From 88ad786c7335a17d2f179a84b52a1500a3a1e68b Mon Sep 17 00:00:00 2001
From: david <david@e0a8ed71-7df4-0310-8962-fdc924857419>
Date: Thu, 26 Mar 2009 18:45:25 +0000
Subject: [PATCH] Remove service fingerprints in anonymize.py because they
 often contain identifying information.

---
 ndiff/test-scans/anonymize.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/ndiff/test-scans/anonymize.py b/ndiff/test-scans/anonymize.py
index 20d0c55b8..b1df5785f 100755
--- a/ndiff/test-scans/anonymize.py
+++ b/ndiff/test-scans/anonymize.py
@@ -2,13 +2,12 @@
 
 # Anonymize an Nmap XML file, replacing host name and IP addresses with random
 # anonymous ones. Anonymized names will be consistent between runs of the
-# program. Give a file name as an argument. The anonymized file is written to
-# stdout.
+# program. Any servicefp attributes are removed. Give a file name as an
+# argument. The anonymized file is written to stdout.
 #
 # The anonymization is not rigorous. This program just matches regular
 # expressions against things that look like address and host names. It is
-# possible that it will leave some identifying information, for example a host
-# name split across lines in a service fingerprint.
+# possible that it will leave some identifying information.
 
 import hashlib
 import random
@@ -99,6 +98,7 @@ def anonymize_file(f):
         line = re.sub(ipv6_re, repl_addr, line)
         line = re.sub(r'<hostname name="([^"]*)"', repl_hostname_name, line)
         line = re.sub(r'\bhostname="([^"]*)"', repl_hostname, line)
+        line = re.sub(r' *\bservicefp="([^"]*)"', r'', line)
         yield line
 
 def main():