[mtv] Fix description extraction (Closes #8962)

[youtube-dl] / youtube_dl / extractor / mtv.py

diff --git a/youtube_dl/extractor/mtv.py b/youtube_dl/extractor/mtv.py
index 56ddcbbf303b1eb1fee3b6f7cc19fba0cbc4c973..824bbcb4e6d58d3856e521c6de055346164dd1db 100644 (file)
--- a/youtube_dl/extractor/mtv.py
+++ b/youtube_dl/extractor/mtv.py
@@ -17,6 +17,7 @@ from ..utils import (
     unescapeHTML,
     url_basename,
     RegexNotFoundError,
+    xpath_text,
 )
 
 
@@ -111,7 +112,8 @@ class MTVServicesInfoExtractor(InfoExtractor):
         uri = itemdoc.find('guid').text
         video_id = self._id_from_uri(uri)
         self.report_extraction(video_id)
-        mediagen_url = itemdoc.find('%s/%s' % (_media_xml_tag('group'), _media_xml_tag('content'))).attrib['url']
+        content_el = itemdoc.find('%s/%s' % (_media_xml_tag('group'), _media_xml_tag('content')))
+        mediagen_url = content_el.attrib['url']
         # Remove the templates, like &device={device}
         mediagen_url = re.sub(r'&[^=]*?={.*?}(?=(&|$))', '', mediagen_url)
         if 'acceptMethods' not in mediagen_url:
@@ -129,11 +131,7 @@ class MTVServicesInfoExtractor(InfoExtractor):
             message += item.text
             raise ExtractorError(message, expected=True)
 
-        description_node = itemdoc.find('description')
-        if description_node is not None:
-            description = description_node.text.strip()
-        else:
-            description = None
+        description = xpath_text(itemdoc, 'description')
 
         title_el = None
         if title_el is None:
@@ -159,9 +157,6 @@ class MTVServicesInfoExtractor(InfoExtractor):
         if mtvn_id_node is not None:
             mtvn_id = mtvn_id_node.text
 
-        content_el = find_xpath_attr(itemdoc, self._xpath_ns('.//content', 'http://search.yahoo.com/mrss/'), 'duration')
-        duration = float_or_none(content_el.attrib.get('duration')) if content_el is not None else None
-
         return {
             'title': title,
             'formats': self._extract_video_formats(mediagen_doc, mtvn_id),
@@ -169,7 +164,7 @@ class MTVServicesInfoExtractor(InfoExtractor):
             'id': video_id,
             'thumbnail': self._get_thumbnail_url(uri, itemdoc),
             'description': description,
-            'duration': duration,
+            'duration': float_or_none(content_el.attrib.get('duration')),
         }
 
     def _get_feed_query(self, uri):