2013-08-27 11:56:48 +02:00
from . common import InfoExtractor
2016-08-28 17:43:15 +02:00
from . turner import TurnerBaseIE
2022-10-04 05:09:23 +02:00
from . . utils import merge_dicts , try_call , url_basename
2013-08-27 11:56:48 +02:00
2013-08-28 00:14:19 +02:00
2016-08-28 17:43:15 +02:00
class CNNIE ( TurnerBaseIE ) :
2016-08-20 20:00:25 +02:00
_VALID_URL = r ''' (?x)https?://(?:(?P<sub_domain>edition|www|money) \ .)?cnn \ .com/(?:video/(?:data/.+?| \ ?)/)?videos?/
2015-05-23 20:04:02 +02:00
( ? P < path > . + ? / ( ? P < title > [ ^ / ] + ? ) ( ? : \. ( ? : [ a - z \- ] + ) | ( ? = & ) ) ) '''
2013-08-27 11:56:48 +02:00
2013-08-28 00:14:19 +02:00
_TESTS = [ {
2014-01-07 10:04:48 +01:00
' url ' : ' http://edition.cnn.com/video/?/video/sports/2013/06/09/nadal-1-on-1.cnn ' ,
' md5 ' : ' 3e6121ea48df7e2259fe73a0628605c4 ' ,
' info_dict ' : {
2016-08-30 11:46:48 +02:00
' id ' : ' sports/2013/06/09/nadal-1-on-1.cnn ' ,
2014-11-04 22:25:08 +01:00
' ext ' : ' mp4 ' ,
2014-01-07 10:04:48 +01:00
' title ' : ' Nadal wins 8th French Open title ' ,
' description ' : ' World Sport \' s Amanda Davies chats with 2013 French Open champion Rafael Nadal. ' ,
' duration ' : 135 ,
' upload_date ' : ' 20130609 ' ,
2013-08-27 11:56:48 +02:00
} ,
2016-08-28 17:43:15 +02:00
' expected_warnings ' : [ ' Failed to download m3u8 information ' ] ,
2014-11-23 21:39:15 +01:00
} , {
2016-02-14 10:37:17 +01:00
' url ' : ' http://edition.cnn.com/video/?/video/us/2013/08/21/sot-student-gives-epic-speech.georgia-institute-of-technology&utm_source=feedburner&utm_medium=feed&utm_campaign=Feed % 3A+rss %2F cnn_topstories+ % 28RSS % 3A+Top+Stories % 29 ' ,
' md5 ' : ' b5cc60c60a3477d185af8f19a2a26f4e ' ,
' info_dict ' : {
2016-08-30 11:46:48 +02:00
' id ' : ' us/2013/08/21/sot-student-gives-epic-speech.georgia-institute-of-technology ' ,
2014-11-04 22:25:08 +01:00
' ext ' : ' mp4 ' ,
2016-02-14 10:37:17 +01:00
' title ' : " Student ' s epic speech stuns new freshmen " ,
2024-06-12 01:09:58 +02:00
' description ' : ' A Georgia Tech student welcomes the incoming freshmen with an epic speech backed by music from " 2001: A Space Odyssey. " ' ,
2016-02-14 10:37:17 +01:00
' upload_date ' : ' 20130821 ' ,
2016-08-28 17:43:15 +02:00
} ,
' expected_warnings ' : [ ' Failed to download m3u8 information ' ] ,
2014-12-29 20:50:28 +01:00
} , {
' url ' : ' http://www.cnn.com/video/data/2.0/video/living/2014/12/22/growing-america-nashville-salemtown-board-episode-1.hln.html ' ,
' md5 ' : ' f14d02ebd264df951feb2400e2c25a1b ' ,
' info_dict ' : {
2016-08-30 11:46:48 +02:00
' id ' : ' living/2014/12/22/growing-america-nashville-salemtown-board-episode-1.hln ' ,
2014-12-29 20:50:28 +01:00
' ext ' : ' mp4 ' ,
' title ' : ' Nashville Ep. 1: Hand crafted skateboards ' ,
' description ' : ' md5:e7223a503315c9f150acac52e76de086 ' ,
' upload_date ' : ' 20141222 ' ,
2016-08-28 17:43:15 +02:00
} ,
' expected_warnings ' : [ ' Failed to download m3u8 information ' ] ,
2016-08-20 20:00:25 +02:00
} , {
' url ' : ' http://money.cnn.com/video/news/2016/08/19/netflix-stunning-stats.cnnmoney/index.html ' ,
' md5 ' : ' 52a515dc1b0f001cd82e4ceda32be9d1 ' ,
' info_dict ' : {
2016-08-30 11:46:48 +02:00
' id ' : ' /video/news/2016/08/19/netflix-stunning-stats.cnnmoney ' ,
2016-08-20 20:00:25 +02:00
' ext ' : ' mp4 ' ,
' title ' : ' 5 stunning stats about Netflix ' ,
' description ' : ' Did you know that Netflix has more than 80 million members? Here are five facts about the online video distributor that you probably didn \' t know. ' ,
' upload_date ' : ' 20160819 ' ,
2016-08-28 17:43:15 +02:00
} ,
' params ' : {
# m3u8 download
' skip_download ' : True ,
} ,
2015-03-28 23:39:41 +01:00
} , {
' url ' : ' http://cnn.com/video/?/video/politics/2015/03/27/pkg-arizona-senator-church-attendance-mandatory.ktvk ' ,
' only_matching ' : True ,
2015-04-07 14:59:13 +02:00
} , {
' url ' : ' http://cnn.com/video/?/video/us/2015/04/06/dnt-baker-refuses-anti-gay-order.wkmg ' ,
' only_matching ' : True ,
2016-08-20 20:00:25 +02:00
} , {
' url ' : ' http://edition.cnn.com/videos/arts/2016/04/21/olympic-games-cultural-a-z-brazil.cnn ' ,
' only_matching ' : True ,
2013-08-28 00:14:19 +02:00
} ]
2013-08-27 11:56:48 +02:00
2016-08-20 20:00:25 +02:00
_CONFIG = {
# http://edition.cnn.com/.element/apps/cvp/3.0/cfg/spider/cnn/expansion/config.xml
' edition ' : {
' data_src ' : ' http://edition.cnn.com/video/data/3.0/video/ %s /index.xml ' ,
' media_src ' : ' http://pmd.cdn.turner.com/cnn/big ' ,
} ,
# http://money.cnn.com/.element/apps/cvp2/cfg/config.xml
' money ' : {
' data_src ' : ' http://money.cnn.com/video/data/4.0/video/ %s .xml ' ,
' media_src ' : ' http://ht3.cdn.turner.com/money/big ' ,
} ,
}
2016-08-29 20:26:53 +02:00
def _extract_timestamp ( self , video_data ) :
# TODO: fix timestamp extraction
return None
2013-08-27 11:56:48 +02:00
def _real_extract ( self , url ) :
2021-08-19 03:41:24 +02:00
sub_domain , path , page_title = self . _match_valid_url ( url ) . groups ( )
2016-08-20 20:00:25 +02:00
if sub_domain not in ( ' money ' , ' edition ' ) :
sub_domain = ' edition '
config = self . _CONFIG [ sub_domain ]
2016-08-28 17:43:15 +02:00
return self . _extract_cvp_info (
config [ ' data_src ' ] % path , page_title , {
' default ' : {
' media_src ' : config [ ' media_src ' ] ,
2021-01-01 13:26:37 +01:00
} ,
' f4m ' : {
' host ' : ' cnn-vh.akamaihd.net ' ,
} ,
2016-08-28 17:43:15 +02:00
} )
2014-02-11 14:38:17 +01:00
class CNNBlogsIE ( InfoExtractor ) :
_VALID_URL = r ' https?://[^ \ .]+ \ .blogs \ .cnn \ .com/.+ '
_TEST = {
' url ' : ' http://reliablesources.blogs.cnn.com/2014/02/09/criminalizing-journalism/ ' ,
' md5 ' : ' 3e56f97b0b6ffb4b79f4ea0749551084 ' ,
' info_dict ' : {
2016-08-30 11:46:48 +02:00
' id ' : ' bestoftv/2014/02/09/criminalizing-journalism.cnn ' ,
2014-02-11 14:38:17 +01:00
' ext ' : ' mp4 ' ,
' title ' : ' Criminalizing journalism? ' ,
' description ' : ' Glenn Greenwald responds to comments made this week on Capitol Hill that journalists could be criminal accessories. ' ,
' upload_date ' : ' 20140209 ' ,
} ,
2016-08-29 19:23:21 +02:00
' expected_warnings ' : [ ' Failed to download m3u8 information ' ] ,
2014-02-11 14:38:17 +01:00
' add_ie ' : [ ' CNN ' ] ,
}
def _real_extract ( self , url ) :
webpage = self . _download_webpage ( url , url_basename ( url ) )
cnn_url = self . _html_search_regex ( r ' data-url= " (.+?) " ' , webpage , ' cnn url ' )
2019-01-01 17:56:05 +01:00
return self . url_result ( cnn_url , CNNIE . ie_key ( ) )
2014-12-22 18:40:36 +01:00
class CNNArticleIE ( InfoExtractor ) :
2016-08-20 20:00:25 +02:00
_VALID_URL = r ' https?://(?:(?:edition|www) \ .)?cnn \ .com/(?!videos?/) '
2014-12-22 18:40:36 +01:00
_TEST = {
' url ' : ' http://www.cnn.com/2014/12/21/politics/obama-north-koreas-hack-not-war-but-cyber-vandalism/ ' ,
2015-01-21 10:27:18 +01:00
' md5 ' : ' 689034c2a3d9c6dc4aa72d65a81efd01 ' ,
2014-12-22 18:40:36 +01:00
' info_dict ' : {
2016-08-30 11:46:48 +02:00
' id ' : ' bestoftv/2014/12/21/ip-north-korea-obama.cnn ' ,
2014-12-22 18:40:36 +01:00
' ext ' : ' mp4 ' ,
2015-01-21 10:27:18 +01:00
' title ' : ' Obama: Cyberattack not an act of war ' ,
2016-08-29 19:23:21 +02:00
' description ' : ' md5:0a802a40d2376f60e6b04c8d5bcebc4b ' ,
2015-01-21 10:27:18 +01:00
' upload_date ' : ' 20141221 ' ,
2014-12-22 18:40:36 +01:00
} ,
2016-08-29 19:23:21 +02:00
' expected_warnings ' : [ ' Failed to download m3u8 information ' ] ,
2014-12-22 18:40:36 +01:00
' add_ie ' : [ ' CNN ' ] ,
}
def _real_extract ( self , url ) :
webpage = self . _download_webpage ( url , url_basename ( url ) )
cnn_url = self . _html_search_regex ( r " video: \ s* ' ([^ ' ]+) ' " , webpage , ' cnn url ' )
2019-01-01 17:56:05 +01:00
return self . url_result ( ' http://cnn.com/video/?/video/ ' + cnn_url , CNNIE . ie_key ( ) )
2022-10-04 05:09:23 +02:00
class CNNIndonesiaIE ( InfoExtractor ) :
_VALID_URL = r ' https?://www \ .cnnindonesia \ .com/[ \ w-]+/(?P<upload_date> \ d {8} ) \ d+- \ d+-(?P<id> \ d+)/(?P<display_id>[ \ w-]+) '
_TESTS = [ {
' url ' : ' https://www.cnnindonesia.com/ekonomi/20220909212635-89-845885/alasan-harga-bbm-di-indonesia-masih-disubsidi ' ,
' info_dict ' : {
' id ' : ' 845885 ' ,
' ext ' : ' mp4 ' ,
' description ' : ' md5:e7954bfa6f1749bc9ef0c079a719c347 ' ,
' upload_date ' : ' 20220909 ' ,
' title ' : ' Alasan Harga BBM di Indonesia Masih Disubsidi ' ,
' timestamp ' : 1662859088 ,
' duration ' : 120.0 ,
' thumbnail ' : r ' re:https://akcdn \ .detik \ .net \ .id/visual/2022/09/09/thumbnail-ekopedia-alasan-harga-bbm-disubsidi_169 \ .jpeg ' ,
' tags ' : [ ' ekopedia ' , ' subsidi bbm ' , ' subsidi ' , ' bbm ' , ' bbm subsidi ' , ' harga pertalite naik ' ] ,
' age_limit ' : 0 ,
' release_timestamp ' : 1662859088 ,
' release_date ' : ' 20220911 ' ,
' uploader ' : ' Asfahan Yahsyi ' ,
2024-06-12 01:09:58 +02:00
} ,
2022-10-04 05:09:23 +02:00
} , {
' url ' : ' https://www.cnnindonesia.com/internasional/20220911104341-139-846189/video-momen-charles-disambut-meriah-usai-dilantik-jadi-raja-inggris ' ,
' info_dict ' : {
' id ' : ' 846189 ' ,
' ext ' : ' mp4 ' ,
' upload_date ' : ' 20220911 ' ,
' duration ' : 76.0 ,
' timestamp ' : 1662869995 ,
' description ' : ' md5:ece7b003b3ee7d81c6a5cfede7d5397d ' ,
' thumbnail ' : r ' re:https://akcdn \ .detik \ .net \ .id/visual/2022/09/11/thumbnail-video-1_169 \ .jpeg ' ,
' title ' : ' VIDEO: Momen Charles Disambut Meriah usai Dilantik jadi Raja Inggris ' ,
' tags ' : [ ' raja charles ' , ' raja charles iii ' , ' ratu elizabeth ' , ' ratu elizabeth meninggal dunia ' , ' raja inggris ' , ' inggris ' ] ,
' age_limit ' : 0 ,
' release_date ' : ' 20220911 ' ,
' uploader ' : ' REUTERS ' ,
' release_timestamp ' : 1662869995 ,
2024-06-12 01:09:58 +02:00
} ,
2022-10-04 05:09:23 +02:00
} ]
def _real_extract ( self , url ) :
upload_date , video_id , display_id = self . _match_valid_url ( url ) . group ( ' upload_date ' , ' id ' , ' display_id ' )
webpage = self . _download_webpage ( url , display_id )
json_ld_list = list ( self . _yield_json_ld ( webpage , display_id ) )
json_ld_data = self . _json_ld ( json_ld_list , display_id )
embed_url = next (
json_ld . get ( ' embedUrl ' ) for json_ld in json_ld_list if json_ld . get ( ' @type ' ) == ' VideoObject ' )
return merge_dicts ( json_ld_data , {
' _type ' : ' url_transparent ' ,
' url ' : embed_url ,
' upload_date ' : upload_date ,
2024-06-12 01:09:58 +02:00
' tags ' : try_call ( lambda : self . _html_search_meta ( ' keywords ' , webpage ) . split ( ' , ' ) ) ,
2022-10-04 05:09:23 +02:00
} )