fixing support for parsing BURP logs

stamparm · stamparm · commit 7ce3af68fc5d · 2011-10-27T17:31:34.000Z
diff --git a/doc/THANKS b/doc/THANKS
@@ -417,6 +417,9 @@ Andrea Rossi <andyroyalbattle@yahoo.it>
     for reporting a minor bug
     for suggesting a feature
 
+Mathew Rowley <mathew.rowley@intrepidusgroup.com>
+    for reporting a bug
+
 Frederic Roy <frederic.roy@telindus.fr>
     for reporting a couple of bugs
 
diff --git a/lib/core/option.py b/lib/core/option.py
@@ -99,7 +99,7 @@
 from lib.core.settings import MAXDB_ALIASES
 from lib.core.settings import SYBASE_ALIASES
 from lib.core.settings import DB2_ALIASES
-from lib.core.settings import BURP_SPLITTER
+from lib.core.settings import BURP_REQUEST_REGEX
 from lib.core.settings import LOCALHOST
 from lib.core.settings import MAX_NUMBER_OF_THREADS
 from lib.core.settings import PARAMETER_SPLITTING_REGEX
@@ -211,20 +211,23 @@ def __parseBurpLog(content):
         port = None
         scheme = None
 
-        reqResList = content.split(BURP_SPLITTER)
+        reqResList = re.findall(BURP_REQUEST_REGEX, content, re.I | re.S)
+
+        if not reqResList:
+            reqResList = [content]
 
         for request in reqResList:
             if scheme is None:
-                schemePort = re.search("\d\d[\:|\.]\d\d[\:|\.]\d\d\s+(http[\w]*)\:\/\/.*?\:([\d]+)", request, re.I)
+                schemePort = re.search("(http[\w]*)\:\/\/.*?\:([\d]+).+?={10,}", request, re.I | re.S)
 
                 if schemePort:
                     scheme = schemePort.group(1)
                     port = schemePort.group(2)
 
-            if not re.search ("^[\n]*(GET|POST).*?\sHTTP\/", request, re.I):
+            if not re.search ("^[\n]*(GET|POST).*?\sHTTP\/", request, re.I | re.M):
                 continue
 
-            if re.search("^[\n]*(GET|POST).*?\.(gif|jpg|png)\sHTTP\/", request, re.I):
+            if re.search("^[\n]*(GET|POST).*?\.(gif|jpg|png)\sHTTP\/", request, re.I | re.M):
                 continue
 
             getPostReq = False
diff --git a/lib/core/settings.py b/lib/core/settings.py
@@ -267,7 +267,7 @@
 WEBSCARAB_SPLITTER = "### Conversation"
 
 # Splitter used between requests in BURP log files
-BURP_SPLITTER = "======================================================"
+BURP_REQUEST_REGEX = r"={10,}\s+[^=]+={10,}\s+.+?={10,}"
 
 # Encoding used for Unicode data
 UNICODE_ENCODING = "utf8"
