2 from __future__ import unicode_literals
4 from .common import InfoExtractor
5 from ..utils import int_or_none
8 class SverigesRadioBaseIE(InfoExtractor):
9 _BASE_URL = 'https://sverigesradio.se/sida/playerajax'
10 _QUALITIES = ['high', 'medium', 'low']
12 5: {'acodec': 'mp3', 'abr': 128},
13 11: {'acodec': 'aac', 'abr': 192},
14 12: {'acodec': 'aac', 'abr': 32},
15 13: {'acodec': 'aac', 'abr': 96},
18 def _extract_formats(self, query, audio_id, audio_type):
20 for quality in self._QUALITIES:
21 audiourl = self._download_json(
22 self._BASE_URL + '/getaudiourl', audio_id,
24 query=dict(query, type=audio_type, quality=quality, format='iis'))
28 # for some reason url can be empty, skip if so
29 # also skip if url has already been seen (quality parameter is ignored?)
30 url = audiourl.get('audioUrl')
31 if url is None or url == "" or url in audiourls:
34 audioformat = {'vcodec': 'none', 'url': url}
35 # add codec and bitrate if known coding format
36 codingformat = audiourl.get('codingFormat')
38 audioformat.update(self._CODING_FORMATS.get(codingformat, {}))
40 audiourls[url] = audioformat
42 return audiourls.values()
44 def _extract_audio(self, audio_type, url):
45 audio_id = self._match_id(url)
46 query = {'id': audio_id, 'type': audio_type}
48 metadata = self._download_json(self._BASE_URL + '/audiometadata', audio_id, query=query)
49 item = metadata['items'][0]
51 formats = self._extract_formats(query, audio_id, audio_type)
52 self._sort_formats(formats)
56 'title': item['subtitle'],
58 'series': item.get('title'),
59 'duration': int_or_none(item.get('duration')),
60 'thumbnail': item.get('displayimageurl'),
61 'description': item.get('description'),
65 class SverigesRadioPublicationIE(SverigesRadioBaseIE):
66 _VALID_URL = r'https?://(?:www\.)?sverigesradio\.se/sida/(?:artikel|gruppsida)\.aspx\?.*artikel=(?P<id>[0-9]+)'
68 'url': 'https://sverigesradio.se/sida/artikel.aspx?programid=83&artikel=7038546',
69 'md5': '6a4917e1923fccb080e5a206a5afa542',
74 'series': 'Nyheter (Ekot)',
75 'title': 'Esa Teittinen: Sanningen har inte kommit fram',
76 'description': 'md5:daf7ce66a8f0a53d5465a5984d3839df',
77 'thumbnail': 're:^https://static-cdn.sr.se/sida/images/',
80 'url': 'https://sverigesradio.se/sida/gruppsida.aspx?programid=3304&grupp=6247&artikel=7146887',
81 'only_matching': True,
84 def _real_extract(self, url):
85 return self._extract_audio('publication', url)
88 class SverigesRadioEpisodeIE(SverigesRadioBaseIE):
89 _VALID_URL = r'https?://(?:www\.)?sverigesradio\.se/(?:sida/)?avsnitt/(?P<id>[0-9]+)'
91 'url': 'https://sverigesradio.se/avsnitt/1140922?programid=1300',
92 'md5': '20dc4d8db24228f846be390b0c59a07c',
98 'title': 'Metoo och valen',
99 'description': 'md5:fcb5c1f667f00badcc702b196f10a27e',
100 'thumbnail': 're:^https://static-cdn.sr.se/sida/images/'
104 def _real_extract(self, url):
105 return self._extract_audio('episode', url)