Some sites put TITLE in HTML without HEAD.

[bookmarks_db.git] / Robots / parse_html_beautifulsoup.py

diff --git a/Robots/parse_html_beautifulsoup.py b/Robots/parse_html_beautifulsoup.py
index 62ef27709a67d27a91c7a86d220e6214d7f94888..209486c4054f015a179d4469ccaa5adc3f8bbe8f 100644 (file)
--- a/Robots/parse_html_beautifulsoup.py
+++ b/Robots/parse_html_beautifulsoup.py
@@ -63,11 +63,14 @@ def parse_html(filename, charset=None):
    except AttributeError:
       return None
 
+   if head is None:
+      head = root.html # Some sites put TITLE in HTML without HEAD
+
    _charset = root.originalEncoding
    try:
       title = head.title.string.encode(_charset)
    except AttributeError:
-      return '' # HEAD but no TITLE
+      title = '' # HEAD but no TITLE
 
    meta = head.find(_find_refresh, recursive=False)
    if meta: