strip_jsonp,
unescapeHTML,
unified_strdate,
+ url_or_none,
)
'params': {
'skip_download': True, # rtsp downloads
},
- '_skip': 'Blocked outside of Austria / Germany',
+ 'skip': 'Blocked outside of Austria / Germany',
}, {
'url': 'http://tvthek.orf.at/topic/Fluechtlingskrise/10463081/Heimat-Fremde-Heimat/13879132/Senioren-betreuen-Migrantenkinder/13879141',
- 'skip_download': True,
+ 'only_matching': True,
}, {
'url': 'http://tvthek.orf.at/profile/Universum/35429',
- 'skip_download': True,
+ 'only_matching': True,
}]
def _real_extract(self, url):
webpage, 'playlist', group='json'),
playlist_id, transform_source=unescapeHTML)['playlist']['videos']
- def quality_to_int(s):
- m = re.search('([0-9]+)', s)
- if m is None:
- return -1
- return int(m.group(1))
-
entries = []
for sd in data_jsb:
video_id, title = sd.get('id'), sd.get('title')
if not video_id or not title:
continue
video_id = compat_str(video_id)
- formats = [{
- 'preference': -10 if fd['delivery'] == 'hls' else None,
- 'format_id': '%s-%s-%s' % (
- fd['delivery'], fd['quality'], fd['quality_string']),
- 'url': fd['src'],
- 'protocol': fd['protocol'],
- 'quality': quality_to_int(fd['quality']),
- } for fd in sd['sources']]
+ formats = []
+ for fd in sd['sources']:
+ src = url_or_none(fd.get('src'))
+ if not src:
+ continue
+ format_id_list = []
+ for key in ('delivery', 'quality', 'quality_string'):
+ value = fd.get(key)
+ if value:
+ format_id_list.append(value)
+ format_id = '-'.join(format_id_list)
+ if determine_ext(fd['src']) == 'm3u8':
+ formats.extend(self._extract_m3u8_formats(
+ fd['src'], video_id, 'mp4', m3u8_id=format_id))
+ elif determine_ext(fd['src']) == 'f4m':
+ formats.extend(self._extract_f4m_formats(
+ fd['src'], video_id, f4m_id=format_id))
+ else:
+ formats.append({
+ 'format_id': format_id,
+ 'url': src,
+ 'protocol': fd.get('protocol'),
+ })
# Check for geoblocking.
# There is a property is_geoprotection, but that's always false
'description': subtitle,
'duration': (info['end'] - info['start']) / 1000,
'timestamp': info['start'] / 1000,
- 'ext': 'mp3'
+ 'ext': 'mp3',
+ 'series': data.get('programTitle')
}
entries = [extract_entry_dict(t, data['title'], data['subtitle']) for t in data['streams']]