2 from __future__ import unicode_literals
6 from .common import InfoExtractor
7 from ..utils import ExtractorError
15 class YoukuIE(InfoExtractor):
19 http://(?:v|player)\.youku\.com/(?:v_show/id_|player\.php/sid/)|
21 (?P<id>[A-Za-z0-9]+)(?:\.html|/v\.swf|)
25 'url': 'http://v.youku.com/v_show/id_XMTc1ODE5Njcy.html',
26 'md5': '5f3af4192eabacc4501508d54a8cabd7',
28 'id': 'XMTc1ODE5Njcy_part1',
29 'title': '★Smile﹗♡ Git Fresh -Booty Music舞蹈.',
33 'url': 'http://player.youku.com/player.php/sid/XNDgyMDQ2NTQw/v.swf',
34 'only_matching': True,
36 'url': 'http://v.youku.com/v_show/id_XODgxNjg1Mzk2_ev_1.html',
38 'id': 'XODgxNjg1Mzk2',
44 def construct_video_urls(self, data1, data2):
50 t = (t + ls[i] + compat_ord(s1[i % len(s1)])) % 256
51 ls[i], ls[t] = ls[t], ls[i]
54 for i in range(len(s2)):
57 ls[x], ls[y] = ls[y], ls[x]
58 s.append(compat_ord(s2[i]) ^ ls[(ls[x] + ls[y]) % 256])
62 b'becaf9be', base64.b64decode(data2['ep'].encode('ascii'))
63 ).decode('ascii').split('_')
70 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ/\:._-1234567890')
71 shuffled_string_ls = []
75 seed = (seed * 0xd3 + 0x754f) % 0x10000
76 idx = seed * len(string_ls) // 0x10000
77 shuffled_string_ls.append(string_ls[idx])
81 for format in data1['streamtypes']:
83 int(i) for i in data1['streamfileids'][format].strip('*').split('*')]
85 [shuffled_string_ls[i] for i in streamfileid])
86 fileid_dict[format] = fileid[:8] + '%s' + fileid[10:]
88 def get_fileid(format, n):
89 fileid = fileid_dict[format] % hex(int(n))[2:].upper().zfill(2)
93 def generate_ep(format, n):
94 fileid = get_fileid(format, n)
97 ('%s_%s_%s' % (sid, fileid, token)).encode('ascii')
99 ep = base64.b64encode(ep_t).decode('ascii')
102 # generate video_urls
104 for format in data1['streamtypes']:
106 for dt in data1['segs'][format]:
107 n = str(int(dt['no']))
110 'hd': self.get_hd(format),
118 'ep': generate_ep(format, n)
121 'http://k.youku.com/player/getFlvPath/' + \
123 '_' + str(int(n) + 1).zfill(2) + \
124 '/st/' + self.parse_ext_l(format) + \
125 '/fileid/' + get_fileid(format, n) + '?' + \
126 compat_urllib_parse.urlencode(param)
127 video_urls.append(video_url)
128 video_urls_dict[format] = video_urls
130 return video_urls_dict
132 def get_hd(self, fm):
141 return hd_id_dict[fm]
143 def parse_ext_l(self, fm):
154 def get_format_name(self, fm):
165 def _real_extract(self, url):
166 video_id = self._match_id(url)
169 data1_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s' % video_id
170 data2_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s/Pf/4/ctype/12/ev/1' % video_id
172 raw_data1 = self._download_json(data1_url, video_id)
173 raw_data2 = self._download_json(data2_url, video_id)
174 data1 = raw_data1['data'][0]
175 data2 = raw_data2['data'][0]
177 error_code = data1.get('error_code')
179 # -8 means blocked outside China.
180 # Chinese and English, separated by newline.
181 error = data1.get('error')
182 raise ExtractorError(
183 error or 'Server reported error %i' %
187 title = data1['title']
189 # generate video_urls_dict
190 video_urls_dict = self.construct_video_urls(data1, data2)
194 for fm in data1['streamtypes']:
195 video_urls = video_urls_dict[fm]
196 for i in range(len(video_urls)):
197 if len(entries) < i + 1:
198 entries.append({'formats': []})
199 entries[i]['formats'].append(
201 'url': video_urls[i],
202 'format_id': self.get_format_name(fm),
203 'ext': self.parse_ext_l(fm),
204 'filesize': int(data1['segs'][fm][i]['size'])
208 for i in range(len(entries)):
211 'id': '%s_part%d' % (video_id, i + 1),
217 '_type': 'multi_video',