2 from __future__ import unicode_literals
9 from .common import InfoExtractor
10 from ..compat import (
20 class YoukuIE(InfoExtractor):
25 http://(?:v|player)\.youku\.com/(?:v_show/id_|player\.php/sid/)|
27 (?P<id>[A-Za-z0-9]+)(?:\.html|/v\.swf|)
32 'url': 'http://v.youku.com/v_show/id_XMTc1ODE5Njcy.html',
34 'id': 'XMTc1ODE5Njcy_part1',
35 'title': '★Smile﹗♡ Git Fresh -Booty Music舞蹈.',
39 'url': 'http://player.youku.com/player.php/sid/XNDgyMDQ2NTQw/v.swf',
40 'only_matching': True,
42 'url': 'http://v.youku.com/v_show/id_XODgxNjg1Mzk2_ev_1.html',
44 'id': 'XODgxNjg1Mzk2',
48 'skip': 'Available in China only',
50 'url': 'http://v.youku.com/v_show/id_XMTI1OTczNDM5Mg==.html',
52 'id': 'XMTI1OTczNDM5Mg',
57 'url': 'http://v.youku.com/v_show/id_XNjA1NzA2Njgw.html',
58 'note': 'Video protected with password',
60 'id': 'XNjA1NzA2Njgw',
61 'title': '邢義田复旦讲座之想象中的胡人—从“左衽孔子”说起',
65 'videopassword': '100600',
69 def construct_video_urls(self, data):
75 t = (t + ls[i] + compat_ord(s1[i % len(s1)])) % 256
76 ls[i], ls[t] = ls[t], ls[i]
79 for i in range(len(s2)):
82 ls[x], ls[y] = ls[y], ls[x]
83 s.append(compat_ord(s2[i]) ^ ls[(ls[x] + ls[y]) % 256])
87 b'becaf9be', base64.b64decode(data['security']['encrypt_string'].encode('ascii'))
88 ).decode('ascii').split('_')
91 oip = data['security']['ip']
94 for stream in data['stream']:
95 format = stream.get('stream_type')
96 fileid = stream['stream_fileid']
97 fileid_dict[format] = fileid
99 def get_fileid(format, n):
100 number = hex(int(str(n), 10))[2:].upper()
102 number = '0' + number
103 streamfileids = fileid_dict[format]
104 fileid = streamfileids[0:8] + number + streamfileids[10:]
108 def generate_ep(format, n):
109 fileid = get_fileid(format, n)
112 ('%s_%s_%s' % (sid, fileid, token)).encode('ascii')
114 ep = base64.b64encode(ep_t).decode('ascii')
117 # generate video_urls
119 for stream in data['stream']:
120 format = stream.get('stream_type')
122 for dt in stream['segs']:
123 n = str(stream['segs'].index(dt))
126 'hd': self.get_hd(format),
133 'ep': generate_ep(format, n)
136 'http://k.youku.com/player/getFlvPath/' + \
139 '/st/' + self.parse_ext_l(format) + \
140 '/fileid/' + get_fileid(format, n) + '?' + \
141 compat_urllib_parse.urlencode(param)
142 video_urls.append(video_url)
143 video_urls_dict[format] = video_urls
145 return video_urls_dict
149 return '%d%s' % (int(time.time()), ''.join([
150 random.choice(string.ascii_letters) for i in range(3)]))
152 def get_hd(self, fm):
165 return hd_id_dict[fm]
167 def parse_ext_l(self, fm):
182 def get_format_name(self, fm):
197 def _real_extract(self, url):
198 video_id = self._match_id(url)
200 self._set_cookie('youku.com', '__ysuid', self.get_ysuid())
202 def retrieve_data(req_url, note):
206 self._set_cookie('youku.com', 'xreferrer', 'http://www.youku.com')
207 req = sanitized_Request(req_url, headers=headers)
209 cn_verification_proxy = self._downloader.params.get('cn_verification_proxy')
210 if cn_verification_proxy:
211 req.add_header('Ytdl-request-proxy', cn_verification_proxy)
213 raw_data = self._download_json(req, video_id, note=note)
215 return raw_data['data']
217 video_password = self._downloader.params.get('videopassword', None)
220 basic_data_url = "http://play.youku.com/play/get.json?vid=%s&ct=12" % video_id
222 basic_data_url += '&pwd=%s' % video_password
224 data = retrieve_data(basic_data_url, 'Downloading JSON metadata')
226 error = data.get('error')
228 error_note = error.get('note')
229 if error_note is not None and '因版权原因无法观看此视频' in error_note:
230 raise ExtractorError(
231 'Youku said: Sorry, this video is available in China only', expected=True)
233 msg = 'Youku server reported error %i' % error.get('code')
234 if error_note is not None:
235 msg += ': ' + error_note
236 raise ExtractorError(msg)
239 title = data['video']['title']
241 # generate video_urls_dict
242 video_urls_dict = self.construct_video_urls(data)
246 'id': '%s_part%d' % (video_id, i + 1),
249 # some formats are not available for all parts, we have to detect
251 } for i in range(max(len(v.get('segs')) for v in data['stream']))]
252 for stream in data['stream']:
253 fm = stream.get('stream_type')
254 video_urls = video_urls_dict[fm]
255 for video_url, seg, entry in zip(video_urls, stream['segs'], entries):
256 entry['formats'].append({
258 'format_id': self.get_format_name(fm),
259 'ext': self.parse_ext_l(fm),
260 'filesize': int(seg['size']),
264 '_type': 'multi_video',