1 from __future__ import unicode_literals
5 from .subtitles import SubtitlesInfoExtractor
6 from ..utils import ExtractorError
9 class BBCCoUkIE(SubtitlesInfoExtractor):
11 IE_DESC = 'BBC iPlayer'
12 _VALID_URL = r'https?://(?:www\.)?bbc\.co\.uk/(?:programmes|iplayer/episode)/(?P<id>[\da-z]{8})'
16 'url': 'http://www.bbc.co.uk/programmes/b039g8p7',
20 'title': 'Kaleidoscope: Leonard Cohen',
21 'description': 'md5:db4755d7a665ae72343779f7dacb402c',
26 'skip_download': True,
30 'url': 'http://www.bbc.co.uk/iplayer/episode/b00yng5w/The_Man_in_Black_Series_3_The_Printed_Name/',
34 'title': 'The Man in Black: Series 3: The Printed Name',
35 'description': "Mark Gatiss introduces Nicholas Pierpan's chilling tale of a writer's devilish pact with a mysterious man. Stars Ewan Bailey.",
40 'skip_download': True,
42 'skip': 'Episode is no longer available on BBC iPlayer Radio',
45 'url': 'http://www.bbc.co.uk/iplayer/episode/b03vhd1f/The_Voice_UK_Series_3_Blind_Auditions_5/',
49 'title': 'The Voice UK: Series 3: Blind Auditions 5',
50 '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.",
55 'skip_download': True,
57 'skip': 'Currently BBC iPlayer TV programmes are available to play in the UK only',
61 def _extract_asx_playlist(self, connection, programme_id):
62 asx = self._download_xml(connection.get('href'), programme_id, 'Downloading ASX playlist')
63 return [ref.get('href') for ref in asx.findall('./Entry/ref')]
65 def _extract_connection(self, connection, programme_id):
67 protocol = connection.get('protocol')
68 supplier = connection.get('supplier')
69 if protocol == 'http':
70 href = connection.get('href')
73 for i, ref in enumerate(self._extract_asx_playlist(connection, programme_id)):
76 'format_id': 'ref%s_%s' % (i, supplier),
82 'format_id': supplier,
84 elif protocol == 'rtmp':
85 application = connection.get('application', 'ondemand')
86 auth_string = connection.get('authString')
87 identifier = connection.get('identifier')
88 server = connection.get('server')
90 'url': '%s://%s/%s?%s' % (protocol, server, application, auth_string),
91 'play_path': identifier,
92 'app': '%s?%s' % (application, auth_string),
93 'page_url': 'http://www.bbc.co.uk',
94 'player_url': 'http://www.bbc.co.uk/emp/releases/iplayer/revisions/617463_618125_4/617463_618125_4_emp.swf',
97 'format_id': supplier,
101 def _extract_items(self, playlist):
102 return playlist.findall('./{http://bbc.co.uk/2008/emp/playlist}item')
104 def _extract_medias(self, media_selection):
105 return media_selection.findall('./{http://bbc.co.uk/2008/mp/mediaselection}media')
107 def _extract_connections(self, media):
108 return media.findall('./{http://bbc.co.uk/2008/mp/mediaselection}connection')
110 def _extract_video(self, media, programme_id):
112 vbr = int(media.get('bitrate'))
113 vcodec = media.get('encoding')
114 service = media.get('service')
115 width = int(media.get('width'))
116 height = int(media.get('height'))
117 file_size = int(media.get('media_file_size'))
118 for connection in self._extract_connections(media):
119 conn_formats = self._extract_connection(connection, programme_id)
120 for format in conn_formats:
122 'format_id': '%s_%s' % (service, format['format_id']),
127 'filesize': file_size,
129 formats.extend(conn_formats)
132 def _extract_audio(self, media, programme_id):
134 abr = int(media.get('bitrate'))
135 acodec = media.get('encoding')
136 service = media.get('service')
137 for connection in self._extract_connections(media):
138 conn_formats = self._extract_connection(connection, programme_id)
139 for format in conn_formats:
141 'format_id': '%s_%s' % (service, format['format_id']),
145 formats.extend(conn_formats)
148 def _extract_captions(self, media, programme_id):
150 for connection in self._extract_connections(media):
151 captions = self._download_xml(connection.get('href'), programme_id, 'Downloading captions')
152 lang = captions.get('{http://www.w3.org/XML/1998/namespace}lang', 'en')
153 ps = captions.findall('./{0}body/{0}div/{0}p'.format('{http://www.w3.org/2006/10/ttaf1}'))
155 for pos, p in enumerate(ps):
156 srt += '%s\r\n%s --> %s\r\n%s\r\n\r\n' % (str(pos), p.get('begin'), p.get('end'),
157 p.text.strip() if p.text is not None else '')
158 subtitles[lang] = srt
161 def _real_extract(self, url):
162 mobj = re.match(self._VALID_URL, url)
163 group_id = mobj.group('id')
165 playlist = self._download_xml('http://www.bbc.co.uk/iplayer/playlist/%s' % group_id, group_id,
166 'Downloading playlist XML')
168 no_items = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}noItems')
169 if no_items is not None:
170 reason = no_items.get('reason')
171 if reason == 'preAvailability':
172 msg = 'Episode %s is not yet available' % group_id
173 elif reason == 'postAvailability':
174 msg = 'Episode %s is no longer available' % group_id
176 msg = 'Episode %s is not available: %s' % (group_id, reason)
177 raise ExtractorError(msg, expected=True)
182 for item in self._extract_items(playlist):
183 kind = item.get('kind')
184 if kind != 'programme' and kind != 'radioProgramme':
186 title = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}title').text
187 description = playlist.find('./{http://bbc.co.uk/2008/emp/playlist}summary').text
189 programme_id = item.get('identifier')
190 duration = int(item.get('duration'))
192 media_selection = self._download_xml(
193 'http://open.live.bbc.co.uk/mediaselector/5/select/version/2.0/mediaset/pc/vpid/%s' % programme_id,
194 programme_id, 'Downloading media selection XML')
196 for media in self._extract_medias(media_selection):
197 kind = media.get('kind')
199 formats.extend(self._extract_audio(media, programme_id))
200 elif kind == 'video':
201 formats.extend(self._extract_video(media, programme_id))
202 elif kind == 'captions':
203 subtitles = self._extract_captions(media, programme_id)
205 if self._downloader.params.get('listsubtitles', False):
206 self._list_available_subtitles(programme_id, subtitles)
209 self._sort_formats(formats)
214 'description': description,
215 'duration': duration,
217 'subtitles': subtitles,