[pbs] add test for videos with undocumented http formats and remove unused import
[youtube-dl] / 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     js_to_json,
12     strip_jsonp,
13     unified_strdate,
14     US_RATINGS,
15 )
16
17
18 class PBSIE(InfoExtractor):
19     _STATIONS = (
20         (r'(?:video|www|player)\.pbs\.org', 'PBS: Public Broadcasting Service'),  # http://www.pbs.org/
21         (r'video\.aptv\.org', 'APT - Alabama Public Television (WBIQ)'),  # http://aptv.org/
22         (r'video\.gpb\.org', 'GPB/Georgia Public Broadcasting (WGTV)'),  # http://www.gpb.org/
23         (r'video\.mpbonline\.org', 'Mississippi Public Broadcasting (WMPN)'),  # http://www.mpbonline.org
24         (r'video\.wnpt\.org', 'Nashville Public Television (WNPT)'),  # http://www.wnpt.org
25         (r'video\.wfsu\.org', 'WFSU-TV (WFSU)'),  # http://wfsu.org/
26         (r'video\.wsre\.org', 'WSRE (WSRE)'),  # http://www.wsre.org
27         (r'video\.wtcitv\.org', 'WTCI (WTCI)'),  # http://www.wtcitv.org
28         (r'video\.pba\.org', 'WPBA/Channel 30 (WPBA)'),  # http://pba.org/
29         (r'video\.alaskapublic\.org', 'Alaska Public Media (KAKM)'),  # http://alaskapublic.org/kakm
30         # (r'kuac\.org', 'KUAC (KUAC)'),  # http://kuac.org/kuac-tv/
31         # (r'ktoo\.org', '360 North (KTOO)'),  # http://www.ktoo.org/
32         # (r'azpm\.org', 'KUAT 6 (KUAT)'),  # http://www.azpm.org/
33         (r'video\.azpbs\.org', 'Arizona PBS (KAET)'),  # http://www.azpbs.org
34         (r'portal\.knme\.org', 'KNME-TV/Channel 5 (KNME)'),  # http://www.newmexicopbs.org/
35         (r'video\.vegaspbs\.org', 'Vegas PBS (KLVX)'),  # http://vegaspbs.org/
36         (r'watch\.aetn\.org', 'AETN/ARKANSAS ETV NETWORK (KETS)'),  # http://www.aetn.org/
37         (r'video\.ket\.org', 'KET (WKLE)'),  # http://www.ket.org/
38         (r'video\.wkno\.org', 'WKNO/Channel 10 (WKNO)'),  # http://www.wkno.org/
39         (r'video\.lpb\.org', 'LPB/LOUISIANA PUBLIC BROADCASTING (WLPB)'),  # http://www.lpb.org/
40         (r'videos\.oeta\.tv', 'OETA (KETA)'),  # http://www.oeta.tv
41         (r'video\.optv\.org', 'Ozarks Public Television (KOZK)'),  # http://www.optv.org/
42         (r'watch\.wsiu\.org', 'WSIU Public Broadcasting (WSIU)'),  # http://www.wsiu.org/
43         (r'video\.keet\.org', 'KEET TV (KEET)'),  # http://www.keet.org
44         (r'pbs\.kixe\.org', 'KIXE/Channel 9 (KIXE)'),  # http://kixe.org/
45         (r'video\.kpbs\.org', 'KPBS San Diego (KPBS)'),  # http://www.kpbs.org/
46         (r'video\.kqed\.org', 'KQED (KQED)'),  # http://www.kqed.org
47         (r'vids\.kvie\.org', 'KVIE Public Television (KVIE)'),  # http://www.kvie.org
48         (r'video\.pbssocal\.org', 'PBS SoCal/KOCE (KOCE)'),  # http://www.pbssocal.org/
49         (r'video\.valleypbs\.org', 'ValleyPBS (KVPT)'),  # http://www.valleypbs.org/
50         (r'video\.cptv\.org', 'CONNECTICUT PUBLIC TELEVISION (WEDH)'),  # http://cptv.org
51         (r'watch\.knpb\.org', 'KNPB Channel 5 (KNPB)'),  # http://www.knpb.org/
52         (r'video\.soptv\.org', 'SOPTV (KSYS)'),  # http://www.soptv.org
53         # (r'klcs\.org', 'KLCS/Channel 58 (KLCS)'),  # http://www.klcs.org
54         # (r'krcb\.org', 'KRCB Television & Radio (KRCB)'),  # http://www.krcb.org
55         # (r'kvcr\.org', 'KVCR TV/DT/FM :: Vision for the Future (KVCR)'),  # http://kvcr.org
56         (r'video\.rmpbs\.org', 'Rocky Mountain PBS (KRMA)'),  # http://www.rmpbs.org
57         (r'video\.kenw\.org', 'KENW-TV3 (KENW)'),  # http://www.kenw.org
58         (r'video\.kued\.org', 'KUED Channel 7 (KUED)'),  # http://www.kued.org
59         (r'video\.wyomingpbs\.org', 'Wyoming PBS (KCWC)'),  # http://www.wyomingpbs.org
60         (r'video\.cpt12\.org', 'Colorado Public Television / KBDI 12 (KBDI)'),  # http://www.cpt12.org/
61         (r'video\.kbyueleven\.org', 'KBYU-TV (KBYU)'),  # http://www.kbyutv.org/
62         (r'video\.thirteen\.org', 'Thirteen/WNET New York (WNET)'),  # http://www.thirteen.org
63         (r'video\.wgbh\.org', 'WGBH/Channel 2 (WGBH)'),  # http://wgbh.org
64         (r'video\.wgby\.org', 'WGBY (WGBY)'),  # http://www.wgby.org
65         (r'watch\.njtvonline\.org', 'NJTV Public Media NJ (WNJT)'),  # http://www.njtvonline.org/
66         # (r'ripbs\.org', 'Rhode Island PBS (WSBE)'),  # http://www.ripbs.org/home/
67         (r'watch\.wliw\.org', 'WLIW21 (WLIW)'),  # http://www.wliw.org/
68         (r'video\.mpt\.tv', 'mpt/Maryland Public Television (WMPB)'),  # http://www.mpt.org
69         (r'watch\.weta\.org', 'WETA Television and Radio (WETA)'),  # http://www.weta.org
70         (r'video\.whyy\.org', 'WHYY (WHYY)'),  # http://www.whyy.org
71         (r'video\.wlvt\.org', 'PBS 39 (WLVT)'),  # http://www.wlvt.org/
72         (r'video\.wvpt\.net', 'WVPT - Your Source for PBS and More! (WVPT)'),  # http://www.wvpt.net
73         (r'video\.whut\.org', 'Howard University Television (WHUT)'),  # http://www.whut.org
74         (r'video\.wedu\.org', 'WEDU PBS (WEDU)'),  # http://www.wedu.org
75         (r'video\.wgcu\.org', 'WGCU Public Media (WGCU)'),  # http://www.wgcu.org/
76         # (r'wjct\.org', 'WJCT Public Broadcasting (WJCT)'),  # http://www.wjct.org
77         (r'video\.wpbt2\.org', 'WPBT2 (WPBT)'),  # http://www.wpbt2.org
78         (r'video\.wucftv\.org', 'WUCF TV (WUCF)'),  # http://wucftv.org
79         (r'video\.wuft\.org', 'WUFT/Channel 5 (WUFT)'),  # http://www.wuft.org
80         (r'watch\.wxel\.org', 'WXEL/Channel 42 (WXEL)'),  # http://www.wxel.org/home/
81         (r'video\.wlrn\.org', 'WLRN/Channel 17 (WLRN)'),  # http://www.wlrn.org/
82         (r'video\.wusf\.usf\.edu', 'WUSF Public Broadcasting (WUSF)'),  # http://wusf.org/
83         (r'video\.scetv\.org', 'ETV (WRLK)'),  # http://www.scetv.org
84         (r'video\.unctv\.org', 'UNC-TV (WUNC)'),  # http://www.unctv.org/
85         # (r'pbsguam\.org', 'PBS Guam (KGTF)'),  # http://www.pbsguam.org/
86         (r'video\.pbshawaii\.org', 'PBS Hawaii - Oceanic Cable Channel 10 (KHET)'),  # http://www.pbshawaii.org/
87         (r'video\.idahoptv\.org', 'Idaho Public Television (KAID)'),  # http://idahoptv.org
88         (r'video\.ksps\.org', 'KSPS (KSPS)'),  # http://www.ksps.org/home/
89         (r'watch\.opb\.org', 'OPB (KOPB)'),  # http://www.opb.org
90         (r'watch\.nwptv\.org', 'KWSU/Channel 10 & KTNW/Channel 31 (KWSU)'),  # http://www.kwsu.org
91         (r'video\.will\.illinois\.edu', 'WILL-TV (WILL)'),  # http://will.illinois.edu/
92         (r'video\.networkknowledge\.tv', 'Network Knowledge - WSEC/Springfield (WSEC)'),  # http://www.wsec.tv
93         (r'video\.wttw\.com', 'WTTW11 (WTTW)'),  # http://www.wttw.com/
94         # (r'wtvp\.org', 'WTVP & WTVP.org, Public Media for Central Illinois (WTVP)'),  # http://www.wtvp.org/
95         (r'video\.iptv\.org', 'Iowa Public Television/IPTV (KDIN)'),  # http://www.iptv.org/
96         (r'video\.ninenet\.org', 'Nine Network (KETC)'),  # http://www.ninenet.org
97         (r'video\.wfwa\.org', 'PBS39 Fort Wayne (WFWA)'),  # http://wfwa.org/
98         (r'video\.wfyi\.org', 'WFYI Indianapolis (WFYI)'),  # http://www.wfyi.org
99         (r'video\.mptv\.org', 'Milwaukee Public Television (WMVS)'),  # http://www.mptv.org
100         (r'video\.wnin\.org', 'WNIN (WNIN)'),  # http://www.wnin.org/
101         (r'video\.wnit\.org', 'WNIT Public Television (WNIT)'),  # http://www.wnit.org/
102         (r'video\.wpt\.org', 'WPT (WPNE)'),  # http://www.wpt.org/
103         (r'video\.wvut\.org', 'WVUT/Channel 22 (WVUT)'),  # http://wvut.org/
104         (r'video\.weiu\.net', 'WEIU/Channel 51 (WEIU)'),  # http://www.weiu.net
105         (r'video\.wqpt\.org', 'WQPT-TV (WQPT)'),  # http://www.wqpt.org
106         (r'video\.wycc\.org', 'WYCC PBS Chicago (WYCC)'),  # http://www.wycc.org
107         # (r'lakeshorepublicmedia\.org', 'Lakeshore Public Television (WYIN)'),  # http://lakeshorepublicmedia.org/
108         (r'video\.wipb\.org', 'WIPB-TV (WIPB)'),  # http://wipb.org
109         (r'video\.indianapublicmedia\.org', 'WTIU (WTIU)'),  # http://indianapublicmedia.org/tv/
110         (r'watch\.cetconnect\.org', 'CET  (WCET)'),  # http://www.cetconnect.org
111         (r'video\.thinktv\.org', 'ThinkTVNetwork (WPTD)'),  # http://www.thinktv.org
112         (r'video\.wbgu\.org', 'WBGU-TV (WBGU)'),  # http://wbgu.org
113         (r'video\.wgvu\.org', 'WGVU TV (WGVU)'),  # http://www.wgvu.org/
114         (r'video\.netnebraska\.org', 'NET1 (KUON)'),  # http://netnebraska.org
115         (r'video\.pioneer\.org', 'Pioneer Public Television (KWCM)'),  # http://www.pioneer.org
116         (r'watch\.sdpb\.org', 'SDPB Television (KUSD)'),  # http://www.sdpb.org
117         (r'video\.tpt\.org', 'TPT (KTCA)'),  # http://www.tpt.org
118         (r'watch\.ksmq\.org', 'KSMQ (KSMQ)'),  # http://www.ksmq.org/
119         (r'watch\.kpts\.org', 'KPTS/Channel 8 (KPTS)'),  # http://www.kpts.org/
120         (r'watch\.ktwu\.org', 'KTWU/Channel 11 (KTWU)'),  # http://ktwu.org
121         # (r'shptv\.org', 'Smoky Hills Public Television (KOOD)'),  # http://www.shptv.org
122         # (r'kcpt\.org', 'KCPT Kansas City Public Television (KCPT)'),  # http://kcpt.org/
123         # (r'blueridgepbs\.org', 'Blue Ridge PBS (WBRA)'),  # http://www.blueridgepbs.org/
124         (r'watch\.easttennesseepbs\.org', 'East Tennessee PBS (WSJK)'),  # http://easttennesseepbs.org
125         (r'video\.wcte\.tv', 'WCTE-TV (WCTE)'),  # http://www.wcte.org
126         (r'video\.wljt\.org', 'WLJT, Channel 11 (WLJT)'),  # http://wljt.org/
127         (r'video\.wosu\.org', 'WOSU TV (WOSU)'),  # http://wosu.org/
128         (r'video\.woub\.org', 'WOUB/WOUC (WOUB)'),  # http://woub.org/tv/index.php?section=5
129         (r'video\.wvpublic\.org', 'WVPB (WVPB)'),  # http://wvpublic.org/
130         (r'video\.wkyupbs\.org', 'WKYU-PBS (WKYU)'),  # http://www.wkyupbs.org
131         # (r'wyes\.org', 'WYES-TV/New Orleans (WYES)'),  # http://www.wyes.org
132         (r'video\.kera\.org', 'KERA 13 (KERA)'),  # http://www.kera.org/
133         (r'video\.mpbn\.net', 'MPBN (WCBB)'),  # http://www.mpbn.net/
134         (r'video\.mountainlake\.org', 'Mountain Lake PBS (WCFE)'),  # http://www.mountainlake.org/
135         (r'video\.nhptv\.org', 'NHPTV (WENH)'),  # http://nhptv.org/
136         (r'video\.vpt\.org', 'Vermont PBS (WETK)'),  # http://www.vpt.org
137         (r'video\.witf\.org', 'witf (WITF)'),  # http://www.witf.org
138         (r'watch\.wqed\.org', 'WQED Multimedia (WQED)'),  # http://www.wqed.org/
139         (r'video\.wmht\.org', 'WMHT Educational Telecommunications (WMHT)'),  # http://www.wmht.org/home/
140         (r'video\.deltabroadcasting\.org', 'Q-TV (WDCQ)'),  # http://www.deltabroadcasting.org
141         (r'video\.dptv\.org', 'WTVS Detroit Public TV (WTVS)'),  # http://www.dptv.org/
142         (r'video\.wcmu\.org', 'CMU Public Television (WCMU)'),  # http://www.wcmu.org
143         (r'video\.wkar\.org', 'WKAR-TV (WKAR)'),  # http://wkar.org/
144         (r'wnmuvideo\.nmu\.edu', 'WNMU-TV Public TV 13 (WNMU)'),  # http://wnmutv.nmu.edu
145         (r'video\.wdse\.org', 'WDSE - WRPT (WDSE)'),  # http://www.wdse.org/
146         (r'video\.wgte\.org', 'WGTE TV (WGTE)'),  # http://www.wgte.org
147         (r'video\.lptv\.org', 'Lakeland Public Television (KAWE)'),  # http://www.lakelandptv.org
148         # (r'prairiepublic\.org', 'PRAIRIE PUBLIC (KFME)'),  # http://www.prairiepublic.org/
149         (r'video\.kmos\.org', 'KMOS-TV - Channels 6.1, 6.2 and 6.3 (KMOS)'),  # http://www.kmos.org/
150         (r'watch\.montanapbs\.org', 'MontanaPBS (KUSM)'),  # http://montanapbs.org
151         (r'video\.krwg\.org', 'KRWG/Channel 22 (KRWG)'),  # http://www.krwg.org
152         (r'video\.kacvtv\.org', 'KACV (KACV)'),  # http://www.panhandlepbs.org/home/
153         (r'video\.kcostv\.org', 'KCOS/Channel 13 (KCOS)'),  # www.kcostv.org
154         (r'video\.wcny\.org', 'WCNY/Channel 24 (WCNY)'),  # http://www.wcny.org
155         (r'video\.wned\.org', 'WNED (WNED)'),  # http://www.wned.org/
156         (r'watch\.wpbstv\.org', 'WPBS (WPBS)'),  # http://www.wpbstv.org
157         (r'video\.wskg\.org', 'WSKG Public TV (WSKG)'),  # http://wskg.org
158         (r'video\.wxxi\.org', 'WXXI (WXXI)'),  # http://wxxi.org
159         (r'video\.wpsu\.org', 'WPSU (WPSU)'),  # http://www.wpsu.org
160         # (r'wqln\.org', 'WQLN/Channel 54 (WQLN)'),  # http://www.wqln.org
161         (r'on-demand\.wvia\.org', 'WVIA Public Media Studios (WVIA)'),  # http://www.wvia.org/
162         (r'video\.wtvi\.org', 'WTVI (WTVI)'),  # http://www.wtvi.org/
163         # (r'whro\.org', 'WHRO (WHRO)'),  # http://whro.org
164         (r'video\.westernreservepublicmedia\.org', 'Western Reserve PBS (WNEO)'),  # http://www.WesternReservePublicMedia.org/
165         (r'video\.ideastream\.org', 'WVIZ/PBS ideastream (WVIZ)'),  # http://www.wviz.org/
166         (r'video\.kcts9\.org', 'KCTS 9 (KCTS)'),  # http://kcts9.org/
167         (r'video\.basinpbs\.org', 'Basin PBS (KPBT)'),  # http://www.basinpbs.org
168         (r'video\.houstonpbs\.org', 'KUHT / Channel 8 (KUHT)'),  # http://www.houstonpublicmedia.org/
169         # (r'tamu\.edu', 'KAMU - TV (KAMU)'),  # http://KAMU.tamu.edu
170         # (r'kedt\.org', 'KEDT/Channel 16 (KEDT)'),  # http://www.kedt.org
171         (r'video\.klrn\.org', 'KLRN (KLRN)'),  # http://www.klrn.org
172         (r'video\.klru\.tv', 'KLRU (KLRU)'),  # http://www.klru.org
173         # (r'kmbh\.org', 'KMBH-TV (KMBH)'),  # http://www.kmbh.org
174         # (r'knct\.org', 'KNCT (KNCT)'),  # http://www.knct.org
175         # (r'ktxt\.org', 'KTTZ-TV (KTXT)'),  # http://www.ktxt.org
176         (r'video\.wtjx\.org', 'WTJX Channel 12 (WTJX)'),  # http://www.wtjx.org/
177         (r'video\.ideastations\.org', 'WCVE PBS (WCVE)'),  # http://ideastations.org/
178         (r'video\.kbtc\.org', 'KBTC Public Television (KBTC)'),  # http://kbtc.org
179     )
180
181     IE_NAME = 'pbs'
182     IE_DESC = 'Public Broadcasting Service (PBS) and member stations: %s' % ', '.join(list(zip(*_STATIONS))[1])
183
184     _VALID_URL = r'''(?x)https?://
185         (?:
186            # Direct video URL
187            (?:%s)/(?:viralplayer|video)/(?P<id>[0-9]+)/? |
188            # Article with embedded player (or direct video)
189            (?:www\.)?pbs\.org/(?:[^/]+/){2,5}(?P<presumptive_id>[^/]+?)(?:\.html)?/?(?:$|[?\#]) |
190            # Player
191            (?:video|player)\.pbs\.org/(?:widget/)?partnerplayer/(?P<player_id>[^/]+)/
192         )
193     ''' % '|'.join(list(zip(*_STATIONS))[0])
194
195     _TESTS = [
196         {
197             'url': 'http://www.pbs.org/tpt/constitution-usa-peter-sagal/watch/a-more-perfect-union/',
198             'md5': '173dc391afd361fa72eab5d3d918968d',
199             'info_dict': {
200                 'id': '2365006249',
201                 'ext': 'mp4',
202                 'title': 'Constitution USA with Peter Sagal - A More Perfect Union',
203                 'description': 'md5:36f341ae62e251b8f5bd2b754b95a071',
204                 'duration': 3190,
205             },
206         },
207         {
208             'url': 'http://www.pbs.org/wgbh/pages/frontline/losing-iraq/',
209             'md5': '6f722cb3c3982186d34b0f13374499c7',
210             'info_dict': {
211                 'id': '2365297690',
212                 'ext': 'mp4',
213                 'title': 'FRONTLINE - Losing Iraq',
214                 'description': 'md5:4d3eaa01f94e61b3e73704735f1196d9',
215                 'duration': 5050,
216             },
217         },
218         {
219             'url': 'http://www.pbs.org/newshour/bb/education-jan-june12-cyberschools_02-23/',
220             'md5': 'b19856d7f5351b17a5ab1dc6a64be633',
221             'info_dict': {
222                 'id': '2201174722',
223                 'ext': 'mp4',
224                 'title': 'PBS NewsHour - Cyber Schools Gain Popularity, but Quality Questions Persist',
225                 'description': 'md5:95a19f568689d09a166dff9edada3301',
226                 'duration': 801,
227             },
228         },
229         {
230             'url': 'http://www.pbs.org/wnet/gperf/dudamel-conducts-verdi-requiem-hollywood-bowl-full-episode/3374/',
231             'md5': 'c62859342be2a0358d6c9eb306595978',
232             'info_dict': {
233                 'id': '2365297708',
234                 'ext': 'mp4',
235                 'title': 'Great Performances - Dudamel Conducts Verdi Requiem at the Hollywood Bowl - Full',
236                 'description': 'md5:657897370e09e2bc6bf0f8d2cd313c6b',
237                 'duration': 6559,
238                 'thumbnail': 're:^https?://.*\.jpg$',
239             },
240         },
241         {
242             'url': 'http://www.pbs.org/wgbh/nova/earth/killer-typhoon.html',
243             'md5': '908f3e5473a693b266b84e25e1cf9703',
244             'info_dict': {
245                 'id': '2365160389',
246                 'display_id': 'killer-typhoon',
247                 'ext': 'mp4',
248                 'description': 'md5:c741d14e979fc53228c575894094f157',
249                 'title': 'NOVA - Killer Typhoon',
250                 'duration': 3172,
251                 'thumbnail': 're:^https?://.*\.jpg$',
252                 'upload_date': '20140122',
253                 'age_limit': 10,
254             },
255         },
256         {
257             'url': 'http://www.pbs.org/wgbh/pages/frontline/united-states-of-secrets/',
258             'info_dict': {
259                 'id': 'united-states-of-secrets',
260             },
261             'playlist_count': 2,
262         },
263         {
264             'url': 'http://www.pbs.org/wgbh/americanexperience/films/death/player/',
265             'info_dict': {
266                 'id': '2276541483',
267                 'display_id': 'player',
268                 'ext': 'mp4',
269                 'title': 'American Experience - Death and the Civil War, Chapter 1',
270                 'description': 'md5:1b80a74e0380ed2a4fb335026de1600d',
271                 'duration': 682,
272                 'thumbnail': 're:^https?://.*\.jpg$',
273             },
274             'params': {
275                 'skip_download': True,  # requires ffmpeg
276             },
277         },
278         {
279             'url': 'http://www.pbs.org/video/2365245528/',
280             'md5': '115223d41bd55cda8ae5cd5ed4e11497',
281             'info_dict': {
282                 'id': '2365245528',
283                 'display_id': '2365245528',
284                 'ext': 'mp4',
285                 'title': 'FRONTLINE - United States of Secrets (Part One)',
286                 'description': 'md5:55756bd5c551519cc4b7703e373e217e',
287                 'duration': 6851,
288                 'thumbnail': 're:^https?://.*\.jpg$',
289             },
290         },
291         {
292             # Video embedded in iframe containing angle brackets as attribute's value (e.g.
293             # "<iframe style='position: absolute;<br />\ntop: 0; left: 0;' ...", see
294             # https://github.com/rg3/youtube-dl/issues/7059)
295             'url': 'http://www.pbs.org/food/features/a-chefs-life-season-3-episode-5-prickly-business/',
296             'md5': '84ced42850d78f1d4650297356e95e6f',
297             'info_dict': {
298                 'id': '2365546844',
299                 'display_id': 'a-chefs-life-season-3-episode-5-prickly-business',
300                 'ext': 'mp4',
301                 'title': "A Chef's Life - Season 3, Ep. 5: Prickly Business",
302                 'description': 'md5:54033c6baa1f9623607c6e2ed245888b',
303                 'duration': 1480,
304                 'thumbnail': 're:^https?://.*\.jpg$',
305             },
306         },
307         {
308             # Frontline video embedded via flp2012.js
309             'url': 'http://www.pbs.org/wgbh/pages/frontline/the-atomic-artists',
310             'info_dict': {
311                 'id': '2070868960',
312                 'display_id': 'the-atomic-artists',
313                 'ext': 'mp4',
314                 'title': 'FRONTLINE - The Atomic Artists',
315                 'description': 'md5:1a2481e86b32b2e12ec1905dd473e2c1',
316                 'duration': 723,
317                 'thumbnail': 're:^https?://.*\.jpg$',
318             },
319             'params': {
320                 'skip_download': True,  # requires ffmpeg
321             },
322         },
323         {
324             # Serves hd only via wigget/partnerplayer page
325             'url': 'http://www.pbs.org/video/2365641075/',
326             'md5': 'acfd4c400b48149a44861cb16dd305cf',
327             'info_dict': {
328                 'id': '2365641075',
329                 'ext': 'mp4',
330                 'title': 'FRONTLINE - Netanyahu at War',
331                 'duration': 6852,
332                 'thumbnail': 're:^https?://.*\.jpg$',
333                 'formats': 'mincount:8',
334             },
335         },
336         {
337             # has undocumented http formats(4500k and 6500k)
338             'url': 'http://www.pbs.org/video/2365815229/',
339             'md5': '94635cd06b7133688e23f4b94e6637a5',
340             'info_dict': {
341                 'id': '2365815229',
342                 'ext': 'mp4',
343                 'title': 'FRONTLINE - Mosquito Hunter',
344             },
345         },
346         {
347             'url': 'http://player.pbs.org/widget/partnerplayer/2365297708/?start=0&end=0&chapterbar=false&endscreen=false&topbar=true',
348             'only_matching': True,
349         },
350         {
351             'url': 'http://watch.knpb.org/video/2365616055/',
352             'only_matching': True,
353         }
354     ]
355     _ERRORS = {
356         101: 'We\'re sorry, but this video is not yet available.',
357         403: 'We\'re sorry, but this video is not available in your region due to right restrictions.',
358         404: 'We are experiencing technical difficulties that are preventing us from playing the video at this time. Please check back again soon.',
359         410: 'This video has expired and is no longer available for online streaming.',
360     }
361
362     def _extract_webpage(self, url):
363         mobj = re.match(self._VALID_URL, url)
364
365         presumptive_id = mobj.group('presumptive_id')
366         display_id = presumptive_id
367         if presumptive_id:
368             webpage = self._download_webpage(url, display_id)
369
370             upload_date = unified_strdate(self._search_regex(
371                 r'<input type="hidden" id="air_date_[0-9]+" value="([^"]+)"',
372                 webpage, 'upload date', default=None))
373
374             # tabbed frontline videos
375             MULTI_PART_REGEXES = (
376                 r'<div[^>]+class="videotab[^"]*"[^>]+vid="(\d+)"',
377                 r'<a[^>]+href=["\']#video-\d+["\'][^>]+data-coveid=["\'](\d+)',
378             )
379             for p in MULTI_PART_REGEXES:
380                 tabbed_videos = re.findall(p, webpage)
381                 if tabbed_videos:
382                     return tabbed_videos, presumptive_id, upload_date
383
384             MEDIA_ID_REGEXES = [
385                 r"div\s*:\s*'videoembed'\s*,\s*mediaid\s*:\s*'(\d+)'",  # frontline video embed
386                 r'class="coveplayerid">([^<]+)<',                       # coveplayer
387                 r'<section[^>]+data-coveid="(\d+)"',                    # coveplayer from http://www.pbs.org/wgbh/frontline/film/real-csi/
388                 r'<input type="hidden" id="pbs_video_id_[0-9]+" value="([0-9]+)"/>',  # jwplayer
389             ]
390
391             media_id = self._search_regex(
392                 MEDIA_ID_REGEXES, webpage, 'media ID', fatal=False, default=None)
393             if media_id:
394                 return media_id, presumptive_id, upload_date
395
396             # Fronline video embedded via flp
397             video_id = self._search_regex(
398                 r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid', default=None)
399             if video_id:
400                 # pkg_id calculation is reverse engineered from
401                 # http://www.pbs.org/wgbh/pages/frontline/js/flp2012.js
402                 prg_id = self._search_regex(
403                     r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid')[7:]
404                 if 'q' in prg_id:
405                     prg_id = prg_id.split('q')[1]
406                 prg_id = int(prg_id, 16)
407                 getdir = self._download_json(
408                     'http://www.pbs.org/wgbh/pages/frontline/.json/getdir/getdir%d.json' % prg_id,
409                     presumptive_id, 'Downloading getdir JSON',
410                     transform_source=strip_jsonp)
411                 return getdir['mid'], presumptive_id, upload_date
412
413             for iframe in re.findall(r'(?s)<iframe(.+?)></iframe>', webpage):
414                 url = self._search_regex(
415                     r'src=(["\'])(?P<url>.+?partnerplayer.+?)\1', iframe,
416                     'player URL', default=None, group='url')
417                 if url:
418                     break
419
420             mobj = re.match(self._VALID_URL, url)
421
422         player_id = mobj.group('player_id')
423         if not display_id:
424             display_id = player_id
425         if player_id:
426             player_page = self._download_webpage(
427                 url, display_id, note='Downloading player page',
428                 errnote='Could not download player page')
429             video_id = self._search_regex(
430                 r'<div\s+id="video_([0-9]+)"', player_page, 'video ID')
431         else:
432             video_id = mobj.group('id')
433             display_id = video_id
434
435         return video_id, display_id, None
436
437     def _real_extract(self, url):
438         video_id, display_id, upload_date = self._extract_webpage(url)
439
440         if isinstance(video_id, list):
441             entries = [self.url_result(
442                 'http://video.pbs.org/video/%s' % vid_id, 'PBS', vid_id)
443                 for vid_id in video_id]
444             return self.playlist_result(entries, display_id)
445
446         info = None
447         redirects = []
448         redirect_urls = set()
449
450         def extract_redirect_urls(info):
451             for encoding_name in ('recommended_encoding', 'alternate_encoding'):
452                 redirect = info.get(encoding_name)
453                 if not redirect:
454                     continue
455                 redirect_url = redirect.get('url')
456                 if redirect_url and redirect_url not in redirect_urls:
457                     redirects.append(redirect)
458                     redirect_urls.add(redirect_url)
459
460         # Player pages may also serve different qualities
461         for page in ('widget/partnerplayer', 'portalplayer'):
462             player = self._download_webpage(
463                 'http://player.pbs.org/%s/%s' % (page, video_id),
464                 display_id, 'Downloading %s page' % page, fatal=False)
465             if player:
466                 video_info = self._parse_json(
467                     self._search_regex(
468                         r'(?s)PBS\.videoData\s*=\s*({.+?});\n',
469                         player, '%s video data' % page, default='{}'),
470                     display_id, transform_source=js_to_json, fatal=False)
471                 if video_info:
472                     extract_redirect_urls(video_info)
473                     if not info:
474                         info = video_info
475
476         formats = []
477         http_url = None
478         for num, redirect in enumerate(redirects):
479             redirect_id = redirect.get('eeid')
480
481             redirect_info = self._download_json(
482                 '%s?format=json' % redirect['url'], display_id,
483                 'Downloading %s video url info' % (redirect_id or num))
484
485             if redirect_info['status'] == 'error':
486                 raise ExtractorError(
487                     '%s said: %s' % (
488                         self.IE_NAME,
489                         self._ERRORS.get(redirect_info['http_code'], redirect_info['message'])),
490                     expected=True)
491
492             format_url = redirect_info.get('url')
493             if not format_url:
494                 continue
495
496             if determine_ext(format_url) == 'm3u8':
497                 formats.extend(self._extract_m3u8_formats(
498                     format_url, display_id, 'mp4', m3u8_id='hls', fatal=False))
499             else:
500                 formats.append({
501                     'url': format_url,
502                     'format_id': redirect_id,
503                 })
504                 if re.search(r'^https?://.*(?:\d+k|baseline)', format_url):
505                     http_url = format_url
506         self._remove_duplicate_formats(formats)
507         m3u8_formats = list(filter(
508             lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none' and f.get('resolution') != 'multiple',
509             formats))
510         if http_url:
511             for m3u8_format in m3u8_formats:
512                 bitrate = self._search_regex(r'(\d+)k', m3u8_format['url'], 'bitrate', default=None)
513                 # lower qualities(150k and 192k) are not available as http formats
514                 # https://github.com/rg3/youtube-dl/commit/cbc032c8b70a038a69259378c92b4ba97b42d491#commitcomment-17313656
515                 # we will try to extract any http format higher than than the lowest quality documented in
516                 # https://projects.pbs.org/confluence/display/coveapi/COVE+Video+Specifications
517                 # as there also undocumented http formats formats(4500k and 6500k)
518                 # http://www.pbs.org/video/2365815229/
519                 if not bitrate or int(bitrate) < 400:
520                     continue
521                 f_url = re.sub(r'\d+k|baseline', bitrate + 'k', http_url)
522                 # This may produce invalid links sometimes (e.g.
523                 # http://www.pbs.org/wgbh/frontline/film/suicide-plan)
524                 if not self._is_valid_url(f_url, display_id, 'http-%sk video' % bitrate):
525                     continue
526                 f = m3u8_format.copy()
527                 f.update({
528                     'url': f_url,
529                     'format_id': m3u8_format['format_id'].replace('hls', 'http'),
530                     'protocol': 'http',
531                 })
532                 formats.append(f)
533         self._sort_formats(formats)
534
535         rating_str = info.get('rating')
536         if rating_str is not None:
537             rating_str = rating_str.rpartition('-')[2]
538         age_limit = US_RATINGS.get(rating_str)
539
540         subtitles = {}
541         closed_captions_url = info.get('closed_captions_url')
542         if closed_captions_url:
543             subtitles['en'] = [{
544                 'ext': 'ttml',
545                 'url': closed_captions_url,
546             }]
547             mobj = re.search(r'/(\d+)_Encoded\.dfxp', closed_captions_url)
548             if mobj:
549                 ttml_caption_suffix, ttml_caption_id = mobj.group(0, 1)
550                 ttml_caption_id = int(ttml_caption_id)
551                 subtitles['en'].extend([{
552                     'url': closed_captions_url.replace(
553                         ttml_caption_suffix, '/%d_Encoded.srt' % (ttml_caption_id + 1)),
554                     'ext': 'srt',
555                 }, {
556                     'url': closed_captions_url.replace(
557                         ttml_caption_suffix, '/%d_Encoded.vtt' % (ttml_caption_id + 2)),
558                     'ext': 'vtt',
559                 }])
560
561         # info['title'] is often incomplete (e.g. 'Full Episode', 'Episode 5', etc)
562         # Try turning it to 'program - title' naming scheme if possible
563         alt_title = info.get('program', {}).get('title')
564         if alt_title:
565             info['title'] = alt_title + ' - ' + re.sub(r'^' + alt_title + '[\s\-:]+', '', info['title'])
566
567         return {
568             'id': video_id,
569             'display_id': display_id,
570             'title': info['title'],
571             'description': info.get('description') or info.get('program', {}).get('description'),
572             'thumbnail': info.get('image_url'),
573             'duration': int_or_none(info.get('duration')),
574             'age_limit': age_limit,
575             'upload_date': upload_date,
576             'formats': formats,
577             'subtitles': subtitles,
578         }