yt_dlp/extractor/ndr.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     determine_ext,
   8     int_or_none,
   9     merge_dicts,
  10     parse_iso8601,
  11     qualities,
  12     try_get,
  13     urljoin,
  14 )
  15
  16
  17 class NDRBaseIE(InfoExtractor):
  18     def _real_extract(self, url):
  19         mobj = self._match_valid_url(url)
  20         display_id = next(group for group in mobj.groups() if group)
  21         id = mobj.group('id')
  22         webpage = self._download_webpage(url, display_id)
  23         return self._extract_embed(webpage, display_id, id)
  24
  25
  26 class NDRIE(NDRBaseIE):
  27     IE_NAME = 'ndr'
  28     IE_DESC = 'NDR.de - Norddeutscher Rundfunk'
  29     _VALID_URL = r'https?://(?:www\.)?(?:daserste\.)?ndr\.de/(?:[^/]+/)*(?P<display_id>[^/?#]+),(?P<id>[\da-z]+)\.html'
  30     _TESTS = [{
  31         # httpVideo, same content id
  32         'url': 'http://www.ndr.de/fernsehen/Party-Poette-und-Parade,hafengeburtstag988.html',
  33         'md5': '6515bc255dc5c5f8c85bbc38e035a659',
  34         'info_dict': {
  35             'id': 'hafengeburtstag988',
  36             'display_id': 'Party-Poette-und-Parade',
  37             'ext': 'mp4',
  38             'title': 'Party, Pötte und Parade',
  39             'description': 'md5:ad14f9d2f91d3040b6930c697e5f6b4c',
  40             'uploader': 'ndrtv',
  41             'timestamp': 1431108900,
  42             'upload_date': '20150510',
  43             'duration': 3498,
  44         },
  45         'params': {
  46             'skip_download': True,
  47         },
  48     }, {
  49         # httpVideo, different content id
  50         'url': 'http://www.ndr.de/sport/fussball/40-Osnabrueck-spielt-sich-in-einen-Rausch,osna270.html',
  51         'md5': '1043ff203eab307f0c51702ec49e9a71',
  52         'info_dict': {
  53             'id': 'osna272',
  54             'display_id': '40-Osnabrueck-spielt-sich-in-einen-Rausch',
  55             'ext': 'mp4',
  56             'title': 'Osnabrück - Wehen Wiesbaden: Die Highlights',
  57             'description': 'md5:32e9b800b3d2d4008103752682d5dc01',
  58             'uploader': 'ndrtv',
  59             'timestamp': 1442059200,
  60             'upload_date': '20150912',
  61             'duration': 510,
  62         },
  63         'params': {
  64             'skip_download': True,
  65         },
  66     }, {
  67         # httpAudio, same content id
  68         'url': 'http://www.ndr.de/info/La-Valette-entgeht-der-Hinrichtung,audio51535.html',
  69         'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
  70         'info_dict': {
  71             'id': 'audio51535',
  72             'display_id': 'La-Valette-entgeht-der-Hinrichtung',
  73             'ext': 'mp3',
  74             'title': 'La Valette entgeht der Hinrichtung',
  75             'description': 'md5:22f9541913a40fe50091d5cdd7c9f536',
  76             'uploader': 'ndrinfo',
  77             'timestamp': 1290626100,
  78             'upload_date': '20140729',
  79             'duration': 884,
  80         },
  81         'params': {
  82             'skip_download': True,
  83         },
  84     }, {
  85         # with subtitles
  86         'url': 'https://www.ndr.de/fernsehen/sendungen/extra_3/extra-3-Satiremagazin-mit-Christian-Ehring,sendung1091858.html',
  87         'info_dict': {
  88             'id': 'extra18674',
  89             'display_id': 'extra-3-Satiremagazin-mit-Christian-Ehring',
  90             'ext': 'mp4',
  91             'title': 'Extra 3 vom 11.11.2020 mit Christian Ehring',
  92             'description': 'md5:42ee53990a715eaaf4dc7f13a3bd56c6',
  93             'uploader': 'ndrtv',
  94             'upload_date': '20201113',
  95             'duration': 1749,
  96             'subtitles': {
  97                 'de': [{
  98                     'ext': 'ttml',
  99                     'url': r're:^https://www\.ndr\.de.+',
 100                 }],
 101             },
 102         },
 103         'params': {
 104             'skip_download': True,
 105         },
 106         'expected_warnings': ['Unable to download f4m manifest'],
 107     }, {
 108         'url': 'https://www.ndr.de/Fettes-Brot-Ferris-MC-und-Thees-Uhlmann-live-on-stage,festivalsommer116.html',
 109         'only_matching': True,
 110     }]
 111
 112     def _extract_embed(self, webpage, display_id, id):
 113         embed_url = self._html_search_meta(
 114             'embedURL', webpage, 'embed URL',
 115             default=None) or self._search_regex(
 116             r'\bembedUrl["\']\s*:\s*(["\'])(?P<url>(?:(?!\1).)+)\1', webpage,
 117             'embed URL', fatal=False, group='url')
 118         if embed_url is None:
 119             return self.url_result('ndr:%s' % id, ie=NDREmbedBaseIE.ie_key())
 120         description = self._search_regex(
 121             r'<p[^>]+itemprop="description">([^<]+)</p>',
 122             webpage, 'description', default=None) or self._og_search_description(webpage)
 123         timestamp = parse_iso8601(
 124             self._search_regex(
 125                 r'<span[^>]+itemprop="(?:datePublished|uploadDate)"[^>]+content="([^"]+)"',
 126                 webpage, 'upload date', default=None))
 127         info = self._search_json_ld(webpage, display_id, default={})
 128         return merge_dicts({
 129             '_type': 'url_transparent',
 130             'url': embed_url,
 131             'display_id': display_id,
 132             'description': description,
 133             'timestamp': timestamp,
 134         }, info)
 135
 136
 137 class NJoyIE(NDRBaseIE):
 138     IE_NAME = 'njoy'
 139     IE_DESC = 'N-JOY'
 140     _VALID_URL = r'https?://(?:www\.)?n-joy\.de/(?:[^/]+/)*(?:(?P<display_id>[^/?#]+),)?(?P<id>[\da-z]+)\.html'
 141     _TESTS = [{
 142         # httpVideo, same content id
 143         'url': 'http://www.n-joy.de/entertainment/comedy/comedy_contest/Benaissa-beim-NDR-Comedy-Contest,comedycontest2480.html',
 144         'md5': 'cb63be60cd6f9dd75218803146d8dc67',
 145         'info_dict': {
 146             'id': 'comedycontest2480',
 147             'display_id': 'Benaissa-beim-NDR-Comedy-Contest',
 148             'ext': 'mp4',
 149             'title': 'Benaissa beim NDR Comedy Contest',
 150             'description': 'md5:f057a6c4e1c728b10d33b5ffd36ddc39',
 151             'uploader': 'ndrtv',
 152             'upload_date': '20141129',
 153             'duration': 654,
 154         },
 155         'params': {
 156             'skip_download': True,
 157         },
 158     }, {
 159         # httpVideo, different content id
 160         'url': 'http://www.n-joy.de/musik/Das-frueheste-DJ-Set-des-Nordens-live-mit-Felix-Jaehn-,felixjaehn168.html',
 161         'md5': '417660fffa90e6df2fda19f1b40a64d8',
 162         'info_dict': {
 163             'id': 'dockville882',
 164             'display_id': 'Das-frueheste-DJ-Set-des-Nordens-live-mit-Felix-Jaehn-',
 165             'ext': 'mp4',
 166             'title': '"Ich hab noch nie" mit Felix Jaehn',
 167             'description': 'md5:85dd312d53be1b99e1f998a16452a2f3',
 168             'uploader': 'njoy',
 169             'upload_date': '20150822',
 170             'duration': 211,
 171         },
 172         'params': {
 173             'skip_download': True,
 174         },
 175     }, {
 176         'url': 'http://www.n-joy.de/radio/webradio/morningshow209.html',
 177         'only_matching': True,
 178     }]
 179
 180     def _extract_embed(self, webpage, display_id, id):
 181         video_id = self._search_regex(
 182             r'<iframe[^>]+id="pp_([\da-z]+)"', webpage, 'embed id')
 183         description = self._search_regex(
 184             r'<div[^>]+class="subline"[^>]*>[^<]+</div>\s*<p>([^<]+)</p>',
 185             webpage, 'description', fatal=False)
 186         return {
 187             '_type': 'url_transparent',
 188             'ie_key': 'NDREmbedBase',
 189             'url': 'ndr:%s' % video_id,
 190             'display_id': display_id,
 191             'description': description,
 192         }
 193
 194
 195 class NDREmbedBaseIE(InfoExtractor):
 196     IE_NAME = 'ndr:embed:base'
 197     _VALID_URL = r'(?:ndr:(?P<id_s>[\da-z]+)|https?://www\.ndr\.de/(?P<id>[\da-z]+)-ppjson\.json)'
 198     _TESTS = [{
 199         'url': 'ndr:soundcheck3366',
 200         'only_matching': True,
 201     }, {
 202         'url': 'http://www.ndr.de/soundcheck3366-ppjson.json',
 203         'only_matching': True,
 204     }]
 205
 206     def _real_extract(self, url):
 207         mobj = self._match_valid_url(url)
 208         video_id = mobj.group('id') or mobj.group('id_s')
 209
 210         ppjson = self._download_json(
 211             'http://www.ndr.de/%s-ppjson.json' % video_id, video_id)
 212
 213         playlist = ppjson['playlist']
 214
 215         formats = []
 216         quality_key = qualities(('xs', 's', 'm', 'l', 'xl'))
 217
 218         for format_id, f in playlist.items():
 219             src = f.get('src')
 220             if not src:
 221                 continue
 222             ext = determine_ext(src, None)
 223             if ext == 'f4m':
 224                 formats.extend(self._extract_f4m_formats(
 225                     src + '?hdcore=3.7.0&plugin=aasp-3.7.0.39.44', video_id,
 226                     f4m_id='hds', fatal=False))
 227             elif ext == 'm3u8':
 228                 formats.extend(self._extract_m3u8_formats(
 229                     src, video_id, 'mp4', m3u8_id='hls',
 230                     entry_protocol='m3u8_native', fatal=False))
 231             else:
 232                 quality = f.get('quality')
 233                 ff = {
 234                     'url': src,
 235                     'format_id': quality or format_id,
 236                     'quality': quality_key(quality),
 237                 }
 238                 type_ = f.get('type')
 239                 if type_ and type_.split('/')[0] == 'audio':
 240                     ff['vcodec'] = 'none'
 241                     ff['ext'] = ext or 'mp3'
 242                 formats.append(ff)
 243         self._sort_formats(formats)
 244
 245         config = playlist['config']
 246
 247         live = playlist.get('config', {}).get('streamType') in ['httpVideoLive', 'httpAudioLive']
 248         title = config['title']
 249         if live:
 250             title = self._live_title(title)
 251         uploader = ppjson.get('config', {}).get('branding')
 252         upload_date = ppjson.get('config', {}).get('publicationDate')
 253         duration = int_or_none(config.get('duration'))
 254
 255         thumbnails = []
 256         poster = try_get(config, lambda x: x['poster'], dict) or {}
 257         for thumbnail_id, thumbnail in poster.items():
 258             thumbnail_url = urljoin(url, thumbnail.get('src'))
 259             if not thumbnail_url:
 260                 continue
 261             thumbnails.append({
 262                 'id': thumbnail.get('quality') or thumbnail_id,
 263                 'url': thumbnail_url,
 264                 'preference': quality_key(thumbnail.get('quality')),
 265             })
 266
 267         subtitles = {}
 268         tracks = config.get('tracks')
 269         if tracks and isinstance(tracks, list):
 270             for track in tracks:
 271                 if not isinstance(track, dict):
 272                     continue
 273                 track_url = urljoin(url, track.get('src'))
 274                 if not track_url:
 275                     continue
 276                 subtitles.setdefault(track.get('srclang') or 'de', []).append({
 277                     'url': track_url,
 278                     'ext': 'ttml',
 279                 })
 280
 281         return {
 282             'id': video_id,
 283             'title': title,
 284             'is_live': live,
 285             'uploader': uploader if uploader != '-' else None,
 286             'upload_date': upload_date[0:8] if upload_date else None,
 287             'duration': duration,
 288             'thumbnails': thumbnails,
 289             'formats': formats,
 290             'subtitles': subtitles,
 291         }
 292
 293
 294 class NDREmbedIE(NDREmbedBaseIE):
 295     IE_NAME = 'ndr:embed'
 296     _VALID_URL = r'https?://(?:www\.)?(?:daserste\.)?ndr\.de/(?:[^/]+/)*(?P<id>[\da-z]+)-(?:player|externalPlayer)\.html'
 297     _TESTS = [{
 298         'url': 'http://www.ndr.de/fernsehen/sendungen/ndr_aktuell/ndraktuell28488-player.html',
 299         'md5': '8b9306142fe65bbdefb5ce24edb6b0a9',
 300         'info_dict': {
 301             'id': 'ndraktuell28488',
 302             'ext': 'mp4',
 303             'title': 'Norddeutschland begrüßt Flüchtlinge',
 304             'is_live': False,
 305             'uploader': 'ndrtv',
 306             'upload_date': '20150907',
 307             'duration': 132,
 308         },
 309     }, {
 310         'url': 'http://www.ndr.de/ndr2/events/soundcheck/soundcheck3366-player.html',
 311         'md5': '002085c44bae38802d94ae5802a36e78',
 312         'info_dict': {
 313             'id': 'soundcheck3366',
 314             'ext': 'mp4',
 315             'title': 'Ella Henderson braucht Vergleiche nicht zu scheuen',
 316             'is_live': False,
 317             'uploader': 'ndr2',
 318             'upload_date': '20150912',
 319             'duration': 3554,
 320         },
 321         'params': {
 322             'skip_download': True,
 323         },
 324     }, {
 325         'url': 'http://www.ndr.de/info/audio51535-player.html',
 326         'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
 327         'info_dict': {
 328             'id': 'audio51535',
 329             'ext': 'mp3',
 330             'title': 'La Valette entgeht der Hinrichtung',
 331             'is_live': False,
 332             'uploader': 'ndrinfo',
 333             'upload_date': '20140729',
 334             'duration': 884,
 335         },
 336         'params': {
 337             'skip_download': True,
 338         },
 339     }, {
 340         'url': 'http://www.ndr.de/fernsehen/sendungen/visite/visite11010-externalPlayer.html',
 341         'md5': 'ae57f80511c1e1f2fd0d0d3d31aeae7c',
 342         'info_dict': {
 343             'id': 'visite11010',
 344             'ext': 'mp4',
 345             'title': 'Visite - die ganze Sendung',
 346             'is_live': False,
 347             'uploader': 'ndrtv',
 348             'upload_date': '20150902',
 349             'duration': 3525,
 350         },
 351         'params': {
 352             'skip_download': True,
 353         },
 354     }, {
 355         # httpVideoLive
 356         'url': 'http://www.ndr.de/fernsehen/livestream/livestream217-externalPlayer.html',
 357         'info_dict': {
 358             'id': 'livestream217',
 359             'ext': 'flv',
 360             'title': r're:^NDR Fernsehen Niedersachsen \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
 361             'is_live': True,
 362             'upload_date': '20150910',
 363         },
 364         'params': {
 365             'skip_download': True,
 366         },
 367     }, {
 368         'url': 'http://www.ndr.de/ndrkultur/audio255020-player.html',
 369         'only_matching': True,
 370     }, {
 371         'url': 'http://www.ndr.de/fernsehen/sendungen/nordtour/nordtour7124-player.html',
 372         'only_matching': True,
 373     }, {
 374         'url': 'http://www.ndr.de/kultur/film/videos/videoimport10424-player.html',
 375         'only_matching': True,
 376     }, {
 377         'url': 'http://www.ndr.de/fernsehen/sendungen/hamburg_journal/hamj43006-player.html',
 378         'only_matching': True,
 379     }, {
 380         'url': 'http://www.ndr.de/fernsehen/sendungen/weltbilder/weltbilder4518-player.html',
 381         'only_matching': True,
 382     }, {
 383         'url': 'http://www.ndr.de/fernsehen/doku952-player.html',
 384         'only_matching': True,
 385     }]
 386
 387
 388 class NJoyEmbedIE(NDREmbedBaseIE):
 389     IE_NAME = 'njoy:embed'
 390     _VALID_URL = r'https?://(?:www\.)?n-joy\.de/(?:[^/]+/)*(?P<id>[\da-z]+)-(?:player|externalPlayer)_[^/]+\.html'
 391     _TESTS = [{
 392         # httpVideo
 393         'url': 'http://www.n-joy.de/events/reeperbahnfestival/doku948-player_image-bc168e87-5263-4d6d-bd27-bb643005a6de_theme-n-joy.html',
 394         'md5': '8483cbfe2320bd4d28a349d62d88bd74',
 395         'info_dict': {
 396             'id': 'doku948',
 397             'ext': 'mp4',
 398             'title': 'Zehn Jahre Reeperbahn Festival - die Doku',
 399             'is_live': False,
 400             'upload_date': '20150807',
 401             'duration': 1011,
 402         },
 403     }, {
 404         # httpAudio
 405         'url': 'http://www.n-joy.de/news_wissen/stefanrichter100-player_image-d5e938b1-f21a-4b9a-86b8-aaba8bca3a13_theme-n-joy.html',
 406         'md5': 'd989f80f28ac954430f7b8a48197188a',
 407         'info_dict': {
 408             'id': 'stefanrichter100',
 409             'ext': 'mp3',
 410             'title': 'Interview mit einem Augenzeugen',
 411             'is_live': False,
 412             'uploader': 'njoy',
 413             'upload_date': '20150909',
 414             'duration': 140,
 415         },
 416         'params': {
 417             'skip_download': True,
 418         },
 419     }, {
 420         # httpAudioLive, no explicit ext
 421         'url': 'http://www.n-joy.de/news_wissen/webradioweltweit100-player_image-3fec0484-2244-4565-8fb8-ed25fd28b173_theme-n-joy.html',
 422         'info_dict': {
 423             'id': 'webradioweltweit100',
 424             'ext': 'mp3',
 425             'title': r're:^N-JOY Weltweit \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
 426             'is_live': True,
 427             'uploader': 'njoy',
 428             'upload_date': '20150810',
 429         },
 430         'params': {
 431             'skip_download': True,
 432         },
 433     }, {
 434         'url': 'http://www.n-joy.de/musik/dockville882-player_image-3905259e-0803-4764-ac72-8b7de077d80a_theme-n-joy.html',
 435         'only_matching': True,
 436     }, {
 437         'url': 'http://www.n-joy.de/radio/sendungen/morningshow/urlaubsfotos190-player_image-066a5df1-5c95-49ec-a323-941d848718db_theme-n-joy.html',
 438         'only_matching': True,
 439     }, {
 440         'url': 'http://www.n-joy.de/entertainment/comedy/krudetv290-player_image-ab261bfe-51bf-4bf3-87ba-c5122ee35b3d_theme-n-joy.html',
 441         'only_matching': True,
 442     }]