X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=parse_html%2Fbkmk_ph_beautifulsoup4.py;h=060f0786d473c17de36ad073c05632c06f30c35d;hb=00ddc855d130a8ce37f1efcd00ca225b656fbfce;hp=d76c05acd3da7f1bc6d09168562adfc1c05fa8c0;hpb=f6c7680eb82ae0db48ce4c17a9b259e159ef8bd8;p=bookmarks_db.git

diff --git a/parse_html/bkmk_ph_beautifulsoup4.py b/parse_html/bkmk_ph_beautifulsoup4.py
index d76c05a..060f078 100644
--- a/parse_html/bkmk_ph_beautifulsoup4.py
+++ b/parse_html/bkmk_ph_beautifulsoup4.py
@@ -71,7 +71,10 @@ def parse_html(html_text, charset=None, log=None):
             for part in title:
                 #if not isinstance(part, string_type):
                 #    part = part.decode()
-                parts.append(part.strip())
+                if part.strip:
+                    parts.append(part.strip())
+                else:
+                    parts.append(' ')  # Skip tags, they're usually `<br>`
             title = ''.join(parts)
 
     meta = head.find(_find_contenttype, recursive=False)
@@ -136,6 +139,7 @@ def _find_refresh(Tag):
 
 
 def _find_icon(Tag):
-    return (Tag.name == "link") and \
-       (Tag.get_attribute_list("rel", '')[0].lower()
-        in ('icon', 'shortcut icon'))
+    if Tag.name != "link":
+        return False
+    rel = ' '.join(Tag.get_attribute_list("rel", ''))
+    return rel in ('icon', 'shortcut icon')