2 from __future__ import unicode_literals
8 from .common import InfoExtractor
9 from ..compat import (compat_urlparse, compat_urllib_parse)
10 from ..utils import (ExtractorError, parse_iso8601)
13 class LetvIE(InfoExtractor):
14 _VALID_URL = r'http://www.letv.com/ptv/vplay/(?P<id>\d+).html'
17 'url': 'http://www.letv.com/ptv/vplay/22005890.html',
18 'md5': 'cab23bd68d5a8db9be31c9a222c1e8df',
22 'title': '第87届奥斯卡颁奖礼完美落幕 《鸟人》成最大赢家',
23 'timestamp': 1424747397,
24 'upload_date': '20150224',
27 'url': 'http://www.letv.com/ptv/vplay/1118082.html',
36 return val >> n if val >= 0 else (val + 0x100000000) >> n
38 # ror() and calcTimeKey() are reversed from a embedded swf file in KLetvPlayer.swf
39 def ror(self, param1, param2):
41 while _loc3_ < param2:
42 param1 = self.urshift(param1, 1) + ((param1 & 1) << 31)
46 def calcTimeKey(self, param1):
48 _loc3_ = self.ror(param1, _loc2_ % 13)
49 _loc3_ = _loc3_ ^ _loc2_
50 _loc3_ = self.ror(_loc3_, _loc2_ % 17)
53 def _real_extract(self, url):
54 media_id = self._match_id(url)
55 page = self._download_webpage(url, media_id)
61 'tkey': self.calcTimeKey(int(time.time())),
62 'domain': 'www.letv.com'
64 play_json = self._download_json(
65 'http://api.letv.com/mms/out/video/playJson?' + compat_urllib_parse.urlencode(params),
66 media_id, 'playJson data')
69 playstatus = play_json['playstatus']
70 if playstatus['status'] == 0:
71 flag = playstatus['flag']
73 msg = 'Country %s auth error' % playstatus['country']
75 msg = 'Generic error. flag = %d' % flag
76 raise ExtractorError(msg, expected=True)
78 playurl = play_json['playurl']
80 formats = ['350', '1000', '1300', '720p', '1080p']
81 dispatch = playurl['dispatch']
84 for format_id in formats:
85 if format_id in dispatch:
86 media_url = playurl['domain'][0] + dispatch[format_id][0]
88 # Mimic what flvxz.com do
89 url_parts = list(compat_urlparse.urlparse(media_url))
90 qs = dict(compat_urlparse.parse_qs(url_parts[4]))
97 url_parts[4] = compat_urllib_parse.urlencode(qs)
98 media_url = compat_urlparse.urlunparse(url_parts)
102 'ext': os.path.splitext(dispatch[format_id][1])[1][1:]
105 if format_id[-1:] == 'p':
106 url_info_dict['height'] = format_id[:-1]
108 urls.append(url_info_dict)
110 publish_time = parse_iso8601(self._html_search_regex(
111 r'发布时间 ([^<>]+) ', page, 'publish time', fatal=False),
112 delimiter=' ', timezone=datetime.timedelta(hours=8))
117 'title': playurl['title'],
118 'thumbnail': playurl['pic'],
119 'timestamp': publish_time,