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',
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,
37 def construct_video_urls(self, data1, data2):
43 t = (t + ls[i] + compat_ord(s1[i % len(s1)])) % 256
44 ls[i], ls[t] = ls[t], ls[i]
47 for i in range(len(s2)):
50 ls[x], ls[y] = ls[y], ls[x]
51 s.append(compat_ord(s2[i]) ^ ls[(ls[x] + ls[y]) % 256])
55 b'becaf9be', base64.b64decode(data2['ep'].encode('ascii'))
56 ).decode('ascii').split('_')
63 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ/\:._-1234567890')
64 shuffled_string_ls = []
68 seed = (seed * 0xd3 + 0x754f) % 0x10000
69 idx = seed * len(string_ls) // 0x10000
70 shuffled_string_ls.append(string_ls[idx])
74 for format in data1['streamtypes']:
76 int(i) for i in data1['streamfileids'][format].strip('*').split('*')]
78 [shuffled_string_ls[i] for i in streamfileid])
79 fileid_dict[format] = fileid[:8] + '%s' + fileid[10:]
81 def get_fileid(format, n):
82 fileid = fileid_dict[format] % hex(int(n))[2:].upper().zfill(2)
86 def generate_ep(format, n):
87 fileid = get_fileid(format, n)
90 ('%s_%s_%s' % (sid, fileid, token)).encode('ascii')
92 ep = base64.b64encode(ep_t).decode('ascii')
97 for format in data1['streamtypes']:
99 for dt in data1['segs'][format]:
100 n = str(int(dt['no']))
103 'hd': self.get_hd(format),
111 'ep': generate_ep(format, n)
114 'http://k.youku.com/player/getFlvPath/' + \
116 '_' + str(int(n) + 1).zfill(2) + \
117 '/st/' + self.parse_ext_l(format) + \
118 '/fileid/' + get_fileid(format, n) + '?' + \
119 compat_urllib_parse.urlencode(param)
120 video_urls.append(video_url)
121 video_urls_dict[format] = video_urls
123 return video_urls_dict
125 def get_hd(self, fm):
134 return hd_id_dict[fm]
136 def parse_ext_l(self, fm):
147 def get_format_name(self, fm):
158 def _real_extract(self, url):
159 video_id = self._match_id(url)
162 data1_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s' % video_id
163 data2_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s/Pf/4/ctype/12/ev/1' % video_id
165 raw_data1 = self._download_json(data1_url, video_id)
166 raw_data2 = self._download_json(data2_url, video_id)
167 data1 = raw_data1['data'][0]
168 data2 = raw_data2['data'][0]
170 error_code = data1.get('error_code')
172 # -8 means blocked outside China.
173 # Chinese and English, separated by newline.
174 error = data1.get('error')
175 raise ExtractorError(
176 error or 'Server reported error %i' %
180 title = data1['title']
182 # generate video_urls_dict
183 video_urls_dict = self.construct_video_urls(data1, data2)
187 for fm in data1['streamtypes']:
188 video_urls = video_urls_dict[fm]
189 for i in range(len(video_urls)):
190 if len(entries) < i + 1:
191 entries.append({'formats': []})
192 entries[i]['formats'].append(
194 'url': video_urls[i],
195 'format_id': self.get_format_name(fm),
196 'ext': self.parse_ext_l(fm),
197 'filesize': int(data1['segs'][fm][i]['size'])
201 for i in range(len(entries)):
204 'id': '_part%d' % (i + 1),
211 '_type': 'multi_video',
218 info['id'] = video_id