_ Git - youtube-dl/blob - youtube_dl/extractor/rtlnow.py

   1 # encoding: utf-8
   2 import re
   3
   4 from .common import InfoExtractor
   5 from ..utils import ExtractorError
   6
   7 class RTLnowIE(InfoExtractor):
   8     """Information Extractor for RTL(2)now"""
   9     _VALID_URL = r'(?:http://)?(?P<url>(?P<base_url>rtl(?:(?P<is_rtl2>2)|-)now\.rtl(?(is_rtl2)2|)\.de/)[a-zA-Z0-9-]+/[a-zA-Z0-9-]+\.php\?(?:container_id|film_id)=(?P<video_id>[0-9]+)&player=1(?:&season=[0-9]+)?(?:&.*)?)'
  10     _TESTS = [{
  11         u'url': u'http://rtl-now.rtl.de/ahornallee/folge-1.php?film_id=90419&player=1&season=1',
  12         u'file': u'90419.flv',
  13         u'info_dict': {
  14             u'upload_date': u'20070416',
  15             u'title': u'Ahornallee - Folge 1 - Der Einzug',
  16             u'description': u'Folge 1 - Der Einzug',
  17         },
  18         u'params': {
  19             u'skip_download': True,
  20         },
  21     },
  22     {
  23         u'url': u'http://rtl2now.rtl2.de/aerger-im-revier/episode-15-teil-1.php?film_id=69756&player=1&season=2&index=5',
  24         u'file': u'69756.flv',
  25         u'info_dict': {
  26             u'upload_date': u'20120519',
  27             u'title': u'Ärger im Revier - Ein junger Ladendieb, ein handfester Streit...',
  28             u'description': u'Ärger im Revier - Ein junger Ladendieb, ein handfester Streit u.a.',
  29             u'thumbnail': u'http://autoimg.static-fra.de/rtl2now/219850/1500x1500/image2.jpg',
  30         },
  31         u'params': {
  32             u'skip_download': True,
  33         },
  34     },]
  35
  36     def _real_extract(self,url):
  37         mobj = re.match(self._VALID_URL, url)
  38
  39         webpage_url = u'http://' + mobj.group('url')
  40         video_page_url = u'http://' + mobj.group('base_url')
  41         video_id = mobj.group(u'video_id')
  42
  43         webpage = self._download_webpage(webpage_url, video_id)
  44         video_title = self._html_search_regex(r'<title>(?P<title>[^<]+)</title>',
  45             webpage, u'title')
  46         playerdata_url = self._html_search_regex(r'\'playerdata\': \'(?P<playerdata_url>[^\']+)\'',
  47             webpage, u'playerdata_url')
  48
  49         playerdata = self._download_webpage(playerdata_url, video_id)
  50         mobj = re.search(r'<title><!\[CDATA\[(?P<description>.+?)\s+- (?:Sendung )?vom (?P<upload_date_d>[0-9]{2})\.(?P<upload_date_m>[0-9]{2})\.(?:(?P<upload_date_Y>[0-9]{4})|(?P<upload_date_y>[0-9]{2})) [0-9]{2}:[0-9]{2} Uhr\]\]></title>', playerdata)
  51         if mobj:
  52             video_description = mobj.group(u'description')
  53             if mobj.group('upload_date_Y'):
  54                 video_upload_date = mobj.group('upload_date_Y')
  55             else:
  56                 video_upload_date = u'20' + mobj.group('upload_date_y')
  57             video_upload_date += mobj.group('upload_date_m')+mobj.group('upload_date_d')
  58         else:
  59             video_description = None
  60             video_upload_date = None
  61             self._downloader.report_warning(u'Unable to extract description and upload date')
  62
  63         # Thumbnail: not every video has an thumbnail
  64         mobj = re.search(r'<meta property="og:image" content="(?P<thumbnail>[^"]+)">', webpage)
  65         if mobj:
  66             video_thumbnail = mobj.group(u'thumbnail')
  67         else:
  68             video_thumbnail = None
  69
  70         mobj = re.search(r'<filename [^>]+><!\[CDATA\[(?P<url>rtmpe://(?:[^/]+/){2})(?P<play_path>[^\]]+)\]\]></filename>', playerdata)
  71         if mobj is None:
  72             raise ExtractorError(u'Unable to extract media URL')
  73         video_url = mobj.group(u'url')
  74         video_play_path = u'mp4:' + mobj.group(u'play_path')
  75         video_player_url = video_page_url + u'includes/vodplayer.swf'
  76
  77         return [{
  78             'id':          video_id,
  79             'url':         video_url,
  80             'play_path':   video_play_path,
  81             'page_url':    video_page_url,
  82             'player_url':  video_player_url,
  83             'ext':         'flv',
  84             'title':       video_title,
  85             'description': video_description,
  86             'upload_date': video_upload_date,
  87             'thumbnail':   video_thumbnail,
  88         }]