Raphaël G. Git Repositories - youtubedl/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 ..compat import compat_HTTPError
   8 from ..utils import (
   9     ExtractorError,
  10     determine_ext,
  11     int_or_none,
  12     js_to_json,
  13     strip_jsonp,
  14     unified_strdate,
  15     US_RATINGS,
  16 )
  17
  18
  19 class PBSIE(InfoExtractor):
  20     _STATIONS = (
  21         (r'(?:video|www|player)\.pbs\.org', 'PBS: Public Broadcasting Service'),  # http://www.pbs.org/
  22         (r'video\.aptv\.org', 'APT - Alabama Public Television (WBIQ)'),  # http://aptv.org/
  23         (r'video\.gpb\.org', 'GPB/Georgia Public Broadcasting (WGTV)'),  # http://www.gpb.org/
  24         (r'video\.mpbonline\.org', 'Mississippi Public Broadcasting (WMPN)'),  # http://www.mpbonline.org
  25         (r'video\.wnpt\.org', 'Nashville Public Television (WNPT)'),  # http://www.wnpt.org
  26         (r'video\.wfsu\.org', 'WFSU-TV (WFSU)'),  # http://wfsu.org/
  27         (r'video\.wsre\.org', 'WSRE (WSRE)'),  # http://www.wsre.org
  28         (r'video\.wtcitv\.org', 'WTCI (WTCI)'),  # http://www.wtcitv.org
  29         (r'video\.pba\.org', 'WPBA/Channel 30 (WPBA)'),  # http://pba.org/
  30         (r'video\.alaskapublic\.org', 'Alaska Public Media (KAKM)'),  # http://alaskapublic.org/kakm
  31         # (r'kuac\.org', 'KUAC (KUAC)'),  # http://kuac.org/kuac-tv/
  32         # (r'ktoo\.org', '360 North (KTOO)'),  # http://www.ktoo.org/
  33         # (r'azpm\.org', 'KUAT 6 (KUAT)'),  # http://www.azpm.org/
  34         (r'video\.azpbs\.org', 'Arizona PBS (KAET)'),  # http://www.azpbs.org
  35         (r'portal\.knme\.org', 'KNME-TV/Channel 5 (KNME)'),  # http://www.newmexicopbs.org/
  36         (r'video\.vegaspbs\.org', 'Vegas PBS (KLVX)'),  # http://vegaspbs.org/
  37         (r'watch\.aetn\.org', 'AETN/ARKANSAS ETV NETWORK (KETS)'),  # http://www.aetn.org/
  38         (r'video\.ket\.org', 'KET (WKLE)'),  # http://www.ket.org/
  39         (r'video\.wkno\.org', 'WKNO/Channel 10 (WKNO)'),  # http://www.wkno.org/
  40         (r'video\.lpb\.org', 'LPB/LOUISIANA PUBLIC BROADCASTING (WLPB)'),  # http://www.lpb.org/
  41         (r'videos\.oeta\.tv', 'OETA (KETA)'),  # http://www.oeta.tv
  42         (r'video\.optv\.org', 'Ozarks Public Television (KOZK)'),  # http://www.optv.org/
  43         (r'watch\.wsiu\.org', 'WSIU Public Broadcasting (WSIU)'),  # http://www.wsiu.org/
  44         (r'video\.keet\.org', 'KEET TV (KEET)'),  # http://www.keet.org
  45         (r'pbs\.kixe\.org', 'KIXE/Channel 9 (KIXE)'),  # http://kixe.org/
  46         (r'video\.kpbs\.org', 'KPBS San Diego (KPBS)'),  # http://www.kpbs.org/
  47         (r'video\.kqed\.org', 'KQED (KQED)'),  # http://www.kqed.org
  48         (r'vids\.kvie\.org', 'KVIE Public Television (KVIE)'),  # http://www.kvie.org
  49         (r'video\.pbssocal\.org', 'PBS SoCal/KOCE (KOCE)'),  # http://www.pbssocal.org/
  50         (r'video\.valleypbs\.org', 'ValleyPBS (KVPT)'),  # http://www.valleypbs.org/
  51         (r'video\.cptv\.org', 'CONNECTICUT PUBLIC TELEVISION (WEDH)'),  # http://cptv.org
  52         (r'watch\.knpb\.org', 'KNPB Channel 5 (KNPB)'),  # http://www.knpb.org/
  53         (r'video\.soptv\.org', 'SOPTV (KSYS)'),  # http://www.soptv.org
  54         # (r'klcs\.org', 'KLCS/Channel 58 (KLCS)'),  # http://www.klcs.org
  55         # (r'krcb\.org', 'KRCB Television & Radio (KRCB)'),  # http://www.krcb.org
  56         # (r'kvcr\.org', 'KVCR TV/DT/FM :: Vision for the Future (KVCR)'),  # http://kvcr.org
  57         (r'video\.rmpbs\.org', 'Rocky Mountain PBS (KRMA)'),  # http://www.rmpbs.org
  58         (r'video\.kenw\.org', 'KENW-TV3 (KENW)'),  # http://www.kenw.org
  59         (r'video\.kued\.org', 'KUED Channel 7 (KUED)'),  # http://www.kued.org
  60         (r'video\.wyomingpbs\.org', 'Wyoming PBS (KCWC)'),  # http://www.wyomingpbs.org
  61         (r'video\.cpt12\.org', 'Colorado Public Television / KBDI 12 (KBDI)'),  # http://www.cpt12.org/
  62         (r'video\.kbyueleven\.org', 'KBYU-TV (KBYU)'),  # http://www.kbyutv.org/
  63         (r'video\.thirteen\.org', 'Thirteen/WNET New York (WNET)'),  # http://www.thirteen.org
  64         (r'video\.wgbh\.org', 'WGBH/Channel 2 (WGBH)'),  # http://wgbh.org
  65         (r'video\.wgby\.org', 'WGBY (WGBY)'),  # http://www.wgby.org
  66         (r'watch\.njtvonline\.org', 'NJTV Public Media NJ (WNJT)'),  # http://www.njtvonline.org/
  67         # (r'ripbs\.org', 'Rhode Island PBS (WSBE)'),  # http://www.ripbs.org/home/
  68         (r'watch\.wliw\.org', 'WLIW21 (WLIW)'),  # http://www.wliw.org/
  69         (r'video\.mpt\.tv', 'mpt/Maryland Public Television (WMPB)'),  # http://www.mpt.org
  70         (r'watch\.weta\.org', 'WETA Television and Radio (WETA)'),  # http://www.weta.org
  71         (r'video\.whyy\.org', 'WHYY (WHYY)'),  # http://www.whyy.org
  72         (r'video\.wlvt\.org', 'PBS 39 (WLVT)'),  # http://www.wlvt.org/
  73         (r'video\.wvpt\.net', 'WVPT - Your Source for PBS and More! (WVPT)'),  # http://www.wvpt.net
  74         (r'video\.whut\.org', 'Howard University Television (WHUT)'),  # http://www.whut.org
  75         (r'video\.wedu\.org', 'WEDU PBS (WEDU)'),  # http://www.wedu.org
  76         (r'video\.wgcu\.org', 'WGCU Public Media (WGCU)'),  # http://www.wgcu.org/
  77         # (r'wjct\.org', 'WJCT Public Broadcasting (WJCT)'),  # http://www.wjct.org
  78         (r'video\.wpbt2\.org', 'WPBT2 (WPBT)'),  # http://www.wpbt2.org
  79         (r'video\.wucftv\.org', 'WUCF TV (WUCF)'),  # http://wucftv.org
  80         (r'video\.wuft\.org', 'WUFT/Channel 5 (WUFT)'),  # http://www.wuft.org
  81         (r'watch\.wxel\.org', 'WXEL/Channel 42 (WXEL)'),  # http://www.wxel.org/home/
  82         (r'video\.wlrn\.org', 'WLRN/Channel 17 (WLRN)'),  # http://www.wlrn.org/
  83         (r'video\.wusf\.usf\.edu', 'WUSF Public Broadcasting (WUSF)'),  # http://wusf.org/
  84         (r'video\.scetv\.org', 'ETV (WRLK)'),  # http://www.scetv.org
  85         (r'video\.unctv\.org', 'UNC-TV (WUNC)'),  # http://www.unctv.org/
  86         # (r'pbsguam\.org', 'PBS Guam (KGTF)'),  # http://www.pbsguam.org/
  87         (r'video\.pbshawaii\.org', 'PBS Hawaii - Oceanic Cable Channel 10 (KHET)'),  # http://www.pbshawaii.org/
  88         (r'video\.idahoptv\.org', 'Idaho Public Television (KAID)'),  # http://idahoptv.org
  89         (r'video\.ksps\.org', 'KSPS (KSPS)'),  # http://www.ksps.org/home/
  90         (r'watch\.opb\.org', 'OPB (KOPB)'),  # http://www.opb.org
  91         (r'watch\.nwptv\.org', 'KWSU/Channel 10 & KTNW/Channel 31 (KWSU)'),  # http://www.kwsu.org
  92         (r'video\.will\.illinois\.edu', 'WILL-TV (WILL)'),  # http://will.illinois.edu/
  93         (r'video\.networkknowledge\.tv', 'Network Knowledge - WSEC/Springfield (WSEC)'),  # http://www.wsec.tv
  94         (r'video\.wttw\.com', 'WTTW11 (WTTW)'),  # http://www.wttw.com/
  95         # (r'wtvp\.org', 'WTVP & WTVP.org, Public Media for Central Illinois (WTVP)'),  # http://www.wtvp.org/
  96         (r'video\.iptv\.org', 'Iowa Public Television/IPTV (KDIN)'),  # http://www.iptv.org/
  97         (r'video\.ninenet\.org', 'Nine Network (KETC)'),  # http://www.ninenet.org
  98         (r'video\.wfwa\.org', 'PBS39 Fort Wayne (WFWA)'),  # http://wfwa.org/
  99         (r'video\.wfyi\.org', 'WFYI Indianapolis (WFYI)'),  # http://www.wfyi.org
 100         (r'video\.mptv\.org', 'Milwaukee Public Television (WMVS)'),  # http://www.mptv.org
 101         (r'video\.wnin\.org', 'WNIN (WNIN)'),  # http://www.wnin.org/
 102         (r'video\.wnit\.org', 'WNIT Public Television (WNIT)'),  # http://www.wnit.org/
 103         (r'video\.wpt\.org', 'WPT (WPNE)'),  # http://www.wpt.org/
 104         (r'video\.wvut\.org', 'WVUT/Channel 22 (WVUT)'),  # http://wvut.org/
 105         (r'video\.weiu\.net', 'WEIU/Channel 51 (WEIU)'),  # http://www.weiu.net
 106         (r'video\.wqpt\.org', 'WQPT-TV (WQPT)'),  # http://www.wqpt.org
 107         (r'video\.wycc\.org', 'WYCC PBS Chicago (WYCC)'),  # http://www.wycc.org
 108         # (r'lakeshorepublicmedia\.org', 'Lakeshore Public Television (WYIN)'),  # http://lakeshorepublicmedia.org/
 109         (r'video\.wipb\.org', 'WIPB-TV (WIPB)'),  # http://wipb.org
 110         (r'video\.indianapublicmedia\.org', 'WTIU (WTIU)'),  # http://indianapublicmedia.org/tv/
 111         (r'watch\.cetconnect\.org', 'CET  (WCET)'),  # http://www.cetconnect.org
 112         (r'video\.thinktv\.org', 'ThinkTVNetwork (WPTD)'),  # http://www.thinktv.org
 113         (r'video\.wbgu\.org', 'WBGU-TV (WBGU)'),  # http://wbgu.org
 114         (r'video\.wgvu\.org', 'WGVU TV (WGVU)'),  # http://www.wgvu.org/
 115         (r'video\.netnebraska\.org', 'NET1 (KUON)'),  # http://netnebraska.org
 116         (r'video\.pioneer\.org', 'Pioneer Public Television (KWCM)'),  # http://www.pioneer.org
 117         (r'watch\.sdpb\.org', 'SDPB Television (KUSD)'),  # http://www.sdpb.org
 118         (r'video\.tpt\.org', 'TPT (KTCA)'),  # http://www.tpt.org
 119         (r'watch\.ksmq\.org', 'KSMQ (KSMQ)'),  # http://www.ksmq.org/
 120         (r'watch\.kpts\.org', 'KPTS/Channel 8 (KPTS)'),  # http://www.kpts.org/
 121         (r'watch\.ktwu\.org', 'KTWU/Channel 11 (KTWU)'),  # http://ktwu.org
 122         # (r'shptv\.org', 'Smoky Hills Public Television (KOOD)'),  # http://www.shptv.org
 123         # (r'kcpt\.org', 'KCPT Kansas City Public Television (KCPT)'),  # http://kcpt.org/
 124         # (r'blueridgepbs\.org', 'Blue Ridge PBS (WBRA)'),  # http://www.blueridgepbs.org/
 125         (r'watch\.easttennesseepbs\.org', 'East Tennessee PBS (WSJK)'),  # http://easttennesseepbs.org
 126         (r'video\.wcte\.tv', 'WCTE-TV (WCTE)'),  # http://www.wcte.org
 127         (r'video\.wljt\.org', 'WLJT, Channel 11 (WLJT)'),  # http://wljt.org/
 128         (r'video\.wosu\.org', 'WOSU TV (WOSU)'),  # http://wosu.org/
 129         (r'video\.woub\.org', 'WOUB/WOUC (WOUB)'),  # http://woub.org/tv/index.php?section=5
 130         (r'video\.wvpublic\.org', 'WVPB (WVPB)'),  # http://wvpublic.org/
 131         (r'video\.wkyupbs\.org', 'WKYU-PBS (WKYU)'),  # http://www.wkyupbs.org
 132         # (r'wyes\.org', 'WYES-TV/New Orleans (WYES)'),  # http://www.wyes.org
 133         (r'video\.kera\.org', 'KERA 13 (KERA)'),  # http://www.kera.org/
 134         (r'video\.mpbn\.net', 'MPBN (WCBB)'),  # http://www.mpbn.net/
 135         (r'video\.mountainlake\.org', 'Mountain Lake PBS (WCFE)'),  # http://www.mountainlake.org/
 136         (r'video\.nhptv\.org', 'NHPTV (WENH)'),  # http://nhptv.org/
 137         (r'video\.vpt\.org', 'Vermont PBS (WETK)'),  # http://www.vpt.org
 138         (r'video\.witf\.org', 'witf (WITF)'),  # http://www.witf.org
 139         (r'watch\.wqed\.org', 'WQED Multimedia (WQED)'),  # http://www.wqed.org/
 140         (r'video\.wmht\.org', 'WMHT Educational Telecommunications (WMHT)'),  # http://www.wmht.org/home/
 141         (r'video\.deltabroadcasting\.org', 'Q-TV (WDCQ)'),  # http://www.deltabroadcasting.org
 142         (r'video\.dptv\.org', 'WTVS Detroit Public TV (WTVS)'),  # http://www.dptv.org/
 143         (r'video\.wcmu\.org', 'CMU Public Television (WCMU)'),  # http://www.wcmu.org
 144         (r'video\.wkar\.org', 'WKAR-TV (WKAR)'),  # http://wkar.org/
 145         (r'wnmuvideo\.nmu\.edu', 'WNMU-TV Public TV 13 (WNMU)'),  # http://wnmutv.nmu.edu
 146         (r'video\.wdse\.org', 'WDSE - WRPT (WDSE)'),  # http://www.wdse.org/
 147         (r'video\.wgte\.org', 'WGTE TV (WGTE)'),  # http://www.wgte.org
 148         (r'video\.lptv\.org', 'Lakeland Public Television (KAWE)'),  # http://www.lakelandptv.org
 149         # (r'prairiepublic\.org', 'PRAIRIE PUBLIC (KFME)'),  # http://www.prairiepublic.org/
 150         (r'video\.kmos\.org', 'KMOS-TV - Channels 6.1, 6.2 and 6.3 (KMOS)'),  # http://www.kmos.org/
 151         (r'watch\.montanapbs\.org', 'MontanaPBS (KUSM)'),  # http://montanapbs.org
 152         (r'video\.krwg\.org', 'KRWG/Channel 22 (KRWG)'),  # http://www.krwg.org
 153         (r'video\.kacvtv\.org', 'KACV (KACV)'),  # http://www.panhandlepbs.org/home/
 154         (r'video\.kcostv\.org', 'KCOS/Channel 13 (KCOS)'),  # www.kcostv.org
 155         (r'video\.wcny\.org', 'WCNY/Channel 24 (WCNY)'),  # http://www.wcny.org
 156         (r'video\.wned\.org', 'WNED (WNED)'),  # http://www.wned.org/
 157         (r'watch\.wpbstv\.org', 'WPBS (WPBS)'),  # http://www.wpbstv.org
 158         (r'video\.wskg\.org', 'WSKG Public TV (WSKG)'),  # http://wskg.org
 159         (r'video\.wxxi\.org', 'WXXI (WXXI)'),  # http://wxxi.org
 160         (r'video\.wpsu\.org', 'WPSU (WPSU)'),  # http://www.wpsu.org
 161         # (r'wqln\.org', 'WQLN/Channel 54 (WQLN)'),  # http://www.wqln.org
 162         (r'on-demand\.wvia\.org', 'WVIA Public Media Studios (WVIA)'),  # http://www.wvia.org/
 163         (r'video\.wtvi\.org', 'WTVI (WTVI)'),  # http://www.wtvi.org/
 164         # (r'whro\.org', 'WHRO (WHRO)'),  # http://whro.org
 165         (r'video\.westernreservepublicmedia\.org', 'Western Reserve PBS (WNEO)'),  # http://www.WesternReservePublicMedia.org/
 166         (r'video\.ideastream\.org', 'WVIZ/PBS ideastream (WVIZ)'),  # http://www.wviz.org/
 167         (r'video\.kcts9\.org', 'KCTS 9 (KCTS)'),  # http://kcts9.org/
 168         (r'video\.basinpbs\.org', 'Basin PBS (KPBT)'),  # http://www.basinpbs.org
 169         (r'video\.houstonpbs\.org', 'KUHT / Channel 8 (KUHT)'),  # http://www.houstonpublicmedia.org/
 170         # (r'tamu\.edu', 'KAMU - TV (KAMU)'),  # http://KAMU.tamu.edu
 171         # (r'kedt\.org', 'KEDT/Channel 16 (KEDT)'),  # http://www.kedt.org
 172         (r'video\.klrn\.org', 'KLRN (KLRN)'),  # http://www.klrn.org
 173         (r'video\.klru\.tv', 'KLRU (KLRU)'),  # http://www.klru.org
 174         # (r'kmbh\.org', 'KMBH-TV (KMBH)'),  # http://www.kmbh.org
 175         # (r'knct\.org', 'KNCT (KNCT)'),  # http://www.knct.org
 176         # (r'ktxt\.org', 'KTTZ-TV (KTXT)'),  # http://www.ktxt.org
 177         (r'video\.wtjx\.org', 'WTJX Channel 12 (WTJX)'),  # http://www.wtjx.org/
 178         (r'video\.ideastations\.org', 'WCVE PBS (WCVE)'),  # http://ideastations.org/
 179         (r'video\.kbtc\.org', 'KBTC Public Television (KBTC)'),  # http://kbtc.org
 180     )
 181
 182     IE_NAME = 'pbs'
 183     IE_DESC = 'Public Broadcasting Service (PBS) and member stations: %s' % ', '.join(list(zip(*_STATIONS))[1])
 184
 185     _VALID_URL = r'''(?x)https?://
 186         (?:
 187            # Direct video URL
 188            (?:%s)/(?:viralplayer|video)/(?P<id>[0-9]+)/? |
 189            # Article with embedded player (or direct video)
 190            (?:www\.)?pbs\.org/(?:[^/]+/){2,5}(?P<presumptive_id>[^/]+?)(?:\.html)?/?(?:$|[?\#]) |
 191            # Player
 192            (?:video|player)\.pbs\.org/(?:widget/)?partnerplayer/(?P<player_id>[^/]+)/
 193         )
 194     ''' % '|'.join(list(zip(*_STATIONS))[0])
 195
 196     _TESTS = [
 197         {
 198             'url': 'http://www.pbs.org/tpt/constitution-usa-peter-sagal/watch/a-more-perfect-union/',
 199             'md5': '173dc391afd361fa72eab5d3d918968d',
 200             'info_dict': {
 201                 'id': '2365006249',
 202                 'ext': 'mp4',
 203                 'title': 'Constitution USA with Peter Sagal - A More Perfect Union',
 204                 'description': 'md5:36f341ae62e251b8f5bd2b754b95a071',
 205                 'duration': 3190,
 206             },
 207         },
 208         {
 209             'url': 'http://www.pbs.org/wgbh/pages/frontline/losing-iraq/',
 210             'md5': '6f722cb3c3982186d34b0f13374499c7',
 211             'info_dict': {
 212                 'id': '2365297690',
 213                 'ext': 'mp4',
 214                 'title': 'FRONTLINE - Losing Iraq',
 215                 'description': 'md5:4d3eaa01f94e61b3e73704735f1196d9',
 216                 'duration': 5050,
 217             },
 218         },
 219         {
 220             'url': 'http://www.pbs.org/newshour/bb/education-jan-june12-cyberschools_02-23/',
 221             'md5': 'b19856d7f5351b17a5ab1dc6a64be633',
 222             'info_dict': {
 223                 'id': '2201174722',
 224                 'ext': 'mp4',
 225                 'title': 'PBS NewsHour - Cyber Schools Gain Popularity, but Quality Questions Persist',
 226                 'description': 'md5:95a19f568689d09a166dff9edada3301',
 227                 'duration': 801,
 228             },
 229         },
 230         {
 231             'url': 'http://www.pbs.org/wnet/gperf/dudamel-conducts-verdi-requiem-hollywood-bowl-full-episode/3374/',
 232             'md5': 'c62859342be2a0358d6c9eb306595978',
 233             'info_dict': {
 234                 'id': '2365297708',
 235                 'ext': 'mp4',
 236                 'title': 'Great Performances - Dudamel Conducts Verdi Requiem at the Hollywood Bowl - Full',
 237                 'description': 'md5:657897370e09e2bc6bf0f8d2cd313c6b',
 238                 'duration': 6559,
 239                 'thumbnail': 're:^https?://.*\.jpg$',
 240             },
 241         },
 242         {
 243             'url': 'http://www.pbs.org/wgbh/nova/earth/killer-typhoon.html',
 244             'md5': '908f3e5473a693b266b84e25e1cf9703',
 245             'info_dict': {
 246                 'id': '2365160389',
 247                 'display_id': 'killer-typhoon',
 248                 'ext': 'mp4',
 249                 'description': 'md5:c741d14e979fc53228c575894094f157',
 250                 'title': 'NOVA - Killer Typhoon',
 251                 'duration': 3172,
 252                 'thumbnail': 're:^https?://.*\.jpg$',
 253                 'upload_date': '20140122',
 254                 'age_limit': 10,
 255             },
 256         },
 257         {
 258             'url': 'http://www.pbs.org/wgbh/pages/frontline/united-states-of-secrets/',
 259             'info_dict': {
 260                 'id': 'united-states-of-secrets',
 261             },
 262             'playlist_count': 2,
 263         },
 264         {
 265             'url': 'http://www.pbs.org/wgbh/americanexperience/films/death/player/',
 266             'info_dict': {
 267                 'id': '2276541483',
 268                 'display_id': 'player',
 269                 'ext': 'mp4',
 270                 'title': 'American Experience - Death and the Civil War, Chapter 1',
 271                 'description': 'md5:1b80a74e0380ed2a4fb335026de1600d',
 272                 'duration': 682,
 273                 'thumbnail': 're:^https?://.*\.jpg$',
 274             },
 275             'params': {
 276                 'skip_download': True,  # requires ffmpeg
 277             },
 278         },
 279         {
 280             'url': 'http://www.pbs.org/video/2365245528/',
 281             'md5': '115223d41bd55cda8ae5cd5ed4e11497',
 282             'info_dict': {
 283                 'id': '2365245528',
 284                 'display_id': '2365245528',
 285                 'ext': 'mp4',
 286                 'title': 'FRONTLINE - United States of Secrets (Part One)',
 287                 'description': 'md5:55756bd5c551519cc4b7703e373e217e',
 288                 'duration': 6851,
 289                 'thumbnail': 're:^https?://.*\.jpg$',
 290             },
 291         },
 292         {
 293             # Video embedded in iframe containing angle brackets as attribute's value (e.g.
 294             # "<iframe style='position: absolute;<br />\ntop: 0; left: 0;' ...", see
 295             # https://github.com/rg3/youtube-dl/issues/7059)
 296             'url': 'http://www.pbs.org/food/features/a-chefs-life-season-3-episode-5-prickly-business/',
 297             'md5': '84ced42850d78f1d4650297356e95e6f',
 298             'info_dict': {
 299                 'id': '2365546844',
 300                 'display_id': 'a-chefs-life-season-3-episode-5-prickly-business',
 301                 'ext': 'mp4',
 302                 'title': "A Chef's Life - Season 3, Ep. 5: Prickly Business",
 303                 'description': 'md5:54033c6baa1f9623607c6e2ed245888b',
 304                 'duration': 1480,
 305                 'thumbnail': 're:^https?://.*\.jpg$',
 306             },
 307         },
 308         {
 309             # Frontline video embedded via flp2012.js
 310             'url': 'http://www.pbs.org/wgbh/pages/frontline/the-atomic-artists',
 311             'info_dict': {
 312                 'id': '2070868960',
 313                 'display_id': 'the-atomic-artists',
 314                 'ext': 'mp4',
 315                 'title': 'FRONTLINE - The Atomic Artists',
 316                 'description': 'md5:1a2481e86b32b2e12ec1905dd473e2c1',
 317                 'duration': 723,
 318                 'thumbnail': 're:^https?://.*\.jpg$',
 319             },
 320             'params': {
 321                 'skip_download': True,  # requires ffmpeg
 322             },
 323         },
 324         {
 325             # Serves hd only via wigget/partnerplayer page
 326             'url': 'http://www.pbs.org/video/2365641075/',
 327             'md5': 'acfd4c400b48149a44861cb16dd305cf',
 328             'info_dict': {
 329                 'id': '2365641075',
 330                 'ext': 'mp4',
 331                 'title': 'FRONTLINE - Netanyahu at War',
 332                 'duration': 6852,
 333                 'thumbnail': 're:^https?://.*\.jpg$',
 334                 'formats': 'mincount:8',
 335             },
 336         },
 337         {
 338             'url': 'http://player.pbs.org/widget/partnerplayer/2365297708/?start=0&end=0&chapterbar=false&endscreen=false&topbar=true',
 339             'only_matching': True,
 340         },
 341         {
 342             'url': 'http://watch.knpb.org/video/2365616055/',
 343             'only_matching': True,
 344         }
 345     ]
 346     _ERRORS = {
 347         101: 'We\'re sorry, but this video is not yet available.',
 348         403: 'We\'re sorry, but this video is not available in your region due to right restrictions.',
 349         404: 'We are experiencing technical difficulties that are preventing us from playing the video at this time. Please check back again soon.',
 350         410: 'This video has expired and is no longer available for online streaming.',
 351     }
 352
 353     def _extract_webpage(self, url):
 354         mobj = re.match(self._VALID_URL, url)
 355
 356         presumptive_id = mobj.group('presumptive_id')
 357         display_id = presumptive_id
 358         if presumptive_id:
 359             webpage = self._download_webpage(url, display_id)
 360
 361             upload_date = unified_strdate(self._search_regex(
 362                 r'<input type="hidden" id="air_date_[0-9]+" value="([^"]+)"',
 363                 webpage, 'upload date', default=None))
 364
 365             # tabbed frontline videos
 366             MULTI_PART_REGEXES = (
 367                 r'<div[^>]+class="videotab[^"]*"[^>]+vid="(\d+)"',
 368                 r'<a[^>]+href=["\']#video-\d+["\'][^>]+data-coveid=["\'](\d+)',
 369             )
 370             for p in MULTI_PART_REGEXES:
 371                 tabbed_videos = re.findall(p, webpage)
 372                 if tabbed_videos:
 373                     return tabbed_videos, presumptive_id, upload_date
 374
 375             MEDIA_ID_REGEXES = [
 376                 r"div\s*:\s*'videoembed'\s*,\s*mediaid\s*:\s*'(\d+)'",  # frontline video embed
 377                 r'class="coveplayerid">([^<]+)<',                       # coveplayer
 378                 r'<section[^>]+data-coveid="(\d+)"',                    # coveplayer from http://www.pbs.org/wgbh/frontline/film/real-csi/
 379                 r'<input type="hidden" id="pbs_video_id_[0-9]+" value="([0-9]+)"/>',  # jwplayer
 380             ]
 381
 382             media_id = self._search_regex(
 383                 MEDIA_ID_REGEXES, webpage, 'media ID', fatal=False, default=None)
 384             if media_id:
 385                 return media_id, presumptive_id, upload_date
 386
 387             # Fronline video embedded via flp
 388             video_id = self._search_regex(
 389                 r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid', default=None)
 390             if video_id:
 391                 # pkg_id calculation is reverse engineered from
 392                 # http://www.pbs.org/wgbh/pages/frontline/js/flp2012.js
 393                 prg_id = self._search_regex(
 394                     r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid')[7:]
 395                 if 'q' in prg_id:
 396                     prg_id = prg_id.split('q')[1]
 397                 prg_id = int(prg_id, 16)
 398                 getdir = self._download_json(
 399                     'http://www.pbs.org/wgbh/pages/frontline/.json/getdir/getdir%d.json' % prg_id,
 400                     presumptive_id, 'Downloading getdir JSON',
 401                     transform_source=strip_jsonp)
 402                 return getdir['mid'], presumptive_id, upload_date
 403
 404             for iframe in re.findall(r'(?s)<iframe(.+?)></iframe>', webpage):
 405                 url = self._search_regex(
 406                     r'src=(["\'])(?P<url>.+?partnerplayer.+?)\1', iframe,
 407                     'player URL', default=None, group='url')
 408                 if url:
 409                     break
 410
 411             mobj = re.match(self._VALID_URL, url)
 412
 413         player_id = mobj.group('player_id')
 414         if not display_id:
 415             display_id = player_id
 416         if player_id:
 417             player_page = self._download_webpage(
 418                 url, display_id, note='Downloading player page',
 419                 errnote='Could not download player page')
 420             video_id = self._search_regex(
 421                 r'<div\s+id="video_([0-9]+)"', player_page, 'video ID')
 422         else:
 423             video_id = mobj.group('id')
 424             display_id = video_id
 425
 426         return video_id, display_id, None
 427
 428     def _real_extract(self, url):
 429         video_id, display_id, upload_date = self._extract_webpage(url)
 430
 431         if isinstance(video_id, list):
 432             entries = [self.url_result(
 433                 'http://video.pbs.org/video/%s' % vid_id, 'PBS', vid_id)
 434                 for vid_id in video_id]
 435             return self.playlist_result(entries, display_id)
 436
 437         info = None
 438         redirects = []
 439         redirect_urls = set()
 440
 441         def extract_redirect_urls(info):
 442             for encoding_name in ('recommended_encoding', 'alternate_encoding'):
 443                 redirect = info.get(encoding_name)
 444                 if not redirect:
 445                     continue
 446                 redirect_url = redirect.get('url')
 447                 if redirect_url and redirect_url not in redirect_urls:
 448                     redirects.append(redirect)
 449                     redirect_urls.add(redirect_url)
 450
 451         try:
 452             video_info = self._download_json(
 453                 'http://player.pbs.org/videoInfo/%s?format=json&type=partner' % video_id,
 454                 display_id, 'Downloading video info JSON')
 455             extract_redirect_urls(video_info)
 456             info = video_info
 457         except ExtractorError as e:
 458             # videoInfo API may not work for some videos
 459             if not isinstance(e.cause, compat_HTTPError) or e.cause.code != 404:
 460                 raise
 461
 462         # Player pages may also serve different qualities
 463         for page in ('widget/partnerplayer', 'portalplayer'):
 464             player = self._download_webpage(
 465                 'http://player.pbs.org/%s/%s' % (page, video_id),
 466                 display_id, 'Downloading %s page' % page, fatal=False)
 467             if player:
 468                 video_info = self._parse_json(
 469                     self._search_regex(
 470                         r'(?s)PBS\.videoData\s*=\s*({.+?});\n',
 471                         player, '%s video data' % page, default='{}'),
 472                     display_id, transform_source=js_to_json, fatal=False)
 473                 if video_info:
 474                     extract_redirect_urls(video_info)
 475                     if not info:
 476                         info = video_info
 477
 478         formats = []
 479         http_url = None
 480         for num, redirect in enumerate(redirects):
 481             redirect_id = redirect.get('eeid')
 482
 483             redirect_info = self._download_json(
 484                 '%s?format=json' % redirect['url'], display_id,
 485                 'Downloading %s video url info' % (redirect_id or num))
 486
 487             if redirect_info['status'] == 'error':
 488                 raise ExtractorError(
 489                     '%s said: %s' % (
 490                         self.IE_NAME,
 491                         self._ERRORS.get(redirect_info['http_code'], redirect_info['message'])),
 492                     expected=True)
 493
 494             format_url = redirect_info.get('url')
 495             if not format_url:
 496                 continue
 497
 498             if determine_ext(format_url) == 'm3u8':
 499                 formats.extend(self._extract_m3u8_formats(
 500                     format_url, display_id, 'mp4', m3u8_id='hls', fatal=False))
 501             else:
 502                 formats.append({
 503                     'url': format_url,
 504                     'format_id': redirect_id,
 505                 })
 506                 if re.search(r'^https?://.*(?:\d+k|baseline)', format_url):
 507                     http_url = format_url
 508         self._remove_duplicate_formats(formats)
 509         m3u8_formats = list(filter(
 510             lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none' and f.get('resolution') != 'multiple',
 511             formats))
 512         if http_url:
 513             for m3u8_format in m3u8_formats:
 514                 bitrate = self._search_regex(r'(\d+k)', m3u8_format['url'], 'bitrate', default=None)
 515                 # extract only the formats that we know that they will be available as http format.
 516                 # https://projects.pbs.org/confluence/display/coveapi/COVE+Video+Specifications
 517                 if not bitrate or bitrate not in ('400k', '800k', '1200k', '2500k'):
 518                     continue
 519                 f = m3u8_format.copy()
 520                 f.update({
 521                     'url': re.sub(r'\d+k|baseline', bitrate, http_url),
 522                     'format_id': m3u8_format['format_id'].replace('hls', 'http'),
 523                     'protocol': 'http',
 524                 })
 525                 formats.append(f)
 526         self._sort_formats(formats)
 527
 528         rating_str = info.get('rating')
 529         if rating_str is not None:
 530             rating_str = rating_str.rpartition('-')[2]
 531         age_limit = US_RATINGS.get(rating_str)
 532
 533         subtitles = {}
 534         closed_captions_url = info.get('closed_captions_url')
 535         if closed_captions_url:
 536             subtitles['en'] = [{
 537                 'ext': 'ttml',
 538                 'url': closed_captions_url,
 539             }]
 540             mobj = re.search(r'/(\d+)_Encoded\.dfxp', closed_captions_url)
 541             if mobj:
 542                 ttml_caption_suffix, ttml_caption_id = mobj.group(0, 1)
 543                 ttml_caption_id = int(ttml_caption_id)
 544                 subtitles['en'].extend([{
 545                     'url': closed_captions_url.replace(
 546                         ttml_caption_suffix, '/%d_Encoded.srt' % (ttml_caption_id + 1)),
 547                     'ext': 'srt',
 548                 }, {
 549                     'url': closed_captions_url.replace(
 550                         ttml_caption_suffix, '/%d_Encoded.vtt' % (ttml_caption_id + 2)),
 551                     'ext': 'vtt',
 552                 }])
 553
 554         # info['title'] is often incomplete (e.g. 'Full Episode', 'Episode 5', etc)
 555         # Try turning it to 'program - title' naming scheme if possible
 556         alt_title = info.get('program', {}).get('title')
 557         if alt_title:
 558             info['title'] = alt_title + ' - ' + re.sub(r'^' + alt_title + '[\s\-:]+', '', info['title'])
 559
 560         return {
 561             'id': video_id,
 562             'display_id': display_id,
 563             'title': info['title'],
 564             'description': info.get('description') or info.get('program', {}).get('description'),
 565             'thumbnail': info.get('image_url'),
 566             'duration': int_or_none(info.get('duration')),
 567             'age_limit': age_limit,
 568             'upload_date': upload_date,
 569             'formats': formats,
 570             'subtitles': subtitles,
 571         }