2 from __future__ import unicode_literals
7 from .common import InfoExtractor
8 from ..utils import ExtractorError
10 class YoukuIE(InfoExtractor):
14 http://(?:v|player)\.youku\.com/(?:v_show/id_|player\.php/sid/)|
16 (?P<id>[A-Za-z0-9]+)(?:\.html|/v\.swf|)
20 'url': 'http://v.youku.com/v_show/id_XMTc1ODE5Njcy.html',
21 'md5': '5f3af4192eabacc4501508d54a8cabd7',
23 'id': 'XMTc1ODE5Njcy',
24 'title': '★Smile﹗♡ Git Fresh -Booty Music舞蹈.',
29 def construct_video_urls(self, data1, data2):
35 t = (t + ls[i] + ord(s1[i%len(s1)])) % 256
36 ls[i], ls[t] = ls[t], ls[i]
38 for i in range(len(s2)):
41 ls[x], ls[y] = ls[y], ls[x]
42 s += chr((s2[i] ^ ls[(ls[x]+ls[y]) % 256]))
46 'becaf9be', base64.b64decode(bytes(data2['ep'], 'ascii'))
54 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ/\:._-1234567890')
55 shuffled_string_ls = []
59 seed = (seed * 0xd3 + 0x754f) % 0x10000
60 idx = seed * len(string_ls) // 0x10000
61 shuffled_string_ls.append(string_ls[idx])
65 for format in data1['streamtypes']:
67 int(i) for i in data1['streamfileids'][format].strip('*').split('*')]
69 [shuffled_string_ls[i] for i in streamfileid])
70 fileid_dict[format] = fileid[:8] + '%s' + fileid[10:]
72 def get_fileid(format, n):
73 fileid = fileid_dict[format] % hex(int(n))[2:].upper().zfill(2)
77 def generate_ep(format, n):
78 fileid = get_fileid(format, n)
81 bytes('%s_%s_%s' % (sid, fileid, token), 'ascii'))
82 ep = base64.b64encode(bytes(ep_t, 'latin')).decode()
83 ep = ep.replace('+', '%2B')
84 ep = ep.replace('/', '%2F')
85 ep = ep.replace('=', '%2D')
90 for format in data1['streamtypes']:
92 for dt in data1['segs'][format]:
93 n = str(int(dt['no']))
95 'http://k.youku.com/player/getFlvPath/' + \
97 '_' + str(int(n)+1).zfill(2) + \
98 '/st/' + self.parse_ext_l(format) + \
99 '/fileid/' + get_fileid(format, n) + '?' + \
100 'K=' + str(dt['k']) + \
101 '&hd=' + self.get_hd(format) + \
103 '&ts=' + str(dt['seconds']) + \
104 '&ypp=0&ctype=12&ev=1' + \
105 '&token=' + str(token) + \
106 '&oip=' + str(oip) + \
107 '&ep=' + generate_ep(format, n)
108 video_urls.append(video_url)
109 video_urls_dict[format] = video_urls
111 return video_urls_dict
113 def get_hd(self, fm):
122 return hd_id_dict[fm]
124 def parse_ext_l(self, fm):
135 def _real_extract(self, url):
136 mobj = re.match(self._VALID_URL, url)
137 video_id = mobj.group('id')
140 data1_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s' % video_id
141 data2_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s/Pf/4/ctype/12/ev/1' % video_id
143 raw_data1 = self._download_json(data1_url, video_id)
144 raw_data2 = self._download_json(data2_url, video_id)
145 data1 = raw_data1['data'][0]
146 data2 = raw_data2['data'][0]
148 error_code = data1.get('error_code')
150 # -8 means blocked outside China.
151 # Chinese and English, separated by newline.
152 error = data1.get('error')
153 raise ExtractorError(
154 error or 'Server reported error %i' %
158 title = data1['title']
160 # generate video_urls_dict
161 video_urls_dict = self.construct_video_urls(data1, data2)
165 for fm in data1['streamtypes']:
167 video_urls = video_urls_dict[fm]
168 for i in range(len(video_urls)):
169 if len(entries) < i+1:
170 entries.append({'formats': []})
171 entries[i]['formats'].append(
173 'url': video_urls[i],
175 'ext': self.parse_ext_l(fm),
176 'filesize': int(data1['segs'][fm][i]['size'])
180 for i in range(len(entries)):
183 'id': '_part%d' % (i+1),
190 '_type': 'multi_video',
197 info['id'] = video_id