class TagesschauIE(InfoExtractor):
- _VALID_URL = r'https?://(?:www\.)?tagesschau\.de/multimedia/(?:sendung/(ts|tsg|tt|nm)|video/video|tsvorzwanzig)(?P<id>-?[0-9]+)\.html'
+ _VALID_URL = r'https?://(?:www\.)?tagesschau\.de/multimedia/(?:sendung/(ts|tsg|tt|nm|bab/bab)|video/video|tsvorzwanzig)(?P<id>-?[0-9]+)(?:~[-_a-zA-Z0-9]*)?\.html'
_TESTS = [{
- 'url': 'http://www.tagesschau.de/multimedia/video/video1399128.html',
- 'md5': 'bcdeac2194fb296d599ce7929dfa4009',
+ 'url': 'http://www.tagesschau.de/multimedia/video/video-102143.html',
+ 'md5': '917a228bc7df7850783bc47979673a09',
'info_dict': {
- 'id': '1399128',
+ 'id': '102143',
'ext': 'mp4',
- 'title': 'Harald Range, Generalbundesanwalt, zu den Ermittlungen',
- 'description': 'md5:69da3c61275b426426d711bde96463ab',
+ 'title': 'Regierungsumbildung in Athen: Neue Minister in Griechenland vereidigt',
+ 'description': 'md5:171feccd9d9b3dd54d05d501568f6359',
'thumbnail': 're:^http:.*\.jpg$',
},
}, {
'info_dict': {
'id': '3771',
'ext': 'mp4',
- 'description': '',
+ 'description': None,
'title': 'Sendung: tagesschau (mit Gebärdensprache) \t14.07.2015 20:00 Uhr',
'thumbnail': 're:^http:.*\.jpg$',
}
'info_dict': {
'id': '959',
'ext': 'mp4',
- 'description': '',
+ 'description': None,
'title': 'Sendung: tagesschau vor 20 Jahren \t14.07.2015 22:45 Uhr',
'thumbnail': 're:^http:.*\.jpg$',
}
+ }, {
+ 'url': 'http://www.tagesschau.de/multimedia/sendung/bab/bab-3299~_bab-sendung-209.html',
+ 'md5': '42e3757018d9908581481a80cc1806da',
+ 'info_dict': {
+ 'id': '3299',
+ 'ext': 'mp4',
+ 'description': None,
+ 'title': 'Nach dem Referendum: Schaltgespräch nach Athen',
+ 'thumbnail': 're:^http:.*\.jpg$',
+ }
}]
_FORMATS = {
webpage, 'thumbnail', fatal=False)
# there are some videos without description
description = ""
- try:
- description = self._html_search_regex(
- r'(?s)<p class="teasertext">(.*?)</p>',
- webpage, 'description', fatal=False)
- except ExtractorError:
- pass
+ description = self._html_search_regex(
+ r'(?s)<p class="teasertext">(.*?)</p>',
+ webpage, 'description', fatal=False, default=None)
title = self._html_search_regex(
r'<span class="headline".*?>(.*?)</span>', webpage, 'title')