2018-02-08 18:05:12 -07:00
import base64
2018-05-23 19:51:47 -06:00
import json
import re
2013-08-27 04:38:30 -06:00
from . common import InfoExtractor
2023-01-16 14:38:33 -07:00
from . theplatform import ThePlatformIE , default_ns
2017-02-08 08:23:42 -07:00
from . adobepass import AdobePassIE
2018-09-08 01:44:06 -06:00
from . . compat import compat_urllib_parse_unquote
2023-07-09 01:53:02 -06:00
from . . networking import HEADRequest
2014-12-13 04:24:42 -07:00
from . . utils import (
2022-09-30 09:43:40 -06:00
ExtractorError ,
2023-01-14 09:40:42 -07:00
RegexNotFoundError ,
UserNotLive ,
clean_html ,
2023-03-26 16:27:39 -06:00
determine_ext ,
float_or_none ,
2019-10-28 12:31:20 -06:00
int_or_none ,
2023-03-26 16:27:39 -06:00
mimetype2ext ,
2021-03-21 09:56:31 -06:00
parse_age_limit ,
2019-10-28 12:31:20 -06:00
parse_duration ,
2023-03-26 16:27:39 -06:00
remove_end ,
2015-12-08 08:16:14 -07:00
smuggle_url ,
2022-09-30 09:43:40 -06:00
traverse_obj ,
2019-10-28 12:31:20 -06:00
try_get ,
2023-01-14 09:40:42 -07:00
unescapeHTML ,
2019-10-28 12:31:20 -06:00
unified_timestamp ,
2016-12-17 10:11:13 -07:00
update_url_query ,
2022-09-30 09:43:40 -06:00
url_basename ,
2014-07-22 17:47:18 -06:00
)
2013-08-27 04:38:30 -06:00
2022-11-15 17:57:43 -07:00
class NBCIE ( ThePlatformIE ) : # XXX: Do not subclass from concrete IE
2023-01-14 09:40:42 -07:00
_VALID_URL = r ' https?(?P<permalink>://(?:www \ .)?nbc \ .com/(?:classic-tv/)?[^/]+/video/[^/]+/(?P<id>(?:NBCE|n)? \ d+)) '
2014-12-12 09:10:32 -07:00
_TESTS = [
{
2017-02-08 08:23:42 -07:00
' url ' : ' http://www.nbc.com/the-tonight-show/video/jimmy-fallon-surprises-fans-at-ben-jerrys/2848237 ' ,
2014-12-12 09:10:32 -07:00
' info_dict ' : {
2017-02-08 08:23:42 -07:00
' id ' : ' 2848237 ' ,
2016-01-30 13:34:46 -07:00
' ext ' : ' mp4 ' ,
2015-02-19 07:00:39 -07:00
' title ' : ' Jimmy Fallon Surprises Fans at Ben & Jerry \' s ' ,
' description ' : ' Jimmy gives out free scoops of his new " Tonight Dough " ice cream flavor by surprising customers at the Ben & Jerry \' s scoop shop. ' ,
2016-04-01 11:06:11 -06:00
' timestamp ' : 1424246400 ,
' upload_date ' : ' 20150218 ' ,
' uploader ' : ' NBCU-COM ' ,
2023-01-14 09:40:42 -07:00
' episode ' : ' Jimmy Fallon Surprises Fans at Ben & Jerry \' s ' ,
' episode_number ' : 86 ,
' season ' : ' Season 2 ' ,
' season_number ' : 2 ,
' series ' : ' Tonight Show: Jimmy Fallon ' ,
' duration ' : 237.0 ,
' chapters ' : ' count:1 ' ,
' tags ' : ' count:4 ' ,
' thumbnail ' : r ' re:https?://.+ \ .jpg ' ,
2014-12-12 09:10:32 -07:00
} ,
2016-01-30 13:34:46 -07:00
' params ' : {
2023-01-14 09:40:42 -07:00
' skip_download ' : ' m3u8 ' ,
2016-01-30 13:34:46 -07:00
} ,
2014-02-25 15:57:54 -07:00
} ,
2015-05-04 05:09:18 -06:00
{
' url ' : ' http://www.nbc.com/saturday-night-live/video/star-wars-teaser/2832821 ' ,
' info_dict ' : {
2016-01-30 13:34:46 -07:00
' id ' : ' 2832821 ' ,
' ext ' : ' mp4 ' ,
2015-05-04 05:09:18 -06:00
' title ' : ' Star Wars Teaser ' ,
' description ' : ' md5:0b40f9cbde5b671a7ff62fceccc4f442 ' ,
2016-04-01 11:06:11 -06:00
' timestamp ' : 1417852800 ,
' upload_date ' : ' 20141206 ' ,
' uploader ' : ' NBCU-COM ' ,
2015-05-04 05:09:18 -06:00
} ,
2023-01-14 09:40:42 -07:00
' skip ' : ' page not found ' ,
2015-05-04 07:53:05 -06:00
} ,
2016-06-08 06:50:01 -06:00
{
# HLS streams requires the 'hdnea3' cookie
' url ' : ' http://www.nbc.com/Kings/video/goliath/n1806 ' ,
' info_dict ' : {
2017-02-08 08:23:42 -07:00
' id ' : ' 101528f5a9e8127b107e98c5e6ce4638 ' ,
2016-06-08 06:50:01 -06:00
' ext ' : ' mp4 ' ,
' title ' : ' Goliath ' ,
' description ' : ' When an unknown soldier saves the life of the King \' s son in battle, he \' s thrust into the limelight and politics of the kingdom. ' ,
' timestamp ' : 1237100400 ,
' upload_date ' : ' 20090315 ' ,
' uploader ' : ' NBCU-COM ' ,
} ,
2023-01-14 09:40:42 -07:00
' skip ' : ' page not found ' ,
} ,
{
# manifest url does not have extension
' url ' : ' https://www.nbc.com/the-golden-globe-awards/video/oprah-winfrey-receives-cecil-b-de-mille-award-at-the-2018-golden-globes/3646439 ' ,
' info_dict ' : {
' id ' : ' 3646439 ' ,
' ext ' : ' mp4 ' ,
' title ' : ' Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes ' ,
' episode ' : ' Oprah Winfrey Receives Cecil B. de Mille Award at the 2018 Golden Globes ' ,
' episode_number ' : 1 ,
' season ' : ' Season 75 ' ,
' season_number ' : 75 ,
' series ' : ' The Golden Globe Awards ' ,
' description ' : ' Oprah Winfrey receives the Cecil B. de Mille Award at the 75th Annual Golden Globe Awards. ' ,
' uploader ' : ' NBCU-COM ' ,
' upload_date ' : ' 20180107 ' ,
' timestamp ' : 1515312000 ,
' duration ' : 570.0 ,
' tags ' : ' count:8 ' ,
' thumbnail ' : r ' re:https?://.+ \ .jpg ' ,
' chapters ' : ' count:1 ' ,
} ,
' params ' : {
' skip_download ' : ' m3u8 ' ,
} ,
} ,
{
# new video_id format
' url ' : ' https://www.nbc.com/quantum-leap/video/bens-first-leap-nbcs-quantum-leap/NBCE125189978 ' ,
' info_dict ' : {
' id ' : ' NBCE125189978 ' ,
' ext ' : ' mp4 ' ,
' title ' : ' Ben \' s First Leap | NBC \' s Quantum Leap ' ,
' description ' : ' md5:a82762449b7ec4bb83291a7b355ebf8e ' ,
' uploader ' : ' NBCU-COM ' ,
' series ' : ' Quantum Leap ' ,
' season ' : ' Season 1 ' ,
' season_number ' : 1 ,
' episode ' : ' Ben \' s First Leap | NBC \' s Quantum Leap ' ,
' episode_number ' : 1 ,
' duration ' : 170.171 ,
' chapters ' : [ ] ,
' timestamp ' : 1663956155 ,
' upload_date ' : ' 20220923 ' ,
' tags ' : ' count:10 ' ,
' age_limit ' : 0 ,
' thumbnail ' : r ' re:https?://.+ \ .jpg ' ,
} ,
2016-06-08 06:50:01 -06:00
' params ' : {
2023-01-14 09:40:42 -07:00
' skip_download ' : ' m3u8 ' ,
2016-06-08 06:50:01 -06:00
} ,
2017-10-25 10:23:27 -06:00
} ,
{
' url ' : ' https://www.nbc.com/classic-tv/charles-in-charge/video/charles-in-charge-pilot/n3310 ' ,
' only_matching ' : True ,
} ,
2018-09-08 01:44:06 -06:00
{
# Percent escaped url
' url ' : ' https://www.nbc.com/up-all-night/video/day-after-valentine %27s -day/n2189 ' ,
' only_matching ' : True ,
}
2014-12-12 09:10:32 -07:00
]
2014-02-25 15:57:54 -07:00
def _real_extract ( self , url ) :
2021-08-18 19:41:24 -06:00
permalink , video_id = self . _match_valid_url ( url ) . groups ( )
2018-09-08 01:44:06 -06:00
permalink = ' http ' + compat_urllib_parse_unquote ( permalink )
2020-01-16 07:37:16 -07:00
video_data = self . _download_json (
2019-10-14 11:36:25 -06:00
' https://friendship.nbc.co/v2/graphql ' , video_id , query = {
2020-01-16 07:37:16 -07:00
' query ' : ''' query bonanzaPage(
$ app : NBCUBrands ! = nbc
$ name : String !
$ oneApp : Boolean
$ platform : SupportedPlatforms ! = web
$ type : EntityPageType ! = VIDEO
$ userId : String !
) {
bonanzaPage (
app : $ app
name : $ name
oneApp : $ oneApp
platform : $ platform
type : $ type
userId : $ userId
) {
metadata {
2019-10-14 11:36:25 -06:00
. . . on VideoPageData {
description
episodeNumber
keywords
locked
mpxAccountId
mpxGuid
rating
2020-01-16 07:37:16 -07:00
resourceId
2019-10-14 11:36:25 -06:00
seasonNumber
secondaryTitle
seriesShortTitle
}
}
}
2020-01-16 07:37:16 -07:00
} ''' ,
' variables ' : json . dumps ( {
' name ' : permalink ,
' oneApp ' : True ,
' userId ' : ' 0 ' ,
} ) ,
} ) [ ' data ' ] [ ' bonanzaPage ' ] [ ' metadata ' ]
2017-05-07 01:58:34 -06:00
query = {
' mbr ' : ' true ' ,
' manifest ' : ' m3u ' ,
2023-01-06 11:44:35 -07:00
' switch ' : ' HLSServiceSecure ' ,
2017-05-07 01:58:34 -06:00
}
2019-10-14 11:36:25 -06:00
video_id = video_data [ ' mpxGuid ' ]
2021-03-21 09:56:31 -06:00
tp_path = ' NnzsPC/media/guid/ %s / %s ' % ( video_data . get ( ' mpxAccountId ' ) or ' 2410887629 ' , video_id )
tpm = self . _download_theplatform_metadata ( tp_path , video_id )
title = tpm . get ( ' title ' ) or video_data . get ( ' secondaryTitle ' )
2019-10-14 11:36:25 -06:00
if video_data . get ( ' locked ' ) :
2017-05-07 01:58:34 -06:00
resource = self . _get_mvpd_resource (
2020-01-16 07:37:16 -07:00
video_data . get ( ' resourceId ' ) or ' nbcentertainment ' ,
title , video_id , video_data . get ( ' rating ' ) )
2017-05-07 01:58:34 -06:00
query [ ' auth ' ] = self . _extract_mvpd_auth (
url , video_id , ' nbcentertainment ' , resource )
theplatform_url = smuggle_url ( update_url_query (
2019-10-14 11:36:25 -06:00
' http://link.theplatform.com/s/NnzsPC/media/guid/ %s / %s ' % ( video_data . get ( ' mpxAccountId ' ) or ' 2410887629 ' , video_id ) ,
2017-05-07 01:58:34 -06:00
query ) , { ' force_smil_url ' : True } )
2021-03-21 09:56:31 -06:00
# Empty string or 0 can be valid values for these. So the check must be `is None`
description = video_data . get ( ' description ' )
if description is None :
description = tpm . get ( ' description ' )
episode_number = int_or_none ( video_data . get ( ' episodeNumber ' ) )
if episode_number is None :
episode_number = int_or_none ( tpm . get ( ' nbcu$airOrder ' ) )
rating = video_data . get ( ' rating ' )
if rating is None :
try_get ( tpm , lambda x : x [ ' ratings ' ] [ 0 ] [ ' rating ' ] )
season_number = int_or_none ( video_data . get ( ' seasonNumber ' ) )
if season_number is None :
season_number = int_or_none ( tpm . get ( ' nbcu$seasonNumber ' ) )
series = video_data . get ( ' seriesShortTitle ' )
if series is None :
series = tpm . get ( ' nbcu$seriesShortTitle ' )
tags = video_data . get ( ' keywords ' )
if tags is None or len ( tags ) == 0 :
tags = tpm . get ( ' keywords ' )
2017-05-07 01:58:34 -06:00
return {
2016-01-30 13:34:46 -07:00
' _type ' : ' url_transparent ' ,
2021-03-21 09:56:31 -06:00
' age_limit ' : parse_age_limit ( rating ) ,
' description ' : description ,
' episode ' : title ,
' episode_number ' : episode_number ,
2016-01-30 13:34:46 -07:00
' id ' : video_id ,
2021-03-21 09:56:31 -06:00
' ie_key ' : ' ThePlatform ' ,
' season_number ' : season_number ,
' series ' : series ,
' tags ' : tags ,
2017-05-07 01:58:34 -06:00
' title ' : title ,
' url ' : theplatform_url ,
2016-01-30 13:34:46 -07:00
}
2014-02-25 15:57:54 -07:00
2015-03-30 12:47:18 -06:00
class NBCSportsVPlayerIE ( InfoExtractor ) :
2021-01-01 05:26:37 -07:00
_VALID_URL_BASE = r ' https?://(?:vplayer \ .nbcsports \ .com|(?:www \ .)?nbcsports \ .com/vplayer)/ '
_VALID_URL = _VALID_URL_BASE + r ' (?:[^/]+/)+(?P<id>[0-9a-zA-Z_]+) '
2022-07-31 19:23:25 -06:00
_EMBED_REGEX = [ r ' (?:iframe[^>]+|var video|div[^>]+data-(?:mpx-)?)[sS]rc \ s?= \ s? " (?P<url> %s [^ \ " ]+) ' % _VALID_URL_BASE ]
2015-03-30 12:21:27 -06:00
2015-03-30 13:38:45 -06:00
_TESTS = [ {
2017-09-20 13:53:06 -06:00
' url ' : ' https://vplayer.nbcsports.com/p/BxmELC/nbcsports_embed/select/9CsDKds0kvHI ' ,
2015-03-30 12:21:27 -06:00
' info_dict ' : {
' id ' : ' 9CsDKds0kvHI ' ,
2017-09-20 13:53:06 -06:00
' ext ' : ' mp4 ' ,
2015-03-30 12:21:27 -06:00
' description ' : ' md5:df390f70a9ba7c95ff1daace988f0d8d ' ,
' title ' : ' Tyler Kalinoski hits buzzer-beater to lift Davidson ' ,
2016-04-01 11:06:11 -06:00
' timestamp ' : 1426270238 ,
' upload_date ' : ' 20150313 ' ,
' uploader ' : ' NBCU-SPORTS ' ,
2022-01-07 21:54:52 -07:00
' duration ' : 72.818 ,
' chapters ' : [ ] ,
' thumbnail ' : r ' re:^https?://.* \ .jpg$ '
2015-03-30 12:21:27 -06:00
}
2015-03-30 13:38:45 -06:00
} , {
2022-01-07 21:54:52 -07:00
' url ' : ' https://vplayer.nbcsports.com/p/BxmELC/nbcsports_embed/select/media/PEgOtlNcC_y2 ' ,
2015-03-30 13:38:45 -06:00
' only_matching ' : True ,
2021-01-01 05:26:37 -07:00
} , {
' url ' : ' https://www.nbcsports.com/vplayer/p/BxmELC/nbcsports/select/PHJSaFWbrTY9?form=html&autoPlay=true ' ,
' only_matching ' : True ,
2015-03-30 13:38:45 -06:00
} ]
2015-03-30 12:21:27 -06:00
def _real_extract ( self , url ) :
video_id = self . _match_id ( url )
webpage = self . _download_webpage ( url , video_id )
2022-01-07 21:54:52 -07:00
theplatform_url = self . _html_search_regex ( r ' tp:releaseUrl= " (.+?) " ' , webpage , ' url ' )
2015-03-30 12:21:27 -06:00
return self . url_result ( theplatform_url , ' ThePlatform ' )
2015-03-30 12:47:18 -06:00
class NBCSportsIE ( InfoExtractor ) :
2021-01-01 05:26:37 -07:00
_VALID_URL = r ' https?://(?:www \ .)?nbcsports \ .com//?(?!vplayer/)(?:[^/]+/)+(?P<id>[0-9a-z-]+) '
2015-03-30 12:47:18 -06:00
2021-01-01 05:26:37 -07:00
_TESTS = [ {
# iframe src
2023-10-06 17:56:19 -06:00
' url ' : ' https://www.nbcsports.com/watch/nfl/profootballtalk/pft-pm/unpacking-addisons-reckless-driving-citation ' ,
2015-03-30 12:47:18 -06:00
' info_dict ' : {
' id ' : ' PHJSaFWbrTY9 ' ,
2021-01-01 05:26:37 -07:00
' ext ' : ' mp4 ' ,
2015-03-30 12:47:18 -06:00
' title ' : ' Tom Izzo, Michigan St. has \' so much respect \' for Duke ' ,
' description ' : ' md5:ecb459c9d59e0766ac9c7d5d0eda8113 ' ,
2016-04-24 06:44:52 -06:00
' uploader ' : ' NBCU-SPORTS ' ,
' upload_date ' : ' 20150330 ' ,
' timestamp ' : 1427726529 ,
2022-01-07 21:54:52 -07:00
' chapters ' : [ ] ,
' thumbnail ' : ' https://hdliveextra-a.akamaihd.net/HD/image_sports/NBCU_Sports_Group_-_nbcsports/253/303/izzodps.jpg ' ,
' duration ' : 528.395 ,
2015-03-30 12:47:18 -06:00
}
2021-01-01 05:26:37 -07:00
} , {
# data-mpx-src
' url ' : ' https://www.nbcsports.com/philadelphia/philadelphia-phillies/bruce-bochy-hector-neris-hes-idiot ' ,
' only_matching ' : True ,
} , {
# data-src
' url ' : ' https://www.nbcsports.com/boston/video/report-card-pats-secondary-no-match-josh-allen ' ,
' only_matching ' : True ,
} ]
2015-03-30 12:47:18 -06:00
def _real_extract ( self , url ) :
video_id = self . _match_id ( url )
webpage = self . _download_webpage ( url , video_id )
return self . url_result (
NBCSportsVPlayerIE . _extract_url ( webpage ) , ' NBCSportsVPlayer ' )
2018-05-23 19:51:47 -06:00
class NBCSportsStreamIE ( AdobePassIE ) :
_VALID_URL = r ' https?://stream \ .nbcsports \ .com/.+? \ bpid=(?P<id> \ d+) '
_TEST = {
' url ' : ' http://stream.nbcsports.com/nbcsn/generic?pid=206559 ' ,
' info_dict ' : {
' id ' : ' 206559 ' ,
' ext ' : ' mp4 ' ,
' title ' : ' Amgen Tour of California Women \' s Recap ' ,
' description ' : ' md5:66520066b3b5281ada7698d0ea2aa894 ' ,
} ,
' params ' : {
# m3u8 download
' skip_download ' : True ,
} ,
' skip ' : ' Requires Adobe Pass Authentication ' ,
}
def _real_extract ( self , url ) :
video_id = self . _match_id ( url )
live_source = self . _download_json (
' http://stream.nbcsports.com/data/live_sources_ %s .json ' % video_id ,
video_id )
video_source = live_source [ ' videoSources ' ] [ 0 ]
title = video_source [ ' title ' ]
source_url = None
for k in ( ' source ' , ' msl4source ' , ' iossource ' , ' hlsv4 ' ) :
sk = k + ' Url '
source_url = video_source . get ( sk ) or video_source . get ( sk + ' Alt ' )
if source_url :
break
else :
source_url = video_source [ ' ottStreamUrl ' ]
is_live = video_source . get ( ' type ' ) == ' live ' or video_source . get ( ' status ' ) == ' Live '
resource = self . _get_mvpd_resource ( ' nbcsports ' , title , video_id , ' ' )
token = self . _extract_mvpd_auth ( url , video_id , ' nbcsports ' , resource )
tokenized_url = self . _download_json (
' https://token.playmakerservices.com/cdn ' ,
video_id , data = json . dumps ( {
' requestorId ' : ' nbcsports ' ,
' pid ' : video_id ,
' application ' : ' NBCSports ' ,
' version ' : ' v1 ' ,
' platform ' : ' desktop ' ,
' cdn ' : ' akamai ' ,
' url ' : video_source [ ' sourceUrl ' ] ,
' token ' : base64 . b64encode ( token . encode ( ) ) . decode ( ) ,
' resourceId ' : base64 . b64encode ( resource . encode ( ) ) . decode ( ) ,
} ) . encode ( ) ) [ ' tokenizedUrl ' ]
formats = self . _extract_m3u8_formats ( tokenized_url , video_id , ' mp4 ' )
return {
' id ' : video_id ,
2021-12-15 09:00:46 -07:00
' title ' : title ,
2018-05-23 19:51:47 -06:00
' description ' : live_source . get ( ' description ' ) ,
' formats ' : formats ,
' is_live ' : is_live ,
}
2022-11-15 17:57:43 -07:00
class NBCNewsIE ( ThePlatformIE ) : # XXX: Do not subclass from concrete IE
2018-11-21 04:10:06 -07:00
_VALID_URL = r ' (?x)https?://(?:www \ .)?(?:nbcnews|today|msnbc) \ .com/([^/]+/)*(?:.*-)?(?P<id>[^/?]+) '
2022-07-31 19:23:25 -06:00
_EMBED_REGEX = [ r ' <iframe[^>]+src=([ " \' ])(?P<url>(?:https?:)?//www \ .nbcnews \ .com/widget/video-embed/[^ " \' ]+) \ 1 ' ]
2013-08-27 04:38:30 -06:00
2014-05-29 16:38:57 -06:00
_TESTS = [
{
2016-03-17 09:07:36 -06:00
' url ' : ' http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880 ' ,
2023-10-06 17:56:19 -06:00
' md5 ' : ' fb3dcd2d7b1dd9804305fa2fc95ab610 ' , # md5 tends to fluctuate
2014-05-29 16:38:57 -06:00
' info_dict ' : {
2018-11-21 04:10:06 -07:00
' id ' : ' 269389891880 ' ,
2014-10-26 18:14:17 -06:00
' ext ' : ' mp4 ' ,
2014-05-29 16:38:57 -06:00
' title ' : ' How Twitter Reacted To The Snowden Interview ' ,
' description ' : ' md5:65a0bd5d76fe114f3c2727aa3a81fe64 ' ,
2016-06-22 18:36:19 -06:00
' timestamp ' : 1401363060 ,
' upload_date ' : ' 20140529 ' ,
2023-10-06 17:56:19 -06:00
' duration ' : 46.0 ,
' thumbnail ' : ' https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/140529/p_tweet_snow_140529.jpg ' ,
2014-05-29 16:38:57 -06:00
} ,
} ,
2015-01-20 08:23:51 -07:00
{
' url ' : ' http://www.nbcnews.com/feature/dateline-full-episodes/full-episode-family-business-n285156 ' ,
' md5 ' : ' fdbf39ab73a72df5896b6234ff98518a ' ,
' info_dict ' : {
2016-06-22 18:36:19 -06:00
' id ' : ' 529953347624 ' ,
2015-01-20 08:23:51 -07:00
' ext ' : ' mp4 ' ,
' title ' : ' FULL EPISODE: Family Business ' ,
' description ' : ' md5:757988edbaae9d7be1d585eb5d55cc04 ' ,
} ,
2016-03-17 09:07:36 -06:00
' skip ' : ' This page is unavailable. ' ,
2015-01-20 08:23:51 -07:00
} ,
2015-02-14 03:10:23 -07:00
{
' url ' : ' http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844 ' ,
2023-10-06 17:56:19 -06:00
' md5 ' : ' 40d0e48c68896359c80372306ece0fc3 ' ,
2015-02-14 03:10:23 -07:00
' info_dict ' : {
2018-11-21 04:10:06 -07:00
' id ' : ' 394064451844 ' ,
2015-02-14 03:10:23 -07:00
' ext ' : ' mp4 ' ,
' title ' : ' Nightly News with Brian Williams Full Broadcast (February 4) ' ,
' description ' : ' md5:1c10c1eccbe84a26e5debb4381e2d3c5 ' ,
2016-06-22 18:36:19 -06:00
' timestamp ' : 1423104900 ,
' upload_date ' : ' 20150205 ' ,
2023-10-06 17:56:19 -06:00
' duration ' : 1236.0 ,
' thumbnail ' : ' https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/nn_netcast_150204.jpg ' ,
2015-02-14 03:10:23 -07:00
} ,
} ,
2016-03-17 09:07:36 -06:00
{
' url ' : ' http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456 ' ,
2023-10-06 17:56:19 -06:00
' md5 ' : ' ffb59bcf0733dc3c7f0ace907f5e3939 ' ,
2016-03-17 09:07:36 -06:00
' info_dict ' : {
2019-10-28 12:31:20 -06:00
' id ' : ' n431456 ' ,
2016-03-17 09:07:36 -06:00
' ext ' : ' mp4 ' ,
2019-10-28 12:31:20 -06:00
' title ' : " Volkswagen U.S. Chief: We ' Totally Screwed Up ' " ,
' description ' : ' md5:d22d1281a24f22ea0880741bb4dd6301 ' ,
2016-06-22 18:36:19 -06:00
' upload_date ' : ' 20150922 ' ,
' timestamp ' : 1442917800 ,
2023-10-06 17:56:19 -06:00
' duration ' : 37.0 ,
' thumbnail ' : ' https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/x_lon_vwhorn_150922.jpg ' ,
2016-03-17 09:07:36 -06:00
} ,
} ,
2016-04-22 11:07:40 -06:00
{
' url ' : ' http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788 ' ,
2023-10-06 17:56:19 -06:00
' md5 ' : ' 693d1fa21d23afcc9b04c66b227ed9ff ' ,
2016-04-22 11:07:40 -06:00
' info_dict ' : {
2018-11-21 04:10:06 -07:00
' id ' : ' 669831235788 ' ,
2016-04-22 11:07:40 -06:00
' ext ' : ' mp4 ' ,
' title ' : ' See the aurora borealis from space in stunning new NASA video ' ,
' description ' : ' md5:74752b7358afb99939c5f8bb2d1d04b1 ' ,
' upload_date ' : ' 20160420 ' ,
' timestamp ' : 1461152093 ,
2023-10-06 17:56:19 -06:00
' duration ' : 69.0 ,
' thumbnail ' : ' https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/201604/2016-04-20T11-35-09-133Z--1280x720.jpg ' ,
2016-06-22 18:36:19 -06:00
} ,
} ,
{
' url ' : ' http://www.msnbc.com/all-in-with-chris-hayes/watch/the-chaotic-gop-immigration-vote-314487875924 ' ,
' md5 ' : ' 6d236bf4f3dddc226633ce6e2c3f814d ' ,
' info_dict ' : {
2018-11-21 04:10:06 -07:00
' id ' : ' 314487875924 ' ,
2016-06-22 18:36:19 -06:00
' ext ' : ' mp4 ' ,
' title ' : ' The chaotic GOP immigration vote ' ,
' description ' : ' The Republican House votes on a border bill that has no chance of getting through the Senate or signed by the President and is drawing criticism from all sides. ' ,
2017-01-02 05:08:07 -07:00
' thumbnail ' : r ' re:^https?://.* \ .jpg$ ' ,
2016-06-22 18:36:19 -06:00
' timestamp ' : 1406937606 ,
' upload_date ' : ' 20140802 ' ,
2023-10-06 17:56:19 -06:00
' duration ' : 940.0 ,
2016-04-22 11:07:40 -06:00
} ,
} ,
2015-08-01 09:43:33 -06:00
{
' url ' : ' http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952 ' ,
' only_matching ' : True ,
} ,
2016-06-09 23:31:55 -06:00
{
# From http://www.vulture.com/2016/06/letterman-couldnt-care-less-about-late-night.html
' url ' : ' http://www.nbcnews.com/widget/video-embed/701714499682 ' ,
' only_matching ' : True ,
} ,
2014-05-29 16:38:57 -06:00
]
2013-08-27 04:38:30 -06:00
def _real_extract ( self , url ) :
2018-11-21 04:10:06 -07:00
video_id = self . _match_id ( url )
2019-10-28 12:31:20 -06:00
webpage = self . _download_webpage ( url , video_id )
2022-01-19 14:55:15 -07:00
data = self . _search_nextjs_data ( webpage , video_id ) [ ' props ' ] [ ' initialState ' ]
2019-10-28 12:31:20 -06:00
video_data = try_get ( data , lambda x : x [ ' video ' ] [ ' current ' ] , dict )
if not video_data :
video_data = data [ ' article ' ] [ ' content ' ] [ 0 ] [ ' primaryMedia ' ] [ ' video ' ]
title = video_data [ ' headline ' ] [ ' primary ' ]
formats = [ ]
for va in video_data . get ( ' videoAssets ' , [ ] ) :
public_url = va . get ( ' publicUrl ' )
if not public_url :
continue
if ' ://link.theplatform.com/ ' in public_url :
public_url = update_url_query ( public_url , { ' format ' : ' redirect ' } )
format_id = va . get ( ' format ' )
if format_id == ' M3U ' :
formats . extend ( self . _extract_m3u8_formats (
public_url , video_id , ' mp4 ' , ' m3u8_native ' ,
m3u8_id = format_id , fatal = False ) )
continue
tbr = int_or_none ( va . get ( ' bitrate ' ) , 1000 )
if tbr :
format_id + = ' - %d ' % tbr
formats . append ( {
' format_id ' : format_id ,
' url ' : public_url ,
' width ' : int_or_none ( va . get ( ' width ' ) ) ,
' height ' : int_or_none ( va . get ( ' height ' ) ) ,
' tbr ' : tbr ,
' ext ' : ' mp4 ' ,
} )
2016-12-17 10:11:13 -07:00
2019-10-28 12:31:20 -06:00
subtitles = { }
closed_captioning = video_data . get ( ' closedCaptioning ' )
if closed_captioning :
for cc_url in closed_captioning . values ( ) :
if not cc_url :
continue
subtitles . setdefault ( ' en ' , [ ] ) . append ( {
' url ' : cc_url ,
} )
2016-12-17 10:11:13 -07:00
2018-11-21 04:10:06 -07:00
return {
' id ' : video_id ,
2019-10-28 12:31:20 -06:00
' title ' : title ,
' description ' : try_get ( video_data , lambda x : x [ ' description ' ] [ ' primary ' ] ) ,
' thumbnail ' : try_get ( video_data , lambda x : x [ ' primaryImage ' ] [ ' url ' ] [ ' primary ' ] ) ,
' duration ' : parse_duration ( video_data . get ( ' duration ' ) ) ,
' timestamp ' : unified_timestamp ( video_data . get ( ' datePublished ' ) ) ,
' formats ' : formats ,
' subtitles ' : subtitles ,
2018-11-21 04:10:06 -07:00
}
2016-09-11 12:55:15 -06:00
class NBCOlympicsIE ( InfoExtractor ) :
2018-02-08 18:05:12 -07:00
IE_NAME = ' nbcolympics '
2021-08-03 22:19:44 -06:00
_VALID_URL = r ' https?://www \ .nbcolympics \ .com/videos?/(?P<id>[0-9a-z-]+) '
2016-09-11 12:55:15 -06:00
_TEST = {
# Geo-restricted to US
' url ' : ' http://www.nbcolympics.com/video/justin-roses-son-leo-was-tears-after-his-dad-won-gold ' ,
' md5 ' : ' 54fecf846d05429fbaa18af557ee523a ' ,
' info_dict ' : {
' id ' : ' WjTBzDXx5AUq ' ,
' display_id ' : ' justin-roses-son-leo-was-tears-after-his-dad-won-gold ' ,
' ext ' : ' mp4 ' ,
' title ' : ' Rose \' s son Leo was in tears after his dad won gold ' ,
' description ' : ' Olympic gold medalist Justin Rose gets emotional talking to the impact his win in men \' s golf has already had on his children. ' ,
' timestamp ' : 1471274964 ,
' upload_date ' : ' 20160815 ' ,
' uploader ' : ' NBCU-SPORTS ' ,
} ,
2023-10-06 17:56:19 -06:00
' skip ' : ' 404 Not Found ' ,
2016-09-11 12:55:15 -06:00
}
def _real_extract ( self , url ) :
display_id = self . _match_id ( url )
webpage = self . _download_webpage ( url , display_id )
2021-08-03 22:19:44 -06:00
try :
drupal_settings = self . _parse_json ( self . _search_regex (
r ' jQuery \ .extend \ (Drupal \ .settings \ s*, \ s*( { .+?}) \ ); ' ,
webpage , ' drupal settings ' ) , display_id )
iframe_url = drupal_settings [ ' vod ' ] [ ' iframe_url ' ]
theplatform_url = iframe_url . replace (
' vplayer.nbcolympics.com ' , ' player.theplatform.com ' )
except RegexNotFoundError :
theplatform_url = self . _search_regex (
r " ([ \" ' ])embedUrl \ 1: *([ \" ' ])(?P<embedUrl>.+) \ 2 " ,
webpage , ' embedding URL ' , group = " embedUrl " )
2016-09-11 12:55:15 -06:00
return {
' _type ' : ' url_transparent ' ,
' url ' : theplatform_url ,
' ie_key ' : ThePlatformIE . ie_key ( ) ,
' display_id ' : display_id ,
}
2018-02-08 18:05:12 -07:00
class NBCOlympicsStreamIE ( AdobePassIE ) :
IE_NAME = ' nbcolympics:stream '
_VALID_URL = r ' https?://stream \ .nbcolympics \ .com/(?P<id>[0-9a-z-]+) '
2021-07-29 09:31:27 -06:00
_TESTS = [
{
' note ' : ' Tokenized m3u8 source URL ' ,
' url ' : ' https://stream.nbcolympics.com/womens-soccer-group-round-11 ' ,
' info_dict ' : {
' id ' : ' 2019740 ' ,
' ext ' : ' mp4 ' ,
' title ' : r " re:Women ' s Group Stage - Netherlands vs \ . Brazil [0-9] {4} -[0-9] {2} -[0-9] {2} [0-9] {2} :[0-9] {2} $ " ,
} ,
' params ' : {
' skip_download ' : ' m3u8 ' ,
} ,
2023-10-06 17:56:19 -06:00
' skip ' : ' Livestream ' ,
2021-07-29 09:31:27 -06:00
} , {
' note ' : ' Plain m3u8 source URL ' ,
' url ' : ' https://stream.nbcolympics.com/gymnastics-event-finals-mens-floor-pommel-horse-womens-vault-bars ' ,
' info_dict ' : {
' id ' : ' 2021729 ' ,
' ext ' : ' mp4 ' ,
' title ' : r ' re:Event Finals: M Floor, W Vault, M Pommel, W Uneven Bars [0-9] {4} -[0-9] {2} -[0-9] {2} [0-9] {2} :[0-9] {2} $ ' ,
} ,
' params ' : {
' skip_download ' : ' m3u8 ' ,
} ,
2023-10-06 17:56:19 -06:00
' skip ' : ' Livestream ' ,
2018-02-08 18:05:12 -07:00
} ,
2021-07-29 09:31:27 -06:00
]
2018-02-08 18:05:12 -07:00
def _real_extract ( self , url ) :
display_id = self . _match_id ( url )
webpage = self . _download_webpage ( url , display_id )
pid = self . _search_regex ( r ' pid \ s*= \ s*( \ d+); ' , webpage , ' pid ' )
2021-07-29 09:31:27 -06:00
2018-02-08 18:05:12 -07:00
event_config = self . _download_json (
2021-07-29 09:31:27 -06:00
f ' http://stream.nbcolympics.com/data/event_config_ { pid } .json ' ,
pid , ' Downloading event config ' ) [ ' eventConfig ' ]
title = event_config [ ' eventTitle ' ]
is_live = { ' live ' : True , ' replay ' : False } . get ( event_config . get ( ' eventStatus ' ) )
2018-02-08 18:05:12 -07:00
source_url = self . _download_json (
2021-07-29 09:31:27 -06:00
f ' https://api-leap.nbcsports.com/feeds/assets/ { pid } ?application=NBCOlympics&platform=desktop&format=nbc-player&env=staging ' ,
pid , ' Downloading leap config '
) [ ' videoSources ' ] [ 0 ] [ ' cdnSources ' ] [ ' primary ' ] [ 0 ] [ ' sourceUrl ' ]
if event_config . get ( ' cdnToken ' ) :
ap_resource = self . _get_mvpd_resource (
event_config . get ( ' resourceId ' , ' NBCOlympics ' ) ,
re . sub ( r ' [^ \ w \ d ]+ ' , ' ' , event_config [ ' eventTitle ' ] ) , pid ,
event_config . get ( ' ratingId ' , ' NO VALUE ' ) )
media_token = self . _extract_mvpd_auth ( url , pid , event_config . get ( ' requestorId ' , ' NBCOlympics ' ) , ap_resource )
source_url = self . _download_json (
' https://tokens.playmakerservices.com/ ' , pid , ' Retrieving tokenized URL ' ,
data = json . dumps ( {
' application ' : ' NBCSports ' ,
' authentication-type ' : ' adobe-pass ' ,
' cdn ' : ' akamai ' ,
' pid ' : pid ,
' platform ' : ' desktop ' ,
' requestorId ' : ' NBCOlympics ' ,
' resourceId ' : base64 . b64encode ( ap_resource . encode ( ) ) . decode ( ) ,
' token ' : base64 . b64encode ( media_token . encode ( ) ) . decode ( ) ,
' url ' : source_url ,
' version ' : ' v1 ' ,
} ) . encode ( ) ,
) [ ' akamai ' ] [ 0 ] [ ' tokenizedUrl ' ]
formats = self . _extract_m3u8_formats ( source_url , pid , ' mp4 ' , live = is_live )
for f in formats :
# -http_seekable requires ffmpeg 4.3+ but it doesnt seem possible to
# download with ffmpeg without this option
2022-04-27 10:22:57 -06:00
f [ ' downloader_options ' ] = { ' ffmpeg_args ' : [ ' -seekable ' , ' 0 ' , ' -http_seekable ' , ' 0 ' , ' -icy ' , ' 0 ' ] }
2018-02-08 18:05:12 -07:00
return {
' id ' : pid ,
' display_id ' : display_id ,
' title ' : title ,
' formats ' : formats ,
2021-07-29 09:31:27 -06:00
' is_live ' : is_live ,
2018-02-08 18:05:12 -07:00
}
2022-09-30 09:43:40 -06:00
class NBCStationsIE ( InfoExtractor ) :
_DOMAIN_RE = ' | ' . join ( map ( re . escape , (
' nbcbayarea ' , ' nbcboston ' , ' nbcchicago ' , ' nbcconnecticut ' , ' nbcdfw ' , ' nbclosangeles ' ,
' nbcmiami ' , ' nbcnewyork ' , ' nbcphiladelphia ' , ' nbcsandiego ' , ' nbcwashington ' ,
' necn ' , ' telemundo52 ' , ' telemundoarizona ' , ' telemundochicago ' , ' telemundonuevainglaterra ' ,
) ) )
_VALID_URL = rf ' https?://(?:www \ .)?(?P<site> { _DOMAIN_RE } ) \ .com/(?:[^/?#]+/)*(?P<id>[^/?#]+)/?(?:$|[#?]) '
_TESTS = [ {
' url ' : ' https://www.nbclosangeles.com/news/local/large-structure-fire-in-downtown-la-prompts-smoke-odor-advisory/2968618/ ' ,
' info_dict ' : {
' id ' : ' 2968618 ' ,
' ext ' : ' mp4 ' ,
' title ' : ' Large Structure Fire in Downtown LA Prompts Smoke Odor Advisory ' ,
2023-01-14 09:40:42 -07:00
' description ' : ' md5:417ed3c2d91fe9d301e6db7b0942f182 ' ,
2023-03-26 16:27:39 -06:00
' duration ' : 112.513 ,
2022-09-30 09:43:40 -06:00
' timestamp ' : 1661135892 ,
2023-01-14 09:40:42 -07:00
' upload_date ' : ' 20220822 ' ,
2022-09-30 09:43:40 -06:00
' uploader ' : ' NBC 4 ' ,
2023-01-14 09:40:42 -07:00
' channel_id ' : ' KNBC ' ,
2022-09-30 09:43:40 -06:00
' channel ' : ' nbclosangeles ' ,
} ,
2023-01-14 09:40:42 -07:00
' params ' : {
' skip_download ' : ' m3u8 ' ,
} ,
2022-09-30 09:43:40 -06:00
} , {
' url ' : ' https://www.telemundoarizona.com/responde/huracan-complica-reembolso-para-televidente-de-tucson/2247002/ ' ,
' info_dict ' : {
' id ' : ' 2247002 ' ,
' ext ' : ' mp4 ' ,
2023-01-14 09:40:42 -07:00
' title ' : ' Huracán complica que televidente de Tucson reciba reembolso ' ,
2022-09-30 09:43:40 -06:00
' description ' : ' md5:af298dc73aab74d4fca6abfb12acb6cf ' ,
2023-03-26 16:27:39 -06:00
' duration ' : 172.406 ,
2022-09-30 09:43:40 -06:00
' timestamp ' : 1660886507 ,
' upload_date ' : ' 20220819 ' ,
' uploader ' : ' Telemundo Arizona ' ,
2023-01-14 09:40:42 -07:00
' channel_id ' : ' KTAZ ' ,
2022-09-30 09:43:40 -06:00
' channel ' : ' telemundoarizona ' ,
} ,
2023-01-14 09:40:42 -07:00
' params ' : {
' skip_download ' : ' m3u8 ' ,
} ,
2023-03-26 16:27:39 -06:00
} , {
# direct mp4 link
' url ' : ' https://www.nbcboston.com/weather/video-weather/highs-near-freezing-in-boston-on-wednesday/2961135/ ' ,
' md5 ' : ' 9bf8c41dc7abbb75b1a44f1491a4cc85 ' ,
' info_dict ' : {
' id ' : ' 2961135 ' ,
' ext ' : ' mp4 ' ,
' title ' : ' Highs Near Freezing in Boston on Wednesday ' ,
' description ' : ' md5:3ec486609a926c99f00a3512e6c0e85b ' ,
' duration ' : 235.669 ,
' timestamp ' : 1675268656 ,
' upload_date ' : ' 20230201 ' ,
' uploader ' : ' ' ,
' channel_id ' : ' WBTS ' ,
' channel ' : ' nbcboston ' ,
} ,
2022-09-30 09:43:40 -06:00
} ]
_RESOLUTIONS = {
' 1080 ' : ' 1920 ' ,
' 720 ' : ' 1280 ' ,
' 540 ' : ' 960 ' ,
' 360 ' : ' 640 ' ,
' 234 ' : ' 416 ' ,
}
def _real_extract ( self , url ) :
channel , video_id = self . _match_valid_url ( url ) . group ( ' site ' , ' id ' )
webpage = self . _download_webpage ( url , video_id )
nbc_data = self . _search_json (
2023-01-16 14:38:33 -07:00
r ' <script> \ s*var \ s+nbc \ s*= ' , webpage , ' NBC JSON data ' , video_id )
2022-09-30 09:43:40 -06:00
pdk_acct = nbc_data . get ( ' pdkAcct ' ) or ' Yh1nAC '
fw_ssid = traverse_obj ( nbc_data , ( ' video ' , ' fwSSID ' ) )
2023-01-14 09:40:42 -07:00
video_data = self . _search_json (
r ' data-videos= " \ [ ' , webpage , ' video data ' , video_id , default = { } , transform_source = unescapeHTML )
video_data . update ( self . _search_json (
r ' data-meta= " ' , webpage , ' metadata ' , video_id , default = { } , transform_source = unescapeHTML ) )
if not video_data :
raise ExtractorError ( ' No video metadata found in webpage ' , expected = True )
2022-09-30 09:43:40 -06:00
2023-03-26 16:27:39 -06:00
info , formats = { } , [ ]
2023-01-14 09:40:42 -07:00
is_live = int_or_none ( video_data . get ( ' mpx_is_livestream ' ) ) == 1
query = {
' formats ' : ' MPEG-DASH none,M3U none,MPEG-DASH none,MPEG4,MP3 ' ,
' format ' : ' SMIL ' ,
' fwsitesection ' : fw_ssid ,
' fwNetworkID ' : traverse_obj ( nbc_data , ( ' video ' , ' fwNetworkID ' ) , default = ' 382114 ' ) ,
' pprofile ' : ' ots_desktop_html ' ,
' sensitive ' : ' false ' ,
' w ' : ' 1920 ' ,
' h ' : ' 1080 ' ,
' mode ' : ' LIVE ' if is_live else ' on-demand ' ,
' vpaid ' : ' script ' ,
' schema ' : ' 2.0 ' ,
' sdk ' : ' PDK 6.1.3 ' ,
}
2022-09-30 09:43:40 -06:00
2023-01-14 09:40:42 -07:00
if is_live :
player_id = traverse_obj ( video_data , ( ( None , ( ' video ' , ' meta ' ) ) , (
' mpx_m3upid ' , ' mpx_pid ' , ' pid_streaming_web_medium ' ) ) , get_all = False )
info [ ' title ' ] = f ' { channel } livestream '
2022-09-30 09:43:40 -06:00
else :
2023-01-14 09:40:42 -07:00
player_id = traverse_obj ( video_data , (
( None , ( ' video ' , ' meta ' ) ) , ( ' pid_streaming_web_high ' , ' mpx_pid ' ) ) , get_all = False )
2022-09-30 09:43:40 -06:00
date_string = traverse_obj ( video_data , ' date_string ' , ' date_gmt ' )
if date_string :
date_string = self . _search_regex (
r ' datetime= " ([^ " ]+) " ' , date_string , ' date string ' , fatal = False )
else :
date_string = traverse_obj (
2023-01-14 09:40:42 -07:00
nbc_data , ( ' dataLayer ' , ' adobe ' , ( ' prop70 ' , ' eVar70 ' , ' eVar59 ' ) ) , get_all = False )
2022-09-30 09:43:40 -06:00
2023-01-14 09:40:42 -07:00
video_url = traverse_obj ( video_data , ( ( None , ( ' video ' , ' meta ' ) ) , ' mp4_url ' ) , get_all = False )
2022-09-30 09:43:40 -06:00
if video_url :
2023-03-26 16:27:39 -06:00
ext = determine_ext ( video_url )
2023-01-14 09:40:42 -07:00
height = self . _search_regex ( r ' \ d+-( \ d+)p ' , url_basename ( video_url ) , ' height ' , default = None )
2022-09-30 09:43:40 -06:00
formats . append ( {
' url ' : video_url ,
2023-03-26 16:27:39 -06:00
' ext ' : ext ,
2022-09-30 09:43:40 -06:00
' width ' : int_or_none ( self . _RESOLUTIONS . get ( height ) ) ,
' height ' : int_or_none ( height ) ,
2023-03-26 16:27:39 -06:00
' format_id ' : f ' http- { ext } ' ,
2022-09-30 09:43:40 -06:00
} )
2023-01-14 09:40:42 -07:00
info . update ( {
' title ' : video_data . get ( ' title ' ) or traverse_obj ( nbc_data , (
' dataLayer ' , ( None , ' adobe ' ) , ( ' contenttitle ' , ' title ' , ' prop22 ' ) ) , get_all = False ) ,
' description ' :
traverse_obj ( video_data , ' summary ' , ' excerpt ' , ' video_hero_text ' )
or clean_html ( traverse_obj ( nbc_data , ( ' dataLayer ' , ' summary ' ) ) ) ,
' timestamp ' : unified_timestamp ( date_string ) ,
} )
2022-09-30 09:43:40 -06:00
2023-01-14 09:40:42 -07:00
smil = None
if player_id and fw_ssid :
smil = self . _download_xml (
f ' https://link.theplatform.com/s/ { pdk_acct } / { player_id } ' , video_id ,
note = ' Downloading SMIL data ' , query = query , fatal = is_live )
2023-03-26 16:27:39 -06:00
subtitles = self . _parse_smil_subtitles ( smil , default_ns ) if smil else { }
for video in smil . findall ( self . _xpath_ns ( ' .//video ' , default_ns ) ) if smil else [ ] :
info [ ' duration ' ] = float_or_none ( remove_end ( video . get ( ' dur ' ) , ' ms ' ) , 1000 )
video_src_url = video . get ( ' src ' )
ext = mimetype2ext ( video . get ( ' type ' ) , default = determine_ext ( video_src_url ) )
if ext == ' m3u8 ' :
fmts , subs = self . _extract_m3u8_formats_and_subtitles (
video_src_url , video_id , ' mp4 ' , m3u8_id = ' hls ' , fatal = is_live ,
live = is_live , errnote = ' No HLS formats found ' )
formats . extend ( fmts )
self . _merge_subtitles ( subs , target = subtitles )
elif video_src_url :
formats . append ( {
' url ' : video_src_url ,
' format_id ' : f ' https- { ext } ' ,
' ext ' : ext ,
' width ' : int_or_none ( video . get ( ' width ' ) ) ,
' height ' : int_or_none ( video . get ( ' height ' ) ) ,
} )
2023-01-14 09:40:42 -07:00
if not formats :
self . raise_no_formats ( ' No video content found in webpage ' , expected = True )
elif is_live :
try :
self . _request_webpage (
HEADRequest ( formats [ 0 ] [ ' url ' ] ) , video_id , note = ' Checking live status ' )
except ExtractorError :
raise UserNotLive ( video_id = channel )
2022-09-30 09:43:40 -06:00
return {
2023-01-14 09:40:42 -07:00
' id ' : video_id ,
2022-09-30 09:43:40 -06:00
' channel ' : channel ,
2023-01-14 09:40:42 -07:00
' channel_id ' : nbc_data . get ( ' callLetters ' ) ,
' uploader ' : nbc_data . get ( ' on_air_name ' ) ,
2022-09-30 09:43:40 -06:00
' formats ' : formats ,
2023-01-14 09:40:42 -07:00
' subtitles ' : subtitles ,
' is_live ' : is_live ,
2022-09-30 09:43:40 -06:00
* * info ,
}