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|)
28 'url': 'http://v.youku.com/v_show/id_XMTc1ODE5Njcy.html',
29 'md5': '5f3af4192eabacc4501508d54a8cabd7',
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',
46 'url': 'http://v.youku.com/v_show/id_XMTI1OTczNDM5Mg==.html',
48 'id': 'XMTI1OTczNDM5Mg',
52 'skip': 'Available in China only',
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, data1, data2):
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(data2['ep'].encode('ascii'))
85 ).decode('ascii').split('_')
92 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ/\:._-1234567890')
93 shuffled_string_ls = []
97 seed = (seed * 0xd3 + 0x754f) % 0x10000
98 idx = seed * len(string_ls) // 0x10000
99 shuffled_string_ls.append(string_ls[idx])
103 for format in data1['streamtypes']:
105 int(i) for i in data1['streamfileids'][format].strip('*').split('*')]
107 [shuffled_string_ls[i] for i in streamfileid])
108 fileid_dict[format] = fileid[:8] + '%s' + fileid[10:]
110 def get_fileid(format, n):
111 fileid = fileid_dict[format] % hex(int(n))[2:].upper().zfill(2)
115 def generate_ep(format, n):
116 fileid = get_fileid(format, n)
119 ('%s_%s_%s' % (sid, fileid, token)).encode('ascii')
121 ep = base64.b64encode(ep_t).decode('ascii')
124 # generate video_urls
126 for format in data1['streamtypes']:
128 for dt in data1['segs'][format]:
129 n = str(int(dt['no']))
132 'hd': self.get_hd(format),
140 'ep': generate_ep(format, n)
143 'http://k.youku.com/player/getFlvPath/' + \
145 '_' + str(int(n) + 1).zfill(2) + \
146 '/st/' + self.parse_ext_l(format) + \
147 '/fileid/' + get_fileid(format, n) + '?' + \
148 compat_urllib_parse.urlencode(param)
149 video_urls.append(video_url)
150 video_urls_dict[format] = video_urls
152 return video_urls_dict
154 def get_hd(self, fm):
163 return hd_id_dict[fm]
165 def parse_ext_l(self, fm):
176 def get_format_name(self, fm):
187 def _real_extract(self, url):
188 video_id = self._match_id(url)
190 def retrieve_data(req_url, note):
191 req = sanitized_Request(req_url)
193 cn_verification_proxy = self._downloader.params.get('cn_verification_proxy')
194 if cn_verification_proxy:
195 req.add_header('Ytdl-request-proxy', cn_verification_proxy)
197 raw_data = self._download_json(req, video_id, note=note)
198 return raw_data['data'][0]
200 video_password = self._downloader.params.get('videopassword', None)
203 basic_data_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s' % video_id
205 basic_data_url += '?password=%s' % video_password
207 data1 = retrieve_data(
209 'Downloading JSON metadata 1')
210 data2 = retrieve_data(
211 'http://v.youku.com/player/getPlayList/VideoIDS/%s/Pf/4/ctype/12/ev/1' % video_id,
212 'Downloading JSON metadata 2')
214 error_code = data1.get('error_code')
216 error = data1.get('error')
217 if error is not None and '因版权原因无法观看此视频' in error:
218 raise ExtractorError(
219 'Youku said: Sorry, this video is available in China only', expected=True)
221 msg = 'Youku server reported error %i' % error_code
222 if error is not None:
224 raise ExtractorError(msg)
226 title = data1['title']
228 # generate video_urls_dict
229 video_urls_dict = self.construct_video_urls(data1, data2)
233 'id': '%s_part%d' % (video_id, i + 1),
236 # some formats are not available for all parts, we have to detect
238 } for i in range(max(len(v) for v in data1['segs'].values()))]
239 for fm in data1['streamtypes']:
240 video_urls = video_urls_dict[fm]
241 for video_url, seg, entry in zip(video_urls, data1['segs'][fm], entries):
242 entry['formats'].append({
244 'format_id': self.get_format_name(fm),
245 'ext': self.parse_ext_l(fm),
246 'filesize': int(seg['size']),
250 '_type': 'multi_video',