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