youtube-dl

commit 68bb2fef9565159eba4a47f464b6b420cf2d5cda
parent 854cc54bc1d0488d8fa88bd5dfed6f7f8981847e
Author: Sergey M․ <dstftw@gmail.com>
Date:   Sun,  1 May 2016 07:15:23 +0600

[tagesschau] Restrict playlist entry regex

Diffstat:
M
youtube_dl/extractor/tagesschau.py
 | 
6
+++++-

1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/youtube_dl/extractor/tagesschau.py b/youtube_dl/extractor/tagesschau.py
@@ -200,6 +200,10 @@ class TagesschauIE(InfoExtractor):
     }, {
         'url': 'http://www.tagesschau.de/100sekunden/index.html',
         'only_matching': True,
+    }, {
+        # playlist article with collapsing sections
+        'url': 'http://www.tagesschau.de/wirtschaft/faq-freihandelszone-eu-usa-101.html',
+        'only_matching': True,
     }]
 
     @classmethod
@@ -275,7 +279,7 @@ class TagesschauIE(InfoExtractor):
         if webpage_type == 'website':  # Article
             entries = []
             for num, (entry_title, media_kind, download_text) in enumerate(re.findall(
-                    r'(?s)<p[^>]+class="infotext"[^>]*>.*?<strong>(.+?)</strong>.*?</p>.*?%s' % DOWNLOAD_REGEX,
+                    r'(?s)<p[^>]+class="infotext"[^>]*>\s*(?:<a[^>]+>)?\s*<strong>(.+?)</strong>.*?</p>.*?%s' % DOWNLOAD_REGEX,
                     webpage), 1):
                 entries.append({
                     'id': '%s-%d' % (display_id, num),