yt_dlp/extractor/threeqsdn.py

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