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))/(?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/iplayer/playlist/p01dvks4',
89 'only_matching': True,
93 def _extract_asx_playlist(self, connection, programme_id):
94 asx = self._download_xml(connection.get('href'), programme_id, 'Downloading ASX playlist')
95 return [ref.get('href') for ref in asx.findall('./Entry/ref')]
97 def _extract_connection(self, connection, programme_id):
99 protocol = connection.get('protocol')
100 supplier = connection.get('supplier')
101 if protocol == 'http':
102 href = connection.get('href')
104 if supplier == 'asx':
105 for i, ref in enumerate(self._extract_asx_playlist(connection, programme_id)):
108 'format_id': 'ref%s_%s' % (i, supplier),
114 'format_id': supplier,
116 elif protocol == 'rtmp':
117 application = connection.get('application', 'ondemand')
118 auth_string = connection.get('authString')
119 identifier = connection.get('identifier')
120 server = connection.get('server')
122 'url': '%s://%s/%s?%s' % (protocol, server, application, auth_string),
123 'play_path': identifier,
124 'app': '%s?%s' % (application, auth_string),
125 'page_url': 'http://www.bbc.co.uk',
126 'player_url': 'http://www.bbc.co.uk/emp/releases/iplayer/revisions/617463_618125_4/617463_618125_4_emp.swf',
129 'format_id': supplier,
133 def _extract_items(self, playlist):
134 return playlist.findall('./{http://bbc.co.uk/2008/emp/playlist}item')
136 def _extract_medias(self, media_selection):
137 error = media_selection.find('./{http://bbc.co.uk/2008/mp/mediaselection}error')
138 if error is not None:
139 raise ExtractorError(
140 '%s returned error: %s' % (self.IE_NAME, error.get('id')), expected=True)
141 return media_selection.findall('./{http://bbc.co.uk/2008/mp/mediaselection}media')
143 def _extract_connections(self, media):
144 return media.findall('./{http://bbc.co.uk/2008/mp/mediaselection}connection')
146 def _extract_video(self, media, programme_id):
148 vbr = int(media.get('bitrate'))
149 vcodec = media.get('encoding')
150 service = media.get('service')
151 width = int(media.get('width'))
152 height = int(media.get('height'))
153 file_size = int(media.get('media_file_size'))
154 for connection in self._extract_connections(media):
155 conn_formats = self._extract_connection(connection, programme_id)
156 for format in conn_formats:
158 'format_id': '%s_%s' % (service, format['format_id']),
163 'filesize': file_size,
165 formats.extend(conn_formats)
168 def _extract_audio(self, media, programme_id):
170 abr = int(media.get('bitrate'))
171 acodec = media.get('encoding')
172 service = media.get('service')
173 for connection in self._extract_connections(media):
174 conn_formats = self._extract_connection(connection, programme_id)
175 for format in conn_formats:
177 'format_id': '%s_%s' % (service, format['format_id']),
181 formats.extend(conn_formats)
184 def _extract_captions(self, media, programme_id):
186 for connection in self._extract_connections(media):
187 captions = self._download_xml(connection.get('href'), programme_id, 'Downloading captions')
188 lang = captions.get('{http://www.w3.org/XML/1998/namespace}lang', 'en')
189 ps = captions.findall('./{0}body/{0}div/{0}p'.format('{http://www.w3.org/2006/10/ttaf1}'))
191 for pos, p in enumerate(ps):
192 srt += '%s\r\n%s --> %s\r\n%s\r\n\r\n' % (str(pos), p.get('begin'), p.get('end'),
193 p.text.strip() if p.text is not None else '')
194 subtitles[lang] = srt
197 def _download_media_selector(self, programme_id):
199 media_selection = self._download_xml(
200 'http://open.live.bbc.co.uk/mediaselector/5/select/version/2.0/mediaset/pc/vpid/%s' % programme_id,
201 programme_id, 'Downloading media selection XML')
202 except ExtractorError as ee:
203 if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 403:
204 media_selection = xml.etree.ElementTree.fromstring(ee.cause.read().encode('utf-8'))
211 for media in self._extract_medias(media_selection):
212 kind = media.get('kind')
214 formats.extend(self._extract_audio(media, programme_id))
215 elif kind == 'video':
216 formats.extend(self._extract_video(media, programme_id))
217 elif kind == 'captions':
218 subtitles = self._extract_captions(media, programme_id)
220 return formats, subtitles
222 def _download_playlist(self, playlist_id):
224 playlist = self._download_json(
225 'http://www.bbc.co.uk/programmes/%s/playlist.json' % playlist_id,
226 playlist_id, 'Downloading playlist JSON')
228 version = playlist.get('defaultAvailableVersion')
230 smp_config = version['smpConfig']
231 title = smp_config['title']
232 description = smp_config['summary']
233 for item in smp_config['items']:
235 if kind != 'programme' and kind != 'radioProgramme':
237 programme_id = item.get('vpid')
238 duration = int(item.get('duration'))
239 formats, subtitles = self._download_media_selector(programme_id)
240 return programme_id, title, description, duration, formats, subtitles
241 except ExtractorError as ee:
242 if not isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 404:
245 # fallback to legacy playlist
246 playlist = self._download_xml(
247 'http://www.bbc.co.uk/iplayer/playlist/%s' % playlist_id,
248 playlist_id, 'Downloading legacy playlist XML')
250 no_items = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}noItems')
251 if no_items is not None:
252 reason = no_items.get('reason')
253 if reason == 'preAvailability':
254 msg = 'Episode %s is not yet available' % playlist_id
255 elif reason == 'postAvailability':
256 msg = 'Episode %s is no longer available' % playlist_id
257 elif reason == 'noMedia':
258 msg = 'Episode %s is not currently available' % playlist_id
260 msg = 'Episode %s is not available: %s' % (playlist_id, reason)
261 raise ExtractorError(msg, expected=True)
263 for item in self._extract_items(playlist):
264 kind = item.get('kind')
265 if kind != 'programme' and kind != 'radioProgramme':
267 title = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}title').text
268 description = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}summary').text
269 programme_id = item.get('identifier')
270 duration = int(item.get('duration'))
271 formats, subtitles = self._download_media_selector(programme_id)
273 return programme_id, title, description, duration, formats, subtitles
275 def _real_extract(self, url):
276 group_id = self._match_id(url)
278 webpage = self._download_webpage(url, group_id, 'Downloading video page')
280 programme_id = self._search_regex(
281 r'"vpid"\s*:\s*"([\da-z]{8})"', webpage, 'vpid', fatal=False, default=None)
283 player = self._download_json(
284 'http://www.bbc.co.uk/iplayer/episode/%s.json' % group_id,
285 group_id)['jsConf']['player']
286 title = player['title']
287 description = player['subtitle']
288 duration = player['duration']
289 formats, subtitles = self._download_media_selector(programme_id)
291 programme_id, title, description, duration, formats, subtitles = self._download_playlist(group_id)
293 if self._downloader.params.get('listsubtitles', False):
294 self._list_available_subtitles(programme_id, subtitles)
297 self._sort_formats(formats)
302 'description': description,
303 'duration': duration,
305 'subtitles': subtitles,