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

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     unified_strdate,
   8     compat_urlparse,
   9     determine_ext,
  10 )
  11
  12
  13 class WDRIE(InfoExtractor):
  14     _PLAYER_REGEX = '-(?:video|audio)player(?:_size-[LMS])?'
  15     _VALID_URL = r'(?P<url>https?://www\d?\.(?:wdr\d?|funkhauseuropa)\.de/)(?P<id>.+?)(?P<player>%s)?\.html' % _PLAYER_REGEX
  16
  17     _TESTS = [
  18         {
  19             'url': 'http://www1.wdr.de/mediathek/video/sendungen/servicezeit/videoservicezeit560-videoplayer_size-L.html',
  20             'info_dict': {
  21                 'id': 'mdb-362427',
  22                 'ext': 'flv',
  23                 'title': 'Servicezeit',
  24                 'description': 'md5:c8f43e5e815eeb54d0b96df2fba906cb',
  25                 'upload_date': '20140310',
  26             },
  27             'params': {
  28                 'skip_download': True,
  29             },
  30         },
  31         {
  32             'url': 'http://www1.wdr.de/themen/av/videomargaspiegelisttot101-videoplayer.html',
  33             'info_dict': {
  34                 'id': 'mdb-363194',
  35                 'ext': 'flv',
  36                 'title': 'Marga Spiegel ist tot',
  37                 'upload_date': '20140311',
  38             },
  39             'params': {
  40                 'skip_download': True,
  41             },
  42         },
  43         {
  44             'url': 'http://www1.wdr.de/themen/kultur/audioerlebtegeschichtenmargaspiegel100-audioplayer.html',
  45             'md5': '83e9e8fefad36f357278759870805898',
  46             'info_dict': {
  47                 'id': 'mdb-194332',
  48                 'ext': 'mp3',
  49                 'title': 'Erlebte Geschichten: Marga Spiegel (29.11.2009)',
  50                 'upload_date': '20091129',
  51             },
  52         },
  53         {
  54             'url': 'http://www.funkhauseuropa.de/av/audiogrenzenlosleckerbaklava101-audioplayer.html',
  55             'md5': 'cfff440d4ee64114083ac44676df5d15',
  56             'info_dict': {
  57                 'id': 'mdb-363068',
  58                 'ext': 'mp3',
  59                 'title': 'Grenzenlos lecker - Baklava',
  60                 'upload_date': '20140311',
  61             },
  62         },
  63     ]
  64
  65     def _real_extract(self, url):
  66         mobj = re.match(self._VALID_URL, url)
  67         page_url = mobj.group('url')
  68         page_id = mobj.group('id')
  69
  70         webpage = self._download_webpage(url, page_id)
  71
  72         if mobj.group('player') is None:
  73             entries = [
  74                 self.url_result(page_url + href, 'WDR')
  75                 for href in re.findall(r'<a href="/?(.+?%s\.html)" rel="nofollow"' % self._PLAYER_REGEX, webpage)
  76             ]
  77             return self.playlist_result(entries, page_id)
  78
  79         flashvars = compat_urlparse.parse_qs(
  80             self._html_search_regex(r'<param name="flashvars" value="([^"]+)"', webpage, 'flashvars'))
  81
  82         page_id = flashvars['trackerClipId'][0]
  83         video_url = flashvars['dslSrc'][0]
  84         title = flashvars['trackerClipTitle'][0]
  85         thumbnail = flashvars['startPicture'][0] if 'startPicture' in flashvars else None
  86
  87         if 'trackerClipAirTime' in flashvars:
  88             upload_date = flashvars['trackerClipAirTime'][0]
  89         else:
  90             upload_date = self._html_search_meta('DC.Date', webpage, 'upload date')
  91
  92         if upload_date:
  93             upload_date = unified_strdate(upload_date)
  94
  95         if video_url.endswith('.f4m'):
  96             video_url += '?hdcore=3.2.0&plugin=aasp-3.2.0.77.18'
  97             ext = 'flv'
  98         else:
  99             ext = determine_ext(video_url)
 100
 101         description = self._html_search_meta('Description', webpage, 'description')
 102
 103         return {
 104             'id': page_id,
 105             'url': video_url,
 106             'ext': ext,
 107             'title': title,
 108             'description': description,
 109             'thumbnail': thumbnail,
 110             'upload_date': upload_date,
 111         }