diff options
| author | vobe <vobe@jackal.fr> | 2017-05-19 20:18:03 +0200 |
|---|---|---|
| committer | Sergey M <dstftw@gmail.com> | 2017-05-20 01:18:03 +0700 |
| commit | 77d682da9dc2e62a1505cccb77408b8fa638a7f1 (patch) | |
| tree | e294d3c4d50816269343e0af0d346f43c9427aac | |
| parent | 8fffac6927b4144bfb3e473b58fa52313042ecbb (diff) | |
| download | youtube-dl-77d682da9dc2e62a1505cccb77408b8fa638a7f1.tar.gz youtube-dl-77d682da9dc2e62a1505cccb77408b8fa638a7f1.tar.xz youtube-dl-77d682da9dc2e62a1505cccb77408b8fa638a7f1.zip | |
[toypics] Fix extraction
| -rw-r--r-- | youtube_dl/extractor/toypics.py | 15 |
1 files changed, 8 insertions, 7 deletions
diff --git a/youtube_dl/extractor/toypics.py b/youtube_dl/extractor/toypics.py index 938e05076..774fa565c 100644 --- a/youtube_dl/extractor/toypics.py +++ b/youtube_dl/extractor/toypics.py @@ -24,15 +24,16 @@ class ToypicsIE(InfoExtractor): mobj = re.match(self._VALID_URL, url) video_id = mobj.group('id') page = self._download_webpage(url, video_id) - video_url = self._html_search_regex( - r'src:\s+"(http://static[0-9]+\.toypics\.net/flvideo/[^"]+)"', page, 'video URL') - title = self._html_search_regex( - r'<title>Toypics - ([^<]+)</title>', page, 'title') + formats = self._parse_html5_media_entries(url, page, video_id)[0]['formats'] + title = self._html_search_regex([ + r'<h1[^>]+class=["\']view-video-title[^>]+>([^<]+)</h', + r'<title>([^<]+) - Toypics</title>', + ], page, 'title') username = self._html_search_regex( - r'toypics.net/([^/"]+)" class="user-name">', page, 'username') + r'More videos from <strong>([^<]+)</strong>', page, 'username') return { 'id': video_id, - 'url': video_url, + 'formats': formats, 'title': title, 'uploader': username, 'age_limit': 18, @@ -71,7 +72,7 @@ class ToypicsUserIE(InfoExtractor): note='Downloading page %d/%d' % (n, page_count)) urls.extend( re.findall( - r'<p class="video-entry-title">\s+<a href="(https?://videos.toypics.net/view/[^"]+)">', + r'<div[^>]+class=["\']preview[^>]+>\s*<a[^>]+href="(https?://videos.toypics.net/view/[^"]+)"', lpage)) return { |