From c6b8f4d0c9413df9c33f3bd67859d1a9675ec8aa Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Sun, 27 Sep 2015 05:53:21 +0600 Subject: [PATCH] [condenast] Add support for JS embeds --- youtube_dl/extractor/condenast.py | 25 +++++++++++++++++++++---- 1 file changed, 21 insertions(+), 4 deletions(-) diff --git a/youtube_dl/extractor/condenast.py b/youtube_dl/extractor/condenast.py index d6949ca28..6f92ae2ed 100644 --- a/youtube_dl/extractor/condenast.py +++ b/youtube_dl/extractor/condenast.py @@ -11,6 +11,7 @@ from ..compat import ( ) from ..utils import ( orderedSet, + remove_end, ) @@ -44,12 +45,12 @@ class CondeNastIE(InfoExtractor): 'wmagazine': 'W Magazine', } - _VALID_URL = r'http://(?:video|www|player)\.(?P%s)\.com/(?Pwatch|series|video|embed)/(?P[^/?#]+)' % '|'.join(_SITES.keys()) + _VALID_URL = r'http://(?:video|www|player)\.(?P%s)\.com/(?Pwatch|series|video|embed(?:js)?)/(?P[^/?#]+)' % '|'.join(_SITES.keys()) IE_DESC = 'Condé Nast media group: %s' % ', '.join(sorted(_SITES.values())) - EMBED_URL = r'(?:https?:)?//player\.(?P%s)\.com/(?Pembed)/.+?' % '|'.join(_SITES.keys()) + EMBED_URL = r'(?:https?:)?//player\.(?P%s)\.com/(?Pembed(?:js)?)/.+?' % '|'.join(_SITES.keys()) - _TEST = { + _TESTS = [{ 'url': 'http://video.wired.com/watch/3d-printed-speakers-lit-with-led', 'md5': '1921f713ed48aabd715691f774c451f7', 'info_dict': { @@ -58,7 +59,16 @@ class CondeNastIE(InfoExtractor): 'title': '3D Printed Speakers Lit With LED', 'description': 'Check out these beautiful 3D printed LED speakers. You can\'t actually buy them, but LumiGeek is working on a board that will let you make you\'re own.', } - } + }, { + # JS embed + 'url': 'http://player.cnevids.com/embedjs/55f9cf8b61646d1acf00000c/5511d76261646d5566020000.js', + 'md5': 'f1a6f9cafb7083bab74a710f65d08999', + 'info_dict': { + 'id': '55f9cf8b61646d1acf00000c', + 'ext': 'mp4', + 'title': '3D printed TSA Travel Sentry keys really do open TSA locks', + } + }] def _extract_series(self, url, webpage): title = self._html_search_regex(r'
.*?

(.+?)

', @@ -122,6 +132,13 @@ class CondeNastIE(InfoExtractor): url_type = mobj.group('type') item_id = mobj.group('id') + # Convert JS embed to regular embed + if url_type == 'embedjs': + parsed_url = compat_urlparse.urlparse(url) + url = compat_urlparse.urlunparse(parsed_url._replace( + path=remove_end(parsed_url.path, '.js').replace('/embedjs/', '/embed/'))) + url_type = 'embed' + self.to_screen('Extracting from %s with the Condé Nast extractor' % self._SITES[site]) webpage = self._download_webpage(url, item_id)