Merge remote-tracking branch 'Boris-de/wdrmaus_fix#8562'
[youtube-dl] / youtube_dl / extractor / twentyfourvideo.py
index cb9e5f1b5ccfbf64d90c6bee96d26c548c4379d7..4025edf02b4ca5b8c42e3324fb094fe43141ce68 100644 (file)
@@ -5,6 +5,8 @@ from .common import InfoExtractor
 from ..utils import (
     parse_iso8601,
     int_or_none,
+    xpath_attr,
+    xpath_element,
 )
 
 
@@ -15,7 +17,7 @@ class TwentyFourVideoIE(InfoExtractor):
     _TESTS = [
         {
             'url': 'http://www.24video.net/video/view/1044982',
-            'md5': 'd041af8b5b4246ea466226a0d6693345',
+            'md5': 'e09fc0901d9eaeedac872f154931deeb',
             'info_dict': {
                 'id': '1044982',
                 'ext': 'mp4',
@@ -45,7 +47,8 @@ class TwentyFourVideoIE(InfoExtractor):
 
         title = self._og_search_title(webpage)
         description = self._html_search_regex(
-            r'<span itemprop="description">([^<]+)</span>', webpage, 'description', fatal=False)
+            r'<(p|span)[^>]+itemprop="description"[^>]*>(?P<description>[^<]+)</\1>',
+            webpage, 'description', fatal=False, group='description')
         thumbnail = self._og_search_thumbnail(webpage)
         duration = int_or_none(self._og_search_property(
             'duration', webpage, 'duration', fatal=False))
@@ -69,12 +72,14 @@ class TwentyFourVideoIE(InfoExtractor):
             r'http://www.24video.net/video/xml/%s?mode=init' % video_id,
             video_id, 'Downloading init XML')
 
-        video = self._download_xml(
+        video_xml = self._download_xml(
             'http://www.24video.net/video/xml/%s?mode=play' % video_id,
-            video_id, 'Downloading video XML').find('.//video')
+            video_id, 'Downloading video XML')
+
+        video = xpath_element(video_xml, './/video', 'video', fatal=True)
 
         formats = [{
-            'url': video.attrib['url'],
+            'url': xpath_attr(video, '', 'url', 'video URL', fatal=True),
         }]
 
         like_count = int_or_none(video.get('ratingPlus'))