youtube_dlc/extractor/mitele.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3 import json
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     int_or_none,
   8     parse_iso8601,
   9     smuggle_url,
  10 )
  11
  12
  13 class MiTeleIE(InfoExtractor):
  14     IE_DESC = 'mitele.es'
  15     _VALID_URL = r'https?://(?:www\.)?mitele\.es/(?:[^/]+/)+(?P<id>[^/]+)/player'
  16
  17     _TESTS = [{
  18         'url': 'http://www.mitele.es/programas-tv/diario-de/57b0dfb9c715da65618b4afa/player',
  19         'info_dict': {
  20             'id': 'FhYW1iNTE6J6H7NkQRIEzfne6t2quqPg',
  21             'ext': 'mp4',
  22             'title': 'Diario de La redacción Programa 144',
  23             'description': 'md5:07c35a7b11abb05876a6a79185b58d27',
  24             'series': 'Diario de',
  25             'season': 'Season 14',
  26             'season_number': 14,
  27             'episode': 'Tor, la web invisible',
  28             'episode_number': 3,
  29             'thumbnail': r're:(?i)^https?://.*\.jpg$',
  30             'duration': 2913,
  31             'age_limit': 16,
  32             'timestamp': 1471209401,
  33             'upload_date': '20160814',
  34         },
  35     }, {
  36         # no explicit title
  37         'url': 'http://www.mitele.es/programas-tv/cuarto-milenio/57b0de3dc915da14058b4876/player',
  38         'info_dict': {
  39             'id': 'oyNG1iNTE6TAPP-JmCjbwfwJqqMMX3Vq',
  40             'ext': 'mp4',
  41             'title': 'Cuarto Milenio Temporada 6 Programa 226',
  42             'description': 'md5:5ff132013f0cd968ffbf1f5f3538a65f',
  43             'series': 'Cuarto Milenio',
  44             'season': 'Season 6',
  45             'season_number': 6,
  46             'episode': 'Episode 24',
  47             'episode_number': 24,
  48             'thumbnail': r're:(?i)^https?://.*\.jpg$',
  49             'duration': 7313,
  50             'age_limit': 12,
  51             'timestamp': 1471209021,
  52             'upload_date': '20160814',
  53         },
  54         'params': {
  55             'skip_download': True,
  56         }
  57     }, {
  58         'url': 'http://www.mitele.es/series-online/la-que-se-avecina/57aac5c1c915da951a8b45ed/player',
  59         'only_matching': True,
  60     }, {
  61         'url': 'https://www.mitele.es/programas-tv/diario-de/la-redaccion/programa-144-40_1006364575251/player/',
  62         'only_matching': True,
  63     }]
  64
  65     def _real_extract(self, url):
  66         display_id = self._match_id(url)
  67         webpage = self._download_webpage(url, display_id)
  68         pre_player = self._parse_json(self._search_regex(
  69             r'window\.\$REACTBASE_STATE\.prePlayer_mtweb\s*=\s*({.+})',
  70             webpage, 'Pre Player'), display_id)['prePlayer']
  71         title = pre_player['title']
  72         video = pre_player['video']
  73         video_id = video['dataMediaId']
  74         content = pre_player.get('content') or {}
  75         info = content.get('info') or {}
  76
  77         info = {
  78             'id': video_id,
  79             'title': title,
  80             'description': info.get('synopsis'),
  81             'series': content.get('title'),
  82             'season_number': int_or_none(info.get('season_number')),
  83             'episode': content.get('subtitle'),
  84             'episode_number': int_or_none(info.get('episode_number')),
  85             'duration': int_or_none(info.get('duration')),
  86             'thumbnail': video.get('dataPoster'),
  87             'age_limit': int_or_none(info.get('rating')),
  88             'timestamp': parse_iso8601(pre_player.get('publishedTime')),
  89         }
  90
  91         if video.get('dataCmsId') == 'ooyala':
  92             info.update({
  93                 '_type': 'url_transparent',
  94                 # for some reason only HLS is supported
  95                 'url': smuggle_url('ooyala:' + video_id, {'supportedformats': 'm3u8,dash'}),
  96             })
  97         else:
  98             config = self._download_json(
  99                 video['dataConfig'], video_id, 'Downloading config JSON')
 100             services = config['services']
 101             gbx = self._download_json(
 102                 services['gbx'], video_id, 'Downloading gbx JSON')
 103             caronte = self._download_json(
 104                 services['caronte'], video_id, 'Downloading caronte JSON')
 105             cerbero = self._download_json(
 106                 caronte['cerbero'], video_id, 'Downloading cerbero JSON',
 107                 headers={
 108                     'Content-Type': 'application/json;charset=UTF-8',
 109                     'Origin': 'https://www.mitele.es'
 110                 },
 111                 data=json.dumps({
 112                     'bbx': caronte['bbx'],
 113                     'gbx': gbx['gbx']
 114                 }).encode('utf-8'))
 115             formats = self._extract_m3u8_formats(
 116                 caronte['dls'][0]['stream'], video_id, 'mp4', 'm3u8_native', m3u8_id='hls',
 117                 query=dict([cerbero['tokens']['1']['cdn'].split('=', 1)]))
 118             info['formats'] = formats
 119
 120         return info