2014-04-03 22:21:21 +08:00
# coding: utf-8
from __future__ import unicode_literals
import json
from . common import InfoExtractor
from . . utils import (
2014-04-04 08:09:04 +08:00
ExtractorError ,
2014-04-03 22:21:21 +08:00
)
class CNETIE ( InfoExtractor ) :
_VALID_URL = r ' https?://(?:www \ .)?cnet \ .com/videos/(?P<id>[^/]+)/ '
2015-06-08 13:32:04 +08:00
_TESTS = [ {
2014-04-03 22:21:21 +08:00
' url ' : ' http://www.cnet.com/videos/hands-on-with-microsofts-windows-8-1-update/ ' ,
' info_dict ' : {
' id ' : ' 56f4ea68-bd21-4852-b08c-4de5b8354c60 ' ,
2014-12-12 22:55:55 +08:00
' ext ' : ' flv ' ,
2014-04-03 22:21:21 +08:00
' 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. ' ,
' thumbnail ' : ' re:^http://.*/flmswindows8.jpg$ ' ,
2014-12-12 22:55:55 +08:00
' uploader_id ' : ' 6085384d-619e-11e3-b231-14feb5ca9861 ' ,
2014-04-03 22:21:21 +08:00
' uploader ' : ' Sarah Mitroff ' ,
2014-12-12 22:55:55 +08:00
} ,
' params ' : {
' skip_download ' : ' requires rtmpdump ' ,
2014-04-03 22:21:21 +08:00
}
2015-06-08 13:32:04 +08:00
} , {
' 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 ' : ' flv ' ,
' 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) ' ,
} ,
' params ' : {
' skip_download ' : True , # requires rtmpdump
} ,
} ]
2014-04-03 22:21:21 +08:00
def _real_extract ( self , url ) :
2014-12-12 22:55:55 +08:00
display_id = self . _match_id ( url )
2014-04-03 22:21:21 +08:00
webpage = self . _download_webpage ( url , display_id )
2014-12-12 22:55:55 +08:00
2014-04-03 22:21:21 +08:00
data_json = self . _html_search_regex (
2014-04-21 13:59:29 +08:00
r " <div class= \" cnetVideoPlayer \" \ s+.*?data-cnet-video-options= ' ([^ ' ]+) ' " ,
2014-04-03 22:21:21 +08:00
webpage , ' data json ' )
data = json . loads ( data_json )
vdata = data [ ' video ' ]
2014-04-04 08:09:04 +08:00
if not vdata :
vdata = data [ ' videos ' ] [ 0 ]
if not vdata :
raise ExtractorError ( ' Cannot find video data ' )
2014-04-03 22:21:21 +08:00
2014-12-12 22:55:55 +08:00
mpx_account = data [ ' config ' ] [ ' players ' ] [ ' default ' ] [ ' mpx_account ' ]
2015-06-08 13:32:04 +08:00
vid = vdata [ ' files ' ] . get ( ' rtmp ' , vdata [ ' files ' ] [ ' hds ' ] )
2014-12-12 22:55:55 +08:00
tp_link = ' http://link.theplatform.com/s/ %s / %s ' % ( mpx_account , vid )
2014-04-03 22:21:21 +08:00
video_id = vdata [ ' id ' ]
2014-07-21 19:03:18 +08:00
title = vdata . get ( ' headline ' )
if title is None :
title = vdata . get ( ' title ' )
if title is None :
raise ExtractorError ( ' Cannot find title! ' )
2014-04-03 22:21:21 +08:00
thumbnail = vdata . get ( ' image ' , { } ) . get ( ' path ' )
author = vdata . get ( ' author ' )
if author :
uploader = ' %s %s ' % ( author [ ' firstName ' ] , author [ ' lastName ' ] )
2014-12-12 22:55:55 +08:00
uploader_id = author . get ( ' id ' )
2014-04-03 22:21:21 +08:00
else :
uploader = None
uploader_id = None
return {
2014-12-12 22:55:55 +08:00
' _type ' : ' url_transparent ' ,
' url ' : tp_link ,
2014-04-03 22:21:21 +08:00
' id ' : video_id ,
' display_id ' : display_id ,
' title ' : title ,
' uploader ' : uploader ,
' uploader_id ' : uploader_id ,
' thumbnail ' : thumbnail ,
}