2 from __future__ import unicode_literals
7 from .common import InfoExtractor
8 from ..utils import ExtractorError
10 from ..compat import compat_urllib_parse
12 bytes_is_str = (bytes == str) # for compatible
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.',
34 def construct_video_urls(self, data1, data2):
40 t = (t + ls[i] + ord(s1[i % len(s1)])) % 256
41 ls[i], ls[t] = ls[t], ls[i]
42 s = '' if not bytes_is_str else b''
44 for i in range(len(s2)):
47 ls[x], ls[y] = ls[y], ls[x]
48 if isinstance(s2[i], int):
49 s += chr(s2[i] ^ ls[(ls[x] + ls[y]) % 256])
51 s += chr(ord(s2[i]) ^ ls[(ls[x] + ls[y]) % 256])
56 base64.b64decode(bytes(data2['ep'], 'ascii'))
58 else base64.b64decode(data2['ep'])
66 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ/\:._-1234567890')
67 shuffled_string_ls = []
71 seed = (seed * 0xd3 + 0x754f) % 0x10000
72 idx = seed * len(string_ls) // 0x10000
73 shuffled_string_ls.append(string_ls[idx])
77 for format in data1['streamtypes']:
79 int(i) for i in data1['streamfileids'][format].strip('*').split('*')]
81 [shuffled_string_ls[i] for i in streamfileid])
82 fileid_dict[format] = fileid[:8] + '%s' + fileid[10:]
84 def get_fileid(format, n):
85 fileid = fileid_dict[format] % hex(int(n))[2:].upper().zfill(2)
89 def generate_ep(format, n):
90 fileid = get_fileid(format, n)
93 bytes('%s_%s_%s' % (sid, fileid, token), 'ascii')
95 else ('%s_%s_%s' % (sid, fileid, token))
97 ep = base64.b64encode(
104 # generate video_urls
106 for format in data1['streamtypes']:
108 for dt in data1['segs'][format]:
109 n = str(int(dt['no']))
112 'hd': self.get_hd(format),
120 'ep': generate_ep(format, n)
123 'http://k.youku.com/player/getFlvPath/' + \
125 '_' + str(int(n) + 1).zfill(2) + \
126 '/st/' + self.parse_ext_l(format) + \
127 '/fileid/' + get_fileid(format, n) + '?' + \
128 compat_urllib_parse.urlencode(param)
129 video_urls.append(video_url)
130 video_urls_dict[format] = video_urls
132 return video_urls_dict
134 def get_hd(self, fm):
143 return hd_id_dict[fm]
145 def parse_ext_l(self, fm):
156 def get_format_name(self, fm):
167 def _real_extract(self, url):
168 mobj = re.match(self._VALID_URL, url)
169 video_id = mobj.group('id')
172 data1_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s' % video_id
173 data2_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s/Pf/4/ctype/12/ev/1' % video_id
175 raw_data1 = self._download_json(data1_url, video_id)
176 raw_data2 = self._download_json(data2_url, video_id)
177 data1 = raw_data1['data'][0]
178 data2 = raw_data2['data'][0]
180 error_code = data1.get('error_code')
182 # -8 means blocked outside China.
183 # Chinese and English, separated by newline.
184 error = data1.get('error')
185 raise ExtractorError(
186 error or 'Server reported error %i' %
190 title = data1['title']
192 # generate video_urls_dict
193 video_urls_dict = self.construct_video_urls(data1, data2)
197 for fm in data1['streamtypes']:
198 video_urls = video_urls_dict[fm]
199 for i in range(len(video_urls)):
200 if len(entries) < i + 1:
201 entries.append({'formats': []})
202 entries[i]['formats'].append(
204 'url': video_urls[i],
205 'format_id': self.get_format_name(fm),
206 'ext': self.parse_ext_l(fm),
207 'filesize': int(data1['segs'][fm][i]['size'])
211 for i in range(len(entries)):
214 'id': '_part%d' % (i + 1),
221 '_type': 'multi_video',
228 info['id'] = video_id