2 from __future__ import unicode_literals
6 from .common import InfoExtractor
17 class YoukuIE(InfoExtractor):
22 http://(?:v|player)\.youku\.com/(?:v_show/id_|player\.php/sid/)|
24 (?P<id>[A-Za-z0-9]+)(?:\.html|/v\.swf|)
29 'url': 'http://v.youku.com/v_show/id_XMTc1ODE5Njcy.html',
31 'id': 'XMTc1ODE5Njcy_part1',
32 'title': '★Smile﹗♡ Git Fresh -Booty Music舞蹈.',
36 'url': 'http://player.youku.com/player.php/sid/XNDgyMDQ2NTQw/v.swf',
37 'only_matching': True,
39 'url': 'http://v.youku.com/v_show/id_XODgxNjg1Mzk2_ev_1.html',
41 'id': 'XODgxNjg1Mzk2',
45 'skip': 'Available in China only',
47 'url': 'http://v.youku.com/v_show/id_XMTI1OTczNDM5Mg==.html',
49 'id': 'XMTI1OTczNDM5Mg',
54 'url': 'http://v.youku.com/v_show/id_XNjA1NzA2Njgw.html',
55 'note': 'Video protected with password',
57 'id': 'XNjA1NzA2Njgw',
58 'title': '邢義田复旦讲座之想象中的胡人—从“左衽孔子”说起',
62 'videopassword': '100600',
66 def construct_video_urls(self, data):
72 t = (t + ls[i] + compat_ord(s1[i % len(s1)])) % 256
73 ls[i], ls[t] = ls[t], ls[i]
76 for i in range(len(s2)):
79 ls[x], ls[y] = ls[y], ls[x]
80 s.append(compat_ord(s2[i]) ^ ls[(ls[x] + ls[y]) % 256])
84 b'becaf9be', base64.b64decode(data['security']['encrypt_string'].encode('ascii'))
85 ).decode('ascii').split('_')
88 oip = data['security']['ip']
91 for stream in data['stream']:
92 format = stream.get('stream_type')
93 fileid = stream['stream_fileid']
94 fileid_dict[format] = fileid
96 def get_fileid(format, n):
97 number = hex(int(str(n), 10))[2:].upper()
100 streamfileids = fileid_dict[format]
101 fileid = streamfileids[0:8] + number + streamfileids[10:]
105 def generate_ep(format, n):
106 fileid = get_fileid(format, n)
109 ('%s_%s_%s' % (sid, fileid, token)).encode('ascii')
111 ep = base64.b64encode(ep_t).decode('ascii')
114 # generate video_urls
116 for stream in data['stream']:
117 format = stream.get('stream_type')
119 for dt in stream['segs']:
120 n = str(stream['segs'].index(dt))
123 'hd': self.get_hd(format),
130 'ep': generate_ep(format, n)
133 'http://k.youku.com/player/getFlvPath/' + \
136 '/st/' + self.parse_ext_l(format) + \
137 '/fileid/' + get_fileid(format, n) + '?' + \
138 compat_urllib_parse.urlencode(param)
139 video_urls.append(video_url)
140 video_urls_dict[format] = video_urls
142 return video_urls_dict
144 def get_hd(self, fm):
157 return hd_id_dict[fm]
159 def parse_ext_l(self, fm):
174 def get_format_name(self, fm):
189 def _real_extract(self, url):
190 video_id = self._match_id(url)
192 def retrieve_data(req_url, note):
196 self._set_cookie('youku.com', 'xreferrer', 'http://www.youku.com')
197 req = sanitized_Request(req_url, headers=headers)
199 cn_verification_proxy = self._downloader.params.get('cn_verification_proxy')
200 if cn_verification_proxy:
201 req.add_header('Ytdl-request-proxy', cn_verification_proxy)
203 raw_data = self._download_json(req, video_id, note=note)
205 return raw_data['data']
207 video_password = self._downloader.params.get('videopassword', None)
210 basic_data_url = "http://play.youku.com/play/get.json?vid=%s&ct=12" % video_id
212 basic_data_url += '&pwd=%s' % video_password
214 data = retrieve_data(basic_data_url, 'Downloading JSON metadata')
216 error = data.get('error')
218 error_note = error.get('note')
219 if error_note is not None and '因版权原因无法观看此视频' in error_note:
220 raise ExtractorError(
221 'Youku said: Sorry, this video is available in China only', expected=True)
223 msg = 'Youku server reported error %i' % error.get('code')
224 if error is not None:
225 msg += ': ' + error_note
226 raise ExtractorError(msg)
229 title = data['video']['title']
231 # generate video_urls_dict
232 video_urls_dict = self.construct_video_urls(data)
236 'id': '%s_part%d' % (video_id, i + 1),
239 # some formats are not available for all parts, we have to detect
241 } for i in range(max(len(v.get('segs')) for v in data['stream']))]
242 for stream in data['stream']:
243 fm = stream.get('stream_type')
244 video_urls = video_urls_dict[fm]
245 for video_url, seg, entry in zip(video_urls, stream['segs'], entries):
246 entry['formats'].append({
248 'format_id': self.get_format_name(fm),
249 'ext': self.parse_ext_l(fm),
250 'filesize': int(seg['size']),
254 '_type': 'multi_video',