From 5c162efbd8f9f8b2fc919638070493c0ea86b7fc Mon Sep 17 00:00:00 2001
From: Miroslav Stampar <miroslav.stampar@gmail.com>
Date: Tue, 12 Jul 2011 23:21:15 +0000
Subject: [PATCH] more optimization

---
 lib/core/common.py   | 5 +++++
 lib/core/settings.py | 3 +++
 2 files changed, 8 insertions(+)

diff --git a/lib/core/common.py b/lib/core/common.py
index d04172fbe..b6a471192 100644
--- a/lib/core/common.py
+++ b/lib/core/common.py
@@ -92,6 +92,7 @@ from lib.core.settings import ML
 from lib.core.settings import MIN_TIME_RESPONSES
 from lib.core.settings import PAYLOAD_DELIMITER
 from lib.core.settings import REFLECTED_NON_ALPHA_NUM_REGEX
+from lib.core.settings import REFLECTED_MAX_REGEX_PARTS
 from lib.core.settings import REFLECTED_VALUE_MARKER
 from lib.core.settings import TIME_STDEV_COEFF
 from lib.core.settings import DYNAMICITY_MARK_LENGTH
@@ -2596,6 +2597,10 @@ def removeReflectiveValues(content, payload, suppressWarning=False):
             regex = regex.replace(2 * REFLECTED_NON_ALPHA_NUM_REGEX, REFLECTED_NON_ALPHA_NUM_REGEX)
 
         if all([part.lower() in content.lower() for part in regex.split(REFLECTED_NON_ALPHA_NUM_REGEX)]): # fast optimization check
+            parts = regex.split(REFLECTED_NON_ALPHA_NUM_REGEX)
+            if len(parts) > REFLECTED_MAX_REGEX_PARTS: # preventing CPU hogs
+                regex = "%s.+?%s" % (REFLECTED_NON_ALPHA_NUM_REGEX.join(parts[:REFLECTED_MAX_REGEX_PARTS/2]), REFLECTED_NON_ALPHA_NUM_REGEX.join(parts[-REFLECTED_MAX_REGEX_PARTS/2:]))
+
             retVal = re.sub(regex, REFLECTED_VALUE_MARKER, content, re.I)
 
         if retVal != content:
diff --git a/lib/core/settings.py b/lib/core/settings.py
index 653b64454..6aea7548a 100644
--- a/lib/core/settings.py
+++ b/lib/core/settings.py
@@ -311,6 +311,9 @@ REFLECTED_VALUE_MARKER = '__REFLECTED_VALUE__'
 # Regular expression used for marking non-alphanum characters
 REFLECTED_NON_ALPHA_NUM_REGEX = r'[^\r\n]+?'
 
+# Maximum number of alpha-numerical parts in reflected regex (for speed purposes)
+REFLECTED_MAX_REGEX_PARTS = 10
+
 # Chars which can be used as a failsafe values in case of too long URL encoding value
 URLENCODE_FAILSAFE_CHARS = '()|,'