diff options
author | Sergey M․ <dstftw@gmail.com> | 2015-04-23 21:44:51 +0600 |
---|---|---|
committer | Sergey M․ <dstftw@gmail.com> | 2015-04-23 21:44:51 +0600 |
commit | b46ed49996669a5e602042ae4d357f2ad952af58 (patch) | |
tree | 608301e6109e824dd43bd25ab45beab898258bfe /youtube_dl | |
parent | cd9fdccde04799910fc3ec36fb1e8baa26a5b3e0 (diff) |
[cracked] Fix extraction
Diffstat (limited to 'youtube_dl')
-rw-r--r-- | youtube_dl/extractor/cracked.py | 18 |
1 files changed, 13 insertions, 5 deletions
diff --git a/youtube_dl/extractor/cracked.py b/youtube_dl/extractor/cracked.py index cf763ee7e..e579863df 100644 --- a/youtube_dl/extractor/cracked.py +++ b/youtube_dl/extractor/cracked.py @@ -33,17 +33,25 @@ class CrackedIE(InfoExtractor): video_url = self._html_search_regex( [r'var\s+CK_vidSrc\s*=\s*"([^"]+)"', r'<video\s+src="([^"]+)"'], webpage, 'video URL') - title = self._og_search_title(webpage) - description = self._og_search_description(webpage) + title = self._search_regex( + [r'property="?og:title"?\s+content="([^"]+)"', r'class="?title"?>([^<]+)'], + webpage, 'title') - timestamp = self._html_search_regex(r'<time datetime="([^"]+)"', webpage, 'upload date', fatal=False) + description = self._search_regex( + r'name="?(?:og:)?description"?\s+content="([^"]+)"', + webpage, 'description', default=None) + + timestamp = self._html_search_regex( + r'"date"\s*:\s*"([^"]+)"', webpage, 'upload date', fatal=False) if timestamp: timestamp = parse_iso8601(timestamp[:-6]) view_count = str_to_int(self._html_search_regex( - r'<span class="views" id="viewCounts">([\d,\.]+) Views</span>', webpage, 'view count', fatal=False)) + r'<span\s+class="?views"? id="?viewCounts"?>([\d,\.]+) Views</span>', + webpage, 'view count', fatal=False)) comment_count = str_to_int(self._html_search_regex( - r'<span id="commentCounts">([\d,\.]+)</span>', webpage, 'comment count', fatal=False)) + r'<span\s+id="?commentCounts"?>([\d,\.]+)</span>', + webpage, 'comment count', fatal=False)) m = re.search(r'_(?P<width>\d+)X(?P<height>\d+)\.mp4$', video_url) if m: |