youtube_dlc/extractor/ndr.py

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