From ca97a56e4bd920971e02a55ee5d42fe83e25b4f9 Mon Sep 17 00:00:00 2001 From: "Sergey M." Date: Fri, 28 Feb 2014 23:51:54 +0700 Subject: [PATCH] [vk] Add support for embedded videos (Closes #2473) --- youtube_dl/extractor/generic.py | 5 +++++ youtube_dl/extractor/vk.py | 21 ++++++++++++++++++--- 2 files changed, 23 insertions(+), 3 deletions(-) diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index 595ba268f..641d9babb 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -396,6 +396,11 @@ def _real_extract(self, url): if mobj is not None: return self.url_result(mobj.group('url'), 'Facebook') + # Look for embedded VK player + mobj = re.search(r']+?src=(["\'])(?Phttps?://vk\.com/video_ext\.php.+?)\1', webpage) + if mobj is not None: + return self.url_result(mobj.group('url'), 'VK') + # Look for embedded Huffington Post player mobj = re.search( r']+?src=(["\'])(?Phttps?://embed\.live\.huffingtonpost\.com/.+?)\1', webpage) diff --git a/youtube_dl/extractor/vk.py b/youtube_dl/extractor/vk.py index a293b8875..3b3bec92f 100644 --- a/youtube_dl/extractor/vk.py +++ b/youtube_dl/extractor/vk.py @@ -16,7 +16,7 @@ class VKIE(InfoExtractor): IE_NAME = 'vk.com' - _VALID_URL = r'https?://vk\.com/(?:videos.*?\?.*?z=)?video(?P.*?)(?:\?|%2F|$)' + _VALID_URL = r'https?://vk\.com/(?:video_ext\.php\?.*?\boid=(?P\d+).*?\bid=(?P\d+)|(?:videos.*?\?.*?z=)?video(?P.*?)(?:\?|%2F|$))' _NETRC_MACHINE = 'vk' _TESTS = [ @@ -42,6 +42,18 @@ class VKIE(InfoExtractor): 'duration': 558, } }, + { + 'note': 'Embedded video', + 'url': 'http://vk.com/video_ext.php?oid=32194266&id=162925554&hash=7d8c2e0d5e05aeaa&hd=1', + 'md5': 'c7ce8f1f87bec05b3de07fdeafe21a0a', + 'info_dict': { + 'id': '162925554', + 'ext': 'mp4', + 'uploader': 'Vladimir Gavrin', + 'title': 'Lin Dan', + 'duration': 101, + } + }, { 'url': 'http://vk.com/video-8871596_164049491', 'md5': 'a590bcaf3d543576c9bd162812387666', @@ -54,7 +66,7 @@ class VKIE(InfoExtractor): 'duration': 8352, }, 'skip': 'Requires vk account credentials', - } + }, ] def _login(self): @@ -82,7 +94,10 @@ def _real_initialize(self): def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('id') + video_id = mobj.group('videoid') + + if not video_id: + video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id')) info_url = 'http://vk.com/al_video.php?act=show&al=1&video=%s' % video_id info_page = self._download_webpage(info_url, video_id)