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     strip_jsonp,
  12     unified_strdate,
  13     US_RATINGS,
  14 )
  15
  16
  17 class PBSIE(InfoExtractor):
  18     _STATIONS = (
  19         (r'(?:video|www)\.pbs\.org', 'PBS: Public Broadcasting Service'),  # http://www.pbs.org/
  20         (r'video\.aptv\.org', 'APT - Alabama Public Television (WBIQ)'),  # http://aptv.org/
  21         (r'video\.gpb\.org', 'GPB/Georgia Public Broadcasting (WGTV)'),  # http://www.gpb.org/
  22         (r'video\.mpbonline\.org', 'Mississippi Public Broadcasting (WMPN)'),  # http://www.mpbonline.org
  23         (r'video\.wnpt\.org', 'Nashville Public Television (WNPT)'),  # http://www.wnpt.org
  24         (r'video\.wfsu\.org', 'WFSU-TV (WFSU)'),  # http://wfsu.org/
  25         (r'video\.wsre\.org', 'WSRE (WSRE)'),  # http://www.wsre.org
  26         (r'video\.wtcitv\.org', 'WTCI (WTCI)'),  # http://www.wtcitv.org
  27         (r'video\.pba\.org', 'WPBA/Channel 30 (WPBA)'),  # http://pba.org/
  28         (r'video\.alaskapublic\.org', 'Alaska Public Media (KAKM)'),  # http://alaskapublic.org/kakm
  29         # (r'kuac\.org', 'KUAC (KUAC)'),  # http://kuac.org/kuac-tv/
  30         # (r'ktoo\.org', '360 North (KTOO)'),  # http://www.ktoo.org/
  31         # (r'azpm\.org', 'KUAT 6 (KUAT)'),  # http://www.azpm.org/
  32         (r'video\.azpbs\.org', 'Arizona PBS (KAET)'),  # http://www.azpbs.org
  33         (r'portal\.knme\.org', 'KNME-TV/Channel 5 (KNME)'),  # http://www.newmexicopbs.org/
  34         (r'video\.vegaspbs\.org', 'Vegas PBS (KLVX)'),  # http://vegaspbs.org/
  35         (r'watch\.aetn\.org', 'AETN/ARKANSAS ETV NETWORK (KETS)'),  # http://www.aetn.org/
  36         (r'video\.ket\.org', 'KET (WKLE)'),  # http://www.ket.org/
  37         (r'video\.wkno\.org', 'WKNO/Channel 10 (WKNO)'),  # http://www.wkno.org/
  38         (r'video\.lpb\.org', 'LPB/LOUISIANA PUBLIC BROADCASTING (WLPB)'),  # http://www.lpb.org/
  39         (r'videos\.oeta\.tv', 'OETA (KETA)'),  # http://www.oeta.tv
  40         (r'video\.optv\.org', 'Ozarks Public Television (KOZK)'),  # http://www.optv.org/
  41         (r'watch\.wsiu\.org', 'WSIU Public Broadcasting (WSIU)'),  # http://www.wsiu.org/
  42         (r'video\.keet\.org', 'KEET TV (KEET)'),  # http://www.keet.org
  43         (r'pbs\.kixe\.org', 'KIXE/Channel 9 (KIXE)'),  # http://kixe.org/
  44         (r'video\.kpbs\.org', 'KPBS San Diego (KPBS)'),  # http://www.kpbs.org/
  45         (r'video\.kqed\.org', 'KQED (KQED)'),  # http://www.kqed.org
  46         (r'vids\.kvie\.org', 'KVIE Public Television (KVIE)'),  # http://www.kvie.org
  47         (r'video\.pbssocal\.org', 'PBS SoCal/KOCE (KOCE)'),  # http://www.pbssocal.org/
  48         (r'video\.valleypbs\.org', 'ValleyPBS (KVPT)'),  # http://www.valleypbs.org/
  49         (r'video\.cptv\.org', 'CONNECTICUT PUBLIC TELEVISION (WEDH)'),  # http://cptv.org
  50         (r'watch\.knpb\.org', 'KNPB Channel 5 (KNPB)'),  # http://www.knpb.org/
  51         (r'video\.soptv\.org', 'SOPTV (KSYS)'),  # http://www.soptv.org
  52         # (r'klcs\.org', 'KLCS/Channel 58 (KLCS)'),  # http://www.klcs.org
  53         # (r'krcb\.org', 'KRCB Television & Radio (KRCB)'),  # http://www.krcb.org
  54         # (r'kvcr\.org', 'KVCR TV/DT/FM :: Vision for the Future (KVCR)'),  # http://kvcr.org
  55         (r'video\.rmpbs\.org', 'Rocky Mountain PBS (KRMA)'),  # http://www.rmpbs.org
  56         (r'video\.kenw\.org', 'KENW-TV3 (KENW)'),  # http://www.kenw.org
  57         (r'video\.kued\.org', 'KUED Channel 7 (KUED)'),  # http://www.kued.org
  58         (r'video\.wyomingpbs\.org', 'Wyoming PBS (KCWC)'),  # http://www.wyomingpbs.org
  59         (r'video\.cpt12\.org', 'Colorado Public Television / KBDI 12 (KBDI)'),  # http://www.cpt12.org/
  60         (r'video\.kbyueleven\.org', 'KBYU-TV (KBYU)'),  # http://www.kbyutv.org/
  61         (r'video\.thirteen\.org', 'Thirteen/WNET New York (WNET)'),  # http://www.thirteen.org
  62         (r'video\.wgbh\.org', 'WGBH/Channel 2 (WGBH)'),  # http://wgbh.org
  63         (r'video\.wgby\.org', 'WGBY (WGBY)'),  # http://www.wgby.org
  64         (r'watch\.njtvonline\.org', 'NJTV Public Media NJ (WNJT)'),  # http://www.njtvonline.org/
  65         # (r'ripbs\.org', 'Rhode Island PBS (WSBE)'),  # http://www.ripbs.org/home/
  66         (r'watch\.wliw\.org', 'WLIW21 (WLIW)'),  # http://www.wliw.org/
  67         (r'video\.mpt\.tv', 'mpt/Maryland Public Television (WMPB)'),  # http://www.mpt.org
  68         (r'watch\.weta\.org', 'WETA Television and Radio (WETA)'),  # http://www.weta.org
  69         (r'video\.whyy\.org', 'WHYY (WHYY)'),  # http://www.whyy.org
  70         (r'video\.wlvt\.org', 'PBS 39 (WLVT)'),  # http://www.wlvt.org/
  71         (r'video\.wvpt\.net', 'WVPT - Your Source for PBS and More! (WVPT)'),  # http://www.wvpt.net
  72         (r'video\.whut\.org', 'Howard University Television (WHUT)'),  # http://www.whut.org
  73         (r'video\.wedu\.org', 'WEDU PBS (WEDU)'),  # http://www.wedu.org
  74         (r'video\.wgcu\.org', 'WGCU Public Media (WGCU)'),  # http://www.wgcu.org/
  75         # (r'wjct\.org', 'WJCT Public Broadcasting (WJCT)'),  # http://www.wjct.org
  76         (r'video\.wpbt2\.org', 'WPBT2 (WPBT)'),  # http://www.wpbt2.org
  77         (r'video\.wucftv\.org', 'WUCF TV (WUCF)'),  # http://wucftv.org
  78         (r'video\.wuft\.org', 'WUFT/Channel 5 (WUFT)'),  # http://www.wuft.org
  79         (r'watch\.wxel\.org', 'WXEL/Channel 42 (WXEL)'),  # http://www.wxel.org/home/
  80         (r'video\.wlrn\.org', 'WLRN/Channel 17 (WLRN)'),  # http://www.wlrn.org/
  81         (r'video\.wusf\.usf\.edu', 'WUSF Public Broadcasting (WUSF)'),  # http://wusf.org/
  82         (r'video\.scetv\.org', 'ETV (WRLK)'),  # http://www.scetv.org
  83         (r'video\.unctv\.org', 'UNC-TV (WUNC)'),  # http://www.unctv.org/
  84         # (r'pbsguam\.org', 'PBS Guam (KGTF)'),  # http://www.pbsguam.org/
  85         (r'video\.pbshawaii\.org', 'PBS Hawaii - Oceanic Cable Channel 10 (KHET)'),  # http://www.pbshawaii.org/
  86         (r'video\.idahoptv\.org', 'Idaho Public Television (KAID)'),  # http://idahoptv.org
  87         (r'video\.ksps\.org', 'KSPS (KSPS)'),  # http://www.ksps.org/home/
  88         (r'watch\.opb\.org', 'OPB (KOPB)'),  # http://www.opb.org
  89         (r'watch\.nwptv\.org', 'KWSU/Channel 10 & KTNW/Channel 31 (KWSU)'),  # http://www.kwsu.org
  90         (r'video\.will\.illinois\.edu', 'WILL-TV (WILL)'),  # http://will.illinois.edu/
  91         (r'video\.networkknowledge\.tv', 'Network Knowledge - WSEC/Springfield (WSEC)'),  # http://www.wsec.tv
  92         (r'video\.wttw\.com', 'WTTW11 (WTTW)'),  # http://www.wttw.com/
  93         # (r'wtvp\.org', 'WTVP & WTVP.org, Public Media for Central Illinois (WTVP)'),  # http://www.wtvp.org/
  94         (r'video\.iptv\.org', 'Iowa Public Television/IPTV (KDIN)'),  # http://www.iptv.org/
  95         (r'video\.ninenet\.org', 'Nine Network (KETC)'),  # http://www.ninenet.org
  96         (r'video\.wfwa\.org', 'PBS39 Fort Wayne (WFWA)'),  # http://wfwa.org/
  97         (r'video\.wfyi\.org', 'WFYI Indianapolis (WFYI)'),  # http://www.wfyi.org
  98         (r'video\.mptv\.org', 'Milwaukee Public Television (WMVS)'),  # http://www.mptv.org
  99         (r'video\.wnin\.org', 'WNIN (WNIN)'),  # http://www.wnin.org/
 100         (r'video\.wnit\.org', 'WNIT Public Television (WNIT)'),  # http://www.wnit.org/
 101         (r'video\.wpt\.org', 'WPT (WPNE)'),  # http://www.wpt.org/
 102         (r'video\.wvut\.org', 'WVUT/Channel 22 (WVUT)'),  # http://wvut.org/
 103         (r'video\.weiu\.net', 'WEIU/Channel 51 (WEIU)'),  # http://www.weiu.net
 104         (r'video\.wqpt\.org', 'WQPT-TV (WQPT)'),  # http://www.wqpt.org
 105         (r'video\.wycc\.org', 'WYCC PBS Chicago (WYCC)'),  # http://www.wycc.org
 106         # (r'lakeshorepublicmedia\.org', 'Lakeshore Public Television (WYIN)'),  # http://lakeshorepublicmedia.org/
 107         (r'video\.wipb\.org', 'WIPB-TV (WIPB)'),  # http://wipb.org
 108         (r'video\.indianapublicmedia\.org', 'WTIU (WTIU)'),  # http://indianapublicmedia.org/tv/
 109         (r'watch\.cetconnect\.org', 'CET  (WCET)'),  # http://www.cetconnect.org
 110         (r'video\.thinktv\.org', 'ThinkTVNetwork (WPTD)'),  # http://www.thinktv.org
 111         (r'video\.wbgu\.org', 'WBGU-TV (WBGU)'),  # http://wbgu.org
 112         (r'video\.wgvu\.org', 'WGVU TV (WGVU)'),  # http://www.wgvu.org/
 113         (r'video\.netnebraska\.org', 'NET1 (KUON)'),  # http://netnebraska.org
 114         (r'video\.pioneer\.org', 'Pioneer Public Television (KWCM)'),  # http://www.pioneer.org
 115         (r'watch\.sdpb\.org', 'SDPB Television (KUSD)'),  # http://www.sdpb.org
 116         (r'video\.tpt\.org', 'TPT (KTCA)'),  # http://www.tpt.org
 117         (r'watch\.ksmq\.org', 'KSMQ (KSMQ)'),  # http://www.ksmq.org/
 118         (r'watch\.kpts\.org', 'KPTS/Channel 8 (KPTS)'),  # http://www.kpts.org/
 119         (r'watch\.ktwu\.org', 'KTWU/Channel 11 (KTWU)'),  # http://ktwu.org
 120         # (r'shptv\.org', 'Smoky Hills Public Television (KOOD)'),  # http://www.shptv.org
 121         # (r'kcpt\.org', 'KCPT Kansas City Public Television (KCPT)'),  # http://kcpt.org/
 122         # (r'blueridgepbs\.org', 'Blue Ridge PBS (WBRA)'),  # http://www.blueridgepbs.org/
 123         (r'watch\.easttennesseepbs\.org', 'East Tennessee PBS (WSJK)'),  # http://easttennesseepbs.org
 124         (r'video\.wcte\.tv', 'WCTE-TV (WCTE)'),  # http://www.wcte.org
 125         (r'video\.wljt\.org', 'WLJT, Channel 11 (WLJT)'),  # http://wljt.org/
 126         (r'video\.wosu\.org', 'WOSU TV (WOSU)'),  # http://wosu.org/
 127         (r'video\.woub\.org', 'WOUB/WOUC (WOUB)'),  # http://woub.org/tv/index.php?section=5
 128         (r'video\.wvpublic\.org', 'WVPB (WVPB)'),  # http://wvpublic.org/
 129         (r'video\.wkyupbs\.org', 'WKYU-PBS (WKYU)'),  # http://www.wkyupbs.org
 130         # (r'wyes\.org', 'WYES-TV/New Orleans (WYES)'),  # http://www.wyes.org
 131         (r'video\.kera\.org', 'KERA 13 (KERA)'),  # http://www.kera.org/
 132         (r'video\.mpbn\.net', 'MPBN (WCBB)'),  # http://www.mpbn.net/
 133         (r'video\.mountainlake\.org', 'Mountain Lake PBS (WCFE)'),  # http://www.mountainlake.org/
 134         (r'video\.nhptv\.org', 'NHPTV (WENH)'),  # http://nhptv.org/
 135         (r'video\.vpt\.org', 'Vermont PBS (WETK)'),  # http://www.vpt.org
 136         (r'video\.witf\.org', 'witf (WITF)'),  # http://www.witf.org
 137         (r'watch\.wqed\.org', 'WQED Multimedia (WQED)'),  # http://www.wqed.org/
 138         (r'video\.wmht\.org', 'WMHT Educational Telecommunications (WMHT)'),  # http://www.wmht.org/home/
 139         (r'video\.deltabroadcasting\.org', 'Q-TV (WDCQ)'),  # http://www.deltabroadcasting.org
 140         (r'video\.dptv\.org', 'WTVS Detroit Public TV (WTVS)'),  # http://www.dptv.org/
 141         (r'video\.wcmu\.org', 'CMU Public Television (WCMU)'),  # http://www.wcmu.org
 142         (r'video\.wkar\.org', 'WKAR-TV (WKAR)'),  # http://wkar.org/
 143         (r'wnmuvideo\.nmu\.edu', 'WNMU-TV Public TV 13 (WNMU)'),  # http://wnmutv.nmu.edu
 144         (r'video\.wdse\.org', 'WDSE - WRPT (WDSE)'),  # http://www.wdse.org/
 145         (r'video\.wgte\.org', 'WGTE TV (WGTE)'),  # http://www.wgte.org
 146         (r'video\.lptv\.org', 'Lakeland Public Television (KAWE)'),  # http://www.lakelandptv.org
 147         # (r'prairiepublic\.org', 'PRAIRIE PUBLIC (KFME)'),  # http://www.prairiepublic.org/
 148         (r'video\.kmos\.org', 'KMOS-TV - Channels 6.1, 6.2 and 6.3 (KMOS)'),  # http://www.kmos.org/
 149         (r'watch\.montanapbs\.org', 'MontanaPBS (KUSM)'),  # http://montanapbs.org
 150         (r'video\.krwg\.org', 'KRWG/Channel 22 (KRWG)'),  # http://www.krwg.org
 151         (r'video\.kacvtv\.org', 'KACV (KACV)'),  # http://www.panhandlepbs.org/home/
 152         (r'video\.kcostv\.org', 'KCOS/Channel 13 (KCOS)'),  # www.kcostv.org
 153         (r'video\.wcny\.org', 'WCNY/Channel 24 (WCNY)'),  # http://www.wcny.org
 154         (r'video\.wned\.org', 'WNED (WNED)'),  # http://www.wned.org/
 155         (r'watch\.wpbstv\.org', 'WPBS (WPBS)'),  # http://www.wpbstv.org
 156         (r'video\.wskg\.org', 'WSKG Public TV (WSKG)'),  # http://wskg.org
 157         (r'video\.wxxi\.org', 'WXXI (WXXI)'),  # http://wxxi.org
 158         (r'video\.wpsu\.org', 'WPSU (WPSU)'),  # http://www.wpsu.org
 159         # (r'wqln\.org', 'WQLN/Channel 54 (WQLN)'),  # http://www.wqln.org
 160         (r'on-demand\.wvia\.org', 'WVIA Public Media Studios (WVIA)'),  # http://www.wvia.org/
 161         (r'video\.wtvi\.org', 'WTVI (WTVI)'),  # http://www.wtvi.org/
 162         # (r'whro\.org', 'WHRO (WHRO)'),  # http://whro.org
 163         (r'video\.westernreservepublicmedia\.org', 'Western Reserve PBS (WNEO)'),  # http://www.WesternReservePublicMedia.org/
 164         (r'video\.ideastream\.org', 'WVIZ/PBS ideastream (WVIZ)'),  # http://www.wviz.org/
 165         (r'video\.kcts9\.org', 'KCTS 9 (KCTS)'),  # http://kcts9.org/
 166         (r'video\.basinpbs\.org', 'Basin PBS (KPBT)'),  # http://www.basinpbs.org
 167         (r'video\.houstonpbs\.org', 'KUHT / Channel 8 (KUHT)'),  # http://www.houstonpublicmedia.org/
 168         # (r'tamu\.edu', 'KAMU - TV (KAMU)'),  # http://KAMU.tamu.edu
 169         # (r'kedt\.org', 'KEDT/Channel 16 (KEDT)'),  # http://www.kedt.org
 170         (r'video\.klrn\.org', 'KLRN (KLRN)'),  # http://www.klrn.org
 171         (r'video\.klru\.tv', 'KLRU (KLRU)'),  # http://www.klru.org
 172         # (r'kmbh\.org', 'KMBH-TV (KMBH)'),  # http://www.kmbh.org
 173         # (r'knct\.org', 'KNCT (KNCT)'),  # http://www.knct.org
 174         # (r'ktxt\.org', 'KTTZ-TV (KTXT)'),  # http://www.ktxt.org
 175         (r'video\.wtjx\.org', 'WTJX Channel 12 (WTJX)'),  # http://www.wtjx.org/
 176         (r'video\.ideastations\.org', 'WCVE PBS (WCVE)'),  # http://ideastations.org/
 177         (r'video\.kbtc\.org', 'KBTC Public Television (KBTC)'),  # http://kbtc.org
 178     )
 179
 180     IE_NAME = 'pbs'
 181     IE_DESC = 'Public Broadcasting Service (PBS) and member stations: %s' % ', '.join(list(zip(*_STATIONS))[1])
 182
 183     _VALID_URL = r'''(?x)https?://
 184         (?:
 185            # Direct video URL
 186            (?:%s)/(?:viralplayer|video)/(?P<id>[0-9]+)/? |
 187            # Article with embedded player (or direct video)
 188            (?:www\.)?pbs\.org/(?:[^/]+/){2,5}(?P<presumptive_id>[^/]+?)(?:\.html)?/?(?:$|[?\#]) |
 189            # Player
 190            (?:video|player)\.pbs\.org/(?:widget/)?partnerplayer/(?P<player_id>[^/]+)/
 191         )
 192     ''' % '|'.join(list(zip(*_STATIONS))[0])
 193
 194     _TESTS = [
 195         {
 196             'url': 'http://www.pbs.org/tpt/constitution-usa-peter-sagal/watch/a-more-perfect-union/',
 197             'md5': 'ce1888486f0908d555a8093cac9a7362',
 198             'info_dict': {
 199                 'id': '2365006249',
 200                 'ext': 'mp4',
 201                 'title': 'Constitution USA with Peter Sagal - A More Perfect Union',
 202                 'description': 'md5:ba0c207295339c8d6eced00b7c363c6a',
 203                 'duration': 3190,
 204             },
 205             'params': {
 206                 'skip_download': True,  # requires ffmpeg
 207             },
 208         },
 209         {
 210             'url': 'http://www.pbs.org/wgbh/pages/frontline/losing-iraq/',
 211             'md5': '143c98aa54a346738a3d78f54c925321',
 212             'info_dict': {
 213                 'id': '2365297690',
 214                 'ext': 'mp4',
 215                 'title': 'FRONTLINE - Losing Iraq',
 216                 'description': 'md5:f5bfbefadf421e8bb8647602011caf8e',
 217                 'duration': 5050,
 218             },
 219             'params': {
 220                 'skip_download': True,  # requires ffmpeg
 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': 'md5:5871c15cba347c1b3d28ac47a73c7c28',
 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                 'description': 'md5:68d87ef760660eb564455eb30ca464fe',
 241                 'title': 'Great Performances - Dudamel Conducts Verdi Requiem at the Hollywood Bowl - Full',
 242                 'duration': 6559,
 243                 'thumbnail': 're:^https?://.*\.jpg$',
 244             },
 245             'params': {
 246                 'skip_download': True,  # requires ffmpeg
 247             },
 248         },
 249         {
 250             'url': 'http://www.pbs.org/wgbh/nova/earth/killer-typhoon.html',
 251             'md5': '908f3e5473a693b266b84e25e1cf9703',
 252             'info_dict': {
 253                 'id': '2365160389',
 254                 'display_id': 'killer-typhoon',
 255                 'ext': 'mp4',
 256                 'description': 'md5:c741d14e979fc53228c575894094f157',
 257                 'title': 'NOVA - Killer Typhoon',
 258                 'duration': 3172,
 259                 'thumbnail': 're:^https?://.*\.jpg$',
 260                 'upload_date': '20140122',
 261                 'age_limit': 10,
 262             },
 263             'params': {
 264                 'skip_download': True,  # requires ffmpeg
 265             },
 266         },
 267         {
 268             'url': 'http://www.pbs.org/wgbh/pages/frontline/united-states-of-secrets/',
 269             'info_dict': {
 270                 'id': 'united-states-of-secrets',
 271             },
 272             'playlist_count': 2,
 273         },
 274         {
 275             'url': 'http://www.pbs.org/wgbh/americanexperience/films/death/player/',
 276             'info_dict': {
 277                 'id': '2276541483',
 278                 'display_id': 'player',
 279                 'ext': 'mp4',
 280                 'title': 'American Experience - Death and the Civil War, Chapter 1',
 281                 'description': 'American Experience, TV’s most-watched history series, brings to life the compelling stories from our past that inform our understanding of the world today.',
 282                 'duration': 682,
 283                 'thumbnail': 're:^https?://.*\.jpg$',
 284             },
 285             'params': {
 286                 'skip_download': True,  # requires ffmpeg
 287             },
 288         },
 289         {
 290             'url': 'http://video.pbs.org/video/2365367186/',
 291             'info_dict': {
 292                 'id': '2365367186',
 293                 'display_id': '2365367186',
 294                 'ext': 'mp4',
 295                 'title': 'To Catch A Comet - Full Episode',
 296                 'description': 'On November 12, 2014, billions of kilometers from Earth, spacecraft orbiter Rosetta and lander Philae did what no other had dared to attempt \u2014 land on the volatile surface of a comet as it zooms around the sun at 67,000 km/hr. The European Space Agency hopes this mission can help peer into our past and unlock secrets of our origins.',
 297                 'duration': 3342,
 298                 'thumbnail': 're:^https?://.*\.jpg$',
 299             },
 300             'params': {
 301                 'skip_download': True,  # requires ffmpeg
 302             },
 303             'skip': 'Expired',
 304         },
 305         {
 306             # Video embedded in iframe containing angle brackets as attribute's value (e.g.
 307             # "<iframe style='position: absolute;<br />\ntop: 0; left: 0;' ...", see
 308             # https://github.com/rg3/youtube-dl/issues/7059)
 309             'url': 'http://www.pbs.org/food/features/a-chefs-life-season-3-episode-5-prickly-business/',
 310             'info_dict': {
 311                 'id': '2365546844',
 312                 'display_id': 'a-chefs-life-season-3-episode-5-prickly-business',
 313                 'ext': 'mp4',
 314                 'title': "A Chef's Life - Season 3, Ep. 5: Prickly Business",
 315                 'description': 'md5:61db2ddf27c9912f09c241014b118ed1',
 316                 'duration': 1480,
 317                 'thumbnail': 're:^https?://.*\.jpg$',
 318             },
 319             'params': {
 320                 'skip_download': True,  # requires ffmpeg
 321             },
 322         },
 323         {
 324             # Frontline video embedded via flp2012.js
 325             'url': 'http://www.pbs.org/wgbh/pages/frontline/the-atomic-artists',
 326             'info_dict': {
 327                 'id': '2070868960',
 328                 'display_id': 'the-atomic-artists',
 329                 'ext': 'mp4',
 330                 'title': 'FRONTLINE - The Atomic Artists',
 331                 'description': 'md5:f5bfbefadf421e8bb8647602011caf8e',
 332                 'duration': 723,
 333                 'thumbnail': 're:^https?://.*\.jpg$',
 334             },
 335             'params': {
 336                 'skip_download': True,  # requires ffmpeg
 337             },
 338         },
 339         {
 340             'url': 'http://player.pbs.org/widget/partnerplayer/2365297708/?start=0&end=0&chapterbar=false&endscreen=false&topbar=true',
 341             'only_matching': True,
 342         },
 343         {
 344             'url': 'http://watch.knpb.org/video/2365616055/',
 345             'only_matching': True,
 346         }
 347     ]
 348     _ERRORS = {
 349         101: 'We\'re sorry, but this video is not yet available.',
 350         403: 'We\'re sorry, but this video is not available in your region due to right restrictions.',
 351         404: 'We are experiencing technical difficulties that are preventing us from playing the video at this time. Please check back again soon.',
 352         410: 'This video has expired and is no longer available for online streaming.',
 353     }
 354
 355     def _extract_webpage(self, url):
 356         mobj = re.match(self._VALID_URL, url)
 357
 358         presumptive_id = mobj.group('presumptive_id')
 359         display_id = presumptive_id
 360         if presumptive_id:
 361             webpage = self._download_webpage(url, display_id)
 362
 363             upload_date = unified_strdate(self._search_regex(
 364                 r'<input type="hidden" id="air_date_[0-9]+" value="([^"]+)"',
 365                 webpage, 'upload date', default=None))
 366
 367             # tabbed frontline videos
 368             tabbed_videos = re.findall(
 369                 r'<div[^>]+class="videotab[^"]*"[^>]+vid="(\d+)"', webpage)
 370             if tabbed_videos:
 371                 return tabbed_videos, presumptive_id, upload_date
 372
 373             MEDIA_ID_REGEXES = [
 374                 r"div\s*:\s*'videoembed'\s*,\s*mediaid\s*:\s*'(\d+)'",  # frontline video embed
 375                 r'class="coveplayerid">([^<]+)<',                       # coveplayer
 376                 r'<section[^>]+data-coveid="(\d+)"',                    # coveplayer from http://www.pbs.org/wgbh/frontline/film/real-csi/
 377                 r'<input type="hidden" id="pbs_video_id_[0-9]+" value="([0-9]+)"/>',  # jwplayer
 378             ]
 379
 380             media_id = self._search_regex(
 381                 MEDIA_ID_REGEXES, webpage, 'media ID', fatal=False, default=None)
 382             if media_id:
 383                 return media_id, presumptive_id, upload_date
 384
 385             # Fronline video embedded via flp
 386             video_id = self._search_regex(
 387                 r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid', default=None)
 388             if video_id:
 389                 # pkg_id calculation is reverse engineered from
 390                 # http://www.pbs.org/wgbh/pages/frontline/js/flp2012.js
 391                 prg_id = self._search_regex(
 392                     r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid')[7:]
 393                 if 'q' in prg_id:
 394                     prg_id = prg_id.split('q')[1]
 395                 prg_id = int(prg_id, 16)
 396                 getdir = self._download_json(
 397                     'http://www.pbs.org/wgbh/pages/frontline/.json/getdir/getdir%d.json' % prg_id,
 398                     presumptive_id, 'Downloading getdir JSON',
 399                     transform_source=strip_jsonp)
 400                 return getdir['mid'], presumptive_id, upload_date
 401
 402             for iframe in re.findall(r'(?s)<iframe(.+?)></iframe>', webpage):
 403                 url = self._search_regex(
 404                     r'src=(["\'])(?P<url>.+?partnerplayer.+?)\1', iframe,
 405                     'player URL', default=None, group='url')
 406                 if url:
 407                     break
 408
 409             mobj = re.match(self._VALID_URL, url)
 410
 411         player_id = mobj.group('player_id')
 412         if not display_id:
 413             display_id = player_id
 414         if player_id:
 415             player_page = self._download_webpage(
 416                 url, display_id, note='Downloading player page',
 417                 errnote='Could not download player page')
 418             video_id = self._search_regex(
 419                 r'<div\s+id="video_([0-9]+)"', player_page, 'video ID')
 420         else:
 421             video_id = mobj.group('id')
 422             display_id = video_id
 423
 424         return video_id, display_id, None
 425
 426     def _real_extract(self, url):
 427         video_id, display_id, upload_date = self._extract_webpage(url)
 428
 429         if isinstance(video_id, list):
 430             entries = [self.url_result(
 431                 'http://video.pbs.org/video/%s' % vid_id, 'PBS', vid_id)
 432                 for vid_id in video_id]
 433             return self.playlist_result(entries, display_id)
 434
 435         info = self._download_json(
 436             'http://player.pbs.org/videoInfo/%s?format=json&type=partner' % video_id,
 437             display_id)
 438
 439         formats = []
 440         for encoding_name in ('recommended_encoding', 'alternate_encoding'):
 441             redirect = info.get(encoding_name)
 442             if not redirect:
 443                 continue
 444             redirect_url = redirect.get('url')
 445             if not redirect_url:
 446                 continue
 447
 448             redirect_info = self._download_json(
 449                 redirect_url + '?format=json', display_id,
 450                 'Downloading %s video url info' % encoding_name)
 451
 452             if redirect_info['status'] == 'error':
 453                 raise ExtractorError(
 454                     '%s said: %s' % (
 455                         self.IE_NAME,
 456                         self._ERRORS.get(redirect_info['http_code'], redirect_info['message'])),
 457                     expected=True)
 458
 459             format_url = redirect_info.get('url')
 460             if not format_url:
 461                 continue
 462
 463             if determine_ext(format_url) == 'm3u8':
 464                 formats.extend(self._extract_m3u8_formats(
 465                     format_url, display_id, 'mp4', preference=1, m3u8_id='hls'))
 466             else:
 467                 formats.append({
 468                     'url': format_url,
 469                     'format_id': redirect.get('eeid'),
 470                 })
 471         self._sort_formats(formats)
 472
 473         rating_str = info.get('rating')
 474         if rating_str is not None:
 475             rating_str = rating_str.rpartition('-')[2]
 476         age_limit = US_RATINGS.get(rating_str)
 477
 478         subtitles = {}
 479         closed_captions_url = info.get('closed_captions_url')
 480         if closed_captions_url:
 481             subtitles['en'] = [{
 482                 'ext': 'ttml',
 483                 'url': closed_captions_url,
 484             }]
 485
 486         # info['title'] is often incomplete (e.g. 'Full Episode', 'Episode 5', etc)
 487         # Try turning it to 'program - title' naming scheme if possible
 488         alt_title = info.get('program', {}).get('title')
 489         if alt_title:
 490             info['title'] = alt_title + ' - ' + re.sub(r'^' + alt_title + '[\s\-:]+', '', info['title'])
 491
 492         return {
 493             'id': video_id,
 494             'display_id': display_id,
 495             'title': info['title'],
 496             'description': info['program'].get('description'),
 497             'thumbnail': info.get('image_url'),
 498             'duration': int_or_none(info.get('duration')),
 499             'age_limit': age_limit,
 500             'upload_date': upload_date,
 501             'formats': formats,
 502             'subtitles': subtitles,
 503         }