diff options
Diffstat (limited to 'youtube_dl/extractor/generic.py')
| -rw-r--r-- | youtube_dl/extractor/generic.py | 46 |
1 files changed, 45 insertions, 1 deletions
diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index d48c84f8d..2c8fcf5ae 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -11,6 +11,8 @@ from ..utils import ( compat_urlparse, ExtractorError, + smuggle_url, + unescapeHTML, ) from .brightcove import BrightcoveIE @@ -23,12 +25,33 @@ class GenericIE(InfoExtractor): { u'url': u'http://www.hodiho.fr/2013/02/regis-plante-sa-jeep.html', u'file': u'13601338388002.mp4', - u'md5': u'85b90ccc9d73b4acd9138d3af4c27f89', + u'md5': u'6e15c93721d7ec9e9ca3fdbf07982cfd', u'info_dict': { u"uploader": u"www.hodiho.fr", u"title": u"R\u00e9gis plante sa Jeep" } }, + # embedded vimeo video + { + u'url': u'http://skillsmatter.com/podcast/home/move-semanticsperfect-forwarding-and-rvalue-references', + u'file': u'22444065.mp4', + u'md5': u'2903896e23df39722c33f015af0666e2', + u'info_dict': { + u'title': u'ACCU 2011: Move Semantics,Perfect Forwarding, and Rvalue references- Scott Meyers- 13/04/2011', + u"uploader_id": u"skillsmatter", + u"uploader": u"Skills Matter", + } + }, + # bandcamp page with custom domain + { + u'url': u'http://bronyrock.com/track/the-pony-mash', + u'file': u'3235767654.mp3', + u'info_dict': { + u'title': u'The Pony Mash', + u'uploader': u'M_Pallante', + }, + u'skip': u'There is a limit of 200 free downloads / month for the test song', + }, ] def report_download_webpage(self, video_id): @@ -127,6 +150,27 @@ class GenericIE(InfoExtractor): bc_url = BrightcoveIE._build_brighcove_url(m_brightcove.group()) return self.url_result(bc_url, 'Brightcove') + # Look for embedded Vimeo player + mobj = re.search( + r'<iframe[^>]+?src="(https?://player.vimeo.com/video/.+?)"', webpage) + if mobj: + player_url = unescapeHTML(mobj.group(1)) + surl = smuggle_url(player_url, {'Referer': url}) + return self.url_result(surl, 'Vimeo') + + # Look for embedded YouTube player + mobj = re.search( + r'<iframe[^>]+?src="(https?://(?:www\.)?youtube.com/embed/.+?)"', webpage) + if mobj: + surl = unescapeHTML(mobj.group(1)) + return self.url_result(surl, 'Youtube') + + # Look for Bandcamp pages with custom domain + mobj = re.search(r'<meta property="og:url"[^>]*?content="(.*?bandcamp\.com.*?)"', webpage) + if mobj is not None: + burl = unescapeHTML(mobj.group(1)) + return self.url_result(burl, 'Bandcamp') + # Start with something easy: JW Player in SWFObject mobj = re.search(r'flashvars: [\'"](?:.*&)?file=(http[^\'"&]*)', webpage) if mobj is None: |