Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/twentymin.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 remove_end
   8
   9
  10 class TwentyMinutenIE(InfoExtractor):
  11     IE_NAME = '20min'
  12     _VALID_URL = r'https?://(?:www\.)?20min\.ch/(?:videotv/*\?.*\bvid=(?P<id>\d+)|(?:[^/]+/)*(?P<display_id>[^/#?]+))'
  13     _TESTS = [{
  14         # regular video
  15         'url': 'http://www.20min.ch/videotv/?vid=469148&cid=2',
  16         'md5': 'b52d6bc6ea6398e6a38f12cfd418149c',
  17         'info_dict': {
  18             'id': '469148',
  19             'ext': 'flv',
  20             'title': '85 000 Franken für 15 perfekte Minuten',
  21             'description': 'Was die Besucher vom Silvesterzauber erwarten können. (Video: Alice Grosjean/Murat Temel)',
  22             'thumbnail': 'http://thumbnails.20min-tv.ch/server063/469148/frame-72-469148.jpg'
  23         }
  24     }, {
  25         # news article with video
  26         'url': 'http://www.20min.ch/schweiz/news/story/-Wir-muessen-mutig-nach-vorne-schauen--22050469',
  27         'md5': 'cd4cbb99b94130cff423e967cd275e5e',
  28         'info_dict': {
  29             'id': '469408',
  30             'display_id': '-Wir-muessen-mutig-nach-vorne-schauen--22050469',
  31             'ext': 'flv',
  32             'title': '«Wir müssen mutig nach vorne schauen»',
  33             'description': 'Kein Land sei innovativer als die Schweiz, sagte Johann Schneider-Ammann in seiner Neujahrsansprache. Das Land müsse aber seine Hausaufgaben machen.',
  34             'thumbnail': 'http://www.20min.ch/images/content/2/2/0/22050469/10/teaserbreit.jpg'
  35         },
  36         'skip': '"This video is no longer available" is shown both on the web page and in the downloaded file.',
  37     }, {
  38         # YouTube embed
  39         'url': 'http://www.20min.ch/ro/sports/football/story/Il-marque-une-bicyclette-de-plus-de-30-metres--21115184',
  40         'md5': 'cec64d59aa01c0ed9dbba9cf639dd82f',
  41         'info_dict': {
  42             'id': 'ivM7A7SpDOs',
  43             'ext': 'mp4',
  44             'title': 'GOLAZO DE CHILENA DE JAVI GÓMEZ, FINALISTA AL BALÓN DE CLM 2016',
  45             'description': 'md5:903c92fbf2b2f66c09de514bc25e9f5a',
  46             'upload_date': '20160424',
  47             'uploader': 'RTVCM Castilla-La Mancha',
  48             'uploader_id': 'RTVCM',
  49         },
  50         'add_ie': ['Youtube'],
  51     }, {
  52         'url': 'http://www.20min.ch/videotv/?cid=44&vid=468738',
  53         'only_matching': True,
  54     }, {
  55         'url': 'http://www.20min.ch/ro/sortir/cinema/story/Grandir-au-bahut--c-est-dur-18927411',
  56         'only_matching': True,
  57     }]
  58
  59     def _real_extract(self, url):
  60         mobj = re.match(self._VALID_URL, url)
  61         video_id = mobj.group('id')
  62         display_id = mobj.group('display_id') or video_id
  63
  64         webpage = self._download_webpage(url, display_id)
  65
  66         youtube_url = self._html_search_regex(
  67             r'<iframe[^>]+src="((?:https?:)?//www\.youtube\.com/embed/[^"]+)"',
  68             webpage, 'YouTube embed URL', default=None)
  69         if youtube_url is not None:
  70             return self.url_result(youtube_url, 'Youtube')
  71
  72         title = self._html_search_regex(
  73             r'<h1>.*?<span>(.+?)</span></h1>',
  74             webpage, 'title', default=None)
  75         if not title:
  76             title = remove_end(re.sub(
  77                 r'^20 [Mm]inuten.*? -', '', self._og_search_title(webpage)), ' - News')
  78
  79         if not video_id:
  80             video_id = self._search_regex(
  81                 r'"file\d?"\s*,\s*\"(\d+)', webpage, 'video id')
  82
  83         description = self._html_search_meta(
  84             'description', webpage, 'description')
  85         thumbnail = self._og_search_thumbnail(webpage)
  86
  87         return {
  88             'id': video_id,
  89             'display_id': display_id,
  90             'url': 'http://speed.20min-tv.ch/%sm.flv' % video_id,
  91             'title': title,
  92             'description': description,
  93             'thumbnail': thumbnail,
  94         }