youtube_dl/extractor/atresplayer.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_HTTPError
   8 from ..utils import (
   9     ExtractorError,
  10     int_or_none,
  11     urlencode_postdata,
  12 )
  13
  14
  15 class AtresPlayerIE(InfoExtractor):
  16     _VALID_URL = r'https?://(?:www\.)?atresplayer\.com/[^/]+/[^/]+/[^/]+/[^/]+/(?P<display_id>.+?)_(?P<id>[0-9a-f]{24})'
  17     _NETRC_MACHINE = 'atresplayer'
  18     _TESTS = [
  19         {
  20             'url': 'https://www.atresplayer.com/antena3/series/pequenas-coincidencias/temporada-1/capitulo-7-asuntos-pendientes_5d4aa2c57ed1a88fc715a615/',
  21             'info_dict': {
  22                 'id': '5d4aa2c57ed1a88fc715a615',
  23                 'ext': 'mp4',
  24                 'title': 'Capítulo 7: Asuntos pendientes',
  25                 'description': 'md5:7634cdcb4d50d5381bedf93efb537fbc',
  26                 'duration': 3413,
  27             },
  28             'params': {
  29                 'format': 'bestvideo',
  30             },
  31             'skip': 'This video is only available for registered users'
  32         },
  33         {
  34             'url': 'https://www.atresplayer.com/lasexta/programas/el-club-de-la-comedia/temporada-4/capitulo-10-especial-solidario-nochebuena_5ad08edf986b2855ed47adc4/',
  35             'only_matching': True,
  36         },
  37         {
  38             'url': 'https://www.atresplayer.com/antena3/series/el-secreto-de-puente-viejo/el-chico-de-los-tres-lunares/capitulo-977-29-12-14_5ad51046986b2886722ccdea/',
  39             'only_matching': True,
  40         },
  41     ]
  42     _API_BASE = 'https://api.atresplayer.com/'
  43
  44     def _real_initialize(self):
  45         self._login()
  46
  47     def _handle_error(self, e, code):
  48         if isinstance(e.cause, compat_HTTPError) and e.cause.code == code:
  49             error = self._parse_json(e.cause.read(), None)
  50             if error.get('error') == 'required_registered':
  51                 self.raise_login_required()
  52             raise ExtractorError(error['error_description'], expected=True)
  53         raise
  54
  55     def _login(self):
  56         username, password = self._get_login_info()
  57         if username is None:
  58             return
  59
  60         self._request_webpage(
  61             self._API_BASE + 'login', None, 'Downloading login page')
  62
  63         try:
  64             target_url = self._download_json(
  65                 'https://account.atresmedia.com/api/login', None,
  66                 'Logging in', headers={
  67                     'Content-Type': 'application/x-www-form-urlencoded'
  68                 }, data=urlencode_postdata({
  69                     'username': username,
  70                     'password': password,
  71                 }))['targetUrl']
  72         except ExtractorError as e:
  73             self._handle_error(e, 400)
  74
  75         self._request_webpage(target_url, None, 'Following Target URL')
  76
  77     def _real_extract(self, url):
  78         display_id, video_id = re.match(self._VALID_URL, url).groups()
  79
  80         try:
  81             episode = self._download_json(
  82                 self._API_BASE + 'client/v1/player/episode/' + video_id, video_id)
  83         except ExtractorError as e:
  84             self._handle_error(e, 403)
  85
  86         title = episode['titulo']
  87
  88         formats = []
  89         for source in episode.get('sources', []):
  90             src = source.get('src')
  91             if not src:
  92                 continue
  93             src_type = source.get('type')
  94             if src_type == 'application/vnd.apple.mpegurl':
  95                 formats.extend(self._extract_m3u8_formats(
  96                     src, video_id, 'mp4', 'm3u8_native',
  97                     m3u8_id='hls', fatal=False))
  98             elif src_type == 'application/dash+xml':
  99                 formats.extend(self._extract_mpd_formats(
 100                     src, video_id, mpd_id='dash', fatal=False))
 101         self._sort_formats(formats)
 102
 103         heartbeat = episode.get('heartbeat') or {}
 104         omniture = episode.get('omniture') or {}
 105         get_meta = lambda x: heartbeat.get(x) or omniture.get(x)
 106
 107         return {
 108             'display_id': display_id,
 109             'id': video_id,
 110             'title': title,
 111             'description': episode.get('descripcion'),
 112             'thumbnail': episode.get('imgPoster'),
 113             'duration': int_or_none(episode.get('duration')),
 114             'formats': formats,
 115             'channel': get_meta('channel'),
 116             'season': get_meta('season'),
 117             'episode_number': int_or_none(get_meta('episodeNumber')),
 118         }