From 0e815177c829e445ca1ce27556ca65f8e8b725a5 Mon Sep 17 00:00:00 2001
From: Miroslav Stampar <miroslav.stampar@gmail.com>
Date: Sat, 1 Jan 2011 19:07:40 +0000
Subject: [PATCH] minor update

---
 lib/core/common.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/core/common.py b/lib/core/common.py
index 8ce51f745..94e561a0d 100644
--- a/lib/core/common.py
+++ b/lib/core/common.py
@@ -1111,11 +1111,11 @@ def sanitizeAsciiString(subject):
     else:
         return None
 
-def getFilteredPageContent(page):
+def getFilteredPageContent(page, onlyText=True):
     retVal = page
 
     if isinstance(page, basestring):
-        retVal = re.sub(r"(?s)<script.+?</script>|<style.+?</style>|<[^>]+>|\t|\n|\r", " ", page)
+        retVal = re.sub(r"(?s)<script.+?</script>|<style.+?</style>%s" % (r"|<[^>]+>|\t|\n|\r" if onlyText else ""), " ", page)
 
         while retVal.find("  ") != -1:
             retVal = retVal.replace("  ", " ")