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)/(?P<id>[\da-z]{8})'
17 'url': 'http://www.bbc.co.uk/programmes/b039g8p7',
21 'title': 'Kaleidoscope: Leonard Cohen',
22 'description': 'md5:db4755d7a665ae72343779f7dacb402c',
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,
90 def _extract_asx_playlist(self, connection, programme_id):
91 asx = self._download_xml(connection.get('href'), programme_id, 'Downloading ASX playlist')
92 return [ref.get('href') for ref in asx.findall('./Entry/ref')]
94 def _extract_connection(self, connection, programme_id):
96 protocol = connection.get('protocol')
97 supplier = connection.get('supplier')
98 if protocol == 'http':
99 href = connection.get('href')
101 if supplier == 'asx':
102 for i, ref in enumerate(self._extract_asx_playlist(connection, programme_id)):
105 'format_id': 'ref%s_%s' % (i, supplier),
111 'format_id': supplier,
113 elif protocol == 'rtmp':
114 application = connection.get('application', 'ondemand')
115 auth_string = connection.get('authString')
116 identifier = connection.get('identifier')
117 server = connection.get('server')
119 'url': '%s://%s/%s?%s' % (protocol, server, application, auth_string),
120 'play_path': identifier,
121 'app': '%s?%s' % (application, auth_string),
122 'page_url': 'http://www.bbc.co.uk',
123 'player_url': 'http://www.bbc.co.uk/emp/releases/iplayer/revisions/617463_618125_4/617463_618125_4_emp.swf',
126 'format_id': supplier,
130 def _extract_items(self, playlist):
131 return playlist.findall('./{http://bbc.co.uk/2008/emp/playlist}item')
133 def _extract_medias(self, media_selection):
134 error = media_selection.find('./{http://bbc.co.uk/2008/mp/mediaselection}error')
135 if error is not None:
136 raise ExtractorError(
137 '%s returned error: %s' % (self.IE_NAME, error.get('id')), expected=True)
138 return media_selection.findall('./{http://bbc.co.uk/2008/mp/mediaselection}media')
140 def _extract_connections(self, media):
141 return media.findall('./{http://bbc.co.uk/2008/mp/mediaselection}connection')
143 def _extract_video(self, media, programme_id):
145 vbr = int(media.get('bitrate'))
146 vcodec = media.get('encoding')
147 service = media.get('service')
148 width = int(media.get('width'))
149 height = int(media.get('height'))
150 file_size = int(media.get('media_file_size'))
151 for connection in self._extract_connections(media):
152 conn_formats = self._extract_connection(connection, programme_id)
153 for format in conn_formats:
155 'format_id': '%s_%s' % (service, format['format_id']),
160 'filesize': file_size,
162 formats.extend(conn_formats)
165 def _extract_audio(self, media, programme_id):
167 abr = int(media.get('bitrate'))
168 acodec = media.get('encoding')
169 service = media.get('service')
170 for connection in self._extract_connections(media):
171 conn_formats = self._extract_connection(connection, programme_id)
172 for format in conn_formats:
174 'format_id': '%s_%s' % (service, format['format_id']),
178 formats.extend(conn_formats)
181 def _extract_captions(self, media, programme_id):
183 for connection in self._extract_connections(media):
184 captions = self._download_xml(connection.get('href'), programme_id, 'Downloading captions')
185 lang = captions.get('{http://www.w3.org/XML/1998/namespace}lang', 'en')
186 ps = captions.findall('./{0}body/{0}div/{0}p'.format('{http://www.w3.org/2006/10/ttaf1}'))
188 for pos, p in enumerate(ps):
189 srt += '%s\r\n%s --> %s\r\n%s\r\n\r\n' % (str(pos), p.get('begin'), p.get('end'),
190 p.text.strip() if p.text is not None else '')
191 subtitles[lang] = srt
194 def _download_media_selector(self, programme_id):
196 media_selection = self._download_xml(
197 'http://open.live.bbc.co.uk/mediaselector/5/select/version/2.0/mediaset/pc/vpid/%s' % programme_id,
198 programme_id, 'Downloading media selection XML')
199 except ExtractorError as ee:
200 if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 403:
201 media_selection = xml.etree.ElementTree.fromstring(ee.cause.read().encode('utf-8'))
208 for media in self._extract_medias(media_selection):
209 kind = media.get('kind')
211 formats.extend(self._extract_audio(media, programme_id))
212 elif kind == 'video':
213 formats.extend(self._extract_video(media, programme_id))
214 elif kind == 'captions':
215 subtitles = self._extract_captions(media, programme_id)
217 return formats, subtitles
219 def _download_playlist(self, playlist_id):
221 playlist = self._download_json(
222 'http://www.bbc.co.uk/programmes/%s/playlist.json' % playlist_id,
223 playlist_id, 'Downloading playlist JSON')
225 version = playlist.get('defaultAvailableVersion')
227 smp_config = version['smpConfig']
228 title = smp_config['title']
229 description = smp_config['summary']
230 for item in smp_config['items']:
232 if kind != 'programme' and kind != 'radioProgramme':
234 programme_id = item.get('vpid')
235 duration = int(item.get('duration'))
236 formats, subtitles = self._download_media_selector(programme_id)
237 return programme_id, title, description, duration, formats, subtitles
238 except ExtractorError as ee:
239 if not isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 404:
242 # fallback to legacy playlist
243 playlist = self._download_xml(
244 'http://www.bbc.co.uk/iplayer/playlist/%s' % playlist_id,
245 playlist_id, 'Downloading legacy playlist XML')
247 no_items = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}noItems')
248 if no_items is not None:
249 reason = no_items.get('reason')
250 if reason == 'preAvailability':
251 msg = 'Episode %s is not yet available' % playlist_id
252 elif reason == 'postAvailability':
253 msg = 'Episode %s is no longer available' % playlist_id
254 elif reason == 'noMedia':
255 msg = 'Episode %s is not currently available' % playlist_id
257 msg = 'Episode %s is not available: %s' % (playlist_id, reason)
258 raise ExtractorError(msg, expected=True)
260 for item in self._extract_items(playlist):
261 kind = item.get('kind')
262 if kind != 'programme' and kind != 'radioProgramme':
264 title = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}title').text
265 description = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}summary').text
266 programme_id = item.get('identifier')
267 duration = int(item.get('duration'))
268 formats, subtitles = self._download_media_selector(programme_id)
270 return programme_id, title, description, duration, formats, subtitles
272 def _real_extract(self, url):
273 group_id = self._match_id(url)
275 webpage = self._download_webpage(url, group_id, 'Downloading video page')
277 programme_id = self._search_regex(
278 r'"vpid"\s*:\s*"([\da-z]{8})"', webpage, 'vpid', fatal=False, default=None)
280 player = self._download_json(
281 'http://www.bbc.co.uk/iplayer/episode/%s.json' % group_id,
282 group_id)['jsConf']['player']
283 title = player['title']
284 description = player['subtitle']
285 duration = player['duration']
286 formats, subtitles = self._download_media_selector(programme_id)
288 programme_id, title, description, duration, formats, subtitles = self._download_playlist(group_id)
290 if self._downloader.params.get('listsubtitles', False):
291 self._list_available_subtitles(programme_id, subtitles)
294 self._sort_formats(formats)
299 'description': description,
300 'duration': duration,
302 'subtitles': subtitles,