2 from __future__ import unicode_literals
6 from .common import InfoExtractor
14 from ..compat import compat_urllib_parse_urlencode
17 class VLiveIE(InfoExtractor):
19 _VALID_URL = r'https?://(?:(?:www|m)\.)?vlive\.tv/video/(?P<id>[0-9]+)'
21 'url': 'http://www.vlive.tv/video/1326',
22 'md5': 'cc7314812855ce56de70a06a27314983',
26 'title': "[V LIVE] Girl's Day's Broadcast",
27 'creator': "Girl's Day",
31 'url': 'http://www.vlive.tv/video/16937',
35 'title': '[V LIVE] 첸백시 걍방',
38 'subtitles': 'mincount:12',
41 'skip_download': True,
45 def _real_extract(self, url):
46 video_id = self._match_id(url)
48 webpage = self._download_webpage(
49 'http://www.vlive.tv/video/%s' % video_id, video_id)
51 video_params = self._parse_json(self._search_regex(
52 r'\bvlive\.video\.init\(([^)]+)\);',
53 webpage, 'video params'), video_id,
54 transform_source=lambda s: '[' + s + ']')
56 status, long_video_id, key = video_params[2], video_params[5], video_params[6]
57 status = remove_start(status, 'PRODUCT_')
59 if status == 'LIVE_ON_AIR' or status == 'BIG_EVENT_ON_AIR':
60 return self._live(video_id, webpage)
61 elif status == 'VOD_ON_AIR' or status == 'BIG_EVENT_INTRO':
62 if long_video_id and key:
63 return self._replay(video_id, webpage, long_video_id, key)
65 status = 'COMING_SOON'
67 if status == 'LIVE_END':
68 raise ExtractorError('Uploading for replay. Please wait...',
70 elif status == 'COMING_SOON':
71 raise ExtractorError('Coming soon!', expected=True)
72 elif status == 'CANCELED':
73 raise ExtractorError('We are sorry, '
74 'but the live broadcast has been canceled.',
77 raise ExtractorError('Unknown status %s' % status)
79 def _get_common_fields(self, webpage):
80 title = self._og_search_title(webpage)
81 creator = self._html_search_regex(
82 r'<div[^>]+class="info_area"[^>]*>\s*<a\s+[^>]*>([^<]+)',
83 webpage, 'creator', fatal=False)
84 thumbnail = self._og_search_thumbnail(webpage)
88 'thumbnail': thumbnail,
91 def _live(self, video_id, webpage):
92 init_page = self._download_webpage(
93 'http://www.vlive.tv/video/init/view',
94 video_id, data="videoSeq=%s" % video_id, headers={
95 'Referer': 'http://www.vlive.tv/video/%s' % video_id,
96 'Content-Type': 'application/x-www-form-urlencoded'
99 live_params = self._search_regex(
100 r'"liveStreamInfo"\s*:\s*(".*"),',
101 init_page, 'video params')
102 live_params = self._parse_json(live_params, video_id)
103 live_params = self._parse_json(live_params, video_id)
106 for vid in live_params.get('resolutions', []):
107 formats.extend(self._extract_m3u8_formats(
108 vid['cdnUrl'], video_id, 'mp4',
109 m3u8_id=vid.get('name'),
110 fatal=False, live=True))
111 self._sort_formats(formats)
113 return dict(self._get_common_fields(webpage),
118 def _replay(self, video_id, webpage, long_video_id, key):
119 playinfo = self._download_json(
120 'http://global.apis.naver.com/rmcnmv/rmcnmv/vod_play_videoInfo.json?%s'
121 % compat_urllib_parse_urlencode({
122 'videoId': long_video_id,
125 'doct': 'json', # document type (xml or json)
126 'cpt': 'vtt', # captions type (vtt or ttml)
130 'url': vid['source'],
131 'format_id': vid.get('encodingOption', {}).get('name'),
132 'abr': float_or_none(vid.get('bitrate', {}).get('audio')),
133 'vbr': float_or_none(vid.get('bitrate', {}).get('video')),
134 'width': int_or_none(vid.get('encodingOption', {}).get('width')),
135 'height': int_or_none(vid.get('encodingOption', {}).get('height')),
136 'filesize': int_or_none(vid.get('size')),
137 } for vid in playinfo.get('videos', {}).get('list', []) if vid.get('source')]
138 self._sort_formats(formats)
140 view_count = int_or_none(playinfo.get('meta', {}).get('count'))
143 for caption in playinfo.get('captions', {}).get('list', []):
144 lang = dict_get(caption, ('locale', 'language', 'country', 'label'))
145 if lang and caption.get('source'):
148 'url': caption['source']}]
150 return dict(self._get_common_fields(webpage),
153 view_count=view_count,