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