1 from __future__ import unicode_literals
5 from .common import InfoExtractor
17 class NPOIE(InfoExtractor):
19 _VALID_URL = r'https?://www\.npo\.nl/[^/]+/[^/]+/(?P<id>[^/?]+)'
23 'url': 'http://www.npo.nl/nieuwsuur/22-06-2014/VPWON_1220719',
24 'md5': '4b3f9c429157ec4775f2c9cb7b911016',
26 'id': 'VPWON_1220719',
29 'description': 'Dagelijks tussen tien en elf: nieuws, sport en achtergronden.',
30 'upload_date': '20140622',
34 'url': 'http://www.npo.nl/de-mega-mike-mega-thomas-show/27-02-2009/VARA_101191800',
35 'md5': 'da50a5787dbfc1603c4ad80f31c5120b',
37 'id': 'VARA_101191800',
39 'title': 'De Mega Mike & Mega Thomas show',
40 'description': 'md5:3b74c97fc9d6901d5a665aac0e5400f4',
41 'upload_date': '20090227',
46 'url': 'http://www.npo.nl/tegenlicht/25-02-2013/VPWON_1169289',
47 'md5': 'f8065e4e5a7824068ed3c7e783178f2c',
49 'id': 'VPWON_1169289',
51 'title': 'Tegenlicht',
52 'description': 'md5:d6476bceb17a8c103c76c3b708f05dd1',
53 'upload_date': '20130225',
58 'url': 'http://www.npo.nl/de-nieuwe-mens-deel-1/21-07-2010/WO_VPRO_043706',
60 'id': 'WO_VPRO_043706',
62 'title': 'De nieuwe mens - Deel 1',
63 'description': 'md5:518ae51ba1293ffb80d8d8ce90b74e4b',
67 # mplayer mms download
68 'skip_download': True,
73 def _real_extract(self, url):
74 mobj = re.match(self._VALID_URL, url)
75 video_id = mobj.group('id')
76 return self._get_info(video_id)
78 def _get_info(self, video_id):
79 metadata = self._download_json(
80 'http://e.omroep.nl/metadata/aflevering/%s' % video_id,
82 # We have to remove the javascript callback
83 transform_source=strip_jsonp,
85 token_page = self._download_webpage(
86 'http://ida.omroep.nl/npoplayer/i.js',
88 note='Downloading token'
90 token = self._search_regex(r'npoplayer\.token = "(.+?)"', token_page, 'token')
94 pubopties = metadata.get('pubopties')
96 quality = qualities(['adaptive', 'wmv_sb', 'h264_sb', 'wmv_bb', 'h264_bb', 'wvc1_std', 'h264_std'])
97 for format_id in pubopties:
98 format_info = self._download_json(
99 'http://ida.omroep.nl/odi/?prid=%s&puboptions=%s&adaptive=yes&token=%s'
100 % (video_id, format_id, token),
101 video_id, 'Downloading %s JSON' % format_id)
102 if format_info.get('error_code', 0) or format_info.get('errorcode', 0):
104 streams = format_info.get('streams')
106 video_info = self._download_json(
107 streams[0] + '&type=json',
108 video_id, 'Downloading %s stream JSON' % format_id)
110 video_info = format_info
111 video_url = video_info.get('url')
114 if format_id == 'adaptive':
115 formats.extend(self._extract_m3u8_formats(video_url, video_id))
119 'format_id': format_id,
120 'quality': quality(format_id),
123 streams = metadata.get('streams')
125 for i, stream in enumerate(streams):
126 stream_url = stream.get('url')
129 if determine_ext(stream_url).lower() != 'asf':
132 'quality': stream.get('kwaliteit'),
135 asx = self._download_xml(
136 stream_url, video_id,
137 'Downloading stream %d ASX playlist' % i,
138 transform_source=fix_xml_ampersands)
139 ref = asx.find('./ENTRY/Ref')
142 video_url = ref.get('href')
147 'ext': stream.get('formaat', 'asf'),
148 'quality': stream.get('kwaliteit'),
151 self._sort_formats(formats)
155 'title': metadata['titel'],
156 'description': metadata['info'],
157 'thumbnail': metadata.get('images', [{'url': None}])[-1]['url'],
158 'upload_date': unified_strdate(metadata.get('gidsdatum')),
159 'duration': parse_duration(metadata.get('tijdsduur')),
164 class TegenlichtVproIE(NPOIE):
165 IE_NAME = 'tegenlicht.vpro.nl'
166 _VALID_URL = r'https?://tegenlicht\.vpro\.nl/afleveringen/.*?'
170 'url': 'http://tegenlicht.vpro.nl/afleveringen/2012-2013/de-toekomst-komt-uit-afrika.html',
171 'md5': 'f8065e4e5a7824068ed3c7e783178f2c',
173 'id': 'VPWON_1169289',
175 'title': 'Tegenlicht',
176 'description': 'md5:d6476bceb17a8c103c76c3b708f05dd1',
177 'upload_date': '20130225',
182 def _real_extract(self, url):
183 name = url_basename(url)
184 webpage = self._download_webpage(url, name)
185 urn = self._html_search_meta('mediaurn', webpage)
186 info_page = self._download_json(
187 'http://rs.vpro.nl/v2/api/media/%s.json' % urn, name)
188 return self._get_info(info_page['mid'])