From fe7ef95e91cec1c1794692029561a68e2aaa7809 Mon Sep 17 00:00:00 2001 From: remitamine Date: Fri, 1 Apr 2016 23:53:32 +0100 Subject: [PATCH] [cbsinteractive] Add support for ZDNet videos --- youtube_dl/extractor/__init__.py | 2 +- .../extractor/{cnet.py => cbsinteractive.py} | 37 ++++++++++++++++--- 2 files changed, 32 insertions(+), 7 deletions(-) rename youtube_dl/extractor/{cnet.py => cbsinteractive.py} (69%) diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py index 438e1cc639..ebf5ca3e93 100644 --- a/youtube_dl/extractor/__init__.py +++ b/youtube_dl/extractor/__init__.py @@ -104,6 +104,7 @@ from .cbc import ( CBCPlayerIE, ) from .cbs import CBSIE +from .cbsinteractive import CBSInteractiveIE from .cbsnews import ( CBSNewsIE, CBSNewsLiveVideoIE, @@ -129,7 +130,6 @@ from .clubic import ClubicIE from .clyp import ClypIE from .cmt import CMTIE from .cnbc import CNBCIE -from .cnet import CNETIE from .cnn import ( CNNIE, CNNBlogsIE, diff --git a/youtube_dl/extractor/cnet.py b/youtube_dl/extractor/cbsinteractive.py similarity index 69% rename from youtube_dl/extractor/cnet.py rename to youtube_dl/extractor/cbsinteractive.py index 58c26f20fe..0011c30296 100644 --- a/youtube_dl/extractor/cnet.py +++ b/youtube_dl/extractor/cbsinteractive.py @@ -1,12 +1,14 @@ # coding: utf-8 from __future__ import unicode_literals +import re + from .theplatform import ThePlatformIE from ..utils import int_or_none -class CNETIE(ThePlatformIE): - _VALID_URL = r'https?://(?:www\.)?cnet\.com/videos/(?P[^/]+)/' +class CBSInteractiveIE(ThePlatformIE): + _VALID_URL = r'https?://(?:www\.)?(?Pcnet|zdnet)\.com/(?:videos|video/share)/(?P[^/?]+)' _TESTS = [{ 'url': 'http://www.cnet.com/videos/hands-on-with-microsofts-windows-8-1-update/', 'info_dict': { @@ -33,15 +35,35 @@ class CNETIE(ThePlatformIE): 'timestamp': 1433289889, 'upload_date': '20150603', }, + }, { + 'url': 'http://www.zdnet.com/video/share/video-keeping-android-smartphones-and-tablets-secure/', + 'info_dict': { + 'id': 'bc1af9f0-a2b5-4e54-880d-0d95525781c0', + 'ext': 'mp4', + 'title': 'Video: Keeping Android smartphones and tablets secure', + 'description': 'Here\'s the best way to keep Android devices secure, and what you do when they\'ve come to the end of their lives.', + 'uploader_id': 'f2d97ea2-8175-11e2-9d12-0018fe8a00b0', + 'uploader': 'Adrian Kingsley-Hughes', + 'timestamp': 1448961720, + 'upload_date': '20151201', + }, + 'params': { + # m3u8 download + 'skip_download': True, + } }] TP_RELEASE_URL_TEMPLATE = 'http://link.theplatform.com/s/kYEXFC/%s?mbr=true' + MPX_ACCOUNTS = { + 'cnet': 2288573011, + 'zdnet': 2387448114, + } def _real_extract(self, url): - display_id = self._match_id(url) + site, display_id = re.match(self._VALID_URL, url).groups() webpage = self._download_webpage(url, display_id) data_json = self._html_search_regex( - r"data-cnet-video(?:-uvp)?-options='([^']+)'", + r"data-(?:cnet|zdnet)-video(?:-uvp)?-options='([^']+)'", webpage, 'data json') data = self._parse_json(data_json, display_id) vdata = data.get('video') or data['videos'][0] @@ -56,8 +78,11 @@ class CNETIE(ThePlatformIE): uploader = None uploader_id = None - media_guid_path = 'media/guid/2288573011/%s' % vdata['mpxRefId'] - formats, subtitles = self._extract_theplatform_smil(self.TP_RELEASE_URL_TEMPLATE % media_guid_path, video_id) + media_guid_path = 'media/guid/%d/%s' % (self.MPX_ACCOUNTS[site], vdata['mpxRefId']) + formats, subtitles = [], {} + if site == 'cnet': + formats, subtitles = self._extract_theplatform_smil( + self.TP_RELEASE_URL_TEMPLATE % media_guid_path, video_id) for (fkey, vid) in vdata['files'].items(): if fkey == 'hls_phone' and 'hls_tablet' in vdata['files']: continue