1 from __future__ import unicode_literals
3 import xml.etree.ElementTree
5 from .common import InfoExtractor
10 from ..compat import compat_HTTPError
13 class BBCCoUkIE(InfoExtractor):
15 IE_DESC = 'BBC iPlayer'
16 _VALID_URL = r'https?://(?:www\.)?bbc\.co\.uk/(?:(?:(?:programmes|iplayer(?:/[^/]+)?/(?:episode|playlist))/)|music/clips[/#])(?P<id>[\da-z]{8})'
20 'url': 'http://www.bbc.co.uk/programmes/b039g8p7',
24 'title': 'Kaleidoscope, Leonard Cohen',
25 'description': 'The Canadian poet and songwriter reflects on his musical career.',
30 'skip_download': True,
34 'url': 'http://www.bbc.co.uk/iplayer/episode/b00yng5w/The_Man_in_Black_Series_3_The_Printed_Name/',
38 'title': 'The Man in Black: Series 3: The Printed Name',
39 'description': "Mark Gatiss introduces Nicholas Pierpan's chilling tale of a writer's devilish pact with a mysterious man. Stars Ewan Bailey.",
44 'skip_download': True,
46 'skip': 'Episode is no longer available on BBC iPlayer Radio',
49 'url': 'http://www.bbc.co.uk/iplayer/episode/b03vhd1f/The_Voice_UK_Series_3_Blind_Auditions_5/',
53 'title': 'The Voice UK: Series 3: Blind Auditions 5',
54 '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.",
59 'skip_download': True,
61 'skip': 'Currently BBC iPlayer TV programmes are available to play in the UK only',
64 'url': 'http://www.bbc.co.uk/iplayer/episode/p026c7jt/tomorrows-worlds-the-unearthly-history-of-science-fiction-2-invasion',
68 'title': "Tomorrow's Worlds: The Unearthly History of Science Fiction",
69 'description': '2. Invasion',
74 'skip_download': True,
76 'skip': 'Currently BBC iPlayer TV programmes are available to play in the UK only',
78 'url': 'http://www.bbc.co.uk/programmes/b04v20dw',
82 'title': 'Pete Tong, The Essential New Tune Special',
83 'description': "Pete has a very special mix - all of 2014's Essential New Tunes!",
88 'skip_download': True,
91 'url': 'http://www.bbc.co.uk/music/clips/p02frcc3',
96 'title': 'Pete Tong, Past, Present and Future Special, Madeon - After Hours mix',
97 'description': 'French house superstar Madeon takes us out of the club and onto the after party.',
102 'skip_download': True,
105 'url': 'http://www.bbc.co.uk/music/clips/p025c0zz',
110 'title': 'Reading and Leeds Festival, 2014, Rae Morris - Closer (Live on BBC Three)',
111 'description': 'Rae Morris performs Closer for BBC Three at Reading 2014',
116 'skip_download': True,
119 'url': 'http://www.bbc.co.uk/iplayer/playlist/p01dvks4',
120 'only_matching': True,
122 'url': 'http://www.bbc.co.uk/music/clips#p02frcc3',
123 'only_matching': True,
125 'url': 'http://www.bbc.co.uk/iplayer/cbeebies/episode/b0480276/bing-14-atchoo',
126 'only_matching': True,
130 def _extract_asx_playlist(self, connection, programme_id):
131 asx = self._download_xml(connection.get('href'), programme_id, 'Downloading ASX playlist')
132 return [ref.get('href') for ref in asx.findall('./Entry/ref')]
134 def _extract_connection(self, connection, programme_id):
136 protocol = connection.get('protocol')
137 supplier = connection.get('supplier')
138 if protocol == 'http':
139 href = connection.get('href')
141 if supplier == 'asx':
142 for i, ref in enumerate(self._extract_asx_playlist(connection, programme_id)):
145 'format_id': 'ref%s_%s' % (i, supplier),
151 'format_id': supplier,
153 elif protocol == 'rtmp':
154 application = connection.get('application', 'ondemand')
155 auth_string = connection.get('authString')
156 identifier = connection.get('identifier')
157 server = connection.get('server')
159 'url': '%s://%s/%s?%s' % (protocol, server, application, auth_string),
160 'play_path': identifier,
161 'app': '%s?%s' % (application, auth_string),
162 'page_url': 'http://www.bbc.co.uk',
163 'player_url': 'http://www.bbc.co.uk/emp/releases/iplayer/revisions/617463_618125_4/617463_618125_4_emp.swf',
166 'format_id': supplier,
170 def _extract_items(self, playlist):
171 return playlist.findall('./{http://bbc.co.uk/2008/emp/playlist}item')
173 def _extract_medias(self, media_selection):
174 error = media_selection.find('./{http://bbc.co.uk/2008/mp/mediaselection}error')
175 if error is not None:
176 raise ExtractorError(
177 '%s returned error: %s' % (self.IE_NAME, error.get('id')), expected=True)
178 return media_selection.findall('./{http://bbc.co.uk/2008/mp/mediaselection}media')
180 def _extract_connections(self, media):
181 return media.findall('./{http://bbc.co.uk/2008/mp/mediaselection}connection')
183 def _extract_video(self, media, programme_id):
185 vbr = int(media.get('bitrate'))
186 vcodec = media.get('encoding')
187 service = media.get('service')
188 width = int(media.get('width'))
189 height = int(media.get('height'))
190 file_size = int(media.get('media_file_size'))
191 for connection in self._extract_connections(media):
192 conn_formats = self._extract_connection(connection, programme_id)
193 for format in conn_formats:
195 'format_id': '%s_%s' % (service, format['format_id']),
200 'filesize': file_size,
202 formats.extend(conn_formats)
205 def _extract_audio(self, media, programme_id):
207 abr = int(media.get('bitrate'))
208 acodec = media.get('encoding')
209 service = media.get('service')
210 for connection in self._extract_connections(media):
211 conn_formats = self._extract_connection(connection, programme_id)
212 for format in conn_formats:
214 'format_id': '%s_%s' % (service, format['format_id']),
218 formats.extend(conn_formats)
221 def _get_subtitles(self, media, programme_id):
223 for connection in self._extract_connections(media):
224 captions = self._download_xml(connection.get('href'), programme_id, 'Downloading captions')
225 lang = captions.get('{http://www.w3.org/XML/1998/namespace}lang', 'en')
226 ps = captions.findall('./{0}body/{0}div/{0}p'.format('{http://www.w3.org/2006/10/ttaf1}'))
229 def _extract_text(p):
230 if p.text is not None:
231 stripped_text = p.text.strip()
234 return ' '.join(span.text.strip() for span in p.findall('{http://www.w3.org/2006/10/ttaf1}span'))
235 for pos, p in enumerate(ps):
236 srt += '%s\r\n%s --> %s\r\n%s\r\n\r\n' % (str(pos), p.get('begin'), p.get('end'), _extract_text(p))
239 'url': connection.get('href'),
249 def _download_media_selector(self, programme_id):
251 media_selection = self._download_xml(
252 'http://open.live.bbc.co.uk/mediaselector/5/select/version/2.0/mediaset/pc/vpid/%s' % programme_id,
253 programme_id, 'Downloading media selection XML')
254 except ExtractorError as ee:
255 if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 403:
256 media_selection = xml.etree.ElementTree.fromstring(ee.cause.read().encode('utf-8'))
263 for media in self._extract_medias(media_selection):
264 kind = media.get('kind')
266 formats.extend(self._extract_audio(media, programme_id))
267 elif kind == 'video':
268 formats.extend(self._extract_video(media, programme_id))
269 elif kind == 'captions':
270 subtitles = self.extract_subtitles(media, programme_id)
272 return formats, subtitles
274 def _download_playlist(self, playlist_id):
276 playlist = self._download_json(
277 'http://www.bbc.co.uk/programmes/%s/playlist.json' % playlist_id,
278 playlist_id, 'Downloading playlist JSON')
280 version = playlist.get('defaultAvailableVersion')
282 smp_config = version['smpConfig']
283 title = smp_config['title']
284 description = smp_config['summary']
285 for item in smp_config['items']:
287 if kind != 'programme' and kind != 'radioProgramme':
289 programme_id = item.get('vpid')
290 duration = int(item.get('duration'))
291 formats, subtitles = self._download_media_selector(programme_id)
292 return programme_id, title, description, duration, formats, subtitles
293 except ExtractorError as ee:
294 if not (isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 404):
297 # fallback to legacy playlist
298 playlist = self._download_xml(
299 'http://www.bbc.co.uk/iplayer/playlist/%s' % playlist_id,
300 playlist_id, 'Downloading legacy playlist XML')
302 no_items = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}noItems')
303 if no_items is not None:
304 reason = no_items.get('reason')
305 if reason == 'preAvailability':
306 msg = 'Episode %s is not yet available' % playlist_id
307 elif reason == 'postAvailability':
308 msg = 'Episode %s is no longer available' % playlist_id
309 elif reason == 'noMedia':
310 msg = 'Episode %s is not currently available' % playlist_id
312 msg = 'Episode %s is not available: %s' % (playlist_id, reason)
313 raise ExtractorError(msg, expected=True)
315 for item in self._extract_items(playlist):
316 kind = item.get('kind')
317 if kind != 'programme' and kind != 'radioProgramme':
319 title = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}title').text
320 description = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}summary').text
321 programme_id = item.get('identifier')
322 duration = int(item.get('duration'))
323 formats, subtitles = self._download_media_selector(programme_id)
325 return programme_id, title, description, duration, formats, subtitles
327 def _real_extract(self, url):
328 group_id = self._match_id(url)
330 webpage = self._download_webpage(url, group_id, 'Downloading video page')
332 thumbnail = self._og_search_thumbnail(webpage)
336 tviplayer = self._search_regex(
337 r'mediator\.bind\(({.+?})\s*,\s*document\.getElementById',
338 webpage, 'player', default=None)
341 player = self._parse_json(tviplayer, group_id).get('player', {})
342 duration = int_or_none(player.get('duration'))
343 programme_id = player.get('vpid')
346 programme_id = self._search_regex(
347 r'"vpid"\s*:\s*"([\da-z]{8})"', webpage, 'vpid', fatal=False, default=None)
350 formats, subtitles = self._download_media_selector(programme_id)
351 title = self._og_search_title(webpage)
352 description = self._search_regex(
353 r'<p class="medium-description">([^<]+)</p>',
354 webpage, 'description', fatal=False)
356 programme_id, title, description, duration, formats, subtitles = self._download_playlist(group_id)
358 self._sort_formats(formats)
363 'description': description,
364 'thumbnail': thumbnail,
365 'duration': duration,
367 'subtitles': subtitles,