# encoding: utf-8
from __future__ import unicode_literals
-import datetime
-import re
-
from .common import InfoExtractor
+from ..utils import int_or_none
class AftonbladetIE(InfoExtractor):
- _VALID_URL = r'^http://tv\.aftonbladet\.se/webbtv.+(?P<video_id>article\d+)\.ab$'
+ _VALID_URL = r'https?://tv\.aftonbladet\.se/abtv/articles/(?P<id>[0-9]+)'
_TEST = {
- 'url': 'http://tv.aftonbladet.se/webbtv/nyheter/vetenskap/rymden/article36015.ab',
+ 'url': 'http://tv.aftonbladet.se/abtv/articles/36015',
'info_dict': {
- 'id': 'article36015',
+ 'id': '36015',
'ext': 'mp4',
'title': 'Vulkanutbrott i rymden - nu släpper NASA bilderna',
'description': 'Jupiters måne mest aktiv av alla himlakroppar',
+ 'timestamp': 1394142732,
'upload_date': '20140306',
},
}
def _real_extract(self, url):
- mobj = re.search(self._VALID_URL, url)
-
- video_id = mobj.group('video_id')
+ video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
# find internal video meta data
- META_URL = 'http://aftonbladet-play.drlib.aptoma.no/video/%s.json'
- internal_meta_id = self._html_search_regex(r'data-aptomaId="([\w\d]+)"', webpage, 'internal_meta_id')
- internal_meta_url = META_URL % internal_meta_id
- internal_meta_json = self._download_json(internal_meta_url, video_id, 'Downloading video meta data')
+ meta_url = 'http://aftonbladet-play-metadata.cdn.drvideo.aptoma.no/video/%s.json'
+ player_config = self._parse_json(self._html_search_regex(
+ r'data-player-config="([^"]+)"', webpage, 'player config'), video_id)
+ internal_meta_id = player_config['aptomaVideoId']
+ internal_meta_url = meta_url % internal_meta_id
+ internal_meta_json = self._download_json(
+ internal_meta_url, video_id, 'Downloading video meta data')
# find internal video formats
- FORMATS_URL = 'http://aftonbladet-play.videodata.drvideo.aptoma.no/actions/video/?id=%s'
+ format_url = 'http://aftonbladet-play.videodata.drvideo.aptoma.no/actions/video/?id=%s'
internal_video_id = internal_meta_json['videoId']
- internal_formats_url = FORMATS_URL % internal_video_id
- internal_formats_json = self._download_json(internal_formats_url, video_id, 'Downloading video formats')
+ internal_formats_url = format_url % internal_video_id
+ internal_formats_json = self._download_json(
+ internal_formats_url, video_id, 'Downloading video formats')
- self.report_extraction(video_id)
formats = []
- for fmt in reversed(internal_formats_json['formats']['http']['pseudostreaming']['mp4']):
+ for fmt in internal_formats_json['formats']['http']['pseudostreaming']['mp4']:
p = fmt['paths'][0]
formats.append({
'url': 'http://%s:%d/%s/%s' % (p['address'], p['port'], p['path'], p['filename']),
'ext': 'mp4',
- 'width': fmt['width'],
- 'height': fmt['height'],
- 'tbr': fmt['bitrate'],
+ 'width': int_or_none(fmt.get('width')),
+ 'height': int_or_none(fmt.get('height')),
+ 'tbr': int_or_none(fmt.get('bitrate')),
'protocol': 'http',
})
+ self._sort_formats(formats)
- timestamp = datetime.datetime.fromtimestamp(internal_meta_json['timePublished'])
- upload_date = timestamp.strftime('%Y%m%d')
-
- return [{
+ return {
'id': video_id,
'title': internal_meta_json['title'],
'formats': formats,
- 'thumbnail': internal_meta_json['imageUrl'],
- 'description': internal_meta_json['shortPreamble'],
- 'upload_date': upload_date,
- 'duration': internal_meta_json['duration'],
- 'view_count': internal_meta_json['views'],
- }]
+ 'thumbnail': internal_meta_json.get('imageUrl'),
+ 'description': internal_meta_json.get('shortPreamble'),
+ 'timestamp': int_or_none(internal_meta_json.get('timePublished')),
+ 'duration': int_or_none(internal_meta_json.get('duration')),
+ 'view_count': int_or_none(internal_meta_json.get('views')),
+ }