]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/pbs.py
f43e3a146e7bd35d9a99ab730289f4a1d4f5b91c
[youtubedl] / 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)?/?(?:$|[?\#]) |
191 # Player
192 (?:video|player)\.pbs\.org/(?:widget/)?partnerplayer/(?P<player_id>[^/]+)/
193 )
194 ''' % '|'.join(list(zip(*_STATIONS))[0])
195
196 _TESTS = [
197 {
198 'url': 'http://www.pbs.org/tpt/constitution-usa-peter-sagal/watch/a-more-perfect-union/',
199 'md5': 'ce1888486f0908d555a8093cac9a7362',
200 'info_dict': {
201 'id': '2365006249',
202 'ext': 'mp4',
203 'title': 'Constitution USA with Peter Sagal - A More Perfect Union',
204 'description': 'md5:36f341ae62e251b8f5bd2b754b95a071',
205 'duration': 3190,
206 },
207 'params': {
208 'skip_download': True, # requires ffmpeg
209 },
210 },
211 {
212 'url': 'http://www.pbs.org/wgbh/pages/frontline/losing-iraq/',
213 'md5': '143c98aa54a346738a3d78f54c925321',
214 'info_dict': {
215 'id': '2365297690',
216 'ext': 'mp4',
217 'title': 'FRONTLINE - Losing Iraq',
218 'description': 'md5:4d3eaa01f94e61b3e73704735f1196d9',
219 'duration': 5050,
220 },
221 'params': {
222 'skip_download': True, # requires ffmpeg
223 }
224 },
225 {
226 'url': 'http://www.pbs.org/newshour/bb/education-jan-june12-cyberschools_02-23/',
227 'md5': 'b19856d7f5351b17a5ab1dc6a64be633',
228 'info_dict': {
229 'id': '2201174722',
230 'ext': 'mp4',
231 'title': 'PBS NewsHour - Cyber Schools Gain Popularity, but Quality Questions Persist',
232 'description': 'md5:95a19f568689d09a166dff9edada3301',
233 'duration': 801,
234 },
235 },
236 {
237 'url': 'http://www.pbs.org/wnet/gperf/dudamel-conducts-verdi-requiem-hollywood-bowl-full-episode/3374/',
238 'md5': 'c62859342be2a0358d6c9eb306595978',
239 'info_dict': {
240 'id': '2365297708',
241 'ext': 'mp4',
242 'title': 'Great Performances - Dudamel Conducts Verdi Requiem at the Hollywood Bowl - Full',
243 'description': 'md5:657897370e09e2bc6bf0f8d2cd313c6b',
244 'duration': 6559,
245 'thumbnail': 're:^https?://.*\.jpg$',
246 },
247 'params': {
248 'skip_download': True, # requires ffmpeg
249 },
250 },
251 {
252 'url': 'http://www.pbs.org/wgbh/nova/earth/killer-typhoon.html',
253 'md5': '908f3e5473a693b266b84e25e1cf9703',
254 'info_dict': {
255 'id': '2365160389',
256 'display_id': 'killer-typhoon',
257 'ext': 'mp4',
258 'description': 'md5:c741d14e979fc53228c575894094f157',
259 'title': 'NOVA - Killer Typhoon',
260 'duration': 3172,
261 'thumbnail': 're:^https?://.*\.jpg$',
262 'upload_date': '20140122',
263 'age_limit': 10,
264 },
265 'params': {
266 'skip_download': True, # requires ffmpeg
267 },
268 },
269 {
270 'url': 'http://www.pbs.org/wgbh/pages/frontline/united-states-of-secrets/',
271 'info_dict': {
272 'id': 'united-states-of-secrets',
273 },
274 'playlist_count': 2,
275 },
276 {
277 'url': 'http://www.pbs.org/wgbh/americanexperience/films/death/player/',
278 'info_dict': {
279 'id': '2276541483',
280 'display_id': 'player',
281 'ext': 'mp4',
282 'title': 'American Experience - Death and the Civil War, Chapter 1',
283 'description': 'md5:1b80a74e0380ed2a4fb335026de1600d',
284 'duration': 682,
285 'thumbnail': 're:^https?://.*\.jpg$',
286 },
287 'params': {
288 'skip_download': True, # requires ffmpeg
289 },
290 },
291 {
292 'url': 'http://www.pbs.org/video/2365245528/',
293 'info_dict': {
294 'id': '2365245528',
295 'display_id': '2365245528',
296 'ext': 'mp4',
297 'title': 'FRONTLINE - United States of Secrets (Part One)',
298 'description': 'md5:55756bd5c551519cc4b7703e373e217e',
299 'duration': 6851,
300 'thumbnail': 're:^https?://.*\.jpg$',
301 },
302 'params': {
303 'skip_download': True, # requires ffmpeg
304 },
305 },
306 {
307 # Video embedded in iframe containing angle brackets as attribute's value (e.g.
308 # "<iframe style='position: absolute;<br />\ntop: 0; left: 0;' ...", see
309 # https://github.com/rg3/youtube-dl/issues/7059)
310 'url': 'http://www.pbs.org/food/features/a-chefs-life-season-3-episode-5-prickly-business/',
311 'info_dict': {
312 'id': '2365546844',
313 'display_id': 'a-chefs-life-season-3-episode-5-prickly-business',
314 'ext': 'mp4',
315 'title': "A Chef's Life - Season 3, Ep. 5: Prickly Business",
316 'description': 'md5:54033c6baa1f9623607c6e2ed245888b',
317 'duration': 1480,
318 'thumbnail': 're:^https?://.*\.jpg$',
319 },
320 'params': {
321 'skip_download': True, # requires ffmpeg
322 },
323 },
324 {
325 # Frontline video embedded via flp2012.js
326 'url': 'http://www.pbs.org/wgbh/pages/frontline/the-atomic-artists',
327 'info_dict': {
328 'id': '2070868960',
329 'display_id': 'the-atomic-artists',
330 'ext': 'mp4',
331 'title': 'FRONTLINE - The Atomic Artists',
332 'description': 'md5:1a2481e86b32b2e12ec1905dd473e2c1',
333 'duration': 723,
334 'thumbnail': 're:^https?://.*\.jpg$',
335 },
336 'params': {
337 'skip_download': True, # requires ffmpeg
338 },
339 },
340 {
341 # Serves hd only via wigget/partnerplayer page
342 'url': 'http://www.pbs.org/video/2365641075/',
343 'info_dict': {
344 'id': '2365641075',
345 'ext': 'mp4',
346 'title': 'FRONTLINE - Netanyahu at War',
347 'duration': 6852,
348 'thumbnail': 're:^https?://.*\.jpg$',
349 'formats': 'mincount:8',
350 },
351 'params': {
352 'skip_download': True, # requires ffmpeg
353 },
354 },
355 {
356 'url': 'http://player.pbs.org/widget/partnerplayer/2365297708/?start=0&end=0&chapterbar=false&endscreen=false&topbar=true',
357 'only_matching': True,
358 },
359 {
360 'url': 'http://watch.knpb.org/video/2365616055/',
361 'only_matching': True,
362 }
363 ]
364 _ERRORS = {
365 101: 'We\'re sorry, but this video is not yet available.',
366 403: 'We\'re sorry, but this video is not available in your region due to right restrictions.',
367 404: 'We are experiencing technical difficulties that are preventing us from playing the video at this time. Please check back again soon.',
368 410: 'This video has expired and is no longer available for online streaming.',
369 }
370
371 def _extract_webpage(self, url):
372 mobj = re.match(self._VALID_URL, url)
373
374 presumptive_id = mobj.group('presumptive_id')
375 display_id = presumptive_id
376 if presumptive_id:
377 webpage = self._download_webpage(url, display_id)
378
379 upload_date = unified_strdate(self._search_regex(
380 r'<input type="hidden" id="air_date_[0-9]+" value="([^"]+)"',
381 webpage, 'upload date', default=None))
382
383 # tabbed frontline videos
384 MULTI_PART_REGEXES = (
385 r'<div[^>]+class="videotab[^"]*"[^>]+vid="(\d+)"',
386 r'<a[^>]+href=["\']#video-\d+["\'][^>]+data-coveid=["\'](\d+)',
387 )
388 for p in MULTI_PART_REGEXES:
389 tabbed_videos = re.findall(p, webpage)
390 if tabbed_videos:
391 return tabbed_videos, presumptive_id, upload_date
392
393 MEDIA_ID_REGEXES = [
394 r"div\s*:\s*'videoembed'\s*,\s*mediaid\s*:\s*'(\d+)'", # frontline video embed
395 r'class="coveplayerid">([^<]+)<', # coveplayer
396 r'<section[^>]+data-coveid="(\d+)"', # coveplayer from http://www.pbs.org/wgbh/frontline/film/real-csi/
397 r'<input type="hidden" id="pbs_video_id_[0-9]+" value="([0-9]+)"/>', # jwplayer
398 ]
399
400 media_id = self._search_regex(
401 MEDIA_ID_REGEXES, webpage, 'media ID', fatal=False, default=None)
402 if media_id:
403 return media_id, presumptive_id, upload_date
404
405 # Fronline video embedded via flp
406 video_id = self._search_regex(
407 r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid', default=None)
408 if video_id:
409 # pkg_id calculation is reverse engineered from
410 # http://www.pbs.org/wgbh/pages/frontline/js/flp2012.js
411 prg_id = self._search_regex(
412 r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid')[7:]
413 if 'q' in prg_id:
414 prg_id = prg_id.split('q')[1]
415 prg_id = int(prg_id, 16)
416 getdir = self._download_json(
417 'http://www.pbs.org/wgbh/pages/frontline/.json/getdir/getdir%d.json' % prg_id,
418 presumptive_id, 'Downloading getdir JSON',
419 transform_source=strip_jsonp)
420 return getdir['mid'], presumptive_id, upload_date
421
422 for iframe in re.findall(r'(?s)<iframe(.+?)></iframe>', webpage):
423 url = self._search_regex(
424 r'src=(["\'])(?P<url>.+?partnerplayer.+?)\1', iframe,
425 'player URL', default=None, group='url')
426 if url:
427 break
428
429 mobj = re.match(self._VALID_URL, url)
430
431 player_id = mobj.group('player_id')
432 if not display_id:
433 display_id = player_id
434 if player_id:
435 player_page = self._download_webpage(
436 url, display_id, note='Downloading player page',
437 errnote='Could not download player page')
438 video_id = self._search_regex(
439 r'<div\s+id="video_([0-9]+)"', player_page, 'video ID')
440 else:
441 video_id = mobj.group('id')
442 display_id = video_id
443
444 return video_id, display_id, None
445
446 def _real_extract(self, url):
447 video_id, display_id, upload_date = self._extract_webpage(url)
448
449 if isinstance(video_id, list):
450 entries = [self.url_result(
451 'http://video.pbs.org/video/%s' % vid_id, 'PBS', vid_id)
452 for vid_id in video_id]
453 return self.playlist_result(entries, display_id)
454
455 info = None
456 redirects = []
457 redirect_urls = set()
458
459 def extract_redirect_urls(info):
460 for encoding_name in ('recommended_encoding', 'alternate_encoding'):
461 redirect = info.get(encoding_name)
462 if not redirect:
463 continue
464 redirect_url = redirect.get('url')
465 if redirect_url and redirect_url not in redirect_urls:
466 redirects.append(redirect)
467 redirect_urls.add(redirect_url)
468
469 try:
470 video_info = self._download_json(
471 'http://player.pbs.org/videoInfo/%s?format=json&type=partner' % video_id,
472 display_id, 'Downloading video info JSON')
473 extract_redirect_urls(video_info)
474 info = video_info
475 except ExtractorError as e:
476 # videoInfo API may not work for some videos
477 if not isinstance(e.cause, compat_HTTPError) or e.cause.code != 404:
478 raise
479
480 # Player pages may also serve different qualities
481 for page in ('widget/partnerplayer', 'portalplayer'):
482 player = self._download_webpage(
483 'http://player.pbs.org/%s/%s' % (page, video_id),
484 display_id, 'Downloading %s page' % page, fatal=False)
485 if player:
486 video_info = self._parse_json(
487 self._search_regex(
488 r'(?s)PBS\.videoData\s*=\s*({.+?});\n',
489 player, '%s video data' % page, default='{}'),
490 display_id, transform_source=js_to_json, fatal=False)
491 if video_info:
492 extract_redirect_urls(video_info)
493 if not info:
494 info = video_info
495
496 formats = []
497 for num, redirect in enumerate(redirects):
498 redirect_id = redirect.get('eeid')
499
500 redirect_info = self._download_json(
501 '%s?format=json' % redirect['url'], display_id,
502 'Downloading %s video url info' % (redirect_id or num))
503
504 if redirect_info['status'] == 'error':
505 raise ExtractorError(
506 '%s said: %s' % (
507 self.IE_NAME,
508 self._ERRORS.get(redirect_info['http_code'], redirect_info['message'])),
509 expected=True)
510
511 format_url = redirect_info.get('url')
512 if not format_url:
513 continue
514
515 if determine_ext(format_url) == 'm3u8':
516 formats.extend(self._extract_m3u8_formats(
517 format_url, display_id, 'mp4', preference=1, m3u8_id='hls'))
518 else:
519 formats.append({
520 'url': format_url,
521 'format_id': redirect_id,
522 })
523 self._remove_duplicate_formats(formats)
524 self._sort_formats(formats)
525
526 rating_str = info.get('rating')
527 if rating_str is not None:
528 rating_str = rating_str.rpartition('-')[2]
529 age_limit = US_RATINGS.get(rating_str)
530
531 subtitles = {}
532 closed_captions_url = info.get('closed_captions_url')
533 if closed_captions_url:
534 subtitles['en'] = [{
535 'ext': 'ttml',
536 'url': closed_captions_url,
537 }]
538
539 # info['title'] is often incomplete (e.g. 'Full Episode', 'Episode 5', etc)
540 # Try turning it to 'program - title' naming scheme if possible
541 alt_title = info.get('program', {}).get('title')
542 if alt_title:
543 info['title'] = alt_title + ' - ' + re.sub(r'^' + alt_title + '[\s\-:]+', '', info['title'])
544
545 return {
546 'id': video_id,
547 'display_id': display_id,
548 'title': info['title'],
549 'description': info.get('description') or info.get('program', {}).get('description'),
550 'thumbnail': info.get('image_url'),
551 'duration': int_or_none(info.get('duration')),
552 'age_limit': age_limit,
553 'upload_date': upload_date,
554 'formats': formats,
555 'subtitles': subtitles,
556 }