X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/9dc487f48b50767cf540fa36c3de2c386fd74c04..1f17a37b9b95db09a420a1f52cf18723ce4eb8b5:/youtube_dl/extractor/mnet.py?ds=inline diff --git a/youtube_dl/extractor/mnet.py b/youtube_dl/extractor/mnet.py index e3f42e7..0e26ca1 100644 --- a/youtube_dl/extractor/mnet.py +++ b/youtube_dl/extractor/mnet.py @@ -22,7 +22,7 @@ class MnetIE(InfoExtractor): 'timestamp': 1451564040, 'age_limit': 0, 'thumbnails': 'mincount:5', - 'thumbnail': 're:^https?://.*\.jpg$', + 'thumbnail': r're:^https?://.*\.jpg$', 'ext': 'flv', }, 'params': { @@ -40,21 +40,29 @@ class MnetIE(InfoExtractor): def _real_extract(self, url): video_id = self._match_id(url) + # TODO: extract rtmp formats + # no stype -> rtmp url + # stype=H -> m3u8 url + # stype=M -> mpd url info = self._download_json( - 'http://content.api.mnet.com/player/vodConfig?id=%s&ctype=CLIP' % video_id, - video_id, 'Downloading vod config JSON')['data']['info'] + 'http://content.api.mnet.com/player/vodConfig', + video_id, 'Downloading vod config JSON', query={ + 'id': video_id, + 'ctype': 'CLIP', + 'stype': 'H', + })['data']['info'] title = info['title'] - rtmp_info = self._download_json( - info['cdn'], video_id, 'Downloading vod cdn JSON') - - formats = [{ - 'url': rtmp_info['serverurl'] + rtmp_info['fileurl'], - 'ext': 'flv', - 'page_url': url, - 'player_url': 'http://flvfile.mnet.com/service/player/201602/cjem_player_tv.swf?v=201602191318', - }] + cdn_data = self._download_json( + info['cdn'], video_id, 'Downloading vod cdn JSON')['data'][0] + m3u8_url = cdn_data['url'] + token = cdn_data.get('token') + if token and token != '-': + m3u8_url += '?' + token + formats = self._extract_wowza_formats( + m3u8_url, video_id, skip_protocols=['rtmp', 'rtsp', 'f4m']) + self._sort_formats(formats) description = info.get('ment') duration = parse_duration(info.get('time'))