yt_dlp/extractor/atresplayer.py

   1 from .common import InfoExtractor
   2 from ..networking.exceptions import HTTPError
   3 from ..utils import (
   4     ExtractorError,
   5     int_or_none,
   6     urlencode_postdata,
   7 )
   8
   9
  10 class AtresPlayerIE(InfoExtractor):
  11     _VALID_URL = r'https?://(?:www\.)?atresplayer\.com/[^/]+/[^/]+/[^/]+/[^/]+/(?P<display_id>.+?)_(?P<id>[0-9a-f]{24})'
  12     _NETRC_MACHINE = 'atresplayer'
  13     _TESTS = [
  14         {
  15             'url': 'https://www.atresplayer.com/antena3/series/pequenas-coincidencias/temporada-1/capitulo-7-asuntos-pendientes_5d4aa2c57ed1a88fc715a615/',
  16             'info_dict': {
  17                 'id': '5d4aa2c57ed1a88fc715a615',
  18                 'ext': 'mp4',
  19                 'title': 'Capítulo 7: Asuntos pendientes',
  20                 'description': 'md5:7634cdcb4d50d5381bedf93efb537fbc',
  21                 'duration': 3413,
  22             },
  23             'skip': 'This video is only available for registered users'
  24         },
  25         {
  26             'url': 'https://www.atresplayer.com/lasexta/programas/el-club-de-la-comedia/temporada-4/capitulo-10-especial-solidario-nochebuena_5ad08edf986b2855ed47adc4/',
  27             'only_matching': True,
  28         },
  29         {
  30             'url': 'https://www.atresplayer.com/antena3/series/el-secreto-de-puente-viejo/el-chico-de-los-tres-lunares/capitulo-977-29-12-14_5ad51046986b2886722ccdea/',
  31             'only_matching': True,
  32         },
  33     ]
  34     _API_BASE = 'https://api.atresplayer.com/'
  35
  36     def _handle_error(self, e, code):
  37         if isinstance(e.cause, HTTPError) and e.cause.status == code:
  38             error = self._parse_json(e.cause.response.read(), None)
  39             if error.get('error') == 'required_registered':
  40                 self.raise_login_required()
  41             raise ExtractorError(error['error_description'], expected=True)
  42         raise
  43
  44     def _perform_login(self, username, password):
  45         self._request_webpage(
  46             self._API_BASE + 'login', None, 'Downloading login page')
  47
  48         try:
  49             target_url = self._download_json(
  50                 'https://account.atresmedia.com/api/login', None,
  51                 'Logging in', headers={
  52                     'Content-Type': 'application/x-www-form-urlencoded'
  53                 }, data=urlencode_postdata({
  54                     'username': username,
  55                     'password': password,
  56                 }))['targetUrl']
  57         except ExtractorError as e:
  58             self._handle_error(e, 400)
  59
  60         self._request_webpage(target_url, None, 'Following Target URL')
  61
  62     def _real_extract(self, url):
  63         display_id, video_id = self._match_valid_url(url).groups()
  64
  65         try:
  66             episode = self._download_json(
  67                 self._API_BASE + 'client/v1/player/episode/' + video_id, video_id)
  68         except ExtractorError as e:
  69             self._handle_error(e, 403)
  70
  71         title = episode['titulo']
  72
  73         formats = []
  74         subtitles = {}
  75         for source in episode.get('sources', []):
  76             src = source.get('src')
  77             if not src:
  78                 continue
  79             src_type = source.get('type')
  80             if src_type == 'application/vnd.apple.mpegurl':
  81                 formats, subtitles = self._extract_m3u8_formats(
  82                     src, video_id, 'mp4', 'm3u8_native',
  83                     m3u8_id='hls', fatal=False)
  84             elif src_type == 'application/dash+xml':
  85                 formats, subtitles = self._extract_mpd_formats(
  86                     src, video_id, mpd_id='dash', fatal=False)
  87
  88         heartbeat = episode.get('heartbeat') or {}
  89         omniture = episode.get('omniture') or {}
  90         get_meta = lambda x: heartbeat.get(x) or omniture.get(x)
  91
  92         return {
  93             'display_id': display_id,
  94             'id': video_id,
  95             'title': title,
  96             'description': episode.get('descripcion'),
  97             'thumbnail': episode.get('imgPoster'),
  98             'duration': int_or_none(episode.get('duration')),
  99             'formats': formats,
 100             'channel': get_meta('channel'),
 101             'season': get_meta('season'),
 102             'episode_number': int_or_none(get_meta('episodeNumber')),
 103             'subtitles': subtitles,
 104         }