1 from __future__ import unicode_literals
3 import xml.etree.ElementTree
5 from .subtitles import SubtitlesInfoExtractor
6 from ..utils import ExtractorError
7 from ..compat import compat_HTTPError
10 class BBCCoUkIE(SubtitlesInfoExtractor):
12 IE_DESC = 'BBC iPlayer'
13 _VALID_URL = r'https?://(?:www\.)?bbc\.co\.uk/(?:(?:(?:programmes|iplayer/(?:episode|playlist))/)|music/clips[/#])(?P<id>[\da-z]{8})'
17 'url': 'http://www.bbc.co.uk/programmes/b039g8p7',
21 'title': 'Kaleidoscope, Leonard Cohen',
22 'description': 'The Canadian poet and songwriter reflects on his musical career.',
27 'skip_download': True,
31 'url': 'http://www.bbc.co.uk/iplayer/episode/b00yng5w/The_Man_in_Black_Series_3_The_Printed_Name/',
35 'title': 'The Man in Black: Series 3: The Printed Name',
36 'description': "Mark Gatiss introduces Nicholas Pierpan's chilling tale of a writer's devilish pact with a mysterious man. Stars Ewan Bailey.",
41 'skip_download': True,
43 'skip': 'Episode is no longer available on BBC iPlayer Radio',
46 'url': 'http://www.bbc.co.uk/iplayer/episode/b03vhd1f/The_Voice_UK_Series_3_Blind_Auditions_5/',
50 'title': 'The Voice UK: Series 3: Blind Auditions 5',
51 'description': "Emma Willis and Marvin Humes present the fifth set of blind auditions in the singing competition, as the coaches continue to build their teams based on voice alone.",
56 'skip_download': True,
58 'skip': 'Currently BBC iPlayer TV programmes are available to play in the UK only',
61 'url': 'http://www.bbc.co.uk/iplayer/episode/p026c7jt/tomorrows-worlds-the-unearthly-history-of-science-fiction-2-invasion',
65 'title': "Tomorrow's Worlds: The Unearthly History of Science Fiction",
66 'description': '2. Invasion',
71 'skip_download': True,
73 'skip': 'Currently BBC iPlayer TV programmes are available to play in the UK only',
75 'url': 'http://www.bbc.co.uk/programmes/b04v20dw',
79 'title': 'Pete Tong, The Essential New Tune Special',
80 'description': "Pete has a very special mix - all of 2014's Essential New Tunes!",
85 'skip_download': True,
88 'url': 'http://www.bbc.co.uk/music/clips/p02frcc3',
93 'title': 'Pete Tong, Past, Present and Future Special, Madeon - After Hours mix',
94 'description': 'French house superstar Madeon takes us out of the club and onto the after party.',
99 'skip_download': True,
102 'url': 'http://www.bbc.co.uk/music/clips/p025c0zz',
107 'title': 'Reading and Leeds Festival, 2014, Rae Morris - Closer (Live on BBC Three)',
108 'description': 'Rae Morris performs Closer for BBC Three at Reading 2014',
113 'skip_download': True,
116 'url': 'http://www.bbc.co.uk/iplayer/playlist/p01dvks4',
117 'only_matching': True,
119 'url': 'http://www.bbc.co.uk/music/clips#p02frcc3',
120 'only_matching': True,
124 def _extract_asx_playlist(self, connection, programme_id):
125 asx = self._download_xml(connection.get('href'), programme_id, 'Downloading ASX playlist')
126 return [ref.get('href') for ref in asx.findall('./Entry/ref')]
128 def _extract_connection(self, connection, programme_id):
130 protocol = connection.get('protocol')
131 supplier = connection.get('supplier')
132 if protocol == 'http':
133 href = connection.get('href')
135 if supplier == 'asx':
136 for i, ref in enumerate(self._extract_asx_playlist(connection, programme_id)):
139 'format_id': 'ref%s_%s' % (i, supplier),
145 'format_id': supplier,
147 elif protocol == 'rtmp':
148 application = connection.get('application', 'ondemand')
149 auth_string = connection.get('authString')
150 identifier = connection.get('identifier')
151 server = connection.get('server')
153 'url': '%s://%s/%s?%s' % (protocol, server, application, auth_string),
154 'play_path': identifier,
155 'app': '%s?%s' % (application, auth_string),
156 'page_url': 'http://www.bbc.co.uk',
157 'player_url': 'http://www.bbc.co.uk/emp/releases/iplayer/revisions/617463_618125_4/617463_618125_4_emp.swf',
160 'format_id': supplier,
164 def _extract_items(self, playlist):
165 return playlist.findall('./{http://bbc.co.uk/2008/emp/playlist}item')
167 def _extract_medias(self, media_selection):
168 error = media_selection.find('./{http://bbc.co.uk/2008/mp/mediaselection}error')
169 if error is not None:
170 raise ExtractorError(
171 '%s returned error: %s' % (self.IE_NAME, error.get('id')), expected=True)
172 return media_selection.findall('./{http://bbc.co.uk/2008/mp/mediaselection}media')
174 def _extract_connections(self, media):
175 return media.findall('./{http://bbc.co.uk/2008/mp/mediaselection}connection')
177 def _extract_video(self, media, programme_id):
179 vbr = int(media.get('bitrate'))
180 vcodec = media.get('encoding')
181 service = media.get('service')
182 width = int(media.get('width'))
183 height = int(media.get('height'))
184 file_size = int(media.get('media_file_size'))
185 for connection in self._extract_connections(media):
186 conn_formats = self._extract_connection(connection, programme_id)
187 for format in conn_formats:
189 'format_id': '%s_%s' % (service, format['format_id']),
194 'filesize': file_size,
196 formats.extend(conn_formats)
199 def _extract_audio(self, media, programme_id):
201 abr = int(media.get('bitrate'))
202 acodec = media.get('encoding')
203 service = media.get('service')
204 for connection in self._extract_connections(media):
205 conn_formats = self._extract_connection(connection, programme_id)
206 for format in conn_formats:
208 'format_id': '%s_%s' % (service, format['format_id']),
212 formats.extend(conn_formats)
215 def _extract_captions(self, media, programme_id):
217 for connection in self._extract_connections(media):
218 captions = self._download_xml(connection.get('href'), programme_id, 'Downloading captions')
219 lang = captions.get('{http://www.w3.org/XML/1998/namespace}lang', 'en')
220 ps = captions.findall('./{0}body/{0}div/{0}p'.format('{http://www.w3.org/2006/10/ttaf1}'))
222 for pos, p in enumerate(ps):
223 srt += '%s\r\n%s --> %s\r\n%s\r\n\r\n' % (str(pos), p.get('begin'), p.get('end'),
224 p.text.strip() if p.text is not None else '')
225 subtitles[lang] = srt
228 def _download_media_selector(self, programme_id):
230 media_selection = self._download_xml(
231 'http://open.live.bbc.co.uk/mediaselector/5/select/version/2.0/mediaset/pc/vpid/%s' % programme_id,
232 programme_id, 'Downloading media selection XML')
233 except ExtractorError as ee:
234 if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 403:
235 media_selection = xml.etree.ElementTree.fromstring(ee.cause.read().encode('utf-8'))
242 for media in self._extract_medias(media_selection):
243 kind = media.get('kind')
245 formats.extend(self._extract_audio(media, programme_id))
246 elif kind == 'video':
247 formats.extend(self._extract_video(media, programme_id))
248 elif kind == 'captions':
249 subtitles = self._extract_captions(media, programme_id)
251 return formats, subtitles
253 def _download_playlist(self, playlist_id):
255 playlist = self._download_json(
256 'http://www.bbc.co.uk/programmes/%s/playlist.json' % playlist_id,
257 playlist_id, 'Downloading playlist JSON')
259 version = playlist.get('defaultAvailableVersion')
261 smp_config = version['smpConfig']
262 title = smp_config['title']
263 description = smp_config['summary']
264 for item in smp_config['items']:
266 if kind != 'programme' and kind != 'radioProgramme':
268 programme_id = item.get('vpid')
269 duration = int(item.get('duration'))
270 formats, subtitles = self._download_media_selector(programme_id)
271 return programme_id, title, description, duration, formats, subtitles
272 except ExtractorError as ee:
273 if not isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 404:
276 # fallback to legacy playlist
277 playlist = self._download_xml(
278 'http://www.bbc.co.uk/iplayer/playlist/%s' % playlist_id,
279 playlist_id, 'Downloading legacy playlist XML')
281 no_items = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}noItems')
282 if no_items is not None:
283 reason = no_items.get('reason')
284 if reason == 'preAvailability':
285 msg = 'Episode %s is not yet available' % playlist_id
286 elif reason == 'postAvailability':
287 msg = 'Episode %s is no longer available' % playlist_id
288 elif reason == 'noMedia':
289 msg = 'Episode %s is not currently available' % playlist_id
291 msg = 'Episode %s is not available: %s' % (playlist_id, reason)
292 raise ExtractorError(msg, expected=True)
294 for item in self._extract_items(playlist):
295 kind = item.get('kind')
296 if kind != 'programme' and kind != 'radioProgramme':
298 title = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}title').text
299 description = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}summary').text
300 programme_id = item.get('identifier')
301 duration = int(item.get('duration'))
302 formats, subtitles = self._download_media_selector(programme_id)
304 return programme_id, title, description, duration, formats, subtitles
306 def _real_extract(self, url):
307 group_id = self._match_id(url)
309 webpage = self._download_webpage(url, group_id, 'Downloading video page')
311 programme_id = self._search_regex(
312 r'"vpid"\s*:\s*"([\da-z]{8})"', webpage, 'vpid', fatal=False, default=None)
314 player = self._download_json(
315 'http://www.bbc.co.uk/iplayer/episode/%s.json' % group_id,
316 group_id)['jsConf']['player']
317 title = player['title']
318 description = player['subtitle']
319 duration = player['duration']
320 formats, subtitles = self._download_media_selector(programme_id)
322 programme_id, title, description, duration, formats, subtitles = self._download_playlist(group_id)
324 if self._downloader.params.get('listsubtitles', False):
325 self._list_available_subtitles(programme_id, subtitles)
328 self._sort_formats(formats)
333 'description': description,
334 'duration': duration,
336 'subtitles': subtitles,