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
14 class YoukuIE(InfoExtractor):
18 http://(?:v|player)\.youku\.com/(?:v_show/id_|player\.php/sid/)|
20 (?P<id>[A-Za-z0-9]+)(?:\.html|/v\.swf|)
24 'url': 'http://v.youku.com/v_show/id_XMTc1ODE5Njcy.html',
25 'md5': '5f3af4192eabacc4501508d54a8cabd7',
27 'id': 'XMTc1ODE5Njcy',
28 'title': '★Smile﹗♡ Git Fresh -Booty Music舞蹈.',
33 def construct_video_urls(self, data1, data2):
39 t = (t + ls[i] + ord(s1[i%len(s1)])) % 256
40 ls[i], ls[t] = ls[t], ls[i]
41 s = '' if not bytes_is_str else b''
43 for i in range(len(s2)):
46 ls[x], ls[y] = ls[y], ls[x]
47 if isinstance(s2[i], int):
48 s += chr(s2[i] ^ ls[(ls[x]+ls[y]) % 256])
50 s += chr(ord(s2[i]) ^ ls[(ls[x]+ls[y]) % 256])
55 base64.b64decode(bytes(data2['ep'], 'ascii')) \
57 else base64.b64decode(data2['ep'])
65 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ/\:._-1234567890')
66 shuffled_string_ls = []
70 seed = (seed * 0xd3 + 0x754f) % 0x10000
71 idx = seed * len(string_ls) // 0x10000
72 shuffled_string_ls.append(string_ls[idx])
76 for format in data1['streamtypes']:
78 int(i) for i in data1['streamfileids'][format].strip('*').split('*')]
80 [shuffled_string_ls[i] for i in streamfileid])
81 fileid_dict[format] = fileid[:8] + '%s' + fileid[10:]
83 def get_fileid(format, n):
84 fileid = fileid_dict[format] % hex(int(n))[2:].upper().zfill(2)
88 def generate_ep(format, n):
89 fileid = get_fileid(format, n)
92 bytes('%s_%s_%s' % (sid, fileid, token), 'ascii') \
94 else ('%s_%s_%s' % (sid, fileid, token))
96 ep = base64.b64encode(
97 bytes(ep_t, 'latin') \
103 # generate video_urls
105 for format in data1['streamtypes']:
107 for dt in data1['segs'][format]:
108 n = str(int(dt['no']))
111 'hd': self.get_hd(format),
119 'ep': generate_ep(format, n)
122 'http://k.youku.com/player/getFlvPath/' + \
124 '_' + str(int(n)+1).zfill(2) + \
125 '/st/' + self.parse_ext_l(format) + \
126 '/fileid/' + get_fileid(format, n) + '?' + \
127 compat_urllib_parse.urlencode(param)
128 video_urls.append(video_url)
129 video_urls_dict[format] = video_urls
131 return video_urls_dict
133 def get_hd(self, fm):
142 return hd_id_dict[fm]
144 def parse_ext_l(self, fm):
155 def _real_extract(self, url):
156 mobj = re.match(self._VALID_URL, url)
157 video_id = mobj.group('id')
160 data1_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s' % video_id
161 data2_url = 'http://v.youku.com/player/getPlayList/VideoIDS/%s/Pf/4/ctype/12/ev/1' % video_id
163 raw_data1 = self._download_json(data1_url, video_id)
164 raw_data2 = self._download_json(data2_url, video_id)
165 data1 = raw_data1['data'][0]
166 data2 = raw_data2['data'][0]
168 error_code = data1.get('error_code')
170 # -8 means blocked outside China.
171 # Chinese and English, separated by newline.
172 error = data1.get('error')
173 raise ExtractorError(
174 error or 'Server reported error %i' %
178 title = data1['title']
180 # generate video_urls_dict
181 video_urls_dict = self.construct_video_urls(data1, data2)
185 for fm in data1['streamtypes']:
187 video_urls = video_urls_dict[fm]
188 for i in range(len(video_urls)):
189 if len(entries) < i+1:
190 entries.append({'formats': []})
191 entries[i]['formats'].append(
193 'url': video_urls[i],
195 'ext': self.parse_ext_l(fm),
196 'filesize': int(data1['segs'][fm][i]['size'])
200 for i in range(len(entries)):
203 'id': '_part%d' % (i+1),
210 '_type': 'multi_video',
217 info['id'] = video_id