Merged changes from cce branch (r3727:HEAD/4008); the branch is now in sync with trunk

1 files changed, 9 insertions, 1 deletions

diff --git a/paste/util/quoting.py b/paste/util/quoting.py
index d029518..5f42e6a 100644
--- a/paste/util/quoting.py
+++ b/paste/util/quoting.py
@@ -6,7 +6,8 @@ import htmlentitydefs
 import urllib
 import re
 
-__all__ = ['html_quote', 'html_unquote', 'url_quote', 'url_unquote']
+__all__ = ['html_quote', 'html_unquote', 'url_quote', 'url_unquote',
+           'strip_html']
 
 default_encoding = 'UTF-8'
 
@@ -59,6 +60,13 @@ def html_unquote(s, encoding=None):
         s = s.decode(encoding or default_encoding)
     return _unquote_re.sub(_entity_subber, s)
 
+def strip_html(s):
+    # should this use html_unquote?
+    s = re.sub('<.*?>', '', s)
+    s = s.replace('&nbsp;', ' ').replace('&lt;', '<')
+    s = s.replace('&gt;', '>').replace('&amp;','&')
+    return s
+
 url_quote = urllib.quote
 url_unquote = urllib.unquote