IE_NAME = 'teachertube'
IE_DESC = 'teachertube.com videos'
- _VALID_URL = r'https?://(?:www\.)?teachertube\.com/(viewVideo\.php\?video_id=|music\.php\?music_id=|video/|audio/)(?P<id>\d+)'
+ _VALID_URL = r'https?://(?:www\.)?teachertube\.com/(viewVideo\.php\?video_id=|music\.php\?music_id=|video/(?:[\da-z-]+-)?|audio/)(?P<id>\d+)'
_TESTS = [{
'url': 'http://www.teachertube.com/viewVideo.php?video_id=339997',
'title': 'PER ASPERA AD ASTRA',
'description': 'RADIJSKA EMISIJA ZRAKOPLOVNE TEHNI?KE ?KOLE P',
},
+ }, {
+ 'url': 'http://www.teachertube.com/video/intro-video-schleicher-297790',
+ 'md5': '9c79fbb2dd7154823996fc28d4a26998',
+ 'info_dict': {
+ 'id': '297790',
+ 'ext': 'mp4',
+ 'title': 'Intro Video - Schleicher',
+ 'description': 'Intro Video - Why to flip, how flipping will',
+ },
}]
def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- video_id = mobj.group('id')
-
+ video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
- title = self._html_search_meta('title', webpage, 'title')
+ title = self._html_search_meta('title', webpage, 'title', fatal=True)
TITLE_SUFFIX = ' - TeacherTube'
if title.endswith(TITLE_SUFFIX):
title = title[:-len(TITLE_SUFFIX)].strip()
_VALID_URL = r'https?://(?:www\.)?teachertube\.com/(user/profile|collection)/(?P<user>[0-9a-zA-Z]+)/?'
- _MEDIA_RE = r'(?s)"sidebar_thumb_time">[0-9:]+</div>.+?<a href="(https?://(?:www\.)?teachertube\.com/(?:video|audio)/[^"]+)">'
+ _MEDIA_RE = r'''(?sx)
+ class="?sidebar_thumb_time"?>[0-9:]+</div>
+ \s*
+ <a\s+href="(https?://(?:www\.)?teachertube\.com/(?:video|audio)/[^"]+)"
+ '''
+ _TEST = {
+ 'url': 'http://www.teachertube.com/user/profile/rbhagwati2',
+ 'info_dict': {
+ 'id': 'rbhagwati2'
+ },
+ 'playlist_mincount': 179,
+ }
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
urls = []
webpage = self._download_webpage(url, user_id)
urls.extend(re.findall(self._MEDIA_RE, webpage))
-
- pages = re.findall(r'/ajax-user/user-videos/%s\?page=([0-9]+)' % user_id, webpage)[1:-1]
+
+ pages = re.findall(r'/ajax-user/user-videos/%s\?page=([0-9]+)' % user_id, webpage)[:-1]
for p in pages:
more = 'http://www.teachertube.com/ajax-user/user-videos/%s?page=%s' % (user_id, p)
- webpage = self._download_webpage(more, user_id, 'Downloading page %s/%s' % (p, len(pages) + 1))
- urls.extend(re.findall(self._MEDIA_RE, webpage))
-
- entries = []
- for url in urls:
- entries.append(self.url_result(url, 'TeacherTube'))
+ webpage = self._download_webpage(more, user_id, 'Downloading page %s/%s' % (p, len(pages)))
+ video_urls = re.findall(self._MEDIA_RE, webpage)
+ urls.extend(video_urls)
+ entries = [self.url_result(vurl, 'TeacherTube') for vurl in urls]
return self.playlist_result(entries, user_id)