diff options
author | Philipp Hagemeister <phihag@phihag.de> | 2014-03-13 16:40:53 +0100 |
---|---|---|
committer | Philipp Hagemeister <phihag@phihag.de> | 2014-03-13 16:40:53 +0100 |
commit | 84769e708c81a1600ff24e7140c889fefc4aa6ef (patch) | |
tree | d3088902c608bc329c722fa8565d58d82343499a /youtube_dl/extractor | |
parent | 9d2ecdbc712d4e500dd0207041392b143082ad00 (diff) |
[ninegag] Fix extraction
Diffstat (limited to 'youtube_dl/extractor')
-rw-r--r-- | youtube_dl/extractor/ninegag.py | 32 |
1 files changed, 20 insertions, 12 deletions
diff --git a/youtube_dl/extractor/ninegag.py b/youtube_dl/extractor/ninegag.py index 2b7236be5..1d7aa40ed 100644 --- a/youtube_dl/extractor/ninegag.py +++ b/youtube_dl/extractor/ninegag.py @@ -15,7 +15,9 @@ class NineGagIE(InfoExtractor): "file": "1912.mp4", "info_dict": { "description": "This 3-minute video will make you smile and then make you feel untalented and insignificant. Anyway, you should share this awesomeness. (Thanks, Dino!)", - "title": "\"People Are Awesome 2013\" Is Absolutely Awesome" + "title": "\"People Are Awesome 2013\" Is Absolutely Awesome", + "view_count": int, + "thumbnail": "re:^https?://", }, 'add_ie': ['Youtube'] } @@ -25,21 +27,27 @@ class NineGagIE(InfoExtractor): video_id = mobj.group('id') webpage = self._download_webpage(url, video_id) - data_json = self._html_search_regex(r'''(?x) - <div\s*id="tv-video"\s*data-video-source="youtube"\s* - data-video-meta="([^"]+)"''', webpage, 'video metadata') - data = json.loads(data_json) + youtube_id = self._html_search_regex( + r'(?s)id="jsid-video-post-container".*?data-external-id="([^"]+)"', + webpage, 'video ID') + description = self._html_search_regex( + r'(?s)<div class="video-caption">.*?<p>(.*?)</p>', webpage, + 'description', fatal=False) + view_count_str = self._html_search_regex( + r'<p><b>([0-9][0-9,]*)</b> views</p>', webpage, 'view count', + fatal=False) + view_count = ( + None if view_count_str is None + else int(view_count_str.replace(',', ''))) return { '_type': 'url_transparent', - 'url': data['youtubeVideoId'], + 'url': youtube_id, 'ie_key': 'Youtube', 'id': video_id, - 'title': data['title'], - 'description': data['description'], - 'view_count': int(data['view_count']), - 'like_count': int(data['statistic']['like']), - 'dislike_count': int(data['statistic']['dislike']), - 'thumbnail': data['thumbnail_url'], + 'title': self._og_search_title(webpage), + 'description': description, + 'view_count': view_count, + 'thumbnail': self._og_search_thumbnail(webpage), } |