yt_dlp/extractor/threeqsdn.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..compat import compat_HTTPError
   7 from ..utils import (
   8     determine_ext,
   9     ExtractorError,
  10     float_or_none,
  11     int_or_none,
  12     join_nonempty,
  13     parse_iso8601,
  14 )
  15
  16
  17 class ThreeQSDNIE(InfoExtractor):
  18     IE_NAME = '3qsdn'
  19     IE_DESC = '3Q SDN'
  20     _VALID_URL = r'https?://playout\.3qsdn\.com/(?P<id>[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})'
  21     _TESTS = [{
  22         # https://player.3qsdn.com/demo.html
  23         'url': 'https://playout.3qsdn.com/7201c779-6b3c-11e7-a40e-002590c750be',
  24         'md5': '64a57396b16fa011b15e0ea60edce918',
  25         'info_dict': {
  26             'id': '7201c779-6b3c-11e7-a40e-002590c750be',
  27             'ext': 'mp4',
  28             'title': 'Video Ads',
  29             'is_live': False,
  30             'description': 'Video Ads Demo',
  31             'timestamp': 1500334803,
  32             'upload_date': '20170717',
  33             'duration': 888.032,
  34             'subtitles': {
  35                 'eng': 'count:1',
  36             },
  37         },
  38         'expected_warnings': ['Unknown MIME type application/mp4 in DASH manifest'],
  39     }, {
  40         # live video stream
  41         'url': 'https://playout.3qsdn.com/66e68995-11ca-11e8-9273-002590c750be',
  42         'info_dict': {
  43             'id': '66e68995-11ca-11e8-9273-002590c750be',
  44             'ext': 'mp4',
  45             'title': 're:^66e68995-11ca-11e8-9273-002590c750be [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
  46             'is_live': True,
  47         },
  48         'params': {
  49             'skip_download': True,  # m3u8 downloads
  50         },
  51     }, {
  52         # live audio stream
  53         'url': 'http://playout.3qsdn.com/9edf36e0-6bf2-11e2-a16a-9acf09e2db48',
  54         'only_matching': True,
  55     }, {
  56         # live audio stream with some 404 URLs
  57         'url': 'http://playout.3qsdn.com/ac5c3186-777a-11e2-9c30-9acf09e2db48',
  58         'only_matching': True,
  59     }, {
  60         # geo restricted with 'This content is not available in your country'
  61         'url': 'http://playout.3qsdn.com/d63a3ffe-75e8-11e2-9c30-9acf09e2db48',
  62         'only_matching': True,
  63     }, {
  64         # geo restricted with 'playout.3qsdn.com/forbidden'
  65         'url': 'http://playout.3qsdn.com/8e330f26-6ae2-11e2-a16a-9acf09e2db48',
  66         'only_matching': True,
  67     }, {
  68         # live video with rtmp link
  69         'url': 'https://playout.3qsdn.com/6092bb9e-8f72-11e4-a173-002590c750be',
  70         'only_matching': True,
  71     }, {
  72         # ondemand from http://www.philharmonie.tv/veranstaltung/26/
  73         'url': 'http://playout.3qsdn.com/0280d6b9-1215-11e6-b427-0cc47a188158?protocol=http',
  74         'only_matching': True,
  75     }, {
  76         # live video stream
  77         'url': 'https://playout.3qsdn.com/d755d94b-4ab9-11e3-9162-0025907ad44f?js=true',
  78         'only_matching': True,
  79     }]
  80
  81     @staticmethod
  82     def _extract_url(webpage):
  83         mobj = re.search(
  84             r'<iframe[^>]+\b(?:data-)?src=(["\'])(?P<url>%s.*?)\1' % ThreeQSDNIE._VALID_URL, webpage)
  85         if mobj:
  86             return mobj.group('url')
  87
  88     def _real_extract(self, url):
  89         video_id = self._match_id(url)
  90
  91         try:
  92             config = self._download_json(
  93                 url.replace('://playout.3qsdn.com/', '://playout.3qsdn.com/config/'), video_id)
  94         except ExtractorError as e:
  95             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 401:
  96                 self.raise_geo_restricted()
  97             raise
  98
  99         live = config.get('streamContent') == 'live'
 100         aspect = float_or_none(config.get('aspect'))
 101
 102         formats = []
 103         subtitles = {}
 104         for source_type, source in (config.get('sources') or {}).items():
 105             if not source:
 106                 continue
 107             if source_type == 'dash':
 108                 fmts, subs = self._extract_mpd_formats_and_subtitles(
 109                     source, video_id, mpd_id='mpd', fatal=False)
 110                 formats.extend(fmts)
 111                 subtitles = self._merge_subtitles(subtitles, subs)
 112             elif source_type == 'hls':
 113                 fmts, subs = self._extract_m3u8_formats_and_subtitles(
 114                     source, video_id, 'mp4', 'm3u8' if live else 'm3u8_native',
 115                     m3u8_id='hls', fatal=False)
 116                 formats.extend(fmts)
 117                 subtitles = self._merge_subtitles(subtitles, subs)
 118             elif source_type == 'progressive':
 119                 for s in source:
 120                     src = s.get('src')
 121                     if not (src and self._is_valid_url(src, video_id)):
 122                         continue
 123                     ext = determine_ext(src)
 124                     height = int_or_none(s.get('height'))
 125                     formats.append({
 126                         'ext': ext,
 127                         'format_id': join_nonempty('http', ext, height and '%dp' % height),
 128                         'height': height,
 129                         'source_preference': 0,
 130                         'url': src,
 131                         'vcodec': 'none' if height == 0 else None,
 132                         'width': int(height * aspect) if height and aspect else None,
 133                     })
 134         # It seems like this would be correctly handled by default
 135         # However, unless someone can confirm this, the old
 136         # behaviour is being kept as-is
 137         self._sort_formats(formats, ('res', 'source_preference'))
 138
 139         for subtitle in (config.get('subtitles') or []):
 140             src = subtitle.get('src')
 141             if not src:
 142                 continue
 143             subtitles.setdefault(subtitle.get('label') or 'eng', []).append({
 144                 'url': src,
 145             })
 146
 147         title = config.get('title') or video_id
 148
 149         return {
 150             'id': video_id,
 151             'title': self._live_title(title) if live else title,
 152             'thumbnail': config.get('poster') or None,
 153             'description': config.get('description') or None,
 154             'timestamp': parse_iso8601(config.get('upload_date')),
 155             'duration': float_or_none(config.get('vlength')) or None,
 156             'is_live': live,
 157             'formats': formats,
 158             'subtitles': subtitles,
 159         }