2014-01-28 20:57:38 +00:00
from __future__ import unicode_literals
2014-01-30 03:52:50 +00:00
import json
2013-06-23 20:23:19 +00:00
import re
from . common import InfoExtractor
2014-02-19 00:27:12 +00:00
from . . utils import int_or_none
2013-06-23 20:23:19 +00:00
class LiveLeakIE ( InfoExtractor ) :
2015-01-23 16:09:26 +00:00
_VALID_URL = r ' https?://(?: \ w+ \ .)?liveleak \ .com/view \ ?(?:.*?)i=(?P<id>[ \ w_]+)(?:.*) '
2014-01-30 05:09:06 +00:00
_TESTS = [ {
2014-01-28 20:57:38 +00:00
' url ' : ' http://www.liveleak.com/view?i=757_1364311680 ' ,
2015-01-23 16:22:14 +00:00
' md5 ' : ' 50f79e05ba149149c1b4ea961223d5b3 ' ,
2014-01-28 20:57:38 +00:00
' info_dict ' : {
2014-02-19 00:27:12 +00:00
' id ' : ' 757_1364311680 ' ,
2015-01-23 16:22:14 +00:00
' ext ' : ' flv ' ,
2014-01-28 20:57:38 +00:00
' description ' : ' extremely bad day for this guy..! ' ,
' uploader ' : ' ljfriel2 ' ,
' title ' : ' Most unlucky car accident '
2013-06-27 18:46:46 +00:00
}
2014-11-23 20:20:46 +00:00
} , {
2014-01-30 05:09:06 +00:00
' url ' : ' http://www.liveleak.com/view?i=f93_1390833151 ' ,
2015-01-23 16:22:14 +00:00
' md5 ' : ' b13a29626183c9d33944e6a04f41aafc ' ,
2014-01-30 05:09:06 +00:00
' info_dict ' : {
2014-02-19 00:27:12 +00:00
' id ' : ' f93_1390833151 ' ,
' ext ' : ' mp4 ' ,
2014-01-30 05:09:06 +00:00
' description ' : ' German Television Channel NDR does an exclusive interview with Edward Snowden. \r \n Uploaded on LiveLeak cause German Television thinks the rest of the world isn \' t intereseted in Edward Snowden. ' ,
' uploader ' : ' ARD_Stinkt ' ,
' title ' : ' German Television does first Edward Snowden Interview (ENGLISH) ' ,
}
2014-11-23 20:20:46 +00:00
} , {
2014-02-19 00:27:12 +00:00
' url ' : ' http://www.liveleak.com/view?i=4f7_1392687779 ' ,
' md5 ' : ' 42c6d97d54f1db107958760788c5f48f ' ,
' info_dict ' : {
' id ' : ' 4f7_1392687779 ' ,
' ext ' : ' mp4 ' ,
' description ' : " The guy with the cigarette seems amazingly nonchalant about the whole thing... I really hope my friends ' reactions would be a bit stronger. \r \n \r \n Action-go to 0:55. " ,
' uploader ' : ' CapObveus ' ,
' title ' : ' Man is Fatally Struck by Reckless Car While Packing up a Moving Truck ' ,
' age_limit ' : 18 ,
}
2015-06-14 20:41:44 +00:00
} , {
2015-06-14 20:54:49 +00:00
# Covers https://github.com/rg3/youtube-dl/pull/5983
2015-06-14 20:41:44 +00:00
' url ' : ' http://www.liveleak.com/view?i=801_1409392012 ' ,
' md5 ' : ' 0b3bec2d888c20728ca2ad3642f0ef15 ' ,
' info_dict ' : {
' id ' : ' 801_1409392012 ' ,
' ext ' : ' mp4 ' ,
' description ' : " Happened on 27.7.2014. \r \n At 0:53 you can see people still swimming at near beach. " ,
' uploader ' : ' bony333 ' ,
' title ' : ' Crazy Hungarian tourist films close call waterspout in Croatia '
}
2014-01-30 05:09:06 +00:00
} ]
2013-06-23 20:23:19 +00:00
def _real_extract ( self , url ) :
2015-01-23 16:09:26 +00:00
v ideo_id = self . _match_id ( url )
2013-06-23 20:23:19 +00:00
webpage = self . _download_webpage ( url , video_id )
2014-02-19 00:27:12 +00:00
video_title = self . _og_search_title ( webpage ) . replace ( ' LiveLeak.com - ' , ' ' ) . strip ( )
video_description = self . _og_search_description ( webpage )
video_uploader = self . _html_search_regex (
r ' By:.*?( \ w+)</a> ' , webpage , ' uploader ' , fatal = False )
age_limit = int_or_none ( self . _search_regex (
r ' you confirm that you are ([0-9]+) years and over. ' ,
webpage , ' age limit ' , default = None ) )
2014-01-30 03:52:50 +00:00
sources_raw = self . _search_regex (
2014-01-30 05:09:06 +00:00
r ' (?s)sources: \ s*( \ [.*? \ ]), ' , webpage , ' video URLs ' , default = None )
if sources_raw is None :
2014-02-19 00:27:12 +00:00
alt_source = self . _search_regex (
r ' (file: " .*? " ), ' , webpage , ' video URL ' , default = None )
if alt_source :
sources_raw = ' [ { %s }] ' % alt_source
else :
# Maybe an embed?
embed_url = self . _search_regex (
r ' <iframe[^>]+src= " (http://www.prochan.com/embed \ ?[^ " ]+) " ' ,
webpage , ' embed URL ' )
return {
' _type ' : ' url_transparent ' ,
' url ' : embed_url ,
' id ' : video_id ,
' title ' : video_title ,
' description ' : video_description ,
' uploader ' : video_uploader ,
' age_limit ' : age_limit ,
}
2014-01-30 05:09:06 +00:00
2014-01-30 03:52:50 +00:00
sources_json = re . sub ( r ' \ s([a-z]+): \ s ' , r ' " \ 1 " : ' , sources_raw )
sources = json . loads ( sources_json )
formats = [ {
2015-01-23 16:22:14 +00:00
' format_id ' : ' %s ' % i ,
2014-01-30 03:52:50 +00:00
' format_note ' : s . get ( ' label ' ) ,
' url ' : s [ ' file ' ] ,
2015-01-23 16:22:14 +00:00
} for i , s in enumerate ( sources ) ]
for i , s in enumerate ( sources ) :
2015-06-14 20:54:05 +00:00
# Removing '.h264_*.mp4' gives the raw video, which is essentially
# the same video without the LiveLeak logo at the top (see
# https://github.com/rg3/youtube-dl/pull/4768)
2015-06-14 20:51:21 +00:00
orig_url = re . sub ( r ' \ .h264_.+? \ .mp4 ' , ' ' , s [ ' file ' ] )
2015-01-23 16:22:14 +00:00
if s [ ' file ' ] != orig_url :
formats . append ( {
' format_id ' : ' original- %s ' % i ,
' format_note ' : s . get ( ' label ' ) ,
' url ' : orig_url ,
' preference ' : 1 ,
} )
2014-01-30 03:52:50 +00:00
self . _sort_formats ( formats )
2013-06-23 20:23:19 +00:00
2014-01-28 20:57:38 +00:00
return {
' id ' : video_id ,
2013-06-23 20:23:19 +00:00
' title ' : video_title ,
' description ' : video_description ,
2014-01-30 03:52:50 +00:00
' uploader ' : video_uploader ,
' formats ' : formats ,
2014-02-19 00:27:12 +00:00
' age_limit ' : age_limit ,
2013-06-23 20:23:19 +00:00
}