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

   1 import re
   2 from ..utils import compat_urllib_parse
   3 from .common import InfoExtractor
   4
   5
   6 class EhowIE(InfoExtractor):
   7     _VALID_URL = r'(?:http://)?(?:www\.)?ehow\.com/([^/]+)'
   8     _TEST = {
   9         u'url': u'http://www.ehow.com/video_12245069_hardwood-flooring-basics.html',
  10         u'file': u'12245069.flv',
  11         u'md5': u'9809b4e3f115ae2088440bcb4efbf371',
  12         u'info_dict': {
  13             u"title": u"Hardwood Flooring Basics",
  14             u"description": u"Hardwood flooring may be time consuming, but its ultimately a pretty straightforward concept. Learn about hardwood flooring basics with help from a hardware flooring business owner in this free video...",
  15                         u"uploader": u"Erick Nathan"
  16         }
  17     }
  18
  19     def _real_extract(self, url):
  20         mobj = re.match(self._VALID_URL, url)
  21         video_id = mobj.group(1).split("_")[1]
  22         webpage = self._download_webpage(url, video_id)
  23         video_url = self._search_regex(r'[^A-Za-z0-9]?(?:file|source)=(http[^\'"&]*)',
  24             webpage, u'video URL')
  25         final_url = compat_urllib_parse.unquote(video_url)
  26         thumbnail_url = self._search_regex(r'<meta property="og:image" content="(.+?)" />',
  27             webpage, u'thumbnail URL')
  28         uploader = self._search_regex(r'<meta name="uploader" content="(.+?)" />',
  29             webpage, u'uploader')
  30         title = self._search_regex(r'<meta property="og:title" content="(.+?)" />',
  31             webpage, u'Video title').replace(' | eHow','')
  32         description = self._search_regex(r'<meta property="og:description" content="(.+?)" />',
  33             webpage, u'video description')
  34         ext = final_url.split('.')[-1]
  35         return [{
  36             'id':          video_id,
  37             'url':         final_url,
  38             'ext':         ext,
  39             'title':       title,
  40             'thumbnail':   thumbnail_url,
  41             'description': description,
  42             'uploader':    uploader,
  43         }]
  44