import re
from .common import InfoExtractor
-from ..utils import determine_ext
+from ..utils import (
+ determine_ext,
+ parse_duration,
+ int_or_none,
+)
class Lecture2GoIE(InfoExtractor):
_VALID_URL = r'https?://lecture2go\.uni-hamburg\.de/veranstaltungen/-/v/(?P<id>\d+)'
_TEST = {
'url': 'https://lecture2go.uni-hamburg.de/veranstaltungen/-/v/17473',
- 'md5': 'a9e76f83b3ef58019c4b7dbc35f406c1',
+ 'md5': 'ac02b570883020d208d405d5a3fd2f7f',
'info_dict': {
'id': '17473',
- 'ext': 'mp4',
- 'url': 'https://fms1.rrz.uni-hamburg.de/abo/64.050_FrankHeitmann_2015-04-13_14-35.mp4',
- 'title': '2 - Endliche Automaten und reguläre Sprachen'
+ 'ext': 'flv',
+ 'title': '2 - Endliche Automaten und reguläre Sprachen',
+ 'creator': 'Frank Heitmann',
+ 'duration': 5220,
}
}
self._sort_formats(formats)
- creator = self._html_search_regex(r'<div[^>]+id="description">([^<]+)</div>', webpage, 'creator')
+ creator = self._html_search_regex(
+ r'<div[^>]+id="description">([^<]+)</div>', webpage, 'creator', fatal=False)
+ duration = parse_duration(self._html_search_regex(
+ r'Duration:\s*</em>\s*<em[^>]*>([^<]+)</em>', webpage, 'duration', fatal=False))
+ view_count = int_or_none(self._html_search_regex(
+ r'Views:\s*</em>\s*<em[^>]+>(\d+)</em>', webpage, 'view count', fatal=False))
return {
'id': video_id,
'title': title,
'formats': formats,
- 'creator': creator
+ 'creator': creator,
+ 'duration': duration,
+ 'view_count': view_count,
}