Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/mitele.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import uuid
   5
   6 from .common import InfoExtractor
   7 from ..compat import (
   8     compat_str,
   9     compat_urllib_parse_urlencode,
  10     compat_urlparse,
  11 )
  12 from ..utils import (
  13     int_or_none,
  14     extract_attributes,
  15     determine_ext,
  16     smuggle_url,
  17     parse_duration,
  18 )
  19
  20
  21 class MiTeleBaseIE(InfoExtractor):
  22     def _get_player_info(self, url, webpage):
  23         player_data = extract_attributes(self._search_regex(
  24             r'(?s)(<ms-video-player.+?</ms-video-player>)',
  25             webpage, 'ms video player'))
  26         video_id = player_data['data-media-id']
  27         config_url = compat_urlparse.urljoin(url, player_data['data-config'])
  28         config = self._download_json(
  29             config_url, video_id, 'Downloading config JSON')
  30         mmc_url = config['services']['mmc']
  31
  32         duration = None
  33         formats = []
  34         for m_url in (mmc_url, mmc_url.replace('/flash.json', '/html5.json')):
  35             mmc = self._download_json(
  36                 m_url, video_id, 'Downloading mmc JSON')
  37             if not duration:
  38                 duration = int_or_none(mmc.get('duration'))
  39             for location in mmc['locations']:
  40                 gat = self._proto_relative_url(location.get('gat'), 'http:')
  41                 bas = location.get('bas')
  42                 loc = location.get('loc')
  43                 ogn = location.get('ogn')
  44                 if None in (gat, bas, loc, ogn):
  45                     continue
  46                 token_data = {
  47                     'bas': bas,
  48                     'icd': loc,
  49                     'ogn': ogn,
  50                     'sta': '0',
  51                 }
  52                 media = self._download_json(
  53                     '%s/?%s' % (gat, compat_urllib_parse_urlencode(token_data)),
  54                     video_id, 'Downloading %s JSON' % location['loc'])
  55                 file_ = media.get('file')
  56                 if not file_:
  57                     continue
  58                 ext = determine_ext(file_)
  59                 if ext == 'f4m':
  60                     formats.extend(self._extract_f4m_formats(
  61                         file_ + '&hdcore=3.2.0&plugin=aasp-3.2.0.77.18',
  62                         video_id, f4m_id='hds', fatal=False))
  63                 elif ext == 'm3u8':
  64                     formats.extend(self._extract_m3u8_formats(
  65                         file_, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
  66         self._sort_formats(formats)
  67
  68         return {
  69             'id': video_id,
  70             'formats': formats,
  71             'thumbnail': player_data.get('data-poster') or config.get('poster', {}).get('imageUrl'),
  72             'duration': duration,
  73         }
  74
  75
  76 class MiTeleIE(InfoExtractor):
  77     IE_DESC = 'mitele.es'
  78     _VALID_URL = r'https?://(?:www\.)?mitele\.es/(?:[^/]+/)+(?P<id>[^/]+)/player'
  79
  80     _TESTS = [{
  81         'url': 'http://www.mitele.es/programas-tv/diario-de/57b0dfb9c715da65618b4afa/player',
  82         'info_dict': {
  83             'id': '57b0dfb9c715da65618b4afa',
  84             'ext': 'mp4',
  85             'title': 'Tor, la web invisible',
  86             'description': 'md5:3b6fce7eaa41b2d97358726378d9369f',
  87             'series': 'Diario de',
  88             'season': 'La redacción',
  89             'season_number': 14,
  90             'season_id': 'diario_de_t14_11981',
  91             'episode': 'Programa 144',
  92             'episode_number': 3,
  93             'thumbnail': r're:(?i)^https?://.*\.jpg$',
  94             'duration': 2913,
  95         },
  96         'add_ie': ['Ooyala'],
  97     }, {
  98         # no explicit title
  99         'url': 'http://www.mitele.es/programas-tv/cuarto-milenio/57b0de3dc915da14058b4876/player',
 100         'info_dict': {
 101             'id': '57b0de3dc915da14058b4876',
 102             'ext': 'mp4',
 103             'title': 'Cuarto Milenio Temporada 6 Programa 226',
 104             'description': 'md5:5ff132013f0cd968ffbf1f5f3538a65f',
 105             'series': 'Cuarto Milenio',
 106             'season': 'Temporada 6',
 107             'season_number': 6,
 108             'season_id': 'cuarto_milenio_t06_12715',
 109             'episode': 'Programa 226',
 110             'episode_number': 24,
 111             'thumbnail': r're:(?i)^https?://.*\.jpg$',
 112             'duration': 7313,
 113         },
 114         'params': {
 115             'skip_download': True,
 116         },
 117         'add_ie': ['Ooyala'],
 118     }, {
 119         'url': 'http://www.mitele.es/series-online/la-que-se-avecina/57aac5c1c915da951a8b45ed/player',
 120         'only_matching': True,
 121     }]
 122
 123     def _real_extract(self, url):
 124         video_id = self._match_id(url)
 125         webpage = self._download_webpage(url, video_id)
 126
 127         gigya_url = self._search_regex(
 128             r'<gigya-api>[^>]*</gigya-api>[^>]*<script\s+src="([^"]*)">[^>]*</script>',
 129             webpage, 'gigya', default=None)
 130         gigya_sc = self._download_webpage(
 131             compat_urlparse.urljoin('http://www.mitele.es/', gigya_url),
 132             video_id, 'Downloading gigya script')
 133
 134         # Get a appKey/uuid for getting the session key
 135         appKey_var = self._search_regex(
 136             r'value\s*\(\s*["\']appGridApplicationKey["\']\s*,\s*([0-9a-f]+)',
 137             gigya_sc, 'appKey variable')
 138         appKey = self._search_regex(
 139             r'var\s+%s\s*=\s*["\']([0-9a-f]+)' % appKey_var, gigya_sc, 'appKey')
 140
 141         session_json = self._download_json(
 142             'https://appgrid-api.cloud.accedo.tv/session',
 143             video_id, 'Downloading session keys', query={
 144                 'appKey': appKey,
 145                 'uuid': compat_str(uuid.uuid4()),
 146             })
 147
 148         paths = self._download_json(
 149             'https://appgrid-api.cloud.accedo.tv/metadata/general_configuration,%20web_configuration',
 150             video_id, 'Downloading paths JSON',
 151             query={'sessionKey': compat_str(session_json['sessionKey'])})
 152
 153         ooyala_s = paths['general_configuration']['api_configuration']['ooyala_search']
 154         source = self._download_json(
 155             'http://%s%s%s/docs/%s' % (
 156                 ooyala_s['base_url'], ooyala_s['full_path'],
 157                 ooyala_s['provider_id'], video_id),
 158             video_id, 'Downloading data JSON', query={
 159                 'include_titles': 'Series,Season',
 160                 'product_name': 'test',
 161                 'format': 'full',
 162             })['hits']['hits'][0]['_source']
 163
 164         embedCode = source['offers'][0]['embed_codes'][0]
 165         titles = source['localizable_titles'][0]
 166
 167         title = titles.get('title_medium') or titles['title_long']
 168
 169         description = titles.get('summary_long') or titles.get('summary_medium')
 170
 171         def get(key1, key2):
 172             value1 = source.get(key1)
 173             if not value1 or not isinstance(value1, list):
 174                 return
 175             if not isinstance(value1[0], dict):
 176                 return
 177             return value1[0].get(key2)
 178
 179         series = get('localizable_titles_series', 'title_medium')
 180
 181         season = get('localizable_titles_season', 'title_medium')
 182         season_number = int_or_none(source.get('season_number'))
 183         season_id = source.get('season_id')
 184
 185         episode = titles.get('title_sort_name')
 186         episode_number = int_or_none(source.get('episode_number'))
 187
 188         duration = parse_duration(get('videos', 'duration'))
 189
 190         return {
 191             '_type': 'url_transparent',
 192             # for some reason only HLS is supported
 193             'url': smuggle_url('ooyala:' + embedCode, {'supportedformats': 'm3u8,dash'}),
 194             'id': video_id,
 195             'title': title,
 196             'description': description,
 197             'series': series,
 198             'season': season,
 199             'season_number': season_number,
 200             'season_id': season_id,
 201             'episode': episode,
 202             'episode_number': episode_number,
 203             'duration': duration,
 204             'thumbnail': get('images', 'url'),
 205         }