[gaskrank] Improve (closes #12493)

youtube_dl/extractor/gaskrank.py

@@ -11,10 +11,8 @@ from ..utils import (
 
 
 class GaskrankIE(InfoExtractor):
-    """InfoExtractor for gaskrank.tv"""
-    _VALID_URL = r'https?://(?:www\.)?gaskrank\.tv/tv/(?P<categories>[^/]+)/(?P<id>[^/]+)\.html?'
-    _TESTS = [
-        {
+    _VALID_URL = r'https?://(?:www\.)?gaskrank\.tv/tv/(?P<categories>[^/]+)/(?P<id>[^/]+)\.htm'
+    _TESTS = [{
         'url': 'http://www.gaskrank.tv/tv/motorrad-fun/strike-einparken-durch-anfaenger-crash-mit-groesserem-flurschaden.htm',
         'md5': '1ae88dbac97887d85ebd1157a95fc4f9',
         'info_dict': {
@@ -28,8 +26,7 @@ class GaskrankIE(InfoExtractor):
             'upload_date': '20170110',
             'uploader_url': None,
         }
-        },
-        {
+    }, {
         'url': 'http://www.gaskrank.tv/tv/racing/isle-of-man-tt-2011-michael-du-15920.htm',
         'md5': 'c33ee32c711bc6c8224bfcbe62b23095',
         'info_dict': {
@@ -40,21 +37,22 @@ class GaskrankIE(InfoExtractor):
             'categories': ['racing'],
             'display_id': 'isle-of-man-tt-2011-michael-du-15920',
             'uploader_id': 'IOM',
-                'upload_date': '20160506',
+            'upload_date': '20170523',
             'uploader_url': 'www.iomtt.com',
         }
-        }
-    ]
+    }]
 
     def _real_extract(self, url):
-        """extract information from gaskrank.tv"""
-
         display_id = self._match_id(url)
+
         webpage = self._download_webpage(url, display_id)
-        categories = [re.match(self._VALID_URL, url).group('categories')]
-        title = self._og_search_title(webpage, default=None) or self._html_search_meta(
+
+        title = self._og_search_title(
+            webpage, default=None) or self._html_search_meta(
             'title', webpage, fatal=True)
 
+        categories = [re.match(self._VALID_URL, url).group('categories')]
+
         mobj = re.search(
             r'Video von:\s*(?P<uploader_id>[^|]*?)\s*\|\s*vom:\s*(?P<upload_date>[0-9][0-9]\.[0-9][0-9]\.[0-9][0-9][0-9][0-9])',
             webpage)
@@ -83,7 +81,7 @@ class GaskrankIE(InfoExtractor):
 
         video_id = self._search_regex(
             r'https?://movies\.gaskrank\.tv/([^-]*?)(-[^\.]*)?\.mp4',
-            webpage, 'video id')
+            webpage, 'video id', default=display_id)
 
         entry = self._parse_html5_media_entries(url, webpage, video_id)[0]
         entry.update({