]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/abc.py
1 from __future__
import unicode_literals
8 from . common
import InfoExtractor
9 from .. compat
import compat_str
22 class ABCIE ( InfoExtractor
):
23 IE_NAME
= 'abc.net.au'
24 _VALID_URL
= r
'https?://(?:www\.)?abc\.net\.au/(?:news|btn)/(?:[^/]+/) {1,4} (?P<id>\d {5,} )'
27 'url' : 'http://www.abc.net.au/news/2014-11-05/australia-to-staff-ebola-treatment-centre-in-sierra-leone/5868334' ,
28 'md5' : 'cb3dd03b18455a661071ee1e28344d9f' ,
32 'title' : 'Australia to help staff Ebola treatment centre in Sierra Leone' ,
33 'description' : 'md5:809ad29c67a05f54eb41f2a105693a67' ,
35 'skip' : 'this video has expired' ,
37 'url' : 'http://www.abc.net.au/news/2015-08-17/warren-entsch-introduces-same-sex-marriage-bill/6702326' ,
38 'md5' : '4ebd61bdc82d9a8b722f64f1f4b4d121' ,
42 'upload_date' : '20150816' ,
43 'uploader' : 'ABC News (Australia)' ,
44 'description' : 'Government backbencher Warren Entsch introduces a cross-party sponsored bill to legalise same-sex marriage, saying the bill is designed to promote "an inclusive Australia, not a divided one.". Read more here: http://ab.co/1Mwc6ef' ,
45 'uploader_id' : 'NewsOnABC' ,
46 'title' : 'Marriage Equality: Warren Entsch introduces same sex marriage bill' ,
48 'add_ie' : [ 'Youtube' ],
49 'skip' : 'Not accessible from Travis CI server' ,
51 'url' : 'http://www.abc.net.au/news/2015-10-23/nab-lifts-interest-rates-following-westpac-and-cba/6880080' ,
52 'md5' : 'b96eee7c9edf4fc5a358a0252881cc1f' ,
56 'title' : 'NAB lifts interest rates, following Westpac and CBA' ,
57 'description' : 'md5:f13d8edc81e462fce4a0437c7dc04728' ,
60 'url' : 'http://www.abc.net.au/news/2015-10-19/6866214' ,
61 'only_matching' : True ,
63 'url' : 'https://www.abc.net.au/btn/classroom/wwi-centenary/10527914' ,
67 'title' : 'WWI Centenary' ,
68 'description' : 'md5:c2379ec0ca84072e86b446e536954546' ,
71 'url' : 'https://www.abc.net.au/news/programs/the-world/2020-06-10/black-lives-matter-protests-spawn-support-for/12342074' ,
75 'title' : 'Black Lives Matter protests spawn support for Papuans in Indonesia' ,
76 'description' : 'md5:2961a17dc53abc558589ccd0fb8edd6f' ,
79 'url' : 'https://www.abc.net.au/btn/newsbreak/btn-newsbreak-20200814/12560476' ,
83 'title' : 'Fortnite Banned From Apple and Google App Stores' ,
84 'description' : 'md5:a6df3f36ce8f816b74af4bd6462f5651' ,
85 'upload_date' : '20200813' ,
86 'uploader' : 'Behind the News' ,
87 'uploader_id' : 'behindthenews' ,
91 def _real_extract ( self
, url
):
92 video_id
= self
._ match
_ id
( url
)
93 webpage
= self
._ download
_ webpage
( url
, video_id
)
95 mobj
= re
. search ( r
'<a\s+href="(?P<url>[^"]+)"\s+data-duration="\d+"\s+title="Download audio directly">' , webpage
)
97 urls_info
= mobj
. groupdict ()
101 mobj
= re
. search ( r
'<a href="(?P<url>http://www\.youtube\.com/watch\?v=[^"]+)"><span><strong>External Link:</strong>' ,
104 mobj
= re
. search ( r
'<iframe width="100%" src="(?P<url>//www\.youtube-nocookie\.com/embed/[^?"]+)' , webpage
)
106 urls_info
= mobj
. groupdict ()
111 mobj
= re
. search ( r
'(?P<type>)"sources": (?P<json_data>\[[^\]]+\]),' , webpage
)
114 r
'inline(?P<type>Video|Audio|YouTube)Data\.push\((?P<json_data>[^)]+)\);' ,
117 expired
= self
._ html
_ search
_ regex
( r
'(?s)class="expired-(?:video|audio)".+?<span>(.+?)</span>' , webpage
, 'expired' , None )
119 raise ExtractorError ( ' %s said: %s ' % ( self
. IE_NAME
, expired
), expected
= True )
120 raise ExtractorError ( 'Unable to extract video urls' )
122 urls_info
= self
._ parse
_ json
(
123 mobj
. group ( 'json_data' ), video_id
, transform_source
= js_to_json
)
124 youtube
= mobj
. group ( 'type' ) == 'YouTube'
125 video
= mobj
. group ( 'type' ) == 'Video' or urls_info
[ 0 ][ 'contentType' ] == 'video/mp4'
127 if not isinstance ( urls_info
, list ):
128 urls_info
= [ urls_info
]
131 return self
. playlist_result ([
132 self
. url_result ( url_info
[ 'url' ]) for url_info
in urls_info
])
135 for url_info
in urls_info
:
136 height
= int_or_none ( url_info
. get ( 'height' ))
137 bitrate
= int_or_none ( url_info
. get ( 'bitrate' ))
138 width
= int_or_none ( url_info
. get ( 'width' ))
140 mobj
= re
. search ( r
'_(?:(?P<height>\d+)|(?P<bitrate>\d+)k)\.mp4$' , url_info
[ 'url' ])
142 height_from_url
= mobj
. group ( 'height' )
144 height
= height
or int_or_none ( height_from_url
)
145 width
= width
or int_or_none ( url_info
. get ( 'label' ))
147 bitrate
= bitrate
or int_or_none ( mobj
. group ( 'bitrate' ))
148 format_id
= str_or_none ( url_info
. get ( 'label' ))
150 'url' : url_info
[ 'url' ],
151 'vcodec' : url_info
. get ( 'codec' ) if video
else 'none' ,
155 'filesize' : int_or_none ( url_info
. get ( 'filesize' )),
156 'format_id' : format_id
159 self
._ sort
_ formats
( formats
)
163 'title' : self
._ og
_ search
_ title
( webpage
),
165 'description' : self
._ og
_ search
_ description
( webpage
),
166 'thumbnail' : self
._ og
_ search
_ thumbnail
( webpage
),
170 class ABCIViewIE ( InfoExtractor
):
171 IE_NAME
= 'abc.net.au:iview'
172 _VALID_URL
= r
'https?://iview\.abc\.net\.au/(?:[^/]+/)*video/(?P<id>[^/?#]+)'
173 _GEO_COUNTRIES
= [ 'AU' ]
175 # ABC iview programs are normally available for 14 days only.
177 'url' : 'https://iview.abc.net.au/show/gruen/series/11/video/LE1927H001S00' ,
178 'md5' : '67715ce3c78426b11ba167d875ac6abf' ,
180 'id' : 'LE1927H001S00' ,
182 'title' : "Series 11 Ep 1" ,
184 'description' : 'md5:52cc744ad35045baf6aded2ce7287f67' ,
185 'upload_date' : '20190925' ,
186 'uploader_id' : 'abc1' ,
187 'timestamp' : 1569445289 ,
190 'skip_download' : True ,
194 def _real_extract ( self
, url
):
195 video_id
= self
._ match
_ id
( url
)
196 video_params
= self
._ download
_ json
(
197 'https://iview.abc.net.au/api/programs/' + video_id
, video_id
)
198 title
= unescapeHTML ( video_params
. get ( 'title' ) or video_params
[ 'seriesTitle' ])
199 stream
= next ( s
for s
in video_params
[ 'playlist' ] if s
. get ( 'type' ) in ( 'program' , 'livestream' ))
201 house_number
= video_params
. get ( 'episodeHouseNumber' ) or video_id
202 path
= '/auth/hls/sign?ts= {0} &hn= {1} &d=android-tablet' . format (
203 int ( time
. time ()), house_number
)
205 b
'android.content.res.Resources' ,
206 path
. encode ( 'utf-8' ), hashlib
. sha256
). hexdigest ()
207 token
= self
._ download
_ webpage
(
208 'http://iview.abc.net.au {0} &sig= {1} ' . format ( path
, sig
), video_id
)
210 def tokenize_url ( url
, token
):
211 return update_url_query ( url
, {
215 for sd
in ( '720' , 'sd' , 'sd-low' ):
217 stream
, lambda x
: x
[ 'streams' ][ 'hls' ][ sd
], compat_str
)
220 formats
= self
._ extract
_ m
3u8_ formats
(
221 tokenize_url ( sd_url
, token
), video_id
, 'mp4' ,
222 entry_protocol
= 'm3u8_native' , m3u8_id
= 'hls' , fatal
= False )
225 self
._ sort
_ formats
( formats
)
228 src_vtt
= stream
. get ( 'captions' , {}). get ( 'src-vtt' )
235 is_live
= video_params
. get ( 'livestream' ) == '1'
237 title
= self
._l ive
_ title
( title
)
242 'description' : video_params
. get ( 'description' ),
243 'thumbnail' : video_params
. get ( 'thumbnail' ),
244 'duration' : int_or_none ( video_params
. get ( 'eventDuration' )),
245 'timestamp' : parse_iso8601 ( video_params
. get ( 'pubDate' ), ' ' ),
246 'series' : unescapeHTML ( video_params
. get ( 'seriesTitle' )),
247 'series_id' : video_params
. get ( 'seriesHouseNumber' ) or video_id
[: 7 ],
248 'season_number' : int_or_none ( self
._ search
_ regex
(
249 r
'\bSeries\s+(\d+)\b' , title
, 'season number' , default
= None )),
250 'episode_number' : int_or_none ( self
._ search
_ regex
(
251 r
'\bEp\s+(\d+)\b' , title
, 'episode number' , default
= None )),
252 'episode_id' : house_number
,
253 'uploader_id' : video_params
. get ( 'channel' ),
255 'subtitles' : subtitles
,