youtube_dl/extractor/helsinki.py

   1 # -*- coding: utf-8 -*-
   2
   3 from __future__ import unicode_literals
   4
   5 import re
   6
   7 from .common import InfoExtractor
   8
   9
  10 class HelsinkiIE(InfoExtractor):
  11     IE_DESC = 'helsinki.fi'
  12     _VALID_URL = r'https?://video\.helsinki\.fi/Arkisto/flash\.php\?id=(?P<id>\d+)'
  13     _TEST = {
  14         'url': 'http://video.helsinki.fi/Arkisto/flash.php?id=20258',
  15         'info_dict': {
  16             'id': '20258',
  17             'ext': 'mp4',
  18             'title': 'Tietotekniikkafoorumi-iltapäivä',
  19             'description': 'md5:f5c904224d43c133225130fe156a5ee0',
  20         },
  21         'params': {
  22             'skip_download': True,  # RTMP
  23         }
  24     }
  25
  26     def _real_extract(self, url):
  27         mobj = re.match(self._VALID_URL, url)
  28         video_id = mobj.group('id')
  29         webpage = self._download_webpage(url, video_id)
  30         formats = []
  31
  32         mobj = re.search(r'file=((\w+):[^&]+)', webpage)
  33         if mobj:
  34             formats.append({
  35                 'ext': mobj.group(2),
  36                 'play_path': mobj.group(1),
  37                 'url': 'rtmp://flashvideo.it.helsinki.fi/vod/',
  38                 'player_url': 'http://video.helsinki.fi/player.swf',
  39                 'format_note': 'sd',
  40                 'quality': 0,
  41             })
  42
  43         mobj = re.search(r'hd\.file=((\w+):[^&]+)', webpage)
  44         if mobj:
  45             formats.append({
  46                 'ext': mobj.group(2),
  47                 'play_path': mobj.group(1),
  48                 'url': 'rtmp://flashvideo.it.helsinki.fi/vod/',
  49                 'player_url': 'http://video.helsinki.fi/player.swf',
  50                 'format_note': 'hd',
  51                 'quality': 1,
  52             })
  53
  54         self._sort_formats(formats)
  55
  56         return {
  57             'id': video_id,
  58             'title': self._og_search_title(webpage).replace('Video: ', ''),
  59             'description': self._og_search_description(webpage),
  60             'thumbnail': self._og_search_thumbnail(webpage),
  61             'formats': formats,
  62         }