]> Raphaƫl G. Git Repositories - youtubedl/blob - youtube_dl/extractor/pbs.py
Prepare new release.
[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 ..utils import (
8 ExtractorError,
9 determine_ext,
10 int_or_none,
11 js_to_json,
12 strip_jsonp,
13 strip_or_none,
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)?/?(?:$|[?\#]) |
191 # Player
192 (?:video|player)\.pbs\.org/(?:widget/)?partnerplayer/(?P<player_id>[^/]+)/
193 )
194 ''' % '|'.join(list(zip(*_STATIONS))[0])
195
196 _GEO_COUNTRIES = ['US']
197
198 _TESTS = [
199 {
200 'url': 'http://www.pbs.org/tpt/constitution-usa-peter-sagal/watch/a-more-perfect-union/',
201 'md5': '173dc391afd361fa72eab5d3d918968d',
202 'info_dict': {
203 'id': '2365006249',
204 'ext': 'mp4',
205 'title': 'Constitution USA with Peter Sagal - A More Perfect Union',
206 'description': 'md5:31b664af3c65fd07fa460d306b837d00',
207 'duration': 3190,
208 },
209 },
210 {
211 'url': 'http://www.pbs.org/wgbh/pages/frontline/losing-iraq/',
212 'md5': '6f722cb3c3982186d34b0f13374499c7',
213 'info_dict': {
214 'id': '2365297690',
215 'ext': 'mp4',
216 'title': 'FRONTLINE - Losing Iraq',
217 'description': 'md5:5979a4d069b157f622d02bff62fbe654',
218 'duration': 5050,
219 },
220 },
221 {
222 'url': 'http://www.pbs.org/newshour/bb/education-jan-june12-cyberschools_02-23/',
223 'md5': 'b19856d7f5351b17a5ab1dc6a64be633',
224 'info_dict': {
225 'id': '2201174722',
226 'ext': 'mp4',
227 'title': 'PBS NewsHour - Cyber Schools Gain Popularity, but Quality Questions Persist',
228 'description': 'md5:86ab9a3d04458b876147b355788b8781',
229 'duration': 801,
230 },
231 },
232 {
233 'url': 'http://www.pbs.org/wnet/gperf/dudamel-conducts-verdi-requiem-hollywood-bowl-full-episode/3374/',
234 'md5': 'c62859342be2a0358d6c9eb306595978',
235 'info_dict': {
236 'id': '2365297708',
237 'ext': 'mp4',
238 'title': 'Great Performances - Dudamel Conducts Verdi Requiem at the Hollywood Bowl - Full',
239 'description': 'md5:657897370e09e2bc6bf0f8d2cd313c6b',
240 'duration': 6559,
241 'thumbnail': r're:^https?://.*\.jpg$',
242 },
243 },
244 {
245 'url': 'http://www.pbs.org/wgbh/nova/earth/killer-typhoon.html',
246 'md5': '908f3e5473a693b266b84e25e1cf9703',
247 'info_dict': {
248 'id': '2365160389',
249 'display_id': 'killer-typhoon',
250 'ext': 'mp4',
251 'description': 'md5:c741d14e979fc53228c575894094f157',
252 'title': 'NOVA - Killer Typhoon',
253 'duration': 3172,
254 'thumbnail': r're:^https?://.*\.jpg$',
255 'upload_date': '20140122',
256 'age_limit': 10,
257 },
258 },
259 {
260 'url': 'http://www.pbs.org/wgbh/pages/frontline/united-states-of-secrets/',
261 'info_dict': {
262 'id': 'united-states-of-secrets',
263 },
264 'playlist_count': 2,
265 },
266 {
267 'url': 'http://www.pbs.org/wgbh/americanexperience/films/death/player/',
268 'info_dict': {
269 'id': '2276541483',
270 'display_id': 'player',
271 'ext': 'mp4',
272 'title': 'American Experience - Death and the Civil War, Chapter 1',
273 'description': 'md5:67fa89a9402e2ee7d08f53b920674c18',
274 'duration': 682,
275 'thumbnail': r're:^https?://.*\.jpg$',
276 },
277 'params': {
278 'skip_download': True, # requires ffmpeg
279 },
280 },
281 {
282 'url': 'http://www.pbs.org/video/2365245528/',
283 'md5': '115223d41bd55cda8ae5cd5ed4e11497',
284 'info_dict': {
285 'id': '2365245528',
286 'display_id': '2365245528',
287 'ext': 'mp4',
288 'title': 'FRONTLINE - United States of Secrets (Part One)',
289 'description': 'md5:55756bd5c551519cc4b7703e373e217e',
290 'duration': 6851,
291 'thumbnail': r're:^https?://.*\.jpg$',
292 },
293 },
294 {
295 # Video embedded in iframe containing angle brackets as attribute's value (e.g.
296 # "<iframe style='position: absolute;<br />\ntop: 0; left: 0;' ...", see
297 # https://github.com/rg3/youtube-dl/issues/7059)
298 'url': 'http://www.pbs.org/food/features/a-chefs-life-season-3-episode-5-prickly-business/',
299 'md5': '59b0ef5009f9ac8a319cc5efebcd865e',
300 'info_dict': {
301 'id': '2365546844',
302 'display_id': 'a-chefs-life-season-3-episode-5-prickly-business',
303 'ext': 'mp4',
304 'title': "A Chef's Life - Season 3, Ep. 5: Prickly Business",
305 'description': 'md5:c0ff7475a4b70261c7e58f493c2792a5',
306 'duration': 1480,
307 'thumbnail': r're:^https?://.*\.jpg$',
308 },
309 },
310 {
311 # Frontline video embedded via flp2012.js
312 'url': 'http://www.pbs.org/wgbh/pages/frontline/the-atomic-artists',
313 'info_dict': {
314 'id': '2070868960',
315 'display_id': 'the-atomic-artists',
316 'ext': 'mp4',
317 'title': 'FRONTLINE - The Atomic Artists',
318 'description': 'md5:f677e4520cfacb4a5ce1471e31b57800',
319 'duration': 723,
320 'thumbnail': r're:^https?://.*\.jpg$',
321 },
322 'params': {
323 'skip_download': True, # requires ffmpeg
324 },
325 },
326 {
327 # Serves hd only via wigget/partnerplayer page
328 'url': 'http://www.pbs.org/video/2365641075/',
329 'md5': 'fdf907851eab57211dd589cf12006666',
330 'info_dict': {
331 'id': '2365641075',
332 'ext': 'mp4',
333 'title': 'FRONTLINE - Netanyahu at War',
334 'duration': 6852,
335 'thumbnail': r're:^https?://.*\.jpg$',
336 'formats': 'mincount:8',
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 _real_initialize(self):
356 cookie = (self._download_json(
357 'http://localization.services.pbs.org/localize/auto/cookie/',
358 None, headers=self.geo_verification_headers(), fatal=False) or {}).get('cookie')
359 if cookie:
360 station = self._search_regex(r'#?s=\["([^"]+)"', cookie, 'station')
361 if station:
362 self._set_cookie('.pbs.org', 'pbsol.station', station)
363
364 def _extract_webpage(self, url):
365 mobj = re.match(self._VALID_URL, url)
366
367 description = None
368
369 presumptive_id = mobj.group('presumptive_id')
370 display_id = presumptive_id
371 if presumptive_id:
372 webpage = self._download_webpage(url, display_id)
373
374 description = strip_or_none(self._og_search_description(
375 webpage, default=None) or self._html_search_meta(
376 'description', webpage, default=None))
377 upload_date = unified_strdate(self._search_regex(
378 r'<input type="hidden" id="air_date_[0-9]+" value="([^"]+)"',
379 webpage, 'upload date', default=None))
380
381 # tabbed frontline videos
382 MULTI_PART_REGEXES = (
383 r'<div[^>]+class="videotab[^"]*"[^>]+vid="(\d+)"',
384 r'<a[^>]+href=["\']#video-\d+["\'][^>]+data-coveid=["\'](\d+)',
385 )
386 for p in MULTI_PART_REGEXES:
387 tabbed_videos = re.findall(p, webpage)
388 if tabbed_videos:
389 return tabbed_videos, presumptive_id, upload_date, description
390
391 MEDIA_ID_REGEXES = [
392 r"div\s*:\s*'videoembed'\s*,\s*mediaid\s*:\s*'(\d+)'", # frontline video embed
393 r'class="coveplayerid">([^<]+)<', # coveplayer
394 r'<section[^>]+data-coveid="(\d+)"', # coveplayer from http://www.pbs.org/wgbh/frontline/film/real-csi/
395 r'<input type="hidden" id="pbs_video_id_[0-9]+" value="([0-9]+)"/>', # jwplayer
396 ]
397
398 media_id = self._search_regex(
399 MEDIA_ID_REGEXES, webpage, 'media ID', fatal=False, default=None)
400 if media_id:
401 return media_id, presumptive_id, upload_date, description
402
403 # Fronline video embedded via flp
404 video_id = self._search_regex(
405 r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid', default=None)
406 if video_id:
407 # pkg_id calculation is reverse engineered from
408 # http://www.pbs.org/wgbh/pages/frontline/js/flp2012.js
409 prg_id = self._search_regex(
410 r'videoid\s*:\s*"([\d+a-z]{7,})"', webpage, 'videoid')[7:]
411 if 'q' in prg_id:
412 prg_id = prg_id.split('q')[1]
413 prg_id = int(prg_id, 16)
414 getdir = self._download_json(
415 'http://www.pbs.org/wgbh/pages/frontline/.json/getdir/getdir%d.json' % prg_id,
416 presumptive_id, 'Downloading getdir JSON',
417 transform_source=strip_jsonp)
418 return getdir['mid'], presumptive_id, upload_date, description
419
420 for iframe in re.findall(r'(?s)<iframe(.+?)></iframe>', webpage):
421 url = self._search_regex(
422 r'src=(["\'])(?P<url>.+?partnerplayer.+?)\1', iframe,
423 'player URL', default=None, group='url')
424 if url:
425 break
426
427 mobj = re.match(self._VALID_URL, url)
428
429 player_id = mobj.group('player_id')
430 if not display_id:
431 display_id = player_id
432 if player_id:
433 player_page = self._download_webpage(
434 url, display_id, note='Downloading player page',
435 errnote='Could not download player page')
436 video_id = self._search_regex(
437 r'<div\s+id="video_([0-9]+)"', player_page, 'video ID')
438 else:
439 video_id = mobj.group('id')
440 display_id = video_id
441
442 return video_id, display_id, None, description
443
444 def _real_extract(self, url):
445 video_id, display_id, upload_date, description = self._extract_webpage(url)
446
447 if isinstance(video_id, list):
448 entries = [self.url_result(
449 'http://video.pbs.org/video/%s' % vid_id, 'PBS', vid_id)
450 for vid_id in video_id]
451 return self.playlist_result(entries, display_id)
452
453 info = None
454 redirects = []
455 redirect_urls = set()
456
457 def extract_redirect_urls(info):
458 for encoding_name in ('recommended_encoding', 'alternate_encoding'):
459 redirect = info.get(encoding_name)
460 if not redirect:
461 continue
462 redirect_url = redirect.get('url')
463 if redirect_url and redirect_url not in redirect_urls:
464 redirects.append(redirect)
465 redirect_urls.add(redirect_url)
466
467 # Player pages may also serve different qualities
468 for page in ('widget/partnerplayer', 'portalplayer'):
469 player = self._download_webpage(
470 'http://player.pbs.org/%s/%s' % (page, video_id),
471 display_id, 'Downloading %s page' % page, fatal=False)
472 if player:
473 video_info = self._parse_json(
474 self._search_regex(
475 r'(?s)PBS\.videoData\s*=\s*({.+?});\n',
476 player, '%s video data' % page, default='{}'),
477 display_id, transform_source=js_to_json, fatal=False)
478 if video_info:
479 extract_redirect_urls(video_info)
480 if not info:
481 info = video_info
482
483 formats = []
484 http_url = None
485 for num, redirect in enumerate(redirects):
486 redirect_id = redirect.get('eeid')
487
488 redirect_info = self._download_json(
489 '%s?format=json' % redirect['url'], display_id,
490 'Downloading %s video url info' % (redirect_id or num),
491 headers=self.geo_verification_headers())
492
493 if redirect_info['status'] == 'error':
494 message = self._ERRORS.get(
495 redirect_info['http_code'], redirect_info['message'])
496 if redirect_info['http_code'] == 403:
497 self.raise_geo_restricted(
498 msg=message, countries=self._GEO_COUNTRIES)
499 raise ExtractorError(
500 '%s said: %s' % (self.IE_NAME, message), expected=True)
501
502 format_url = redirect_info.get('url')
503 if not format_url:
504 continue
505
506 if determine_ext(format_url) == 'm3u8':
507 formats.extend(self._extract_m3u8_formats(
508 format_url, display_id, 'mp4', m3u8_id='hls', fatal=False))
509 else:
510 formats.append({
511 'url': format_url,
512 'format_id': redirect_id,
513 })
514 if re.search(r'^https?://.*(?:\d+k|baseline)', format_url):
515 http_url = format_url
516 self._remove_duplicate_formats(formats)
517 m3u8_formats = list(filter(
518 lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none' and f.get('resolution') != 'multiple',
519 formats))
520 if http_url:
521 for m3u8_format in m3u8_formats:
522 bitrate = self._search_regex(r'(\d+)k', m3u8_format['url'], 'bitrate', default=None)
523 # Lower qualities (150k and 192k) are not available as HTTP formats (see [1]),
524 # we won't try extracting them.
525 # Since summer 2016 higher quality formats (4500k and 6500k) are also available
526 # albeit they are not documented in [2].
527 # 1. https://github.com/rg3/youtube-dl/commit/cbc032c8b70a038a69259378c92b4ba97b42d491#commitcomment-17313656
528 # 2. https://projects.pbs.org/confluence/display/coveapi/COVE+Video+Specifications
529 if not bitrate or int(bitrate) < 400:
530 continue
531 f_url = re.sub(r'\d+k|baseline', bitrate + 'k', http_url)
532 # This may produce invalid links sometimes (e.g.
533 # http://www.pbs.org/wgbh/frontline/film/suicide-plan)
534 if not self._is_valid_url(f_url, display_id, 'http-%sk video' % bitrate):
535 continue
536 f = m3u8_format.copy()
537 f.update({
538 'url': f_url,
539 'format_id': m3u8_format['format_id'].replace('hls', 'http'),
540 'protocol': 'http',
541 })
542 formats.append(f)
543 self._sort_formats(formats)
544
545 rating_str = info.get('rating')
546 if rating_str is not None:
547 rating_str = rating_str.rpartition('-')[2]
548 age_limit = US_RATINGS.get(rating_str)
549
550 subtitles = {}
551 closed_captions_url = info.get('closed_captions_url')
552 if closed_captions_url:
553 subtitles['en'] = [{
554 'ext': 'ttml',
555 'url': closed_captions_url,
556 }]
557 mobj = re.search(r'/(\d+)_Encoded\.dfxp', closed_captions_url)
558 if mobj:
559 ttml_caption_suffix, ttml_caption_id = mobj.group(0, 1)
560 ttml_caption_id = int(ttml_caption_id)
561 subtitles['en'].extend([{
562 'url': closed_captions_url.replace(
563 ttml_caption_suffix, '/%d_Encoded.srt' % (ttml_caption_id + 1)),
564 'ext': 'srt',
565 }, {
566 'url': closed_captions_url.replace(
567 ttml_caption_suffix, '/%d_Encoded.vtt' % (ttml_caption_id + 2)),
568 'ext': 'vtt',
569 }])
570
571 # info['title'] is often incomplete (e.g. 'Full Episode', 'Episode 5', etc)
572 # Try turning it to 'program - title' naming scheme if possible
573 alt_title = info.get('program', {}).get('title')
574 if alt_title:
575 info['title'] = alt_title + ' - ' + re.sub(r'^' + alt_title + r'[\s\-:]+', '', info['title'])
576
577 description = info.get('description') or info.get(
578 'program', {}).get('description') or description
579
580 return {
581 'id': video_id,
582 'display_id': display_id,
583 'title': info['title'],
584 'description': description,
585 'thumbnail': info.get('image_url'),
586 'duration': int_or_none(info.get('duration')),
587 'age_limit': age_limit,
588 'upload_date': upload_date,
589 'formats': formats,
590 'subtitles': subtitles,
591 }