youtube_dl/extractor/tvplay.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..compat import compat_str
   8 from ..utils import (
   9     parse_iso8601,
  10     qualities,
  11     determine_ext,
  12     update_url_query,
  13     int_or_none,
  14 )
  15
  16
  17 class TVPlayIE(InfoExtractor):
  18     IE_DESC = 'TV3Play and related services'
  19     _VALID_URL = r'''(?x)https?://(?:www\.)?
  20         (?:tvplay(?:\.skaties)?\.lv/parraides|
  21            (?:tv3play|play\.tv3)\.lt/programos|
  22            tv3play(?:\.tv3)?\.ee/sisu|
  23            tv(?:3|6|8|10)play\.se/program|
  24            (?:(?:tv3play|viasat4play|tv6play)\.no|tv3play\.dk)/programmer|
  25            play\.novatv\.bg/programi
  26         )/[^/]+/(?P<id>\d+)
  27         '''
  28     _TESTS = [
  29         {
  30             'url': 'http://www.tvplay.lv/parraides/vinas-melo-labak/418113?autostart=true',
  31             'md5': 'a1612fe0849455423ad8718fe049be21',
  32             'info_dict': {
  33                 'id': '418113',
  34                 'ext': 'mp4',
  35                 'title': 'Kādi ir īri? - Viņas melo labāk',
  36                 'description': 'Baiba apsmej īrus, kādi tie ir un ko viņi dara.',
  37                 'duration': 25,
  38                 'timestamp': 1406097056,
  39                 'upload_date': '20140723',
  40             },
  41         },
  42         {
  43             'url': 'http://play.tv3.lt/programos/moterys-meluoja-geriau/409229?autostart=true',
  44             'info_dict': {
  45                 'id': '409229',
  46                 'ext': 'flv',
  47                 'title': 'Moterys meluoja geriau',
  48                 'description': 'md5:9aec0fc68e2cbc992d2a140bd41fa89e',
  49                 'duration': 1330,
  50                 'timestamp': 1403769181,
  51                 'upload_date': '20140626',
  52             },
  53             'params': {
  54                 # rtmp download
  55                 'skip_download': True,
  56             },
  57         },
  58         {
  59             'url': 'http://www.tv3play.ee/sisu/kodu-keset-linna/238551?autostart=true',
  60             'info_dict': {
  61                 'id': '238551',
  62                 'ext': 'flv',
  63                 'title': 'Kodu keset linna 398537',
  64                 'description': 'md5:7df175e3c94db9e47c0d81ffa5d68701',
  65                 'duration': 1257,
  66                 'timestamp': 1292449761,
  67                 'upload_date': '20101215',
  68             },
  69             'params': {
  70                 # rtmp download
  71                 'skip_download': True,
  72             },
  73         },
  74         {
  75             'url': 'http://www.tv3play.se/program/husraddarna/395385?autostart=true',
  76             'info_dict': {
  77                 'id': '395385',
  78                 'ext': 'mp4',
  79                 'title': 'Husräddarna S02E07',
  80                 'description': 'md5:f210c6c89f42d4fc39faa551be813777',
  81                 'duration': 2574,
  82                 'timestamp': 1400596321,
  83                 'upload_date': '20140520',
  84             },
  85             'params': {
  86                 'skip_download': True,
  87             },
  88         },
  89         {
  90             'url': 'http://www.tv6play.se/program/den-sista-dokusapan/266636?autostart=true',
  91             'info_dict': {
  92                 'id': '266636',
  93                 'ext': 'mp4',
  94                 'title': 'Den sista dokusåpan S01E08',
  95                 'description': 'md5:295be39c872520221b933830f660b110',
  96                 'duration': 1492,
  97                 'timestamp': 1330522854,
  98                 'upload_date': '20120229',
  99                 'age_limit': 18,
 100             },
 101             'params': {
 102                 'skip_download': True,
 103             },
 104         },
 105         {
 106             'url': 'http://www.tv8play.se/program/antikjakten/282756?autostart=true',
 107             'info_dict': {
 108                 'id': '282756',
 109                 'ext': 'mp4',
 110                 'title': 'Antikjakten S01E10',
 111                 'description': 'md5:1b201169beabd97e20c5ad0ad67b13b8',
 112                 'duration': 2646,
 113                 'timestamp': 1348575868,
 114                 'upload_date': '20120925',
 115             },
 116             'params': {
 117                 'skip_download': True,
 118             },
 119         },
 120         {
 121             'url': 'http://www.tv3play.no/programmer/anna-anka-soker-assistent/230898?autostart=true',
 122             'info_dict': {
 123                 'id': '230898',
 124                 'ext': 'mp4',
 125                 'title': 'Anna Anka søker assistent - Ep. 8',
 126                 'description': 'md5:f80916bf5bbe1c5f760d127f8dd71474',
 127                 'duration': 2656,
 128                 'timestamp': 1277720005,
 129                 'upload_date': '20100628',
 130             },
 131             'params': {
 132                 'skip_download': True,
 133             },
 134         },
 135         {
 136             'url': 'http://www.viasat4play.no/programmer/budbringerne/21873?autostart=true',
 137             'info_dict': {
 138                 'id': '21873',
 139                 'ext': 'mp4',
 140                 'title': 'Budbringerne program 10',
 141                 'description': 'md5:4db78dc4ec8a85bb04fd322a3ee5092d',
 142                 'duration': 1297,
 143                 'timestamp': 1254205102,
 144                 'upload_date': '20090929',
 145             },
 146             'params': {
 147                 'skip_download': True,
 148             },
 149         },
 150         {
 151             'url': 'http://www.tv6play.no/programmer/hotelinspektor-alex-polizzi/361883?autostart=true',
 152             'info_dict': {
 153                 'id': '361883',
 154                 'ext': 'mp4',
 155                 'title': 'Hotelinspektør Alex Polizzi - Ep. 10',
 156                 'description': 'md5:3ecf808db9ec96c862c8ecb3a7fdaf81',
 157                 'duration': 2594,
 158                 'timestamp': 1393236292,
 159                 'upload_date': '20140224',
 160             },
 161             'params': {
 162                 'skip_download': True,
 163             },
 164         },
 165         {
 166             'url': 'http://play.novatv.bg/programi/zdravei-bulgariya/624952?autostart=true',
 167             'info_dict': {
 168                 'id': '624952',
 169                 'ext': 'flv',
 170                 'title': 'Здравей, България (12.06.2015 г.) ',
 171                 'description': 'md5:99f3700451ac5bb71a260268b8daefd7',
 172                 'duration': 8838,
 173                 'timestamp': 1434100372,
 174                 'upload_date': '20150612',
 175             },
 176             'params': {
 177                 # rtmp download
 178                 'skip_download': True,
 179             },
 180         },
 181         {
 182             'url': 'http://tvplay.skaties.lv/parraides/vinas-melo-labak/418113?autostart=true',
 183             'only_matching': True,
 184         },
 185         {
 186             'url': 'http://tv3play.tv3.ee/sisu/kodu-keset-linna/238551?autostart=true',
 187             'only_matching': True,
 188         }
 189     ]
 190
 191     def _real_extract(self, url):
 192         video_id = self._match_id(url)
 193
 194         video = self._download_json(
 195             'http://playapi.mtgx.tv/v1/videos/%s' % video_id, video_id, 'Downloading video JSON')
 196
 197         title = video['title']
 198
 199         if video.get('is_geo_blocked'):
 200             self.report_warning(
 201                 'This content might not be available in your country due to copyright reasons')
 202
 203         streams = self._download_json(
 204             'http://playapi.mtgx.tv/v1/videos/stream/%s' % video_id, video_id, 'Downloading streams JSON')
 205
 206         quality = qualities(['hls', 'medium', 'high'])
 207         formats = []
 208         for format_id, video_url in streams.get('streams', {}).items():
 209             if not video_url or not isinstance(video_url, compat_str):
 210                 continue
 211             ext = determine_ext(video_url)
 212             if ext == 'f4m':
 213                 formats.extend(self._extract_f4m_formats(
 214                     update_url_query(video_url, {
 215                         'hdcore': '3.5.0',
 216                         'plugin': 'aasp-3.5.0.151.81'
 217                     }), video_id, f4m_id='hds', fatal=False))
 218             elif ext == 'm3u8':
 219                 formats.extend(self._extract_m3u8_formats(
 220                     video_url, video_id, 'mp4', 'm3u8_native',
 221                     m3u8_id='hls', fatal=False))
 222             else:
 223                 fmt = {
 224                     'format_id': format_id,
 225                     'quality': quality(format_id),
 226                     'ext': ext,
 227                 }
 228                 if video_url.startswith('rtmp'):
 229                     m = re.search(r'^(?P<url>rtmp://[^/]+/(?P<app>[^/]+))/(?P<playpath>.+)$', video_url)
 230                     if not m:
 231                         continue
 232                     fmt.update({
 233                         'ext': 'flv',
 234                         'url': m.group('url'),
 235                         'app': m.group('app'),
 236                         'play_path': m.group('playpath'),
 237                     })
 238                 else:
 239                     fmt.update({
 240                         'url': video_url,
 241                     })
 242                 formats.append(fmt)
 243         self._sort_formats(formats)
 244
 245         return {
 246             'id': video_id,
 247             'title': title,
 248             'description': video.get('description'),
 249             'duration': int_or_none(video.get('duration')),
 250             'timestamp': parse_iso8601(video.get('created_at')),
 251             'view_count': int_or_none(video.get('views', {}).get('total')),
 252             'age_limit': int_or_none(video.get('age_limit', 0)),
 253             'formats': formats,
 254         }