From: remitamine Date: Sun, 20 Dec 2015 11:43:00 +0000 (+0100) Subject: [cnet] improve extraction X-Git-Url: http://git.oshgnacknak.de/?a=commitdiff_plain;h=c6ed6fadc248bd4957fd7cccc1eeb0f109ac3ffa;p=youtube-dl [cnet] improve extraction - relex data json regex - extract the platform metadata once - extract hds formats - extract duration - extract thumbnail --- diff --git a/youtube_dl/extractor/cnet.py b/youtube_dl/extractor/cnet.py index 3ecf0efd4..5c3908f72 100644 --- a/youtube_dl/extractor/cnet.py +++ b/youtube_dl/extractor/cnet.py @@ -1,33 +1,33 @@ # coding: utf-8 from __future__ import unicode_literals -import json - -from .common import InfoExtractor from .theplatform import ThePlatformIE +from ..utils import int_or_none -class CNETIE(InfoExtractor): +class CNETIE(ThePlatformIE): _VALID_URL = r'https?://(?:www\.)?cnet\.com/videos/(?P[^/]+)/' _TESTS = [{ 'url': 'http://www.cnet.com/videos/hands-on-with-microsofts-windows-8-1-update/', 'info_dict': { 'id': '56f4ea68-bd21-4852-b08c-4de5b8354c60', - 'ext': 'mp4', + 'ext': 'flv', 'title': 'Hands-on with Microsoft Windows 8.1 Update', 'description': 'The new update to the Windows 8 OS brings improved performance for mouse and keyboard users.', 'uploader_id': '6085384d-619e-11e3-b231-14feb5ca9861', 'uploader': 'Sarah Mitroff', + 'duration': 70, }, }, { 'url': 'http://www.cnet.com/videos/whiny-pothole-tweets-at-local-government-when-hit-by-cars-tomorrow-daily-187/', 'info_dict': { 'id': '56527b93-d25d-44e3-b738-f989ce2e49ba', - 'ext': 'mp4', + 'ext': 'flv', + 'title': 'Whiny potholes tweet at local government when hit by cars (Tomorrow Daily 187)', 'description': 'Khail and Ashley wonder what other civic woes can be solved by self-tweeting objects, investigate a new kind of VR camera and watch an origami robot self-assemble, walk, climb, dig and dissolve. #TDPothole', 'uploader_id': 'b163284d-6b73-44fc-b3e6-3da66c392d40', 'uploader': 'Ashley Esqueda', - 'title': 'Whiny potholes tweet at local government when hit by cars (Tomorrow Daily 187)', + 'duration': 1482, }, }] @@ -36,10 +36,10 @@ class CNETIE(InfoExtractor): webpage = self._download_webpage(url, display_id) data_json = self._html_search_regex( - r"