diff options
Diffstat (limited to 'youtube_dl/extractor')
-rw-r--r-- | youtube_dl/extractor/howcast.py | 40 |
1 files changed, 18 insertions, 22 deletions
diff --git a/youtube_dl/extractor/howcast.py b/youtube_dl/extractor/howcast.py index bafc5826f..6ae04782c 100644 --- a/youtube_dl/extractor/howcast.py +++ b/youtube_dl/extractor/howcast.py @@ -1,17 +1,20 @@ +from __future__ import unicode_literals + import re from .common import InfoExtractor class HowcastIE(InfoExtractor): - _VALID_URL = r'(?:https?://)?(?:www\.)?howcast\.com/videos/(?P<id>\d+)' + _VALID_URL = r'https?://(?:www\.)?howcast\.com/videos/(?P<id>\d+)' _TEST = { - u'url': u'http://www.howcast.com/videos/390161-How-to-Tie-a-Square-Knot-Properly', - u'file': u'390161.mp4', - u'md5': u'8b743df908c42f60cf6496586c7f12c3', - u'info_dict': { - u"description": u"The square knot, also known as the reef knot, is one of the oldest, most basic knots to tie, and can be used in many different ways. Here's the proper way to tie a square knot.", - u"title": u"How to Tie a Square Knot Properly" + 'url': 'http://www.howcast.com/videos/390161-How-to-Tie-a-Square-Knot-Properly', + 'md5': '8b743df908c42f60cf6496586c7f12c3', + 'info_dict': { + 'id': '390161', + 'ext': 'mp4', + 'description': 'The square knot, also known as the reef knot, is one of the oldest, most basic knots to tie, and can be used in many different ways. Here\'s the proper way to tie a square knot.', + 'title': 'How to Tie a Square Knot Properly', } } @@ -24,22 +27,15 @@ class HowcastIE(InfoExtractor): 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') + webpage, 'video URL') video_description = self._html_search_regex(r'<meta content=(?:"([^"]+)"|\'([^\']+)\') name=\'description\'', - webpage, u'description', fatal=False) + webpage, '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, + return { + 'id': video_id, + 'url': video_url, + 'title': self._og_search_title(webpage), 'description': video_description, - 'thumbnail': thumbnail, - }] + 'thumbnail': self._og_search_thumbnail(webpage), + } |