1 from __future__ import unicode_literals
5 from .common import InfoExtractor
12 class NPOIE(InfoExtractor):
14 _VALID_URL = r'https?://www\.npo\.nl/[^/]+/[^/]+/(?P<id>[^/?]+)'
17 'url': 'http://www.npo.nl/nieuwsuur/22-06-2014/VPWON_1220719',
18 'md5': '4b3f9c429157ec4775f2c9cb7b911016',
20 'id': 'VPWON_1220719',
23 'description': 'Dagelijks tussen tien en elf: nieuws, sport en achtergronden.',
24 'upload_date': '20140622',
28 def _real_extract(self, url):
29 mobj = re.match(self._VALID_URL, url)
30 video_id = mobj.group('id')
32 metadata = self._download_json(
33 'http://e.omroep.nl/metadata/aflevering/%s' % video_id,
35 # We have to remove the javascript callback
36 transform_source=lambda j: re.sub(r'parseMetadata\((.*?)\);\n//.*$', r'\1', j)
38 token_page = self._download_webpage(
39 'http://ida.omroep.nl/npoplayer/i.js',
41 note='Downloading token'
43 token = self._search_regex(r'npoplayer\.token = "(.+?)"', token_page, 'token')
46 quality = qualities(['adaptive', 'h264_sb', 'h264_bb', 'h264_std'])
47 for format_id in metadata['pubopties']:
48 streams_info = self._download_json(
49 'http://ida.omroep.nl/odi/?prid=%s&puboptions=%s&adaptive=yes&token=%s' % (video_id, format_id, token),
50 video_id, 'Downloading %s streams info' % format_id)
51 stream_info = self._download_json(
52 streams_info['streams'][0] + '&type=json',
53 video_id, 'Downloading %s stream info' % format_id)
54 if format_id == 'adaptive':
55 formats.extend(self._extract_m3u8_formats(stream_info['url'], video_id))
58 'url': stream_info['url'],
59 'format_id': format_id,
60 'quality': quality(format_id),
62 self._sort_formats(formats)
66 'title': metadata['titel'],
67 'description': metadata['info'],
68 'thumbnail': metadata['images'][-1]['url'],
69 'upload_date': unified_strdate(metadata['gidsdatum']),