diff options
author | Sergey M․ <dstftw@gmail.com> | 2015-05-22 00:45:33 +0600 |
---|---|---|
committer | Sergey M․ <dstftw@gmail.com> | 2015-05-22 00:45:33 +0600 |
commit | 77d9cb2f04462677f2a36f487c20e7a7992a0a32 (patch) | |
tree | 968a4d9a4bfc4977a70cf8ec158c170595c7bbf8 | |
parent | 0459432d962bf358566340eed00f6c1c56b7b732 (diff) |
[sportbox] Fix extraction
-rw-r--r-- | youtube_dl/extractor/sportbox.py | 25 |
1 files changed, 9 insertions, 16 deletions
diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py index 8686f9d11..2ab3489e4 100644 --- a/youtube_dl/extractor/sportbox.py +++ b/youtube_dl/extractor/sportbox.py @@ -7,7 +7,7 @@ from .common import InfoExtractor from ..compat import compat_urlparse from ..utils import ( parse_duration, - parse_iso8601, + unified_strdate, ) @@ -20,11 +20,9 @@ class SportBoxIE(InfoExtractor): 'id': '80822', 'ext': 'mp4', 'title': 'Гонка 2 заезд ««Объединенный 2000»: классы Туринг и Супер-продакшн', - 'description': 'md5:81715fa9c4ea3d9e7915dc8180c778ed', + 'description': 'md5:3d72dc4a006ab6805d82f037fdc637ad', 'thumbnail': 're:^https?://.*\.jpg$', - 'timestamp': 1411896237, 'upload_date': '20140928', - 'duration': 4846, }, 'params': { # m3u8 download @@ -48,17 +46,13 @@ class SportBoxIE(InfoExtractor): r'src="/?(vdl/player/[^"]+)"', webpage, 'player') title = self._html_search_regex( - r'<h1 itemprop="name">([^<]+)</h1>', webpage, 'title') - description = self._html_search_regex( - r'(?s)<div itemprop="description">(.+?)</div>', - webpage, 'description', fatal=False) + [r'"nodetitle"\s*:\s*"([^"]+)"', r'class="node-header_{1,2}title">([^<]+)'], + webpage, 'title') + description = self._og_search_description(webpage) or self._html_search_meta( + 'description', webpage, 'description') thumbnail = self._og_search_thumbnail(webpage) - timestamp = parse_iso8601(self._search_regex( - r'<span itemprop="uploadDate">([^<]+)</span>', - webpage, 'timestamp', fatal=False)) - duration = parse_duration(self._html_search_regex( - r'<meta itemprop="duration" content="PT([^"]+)">', - webpage, 'duration', fatal=False)) + upload_date = unified_strdate(self._html_search_meta( + 'dateCreated', webpage, 'upload date')) return { '_type': 'url_transparent', @@ -67,8 +61,7 @@ class SportBoxIE(InfoExtractor): 'title': title, 'description': description, 'thumbnail': thumbnail, - 'timestamp': timestamp, - 'duration': duration, + 'upload_date': upload_date, } |