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