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