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     int_or_none,
  11     parse_iso8601,
  12     parse_duration,
  13     try_get,
  14     update_url_query,
  15 )
  16
  17
  18 class TVNowBaseIE(InfoExtractor):
  19     _VIDEO_FIELDS = (
  20         'id', 'title', 'free', 'geoblocked', 'articleLong', 'articleShort',
  21         'broadcastStartDate', 'isDrm', 'duration', 'season', 'episode',
  22         'manifest.dashclear', 'format.title', 'format.defaultImage169Format',
  23         'format.defaultImage169Logo')
  24
  25     def _call_api(self, path, video_id, query):
  26         return self._download_json(
  27             'https://api.tvnow.de/v3/' + path,
  28             video_id, query=query)
  29
  30     def _extract_video(self, info, display_id):
  31         video_id = compat_str(info['id'])
  32         title = info['title']
  33
  34         mpd_url = info['manifest']['dashclear']
  35         if not mpd_url:
  36             if info.get('isDrm'):
  37                 raise ExtractorError(
  38                     'Video %s is DRM protected' % video_id, expected=True)
  39             if info.get('geoblocked'):
  40                 raise ExtractorError(
  41                     'Video %s is not available from your location due to geo restriction' % video_id,
  42                     expected=True)
  43             if not info.get('free', True):
  44                 raise ExtractorError(
  45                     'Video %s is not available for free' % video_id, expected=True)
  46
  47         mpd_url = update_url_query(mpd_url, {'filter': ''})
  48         formats = self._extract_mpd_formats(mpd_url, video_id, mpd_id='dash', fatal=False)
  49         formats.extend(self._extract_ism_formats(
  50             mpd_url.replace('dash.', 'hss.').replace('/.mpd', '/Manifest'),
  51             video_id, ism_id='mss', fatal=False))
  52         formats.extend(self._extract_m3u8_formats(
  53             mpd_url.replace('dash.', 'hls.').replace('/.mpd', '/.m3u8'),
  54             video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
  55         self._sort_formats(formats)
  56
  57         description = info.get('articleLong') or info.get('articleShort')
  58         timestamp = parse_iso8601(info.get('broadcastStartDate'), ' ')
  59         duration = parse_duration(info.get('duration'))
  60
  61         f = info.get('format', {})
  62
  63         thumbnails = [{
  64             'url': 'https://aistvnow-a.akamaihd.net/tvnow/movie/%s' % video_id,
  65         }]
  66         thumbnail = f.get('defaultImage169Format') or f.get('defaultImage169Logo')
  67         if thumbnail:
  68             thumbnails.append({
  69                 'url': thumbnail,
  70             })
  71
  72         return {
  73             'id': video_id,
  74             'display_id': display_id,
  75             'title': title,
  76             'description': description,
  77             'thumbnails': thumbnails,
  78             'timestamp': timestamp,
  79             'duration': duration,
  80             'series': f.get('title'),
  81             'season_number': int_or_none(info.get('season')),
  82             'episode_number': int_or_none(info.get('episode')),
  83             'episode': title,
  84             'formats': formats,
  85         }
  86
  87
  88 class TVNowIE(TVNowBaseIE):
  89     _VALID_URL = r'''(?x)
  90                     https?://
  91                         (?:www\.)?tvnow\.(?:de|at|ch)/[^/]+/
  92                         (?P<show_id>[^/]+)/
  93                         (?!(?:list|jahr)(?:/|$))(?P<id>[^/?\#&]+)
  94                     '''
  95
  96     _TESTS = [{
  97         'url': 'https://www.tvnow.de/rtl2/grip-das-motormagazin/der-neue-porsche-911-gt-3/player',
  98         'info_dict': {
  99             'id': '331082',
 100             'display_id': 'grip-das-motormagazin/der-neue-porsche-911-gt-3',
 101             'ext': 'mp4',
 102             'title': 'Der neue Porsche 911 GT 3',
 103             'description': 'md5:6143220c661f9b0aae73b245e5d898bb',
 104             'thumbnail': r're:^https?://.*\.jpg$',
 105             'timestamp': 1495994400,
 106             'upload_date': '20170528',
 107             'duration': 5283,
 108             'series': 'GRIP - Das Motormagazin',
 109             'season_number': 14,
 110             'episode_number': 405,
 111             'episode': 'Der neue Porsche 911 GT 3',
 112         },
 113     }, {
 114         # rtl2
 115         'url': 'https://www.tvnow.de/rtl2/armes-deutschland/episode-0008/player',
 116         'only_matching': True,
 117     }, {
 118         # rtlnitro
 119         'url': 'https://www.tvnow.de/nitro/alarm-fuer-cobra-11-die-autobahnpolizei/auf-eigene-faust-pilot/player',
 120         'only_matching': True,
 121     }, {
 122         # superrtl
 123         'url': 'https://www.tvnow.de/superrtl/die-lustigsten-schlamassel-der-welt/u-a-ketchup-effekt/player',
 124         'only_matching': True,
 125     }, {
 126         # ntv
 127         'url': 'https://www.tvnow.de/ntv/startup-news/goetter-in-weiss/player',
 128         'only_matching': True,
 129     }, {
 130         # vox
 131         'url': 'https://www.tvnow.de/vox/auto-mobil/neues-vom-automobilmarkt-2017-11-19-17-00-00/player',
 132         'only_matching': True,
 133     }, {
 134         # rtlplus
 135         'url': 'https://www.tvnow.de/rtlplus/op-ruft-dr-bruckner/die-vernaehte-frau/player',
 136         'only_matching': True,
 137     }, {
 138         'url': 'https://www.tvnow.de/rtl2/grip-das-motormagazin/der-neue-porsche-911-gt-3',
 139         'only_matching': True,
 140     }]
 141
 142     def _real_extract(self, url):
 143         display_id = '%s/%s' % re.match(self._VALID_URL, url).groups()
 144
 145         info = self._call_api(
 146             'movies/' + display_id, display_id, query={
 147                 'fields': ','.join(self._VIDEO_FIELDS),
 148             })
 149
 150         return self._extract_video(info, display_id)
 151
 152
 153 class TVNowListBaseIE(TVNowBaseIE):
 154     _SHOW_VALID_URL = r'''(?x)
 155                     (?P<base_url>
 156                         https?://
 157                             (?:www\.)?tvnow\.(?:de|at|ch)/[^/]+/
 158                             (?P<show_id>[^/]+)
 159                     )
 160                     '''
 161
 162     def _extract_list_info(self, display_id, show_id):
 163         fields = list(self._SHOW_FIELDS)
 164         fields.extend('formatTabs.%s' % field for field in self._SEASON_FIELDS)
 165         fields.extend(
 166             'formatTabs.formatTabPages.container.movies.%s' % field
 167             for field in self._VIDEO_FIELDS)
 168         return self._call_api(
 169             'formats/seo', display_id, query={
 170                 'fields': ','.join(fields),
 171                 'name': show_id + '.php'
 172             })
 173
 174
 175 class TVNowListIE(TVNowListBaseIE):
 176     _VALID_URL = r'%s/(?:list|jahr)/(?P<id>[^?\#&]+)' % TVNowListBaseIE._SHOW_VALID_URL
 177
 178     _SHOW_FIELDS = ('title', )
 179     _SEASON_FIELDS = ('id', 'headline', 'seoheadline', )
 180     _VIDEO_FIELDS = ('id', 'headline', 'seoUrl', )
 181
 182     _TESTS = [{
 183         'url': 'https://www.tvnow.de/rtl/30-minuten-deutschland/list/aktuell',
 184         'info_dict': {
 185             'id': '28296',
 186             'title': '30 Minuten Deutschland - Aktuell',
 187         },
 188         'playlist_mincount': 1,
 189     }, {
 190         'url': 'https://www.tvnow.de/vox/ab-ins-beet/list/staffel-14',
 191         'only_matching': True,
 192     }, {
 193         'url': 'https://www.tvnow.de/rtl2/grip-das-motormagazin/jahr/2018/3',
 194         'only_matching': True,
 195     }]
 196
 197     @classmethod
 198     def suitable(cls, url):
 199         return (False if TVNowIE.suitable(url)
 200                 else super(TVNowListIE, cls).suitable(url))
 201
 202     def _real_extract(self, url):
 203         base_url, show_id, season_id = re.match(self._VALID_URL, url).groups()
 204
 205         list_info = self._extract_list_info(season_id, show_id)
 206
 207         season = next(
 208             season for season in list_info['formatTabs']['items']
 209             if season.get('seoheadline') == season_id)
 210
 211         title = list_info.get('title')
 212         headline = season.get('headline')
 213         if title and headline:
 214             title = '%s - %s' % (title, headline)
 215         else:
 216             title = headline or title
 217
 218         entries = []
 219         for container in season['formatTabPages']['items']:
 220             items = try_get(
 221                 container, lambda x: x['container']['movies']['items'],
 222                 list) or []
 223             for info in items:
 224                 seo_url = info.get('seoUrl')
 225                 if not seo_url:
 226                     continue
 227                 video_id = info.get('id')
 228                 entries.append(self.url_result(
 229                     '%s/%s/player' % (base_url, seo_url), TVNowIE.ie_key(),
 230                     compat_str(video_id) if video_id else None))
 231
 232         return self.playlist_result(
 233             entries, compat_str(season.get('id') or season_id), title)
 234
 235
 236 class TVNowShowIE(TVNowListBaseIE):
 237     _VALID_URL = TVNowListBaseIE._SHOW_VALID_URL
 238
 239     _SHOW_FIELDS = ('id', 'title', )
 240     _SEASON_FIELDS = ('id', 'headline', 'seoheadline', )
 241     _VIDEO_FIELDS = ()
 242
 243     _TESTS = [{
 244         'url': 'https://www.tvnow.at/vox/ab-ins-beet',
 245         'info_dict': {
 246             'id': 'ab-ins-beet',
 247             'title': 'Ab ins Beet!',
 248         },
 249         'playlist_mincount': 7,
 250     }, {
 251         'url': 'https://www.tvnow.at/vox/ab-ins-beet/list',
 252         'only_matching': True,
 253     }, {
 254         'url': 'https://www.tvnow.de/rtl2/grip-das-motormagazin/jahr/',
 255         'only_matching': True,
 256     }]
 257
 258     @classmethod
 259     def suitable(cls, url):
 260         return (False if TVNowIE.suitable(url) or TVNowListIE.suitable(url)
 261                 else super(TVNowShowIE, cls).suitable(url))
 262
 263     def _real_extract(self, url):
 264         base_url, show_id = re.match(self._VALID_URL, url).groups()
 265
 266         list_info = self._extract_list_info(show_id, show_id)
 267
 268         entries = []
 269         for season_info in list_info['formatTabs']['items']:
 270             season_url = season_info.get('seoheadline')
 271             if not season_url:
 272                 continue
 273             season_id = season_info.get('id')
 274             entries.append(self.url_result(
 275                 '%s/list/%s' % (base_url, season_url), TVNowListIE.ie_key(),
 276                 compat_str(season_id) if season_id else None,
 277                 season_info.get('headline')))
 278
 279         return self.playlist_result(entries, show_id, list_info.get('title'))