_ Git - youtube-dl/blob - youtube_dl/extractor/pbs.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     ExtractorError,
   9     determine_ext,
  10     int_or_none,
  11     float_or_none,
  12     js_to_json,
  13     orderedSet,
  14     strip_jsonp,
  15     strip_or_none,
  16     unified_strdate,
  17     US_RATINGS,
  18 )
  19
  20
  21 class PBSIE(InfoExtractor):
  22     _STATIONS = (
  23         (r'(?:video|www|player)\.pbs\.org', 'PBS: Public Broadcasting Service'),  # http://www.pbs.org/
  24         (r'video\.aptv\.org', 'APT - Alabama Public Television (WBIQ)'),  # http://aptv.org/
  25         (r'video\.gpb\.org', 'GPB/Georgia Public Broadcasting (WGTV)'),  # http://www.gpb.org/
  26         (r'video\.mpbonline\.org', 'Mississippi Public Broadcasting (WMPN)'),  # http://www.mpbonline.org
  27         (r'video\.wnpt\.org', 'Nashville Public Television (WNPT)'),  # http://www.wnpt.org
  28         (r'video\.wfsu\.org', 'WFSU-TV (WFSU)'),  # http://wfsu.org/
  29         (r'video\.wsre\.org', 'WSRE (WSRE)'),  # http://www.wsre.org
  30         (r'video\.wtcitv\.org', 'WTCI (WTCI)'),  # http://www.wtcitv.org
  31         (r'video\.pba\.org', 'WPBA/Channel 30 (WPBA)'),  # http://pba.org/
  32         (r'video\.alaskapublic\.org', 'Alaska Public Media (KAKM)'),  # http://alaskapublic.org/kakm
  33         # (r'kuac\.org', 'KUAC (KUAC)'),  # http://kuac.org/kuac-tv/
  34         # (r'ktoo\.org', '360 North (KTOO)'),  # http://www.ktoo.org/
  35         # (r'azpm\.org', 'KUAT 6 (KUAT)'),  # http://www.azpm.org/
  36         (r'video\.azpbs\.org', 'Arizona PBS (KAET)'),  # http://www.azpbs.org
  37         (r'portal\.knme\.org', 'KNME-TV/Channel 5 (KNME)'),  # http://www.newmexicopbs.org/
  38         (r'video\.vegaspbs\.org', 'Vegas PBS (KLVX)'),  # http://vegaspbs.org/
  39         (r'watch\.aetn\.org', 'AETN/ARKANSAS ETV NETWORK (KETS)'),  # http://www.aetn.org/
  40         (r'video\.ket\.org', 'KET (WKLE)'),  # http://www.ket.org/
  41         (r'video\.wkno\.org', 'WKNO/Channel 10 (WKNO)'),  # http://www.wkno.org/
  42         (r'video\.lpb\.org', 'LPB/LOUISIANA PUBLIC BROADCASTING (WLPB)'),  # http://www.lpb.org/
  43         (r'videos\.oeta\.tv', 'OETA (KETA)'),  # http://www.oeta.tv
  44         (r'video\.optv\.org', 'Ozarks Public Television (KOZK)'),  # http://www.optv.org/
  45         (r'watch\.wsiu\.org', 'WSIU Public Broadcasting (WSIU)'),  # http://www.wsiu.org/
  46         (r'video\.keet\.org', 'KEET TV (KEET)'),  # http://www.keet.org
  47         (r'pbs\.kixe\.org', 'KIXE/Channel 9 (KIXE)'),  # http://kixe.org/
  48         (r'video\.kpbs\.org', 'KPBS San Diego (KPBS)'),  # http://www.kpbs.org/
  49         (r'video\.kqed\.org', 'KQED (KQED)'),  # http://www.kqed.org
  50         (r'vids\.kvie\.org', 'KVIE Public Television (KVIE)'),  # http://www.kvie.org
  51         (r'video\.pbssocal\.org', 'PBS SoCal/KOCE (KOCE)'),  # http://www.pbssocal.org/
  52         (r'video\.valleypbs\.org', 'ValleyPBS (KVPT)'),  # http://www.valleypbs.org/
  53         (r'video\.cptv\.org', 'CONNECTICUT PUBLIC TELEVISION (WEDH)'),  # http://cptv.org
  54         (r'watch\.knpb\.org', 'KNPB Channel 5 (KNPB)'),  # http://www.knpb.org/
  55         (r'video\.soptv\.org', 'SOPTV (KSYS)'),  # http://www.soptv.org
  56         # (r'klcs\.org', 'KLCS/Channel 58 (KLCS)'),  # http://www.klcs.org
  57         # (r'krcb\.org', 'KRCB Television & Radio (KRCB)'),  # http://www.krcb.org
  58         # (r'kvcr\.org', 'KVCR TV/DT/FM :: Vision for the Future (KVCR)'),  # http://kvcr.org
  59         (r'video\.rmpbs\.org', 'Rocky Mountain PBS (KRMA)'),  # http://www.rmpbs.org
  60         (r'video\.kenw\.org', 'KENW-TV3 (KENW)'),  # http://www.kenw.org
  61         (r'video\.kued\.org', 'KUED Channel 7 (KUED)'),  # http://www.kued.org
  62         (r'video\.wyomingpbs\.org', 'Wyoming PBS (KCWC)'),  # http://www.wyomingpbs.org
  63         (r'video\.cpt12\.org', 'Colorado Public Television / KBDI 12 (KBDI)'),  # http://www.cpt12.org/
  64         (r'video\.kbyueleven\.org', 'KBYU-TV (KBYU)'),  # http://www.kbyutv.org/
  65         (r'video\.thirteen\.org', 'Thirteen/WNET New York (WNET)'),  # http://www.thirteen.org
  66         (r'video\.wgbh\.org', 'WGBH/Channel 2 (WGBH)'),  # http://wgbh.org
  67         (r'video\.wgby\.org', 'WGBY (WGBY)'),  # http://www.wgby.org
  68         (r'watch\.njtvonline\.org', 'NJTV Public Media NJ (WNJT)'),  # http://www.njtvonline.org/
  69         # (r'ripbs\.org', 'Rhode Island PBS (WSBE)'),  # http://www.ripbs.org/home/
  70         (r'watch\.wliw\.org', 'WLIW21 (WLIW)'),  # http://www.wliw.org/
  71         (r'video\.mpt\.tv', 'mpt/Maryland Public Television (WMPB)'),  # http://www.mpt.org
  72         (r'watch\.weta\.org', 'WETA Television and Radio (WETA)'),  # http://www.weta.org
  73         (r'video\.whyy\.org', 'WHYY (WHYY)'),  # http://www.whyy.org
  74         (r'video\.wlvt\.org', 'PBS 39 (WLVT)'),  # http://www.wlvt.org/
  75         (r'video\.wvpt\.net', 'WVPT - Your Source for PBS and More! (WVPT)'),  # http://www.wvpt.net
  76         (r'video\.whut\.org', 'Howard University Television (WHUT)'),  # http://www.whut.org
  77         (r'video\.wedu\.org', 'WEDU PBS (WEDU)'),  # http://www.wedu.org
  78         (r'video\.wgcu\.org', 'WGCU Public Media (WGCU)'),  # http://www.wgcu.org/
  79         # (r'wjct\.org', 'WJCT Public Broadcasting (WJCT)'),  # http://www.wjct.org
  80         (r'video\.wpbt2\.org', 'WPBT2 (WPBT)'),  # http://www.wpbt2.org
  81         (r'video\.wucftv\.org', 'WUCF TV (WUCF)'),  # http://wucftv.org
  82         (r'video\.wuft\.org', 'WUFT/Channel 5 (WUFT)'),  # http://www.wuft.org
  83         (r'watch\.wxel\.org', 'WXEL/Channel 42 (WXEL)'),  # http://www.wxel.org/home/
  84         (r'video\.wlrn\.org', 'WLRN/Channel 17 (WLRN)'),  # http://www.wlrn.org/
  85         (r'video\.wusf\.usf\.edu', 'WUSF Public Broadcasting (WUSF)'),  # http://wusf.org/
  86         (r'video\.scetv\.org', 'ETV (WRLK)'),  # http://www.scetv.org
  87         (r'video\.unctv\.org', 'UNC-TV (WUNC)'),  # http://www.unctv.org/
  88         # (r'pbsguam\.org', 'PBS Guam (KGTF)'),  # http://www.pbsguam.org/
  89         (r'video\.pbshawaii\.org', 'PBS Hawaii - Oceanic Cable Channel 10 (KHET)'),  # http://www.pbshawaii.org/
  90         (r'video\.idahoptv\.org', 'Idaho Public Television (KAID)'),  # http://idahoptv.org
  91         (r'video\.ksps\.org', 'KSPS (KSPS)'),  # http://www.ksps.org/home/
  92         (r'watch\.opb\.org', 'OPB (KOPB)'),  # http://www.opb.org
  93         (r'watch\.nwptv\.org', 'KWSU/Channel 10 & KTNW/Channel 31 (KWSU)'),  # http://www.kwsu.org
  94         (r'video\.will\.illinois\.edu', 'WILL-TV (WILL)'),  # http://will.illinois.edu/
  95         (r'video\.networkknowledge\.tv', 'Network Knowledge - WSEC/Springfield (WSEC)'),  # http://www.wsec.tv
  96         (r'video\.wttw\.com', 'WTTW11 (WTTW)'),  # http://www.wttw.com/
  97         # (r'wtvp\.org', 'WTVP & WTVP.org, Public Media for Central Illinois (WTVP)'),  # http://www.wtvp.org/
  98         (r'video\.iptv\.org', 'Iowa Public Television/IPTV (KDIN)'),  # http://www.iptv.org/
  99         (r'video\.ninenet\.org', 'Nine Network (KETC)'),  # http://www.ninenet.org
 100         (r'video\.wfwa\.org', 'PBS39 Fort Wayne (WFWA)'),  # http://wfwa.org/
 101         (r'video\.wfyi\.org', 'WFYI Indianapolis (WFYI)'),  # http://www.wfyi.org
 102         (r'video\.mptv\.org', 'Milwaukee Public Television (WMVS)'),  # http://www.mptv.org
 103         (r'video\.wnin\.org', 'WNIN (WNIN)'),  # http://www.wnin.org/
 104         (r'video\.wnit\.org', 'WNIT Public Television (WNIT)'),  # http://www.wnit.org/
 105         (r'video\.wpt\.org', 'WPT (WPNE)'),  # http://www.wpt.org/
 106         (r'video\.wvut\.org', 'WVUT/Channel 22 (WVUT)'),  # http://wvut.org/
 107         (r'video\.weiu\.net', 'WEIU/Channel 51 (WEIU)'),  # http://www.weiu.net
 108         (r'video\.wqpt\.org', 'WQPT-TV (WQPT)'),  # http://www.wqpt.org
 109         (r'video\.wycc\.org', 'WYCC PBS Chicago (WYCC)'),  # http://www.wycc.org
 110         # (r'lakeshorepublicmedia\.org', 'Lakeshore Public Television (WYIN)'),  # http://lakeshorepublicmedia.org/
 111         (r'video\.wipb\.org', 'WIPB-TV (WIPB)'),  # http://wipb.org
 112         (r'video\.indianapublicmedia\.org', 'WTIU (WTIU)'),  # http://indianapublicmedia.org/tv/
 113         (r'watch\.cetconnect\.org', 'CET  (WCET)'),  # http://www.cetconnect.org
 114         (r'video\.thinktv\.org', 'ThinkTVNetwork (WPTD)'),  # http://www.thinktv.org
 115         (r'video\.wbgu\.org', 'WBGU-TV (WBGU)'),  # http://wbgu.org
 116         (r'video\.wgvu\.org', 'WGVU TV (WGVU)'),  # http://www.wgvu.org/
 117         (r'video\.netnebraska\.org', 'NET1 (KUON)'),  # http://netnebraska.org
 118         (r'video\.pioneer\.org', 'Pioneer Public Television (KWCM)'),  # http://www.pioneer.org
 119         (r'watch\.sdpb\.org', 'SDPB Television (KUSD)'),  # http://www.sdpb.org
 120         (r'video\.tpt\.org', 'TPT (KTCA)'),  # http://www.tpt.org
 121         (r'watch\.ksmq\.org', 'KSMQ (KSMQ)'),  # http://www.ksmq.org/
 122         (r'watch\.kpts\.org', 'KPTS/Channel 8 (KPTS)'),  # http://www.kpts.org/
 123         (r'watch\.ktwu\.org', 'KTWU/Channel 11 (KTWU)'),  # http://ktwu.org
 124         # (r'shptv\.org', 'Smoky Hills Public Television (KOOD)'),  # http://www.shptv.org
 125         # (r'kcpt\.org', 'KCPT Kansas City Public Television (KCPT)'),  # http://kcpt.org/
 126         # (r'blueridgepbs\.org', 'Blue Ridge PBS (WBRA)'),  # http://www.blueridgepbs.org/
 127         (r'watch\.easttennesseepbs\.org', 'East Tennessee PBS (WSJK)'),  # http://easttennesseepbs.org
 128         (r'video\.wcte\.tv', 'WCTE-TV (WCTE)'),  # http://www.wcte.org
 129         (r'video\.wljt\.org', 'WLJT, Channel 11 (WLJT)'),  # http://wljt.org/
 130         (r'video\.wosu\.org', 'WOSU TV (WOSU)'),  # http://wosu.org/
 131         (r'video\.woub\.org', 'WOUB/WOUC (WOUB)'),  # http://woub.org/tv/index.php?section=5
 132         (r'video\.wvpublic\.org', 'WVPB (WVPB)'),  # http://wvpublic.org/
 133         (r'video\.wkyupbs\.org', 'WKYU-PBS (WKYU)'),  # http://www.wkyupbs.org
 134         # (r'wyes\.org', 'WYES-TV/New Orleans (WYES)'),  # http://www.wyes.org
 135         (r'video\.kera\.org', 'KERA 13 (KERA)'),  # http://www.kera.org/
 136         (r'video\.mpbn\.net', 'MPBN (WCBB)'),  # http://www.mpbn.net/
 137         (r'video\.mountainlake\.org', 'Mountain Lake PBS (WCFE)'),  # http://www.mountainlake.org/
 138         (r'video\.nhptv\.org', 'NHPTV (WENH)'),  # http://nhptv.org/
 139         (r'video\.vpt\.org', 'Vermont PBS (WETK)'),  # http://www.vpt.org
 140         (r'video\.witf\.org', 'witf (WITF)'),  # http://www.witf.org
 141         (r'watch\.wqed\.org', 'WQED Multimedia (WQED)'),  # http://www.wqed.org/
 142         (r'video\.wmht\.org', 'WMHT Educational Telecommunications (WMHT)'),  # http://www.wmht.org/home/
 143         (r'video\.deltabroadcasting\.org', 'Q-TV (WDCQ)'),  # http://www.deltabroadcasting.org
 144         (r'video\.dptv\.org', 'WTVS Detroit Public TV (WTVS)'),  # http://www.dptv.org/
 145         (r'video\.wcmu\.org', 'CMU Public Television (WCMU)'),  # http://www.wcmu.org
 146         (r'video\.wkar\.org', 'WKAR-TV (WKAR)'),  # http://wkar.org/
 147         (r'wnmuvideo\.nmu\.edu', 'WNMU-TV Public TV 13 (WNMU)'),  # http://wnmutv.nmu.edu
 148         (r'video\.wdse\.org', 'WDSE - WRPT (WDSE)'),  # http://www.wdse.org/
 149         (r'video\.wgte\.org', 'WGTE TV (WGTE)'),  # http://www.wgte.org
 150         (r'video\.lptv\.org', 'Lakeland Public Television (KAWE)'),  # http://www.lakelandptv.org
 151         # (r'prairiepublic\.org', 'PRAIRIE PUBLIC (KFME)'),  # http://www.prairiepublic.org/
 152         (r'video\.kmos\.org', 'KMOS-TV - Channels 6.1, 6.2 and 6.3 (KMOS)'),  # http://www.kmos.org/
 153         (r'watch\.montanapbs\.org', 'MontanaPBS (KUSM)'),  # http://montanapbs.org
 154         (r'video\.krwg\.org', 'KRWG/Channel 22 (KRWG)'),  # http://www.krwg.org
 155         (r'video\.kacvtv\.org', 'KACV (KACV)'),  # http://www.panhandlepbs.org/home/
 156         (r'video\.kcostv\.org', 'KCOS/Channel 13 (KCOS)'),  # www.kcostv.org
 157         (r'video\.wcny\.org', 'WCNY/Channel 24 (WCNY)'),  # http://www.wcny.org
 158         (r'video\.wned\.org', 'WNED (WNED)'),  # http://www.wned.org/
 159         (r'watch\.wpbstv\.org', 'WPBS (WPBS)'),  # http://www.wpbstv.org
 160         (r'video\.wskg\.org', 'WSKG Public TV (WSKG)'),  # http://wskg.org
 161         (r'video\.wxxi\.org', 'WXXI (WXXI)'),  # http://wxxi.org
 162         (r'video\.wpsu\.org', 'WPSU (WPSU)'),  # http://www.wpsu.org
 163         # (r'wqln\.org', 'WQLN/Channel 54 (WQLN)'),  # http://www.wqln.org
 164         (r'on-demand\.wvia\.org', 'WVIA Public Media Studios (WVIA)'),  # http://www.wvia.org/
 165         (r'video\.wtvi\.org', 'WTVI (WTVI)'),  # http://www.wtvi.org/
 166         # (r'whro\.org', 'WHRO (WHRO)'),  # http://whro.org
 167         (r'video\.westernreservepublicmedia\.org', 'Western Reserve PBS (WNEO)'),  # http://www.WesternReservePublicMedia.org/
 168         (r'video\.ideastream\.org', 'WVIZ/PBS ideastream (WVIZ)'),  # http://www.wviz.org/
 169         (r'video\.kcts9\.org', 'KCTS 9 (KCTS)'),  # http://kcts9.org/
 170         (r'video\.basinpbs\.org', 'Basin PBS (KPBT)'),  # http://www.basinpbs.org
 171         (r'video\.houstonpbs\.org', 'KUHT / Channel 8 (KUHT)'),  # http://www.houstonpublicmedia.org/
 172         # (r'tamu\.edu', 'KAMU - TV (KAMU)'),  # http://KAMU.tamu.edu
 173         # (r'kedt\.org', 'KEDT/Channel 16 (KEDT)'),  # http://www.kedt.org
 174         (r'video\.klrn\.org', 'KLRN (KLRN)'),  # http://www.klrn.org
 175         (r'video\.klru\.tv', 'KLRU (KLRU)'),  # http://www.klru.org
 176         # (r'kmbh\.org', 'KMBH-TV (KMBH)'),  # http://www.kmbh.org
 177         # (r'knct\.org', 'KNCT (KNCT)'),  # http://www.knct.org
 178         # (r'ktxt\.org', 'KTTZ-TV (KTXT)'),  # http://www.ktxt.org
 179         (r'video\.wtjx\.org', 'WTJX Channel 12 (WTJX)'),  # http://www.wtjx.org/
 180         (r'video\.ideastations\.org', 'WCVE PBS (WCVE)'),  # http://ideastations.org/
 181         (r'video\.kbtc\.org', 'KBTC Public Television (KBTC)'),  # http://kbtc.org
 182     )
 183
 184     IE_NAME = 'pbs'
 185     IE_DESC = 'Public Broadcasting Service (PBS) and member stations: %s' % ', '.join(list(zip(*_STATIONS))[1])
 186
 187     _VALID_URL = r'''(?x)https?://
 188         (?:
 189            # Direct video URL
 190            (?:%s)/(?:(?:vir|port)alplayer|video)/(?P<id>[0-9]+)(?:[?/]|$) |
 191            # Article with embedded player (or direct video)
 192            (?:www\.)?pbs\.org/(?:[^/]+/){1,5}(?P<presumptive_id>[^/]+?)(?:\.html)?/?(?:$|[?\#]) |
 193            # Player
 194            (?:video|player)\.pbs\.org/(?:widget/)?partnerplayer/(?P<player_id>[^/]+)/
 195         )
 196     ''' % '|'.join(list(zip(*_STATIONS))[0])
 197
 198     _GEO_COUNTRIES = ['US']
 199
 200     _TESTS = [
 201         {
 202             'url': 'http://www.pbs.org/tpt/constitution-usa-peter-sagal/watch/a-more-perfect-union/',
 203             'md5': '173dc391afd361fa72eab5d3d918968d',
 204             'info_dict': {
 205                 'id': '2365006249',
 206                 'ext': 'mp4',
 207                 'title': 'Constitution USA with Peter Sagal - A More Perfect Union',
 208                 'description': 'md5:31b664af3c65fd07fa460d306b837d00',
 209                 'duration': 3190,
 210             },
 211         },
 212         {
 213             'url': 'http://www.pbs.org/wgbh/pages/frontline/losing-iraq/',
 214             'md5': '6f722cb3c3982186d34b0f13374499c7',
 215             'info_dict': {
 216                 'id': '2365297690',
 217                 'ext': 'mp4',
 218                 'title': 'FRONTLINE - Losing Iraq',
 219                 'description': 'md5:5979a4d069b157f622d02bff62fbe654',
 220                 'duration': 5050,
 221             },
 222         },
 223         {
 224             'url': 'http://www.pbs.org/newshour/bb/education-jan-june12-cyberschools_02-23/',
 225             'md5': 'b19856d7f5351b17a5ab1dc6a64be633',
 226             'info_dict': {
 227                 'id': '2201174722',
 228                 'ext': 'mp4',
 229                 'title': 'PBS NewsHour - Cyber Schools Gain Popularity, but Quality Questions Persist',
 230                 'description': 'md5:86ab9a3d04458b876147b355788b8781',
 231                 'duration': 801,
 232             },
 233         },
 234         {
 235             'url': 'http://www.pbs.org/wnet/gperf/dudamel-conducts-verdi-requiem-hollywood-bowl-full-episode/3374/',
 236             'md5': 'c62859342be2a0358d6c9eb306595978',
 237             'info_dict': {
 238                 'id': '2365297708',
 239                 'ext': 'mp4',
 240                 'title': 'Great Performances - Dudamel Conducts Verdi Requiem at the Hollywood Bowl - Full',
 241                 'description': 'md5:657897370e09e2bc6bf0f8d2cd313c6b',
 242                 'duration': 6559,
 243                 'thumbnail': r're:^https?://.*\.jpg$',
 244             },
 245         },
 246         {
 247             'url': 'http://www.pbs.org/wgbh/nova/earth/killer-typhoon.html',
 248             'md5': '908f3e5473a693b266b84e25e1cf9703',
 249             'info_dict': {
 250                 'id': '2365160389',
 251                 'display_id': 'killer-typhoon',
 252                 'ext': 'mp4',
 253                 'description': 'md5:c741d14e979fc53228c575894094f157',
 254                 'title': 'NOVA - Killer Typhoon',
 255                 'duration': 3172,
 256                 'thumbnail': r're:^https?://.*\.jpg$',
 257                 'upload_date': '20140122',
 258                 'age_limit': 10,
 259             },
 260         },
 261         {
 262             'url': 'http://www.pbs.org/wgbh/pages/frontline/united-states-of-secrets/',
 263             'info_dict': {
 264                 'id': 'united-states-of-secrets',
 265             },
 266             'playlist_count': 2,
 267         },
 268         {
 269             'url': 'http://www.pbs.org/wgbh/americanexperience/films/great-war/',
 270             'info_dict': {
 271                 'id': 'great-war',
 272             },
 273             'playlist_count': 3,
 274         },
 275         {
 276             'url': 'http://www.pbs.org/wgbh/americanexperience/films/death/player/',
 277             'info_dict': {
 278                 'id': '2276541483',
 279                 'display_id': 'player',
 280                 'ext': 'mp4',
 281                 'title': 'American Experience - Death and the Civil War, Chapter 1',
 282                 'description': 'md5:67fa89a9402e2ee7d08f53b920674c18',
 283                 'duration': 682,
 284                 'thumbnail': r're:^https?://.*\.jpg$',
 285             },
 286             'params': {
 287                 'skip_download': True,  # requires ffmpeg
 288             },
 289         },
 290         {
 291             'url': 'http://www.pbs.org/video/2365245528/',
 292             'md5': '115223d41bd55cda8ae5cd5ed4e11497',
 293             'info_dict': {
 294                 'id': '2365245528',
 295                 'display_id': '2365245528',
 296                 'ext': 'mp4',
 297                 'title': 'FRONTLINE - United States of Secrets (Part One)',
 298                 'description': 'md5:55756bd5c551519cc4b7703e373e217e',
 299                 'duration': 6851,
 300                 'thumbnail': r're:^https?://.*\.jpg$',
 301             },
 302         },
 303         {
 304             # Video embedded in iframe containing angle brackets as attribute's value (e.g.
 305             # "<iframe style='position: absolute;<br />\ntop: 0; left: 0;' ...", see
 306             # https://github.com/rg3/youtube-dl/issues/7059)
 307             'url': 'http://www.pbs.org/food/features/a-chefs-life-season-3-episode-5-prickly-business/',
 308             'md5': '59b0ef5009f9ac8a319cc5efebcd865e',
 309             'info_dict': {
 310                 'id': '2365546844',
 311                 'display_id': 'a-chefs-life-season-3-episode-5-prickly-business',
 312                 'ext': 'mp4',
 313                 'title': "A Chef's Life - Season 3, Ep. 5: Prickly Business",
 314                 'description': 'md5:c0ff7475a4b70261c7e58f493c2792a5',
 315                 'duration': 1480,
 316                 'thumbnail': r're:^https?://.*\.jpg$',
 317             },
 318         },
 319         {
 320             # Frontline video embedded via flp2012.js
 321             'url': 'http://www.pbs.org/wgbh/pages/frontline/the-atomic-artists',
 322             'info_dict': {
 323                 'id': '2070868960',
 324                 'display_id': 'the-atomic-artists',
 325                 'ext': 'mp4',
 326                 'title': 'FRONTLINE - The Atomic Artists',
 327                 'description': 'md5:f677e4520cfacb4a5ce1471e31b57800',
 328                 'duration': 723,
 329                 'thumbnail': r're:^https?://.*\.jpg$',
 330             },
 331             'params': {
 332                 'skip_download': True,  # requires ffmpeg
 333             },
 334         },
 335         {
 336             # Serves hd only via wigget/partnerplayer page
 337             'url': 'http://www.pbs.org/video/2365641075/',
 338             'md5': 'fdf907851eab57211dd589cf12006666',
 339             'info_dict': {
 340                 'id': '2365641075',
 341                 'ext': 'mp4',
 342                 'title': 'FRONTLINE - Netanyahu at War',
 343                 'duration': 6852,
 344                 'thumbnail': r're:^https?://.*\.jpg$',
 345                 'formats': 'mincount:8',
 346             },
 347         },
 348         {
 349             # https://github.com/rg3/youtube-dl/issues/13801
 350             'url': 'https://www.pbs.org/video/pbs-newshour-full-episode-july-31-2017-1501539057/',
 351             'info_dict': {
 352                 'id': '3003333873',
 353                 'ext': 'mp4',
 354                 'title': 'PBS NewsHour - full episode July 31, 2017',
 355                 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
 356                 'duration': 3265,
 357                 'thumbnail': r're:^https?://.*\.jpg$',
 358             },
 359             'params': {
 360                 'skip_download': True,
 361             },
 362         },
 363         {
 364             'url': 'http://player.pbs.org/widget/partnerplayer/2365297708/?start=0&end=0&chapterbar=false&endscreen=false&topbar=true',
 365             'only_matching': True,
 366         },
 367         {
 368             'url': 'http://watch.knpb.org/video/2365616055/',
 369             'only_matching': True,
 370         },
 371         {
 372             'url': 'https://player.pbs.org/portalplayer/3004638221/?uid=',
 373             'only_matching': True,
 374         }
 375     ]
 376     _ERRORS = {
 377         101: 'We\'re sorry, but this video is not yet available.',
 378         403: 'We\'re sorry, but this video is not available in your region due to right restrictions.',
 379         404: 'We are experiencing technical difficulties that are preventing us from playing the video at this time. Please check back again soon.',
 380         410: 'This video has expired and is no longer available for online streaming.',
 381     }
 382
 383     def _real_initialize(self):
 384         cookie = (self._download_json(
 385             'http://localization.services.pbs.org/localize/auto/cookie/',
 386             None, headers=self.geo_verification_headers(), fatal=False) or {}).get('cookie')
 387         if cookie:
 388             station = self._search_regex(r'#?s=\["([^"]+)"', cookie, 'station')
 389             if station:
 390                 self._set_cookie('.pbs.org', 'pbsol.station', station)
 391
 392     def _extract_webpage(self, url):
 393         mobj = re.match(self._VALID_URL, url)
 394
 395         description = None
 396
 397         presumptive_id = mobj.group('presumptive_id')
 398         display_id = presumptive_id
 399         if presumptive_id:
 400             webpage = self._download_webpage(url, display_id)
 401
 402             description = strip_or_none(self._og_search_description(
 403                 webpage, default=None) or self._html_search_meta(
 404                 'description', webpage, default=None))
 405             upload_date = unified_strdate(self._search_regex(
 406                 r'<input type="hidden" id="air_date_[0-9]+" value="([^"]+)"',
 407                 webpage, 'upload date', default=None))
 408
 409             # tabbed frontline videos
 410             MULTI_PART_REGEXES = (
 411                 r'<div[^>]+class="videotab[^"]*"[^>]+vid="(\d+)"',
 412                 r'<a[^>]+href=["\']#(?:video-|part)\d+["\'][^>]+data-cove[Ii]d=["\'](\d+)',
 413             )
 414             for p in MULTI_PART_REGEXES:
 415                 tabbed_videos = orderedSet(re.findall(p, webpage))
 416                 if tabbed_videos:
 417                     return tabbed_videos, presumptive_id, upload_date, description
 418
 419             MEDIA_ID_REGEXES = [
 420                 r"div\s*:\s*'videoembed'\s*,\s*mediaid\s*:\s*'(\d+)'",  # frontline video embed
 421                 r'class="coveplayerid">([^<]+)<',                       # coveplayer
 422                 r'<section[^>]+data-coveid="(\d+)"',                    # coveplayer from http://www.pbs.org/wgbh/frontline/film/real-csi/
 423                 r'<input type="hidden" id="pbs_video_id_[0-9]+" value="([0-9]+)"/>',  # jwplayer
 424                 r"(?s)window\.PBS\.playerConfig\s*=\s*{.*?id\s*:\s*'([0-9]+)',",
 425             ]
 426
 427             media_id = self._search_regex(
 428                 MEDIA_ID_REGEXES, webpage, 'media ID', fatal=False, default=None)
 429             if media_id:
 430                 return media_id, presumptive_id, upload_date, description
 431
 432             # Fronline video embedded via flp
 433             video_id = self._search_regex(
 434                 r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid', default=None)
 435             if video_id:
 436                 # pkg_id calculation is reverse engineered from
 437                 # http://www.pbs.org/wgbh/pages/frontline/js/flp2012.js
 438                 prg_id = self._search_regex(
 439                     r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid')[7:]
 440                 if 'q' in prg_id:
 441                     prg_id = prg_id.split('q')[1]
 442                 prg_id = int(prg_id, 16)
 443                 getdir = self._download_json(
 444                     'http://www.pbs.org/wgbh/pages/frontline/.json/getdir/getdir%d.json' % prg_id,
 445                     presumptive_id, 'Downloading getdir JSON',
 446                     transform_source=strip_jsonp)
 447                 return getdir['mid'], presumptive_id, upload_date, description
 448
 449             for iframe in re.findall(r'(?s)<iframe(.+?)></iframe>', webpage):
 450                 url = self._search_regex(
 451                     r'src=(["\'])(?P<url>.+?partnerplayer.+?)\1', iframe,
 452                     'player URL', default=None, group='url')
 453                 if url:
 454                     break
 455
 456             if not url:
 457                 url = self._og_search_url(webpage)
 458
 459             mobj = re.match(self._VALID_URL, url)
 460
 461         player_id = mobj.group('player_id')
 462         if not display_id:
 463             display_id = player_id
 464         if player_id:
 465             player_page = self._download_webpage(
 466                 url, display_id, note='Downloading player page',
 467                 errnote='Could not download player page')
 468             video_id = self._search_regex(
 469                 r'<div\s+id="video_([0-9]+)"', player_page, 'video ID')
 470         else:
 471             video_id = mobj.group('id')
 472             display_id = video_id
 473
 474         return video_id, display_id, None, description
 475
 476     def _real_extract(self, url):
 477         video_id, display_id, upload_date, description = self._extract_webpage(url)
 478
 479         if isinstance(video_id, list):
 480             entries = [self.url_result(
 481                 'http://video.pbs.org/video/%s' % vid_id, 'PBS', vid_id)
 482                 for vid_id in video_id]
 483             return self.playlist_result(entries, display_id)
 484
 485         info = None
 486         redirects = []
 487         redirect_urls = set()
 488
 489         def extract_redirect_urls(info):
 490             for encoding_name in ('recommended_encoding', 'alternate_encoding'):
 491                 redirect = info.get(encoding_name)
 492                 if not redirect:
 493                     continue
 494                 redirect_url = redirect.get('url')
 495                 if redirect_url and redirect_url not in redirect_urls:
 496                     redirects.append(redirect)
 497                     redirect_urls.add(redirect_url)
 498
 499         chapters = []
 500         # Player pages may also serve different qualities
 501         for page in ('widget/partnerplayer', 'portalplayer'):
 502             player = self._download_webpage(
 503                 'http://player.pbs.org/%s/%s' % (page, video_id),
 504                 display_id, 'Downloading %s page' % page, fatal=False)
 505             if player:
 506                 video_info = self._parse_json(
 507                     self._search_regex(
 508                         r'(?s)PBS\.videoData\s*=\s*({.+?});\n',
 509                         player, '%s video data' % page, default='{}'),
 510                     display_id, transform_source=js_to_json, fatal=False)
 511                 if video_info:
 512                     extract_redirect_urls(video_info)
 513                     if not info:
 514                         info = video_info
 515                 if not chapters:
 516                     for chapter_data in re.findall(r'(?s)chapters\.push\(({.*?})\)', player):
 517                         chapter = self._parse_json(chapter_data, video_id, js_to_json, fatal=False)
 518                         if not chapter:
 519                             continue
 520                         start_time = float_or_none(chapter.get('start_time'), 1000)
 521                         duration = float_or_none(chapter.get('duration'), 1000)
 522                         if start_time is None or duration is None:
 523                             continue
 524                         chapters.append({
 525                             'start_time': start_time,
 526                             'end_time': start_time + duration,
 527                             'title': chapter.get('title'),
 528                         })
 529
 530         formats = []
 531         http_url = None
 532         for num, redirect in enumerate(redirects):
 533             redirect_id = redirect.get('eeid')
 534
 535             redirect_info = self._download_json(
 536                 '%s?format=json' % redirect['url'], display_id,
 537                 'Downloading %s video url info' % (redirect_id or num),
 538                 headers=self.geo_verification_headers())
 539
 540             if redirect_info['status'] == 'error':
 541                 message = self._ERRORS.get(
 542                     redirect_info['http_code'], redirect_info['message'])
 543                 if redirect_info['http_code'] == 403:
 544                     self.raise_geo_restricted(
 545                         msg=message, countries=self._GEO_COUNTRIES)
 546                 raise ExtractorError(
 547                     '%s said: %s' % (self.IE_NAME, message), expected=True)
 548
 549             format_url = redirect_info.get('url')
 550             if not format_url:
 551                 continue
 552
 553             if determine_ext(format_url) == 'm3u8':
 554                 formats.extend(self._extract_m3u8_formats(
 555                     format_url, display_id, 'mp4', m3u8_id='hls', fatal=False))
 556             else:
 557                 formats.append({
 558                     'url': format_url,
 559                     'format_id': redirect_id,
 560                 })
 561                 if re.search(r'^https?://.*(?:\d+k|baseline)', format_url):
 562                     http_url = format_url
 563         self._remove_duplicate_formats(formats)
 564         m3u8_formats = list(filter(
 565             lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none',
 566             formats))
 567         if http_url:
 568             for m3u8_format in m3u8_formats:
 569                 bitrate = self._search_regex(r'(\d+)k', m3u8_format['url'], 'bitrate', default=None)
 570                 # Lower qualities (150k and 192k) are not available as HTTP formats (see [1]),
 571                 # we won't try extracting them.
 572                 # Since summer 2016 higher quality formats (4500k and 6500k) are also available
 573                 # albeit they are not documented in [2].
 574                 # 1. https://github.com/rg3/youtube-dl/commit/cbc032c8b70a038a69259378c92b4ba97b42d491#commitcomment-17313656
 575                 # 2. https://projects.pbs.org/confluence/display/coveapi/COVE+Video+Specifications
 576                 if not bitrate or int(bitrate) < 400:
 577                     continue
 578                 f_url = re.sub(r'\d+k|baseline', bitrate + 'k', http_url)
 579                 # This may produce invalid links sometimes (e.g.
 580                 # http://www.pbs.org/wgbh/frontline/film/suicide-plan)
 581                 if not self._is_valid_url(f_url, display_id, 'http-%sk video' % bitrate):
 582                     continue
 583                 f = m3u8_format.copy()
 584                 f.update({
 585                     'url': f_url,
 586                     'format_id': m3u8_format['format_id'].replace('hls', 'http'),
 587                     'protocol': 'http',
 588                 })
 589                 formats.append(f)
 590         self._sort_formats(formats)
 591
 592         rating_str = info.get('rating')
 593         if rating_str is not None:
 594             rating_str = rating_str.rpartition('-')[2]
 595         age_limit = US_RATINGS.get(rating_str)
 596
 597         subtitles = {}
 598         closed_captions_url = info.get('closed_captions_url')
 599         if closed_captions_url:
 600             subtitles['en'] = [{
 601                 'ext': 'ttml',
 602                 'url': closed_captions_url,
 603             }]
 604             mobj = re.search(r'/(\d+)_Encoded\.dfxp', closed_captions_url)
 605             if mobj:
 606                 ttml_caption_suffix, ttml_caption_id = mobj.group(0, 1)
 607                 ttml_caption_id = int(ttml_caption_id)
 608                 subtitles['en'].extend([{
 609                     'url': closed_captions_url.replace(
 610                         ttml_caption_suffix, '/%d_Encoded.srt' % (ttml_caption_id + 1)),
 611                     'ext': 'srt',
 612                 }, {
 613                     'url': closed_captions_url.replace(
 614                         ttml_caption_suffix, '/%d_Encoded.vtt' % (ttml_caption_id + 2)),
 615                     'ext': 'vtt',
 616                 }])
 617
 618         # info['title'] is often incomplete (e.g. 'Full Episode', 'Episode 5', etc)
 619         # Try turning it to 'program - title' naming scheme if possible
 620         alt_title = info.get('program', {}).get('title')
 621         if alt_title:
 622             info['title'] = alt_title + ' - ' + re.sub(r'^' + alt_title + r'[\s\-:]+', '', info['title'])
 623
 624         description = info.get('description') or info.get(
 625             'program', {}).get('description') or description
 626
 627         return {
 628             'id': video_id,
 629             'display_id': display_id,
 630             'title': info['title'],
 631             'description': description,
 632             'thumbnail': info.get('image_url'),
 633             'duration': int_or_none(info.get('duration')),
 634             'age_limit': age_limit,
 635             'upload_date': upload_date,
 636             'formats': formats,
 637             'subtitles': subtitles,
 638             'chapters': chapters,
 639         }