playlist_data) or
250 get_element_by_class('p-drama-half-row', playlist_data))
251 if drama_list is None:
252 raise ExtractorError('No episodes found')
253 video_urls = re.findall(r'<a[^>]+href="([^"]+)"', drama_list)
254 return playlist_data, [
255 self.url_result(urljoin(url, video_url), YoukuIE.ie_key())
256 for video_url in video_urls]
258 def _real_extract(self, url):
259 show_id = self._match_id(url)
260 webpage = self._download_webpage(url, show_id)
262 page_config = self._parse_json(self._search_regex(
263 r'var\s+PageConfig\s*=\s*({.+});', webpage, 'page config'),
264 show_id, transform_source=js_to_json)
265 first_page, entries = self._extract_entries(
266 'http://list.youku.com/show/module', show_id, 0, {
267 'id': page_config['showid'],
270 # The first reload_id has the same items as first_page
271 reload_ids = re.findall('<li[^>]+data-id="([^"]+)">', first_page)[1:]
272 for idx, reload_id in enumerate(reload_ids):
273 _, new_entries = self._extract_entries(
274 'http://list.youku.com/show/episode', show_id, idx + 1, {
275 'id': page_config['showid'],
278 entries.extend(new_entries)
280 desc = self._html_search_meta('description', webpage, fatal=False)
281 playlist_title = desc.split(',')[0] if desc else None
282 detail_li = get_element_by_class('p-intro', webpage)
283 playlist_description = get_element_by_class(
284 'intro-more', detail_li) if detail_li else None
286 return self.playlist_result(
287 entries, show_id, playlist_title, playlist_description)