diff options
author | Philipp Hagemeister <phihag@phihag.de> | 2013-06-23 22:30:16 +0200 |
---|---|---|
committer | Philipp Hagemeister <phihag@phihag.de> | 2013-06-23 22:30:16 +0200 |
commit | e17d368ae2473f83c4d88cdcc08f1e74af598561 (patch) | |
tree | 06fde1c69bb5731cf093a35da79dbe7f391e0a7c /youtube_dl/extractor | |
parent | 27110b0567572ea1fe1ffeb46ea07b13b76a2be3 (diff) |
[howcast] Move into own file
Diffstat (limited to 'youtube_dl/extractor')
-rw-r--r-- | youtube_dl/extractor/howcast.py | 37 |
1 files changed, 37 insertions, 0 deletions
diff --git a/youtube_dl/extractor/howcast.py b/youtube_dl/extractor/howcast.py new file mode 100644 index 000000000..7b94f85ad --- /dev/null +++ b/youtube_dl/extractor/howcast.py @@ -0,0 +1,37 @@ +import re + +from .common import InfoExtractor + + +class HowcastIE(InfoExtractor): + _VALID_URL = r'(?:https?://)?(?:www\.)?howcast\.com/videos/(?P<id>\d+)' + + def _real_extract(self, url): + mobj = re.match(self._VALID_URL, url) + + video_id = mobj.group('id') + webpage_url = 'http://www.howcast.com/videos/' + video_id + webpage = self._download_webpage(webpage_url, video_id) + + self.report_extraction(video_id) + + video_url = self._search_regex(r'\'?file\'?: "(http://mobile-media\.howcast\.com/[0-9]+\.mp4)', + webpage, u'video URL') + + video_title = self._html_search_regex(r'<meta content=(?:"([^"]+)"|\'([^\']+)\') property=\'og:title\'', + webpage, u'title') + + video_description = self._html_search_regex(r'<meta content=(?:"([^"]+)"|\'([^\']+)\') name=\'description\'', + webpage, u'description', fatal=False) + + thumbnail = self._html_search_regex(r'<meta content=\'(.+?)\' property=\'og:image\'', + webpage, u'thumbnail', fatal=False) + + return [{ + 'id': video_id, + 'url': video_url, + 'ext': 'mp4', + 'title': video_title, + 'description': video_description, + 'thumbnail': thumbnail, + }] |