Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/tvnow.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_str
   8 from ..utils import (
   9     ExtractorError,
  10     parse_iso8601,
  11     parse_duration,
  12     update_url_query,
  13 )
  14
  15
  16 class TVNowBaseIE(InfoExtractor):
  17     _VIDEO_FIELDS = (
  18         'id', 'title', 'free', 'geoblocked', 'articleLong', 'articleShort',
  19         'broadcastStartDate', 'isDrm', 'duration', 'manifest.dashclear',
  20         'format.defaultImage169Format', 'format.defaultImage169Logo')
  21
  22     def _call_api(self, path, video_id, query):
  23         return self._download_json(
  24             'https://api.tvnow.de/v3/' + path,
  25             video_id, query=query)
  26
  27     def _extract_video(self, info, display_id):
  28         video_id = compat_str(info['id'])
  29         title = info['title']
  30
  31         mpd_url = info['manifest']['dashclear']
  32         if not mpd_url:
  33             if info.get('isDrm'):
  34                 raise ExtractorError(
  35                     'Video %s is DRM protected' % video_id, expected=True)
  36             if info.get('geoblocked'):
  37                 raise ExtractorError(
  38                     'Video %s is not available from your location due to geo restriction' % video_id,
  39                     expected=True)
  40             if not info.get('free', True):
  41                 raise ExtractorError(
  42                     'Video %s is not available for free' % video_id, expected=True)
  43
  44         mpd_url = update_url_query(mpd_url, {'filter': ''})
  45         formats = self._extract_mpd_formats(mpd_url, video_id, mpd_id='dash', fatal=False)
  46         formats.extend(self._extract_ism_formats(
  47             mpd_url.replace('dash.', 'hss.').replace('/.mpd', '/Manifest'),
  48             video_id, ism_id='mss', fatal=False))
  49         formats.extend(self._extract_m3u8_formats(
  50             mpd_url.replace('dash.', 'hls.').replace('/.mpd', '/.m3u8'),
  51             video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
  52         self._sort_formats(formats)
  53
  54         description = info.get('articleLong') or info.get('articleShort')
  55         timestamp = parse_iso8601(info.get('broadcastStartDate'), ' ')
  56         duration = parse_duration(info.get('duration'))
  57
  58         f = info.get('format', {})
  59         thumbnail = f.get('defaultImage169Format') or f.get('defaultImage169Logo')
  60
  61         return {
  62             'id': video_id,
  63             'display_id': display_id,
  64             'title': title,
  65             'description': description,
  66             'thumbnail': thumbnail,
  67             'timestamp': timestamp,
  68             'duration': duration,
  69             'formats': formats,
  70         }
  71
  72
  73 class TVNowIE(TVNowBaseIE):
  74     _VALID_URL = r'https?://(?:www\.)?tvnow\.(?:de|at|ch)/(?:rtl(?:2|plus)?|nitro|superrtl|ntv|vox)/(?P<show_id>[^/]+)/(?:(?:list/[^/]+|jahr/\d{4}/\d{1,2})/)?(?P<id>[^/]+)/(?:player|preview)'
  75
  76     _TESTS = [{
  77         # rtl
  78         'url': 'https://www.tvnow.de/rtl/alarm-fuer-cobra-11/freier-fall/player?return=/rtl',
  79         'info_dict': {
  80             'id': '385314',
  81             'display_id': 'alarm-fuer-cobra-11/freier-fall',
  82             'ext': 'mp4',
  83             'title': 'Freier Fall',
  84             'description': 'md5:8c2d8f727261adf7e0dc18366124ca02',
  85             'thumbnail': r're:^https?://.*\.jpg$',
  86             'timestamp': 1512677700,
  87             'upload_date': '20171207',
  88             'duration': 2862.0,
  89         },
  90     }, {
  91         # rtl2
  92         'url': 'https://www.tvnow.de/rtl2/armes-deutschland/episode-0008/player',
  93         'only_matching': 'True',
  94     }, {
  95         # rtlnitro
  96         'url': 'https://www.tvnow.de/nitro/alarm-fuer-cobra-11-die-autobahnpolizei/auf-eigene-faust-pilot/player',
  97         'only_matching': 'True',
  98     }, {
  99         # superrtl
 100         'url': 'https://www.tvnow.de/superrtl/die-lustigsten-schlamassel-der-welt/u-a-ketchup-effekt/player',
 101         'only_matching': 'True',
 102     }, {
 103         # ntv
 104         'url': 'https://www.tvnow.de/ntv/startup-news/goetter-in-weiss/player',
 105         'only_matching': 'True',
 106     }, {
 107         # vox
 108         'url': 'https://www.tvnow.de/vox/auto-mobil/neues-vom-automobilmarkt-2017-11-19-17-00-00/player',
 109         'only_matching': 'True',
 110     }, {
 111         # rtlplus
 112         'url': 'https://www.tvnow.de/rtlplus/op-ruft-dr-bruckner/die-vernaehte-frau/player',
 113         'only_matching': 'True',
 114     }]
 115
 116     def _real_extract(self, url):
 117         display_id = '%s/%s' % re.match(self._VALID_URL, url).groups()
 118
 119         info = self._call_api(
 120             'movies/' + display_id, display_id, query={
 121                 'fields': ','.join(self._VIDEO_FIELDS),
 122             })
 123
 124         return self._extract_video(info, display_id)
 125
 126
 127 class TVNowListIE(TVNowBaseIE):
 128     _VALID_URL = r'(?P<base_url>https?://(?:www\.)?tvnow\.(?:de|at|ch)/(?:rtl(?:2|plus)?|nitro|superrtl|ntv|vox)/(?P<show_id>[^/]+)/)list/(?P<id>[^?/#&]+)$'
 129
 130     _SHOW_FIELDS = ('title', )
 131     _SEASON_FIELDS = ('id', 'headline', 'seoheadline', )
 132     _VIDEO_FIELDS = ('id', 'headline', 'seoUrl', )
 133
 134     _TESTS = [{
 135         'url': 'https://www.tvnow.de/rtl/30-minuten-deutschland/list/aktuell',
 136         'info_dict': {
 137             'id': '28296',
 138             'title': '30 Minuten Deutschland - Aktuell',
 139         },
 140         'playlist_mincount': 1,
 141     }]
 142
 143     def _real_extract(self, url):
 144         base_url, show_id, season_id = re.match(self._VALID_URL, url).groups()
 145
 146         fields = []
 147         fields.extend(self._SHOW_FIELDS)
 148         fields.extend('formatTabs.%s' % field for field in self._SEASON_FIELDS)
 149         fields.extend(
 150             'formatTabs.formatTabPages.container.movies.%s' % field
 151             for field in self._VIDEO_FIELDS)
 152
 153         list_info = self._call_api(
 154             'formats/seo', season_id, query={
 155                 'fields': ','.join(fields),
 156                 'name': show_id + '.php'
 157             })
 158
 159         season = next(
 160             season for season in list_info['formatTabs']['items']
 161             if season.get('seoheadline') == season_id)
 162
 163         title = '%s - %s' % (list_info['title'], season['headline'])
 164
 165         entries = []
 166         for container in season['formatTabPages']['items']:
 167             for info in ((container.get('container') or {}).get('movies') or {}).get('items') or []:
 168                 seo_url = info.get('seoUrl')
 169                 if not seo_url:
 170                     continue
 171                 entries.append(self.url_result(
 172                     base_url + seo_url + '/player', 'TVNow', info.get('id')))
 173
 174         return self.playlist_result(
 175             entries, compat_str(season.get('id') or season_id), title)