From 6ecb3acc5144aea80821a11bbb5ca715e2e9bdac Mon Sep 17 00:00:00 2001
From: Tony Murray <murraytony@gmail.com>
Date: Wed, 1 Mar 2017 04:23:38 -0600
Subject: [PATCH] Improve stats data scrubbing (#6041)

Include Darwin (OSX) in the Linux hostname scrubbing, add underscores too.
Way more generic serial number replacement.  Preserves the format of the serial number, but removes the data.
---
 includes/callback.php | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/includes/callback.php b/includes/callback.php
index 853a39423c..0462461fcb 100644
--- a/includes/callback.php
+++ b/includes/callback.php
@@ -79,8 +79,25 @@ if ($enabled == 1) {
 
     // sanitize sysDescr
     $device_info = array_map(function ($entry) {
-        $entry['sysDescr'] = preg_replace('/^Linux [A-Za-z\-0-9\.]+ (?=[0-9\.]{5,9})/', 'Linux hostname ', $entry['sysDescr']);
-        $entry['sysDescr'] = preg_replace('/ SN: [A-Z0-9]{12}/', ' SN: 0A0A0A0A0A0A ', $entry['sysDescr']);
+        // remove hostnames from linux and macosx
+        $entry['sysDescr'] = preg_replace_callback('/^(Linux |Darwin )[A-Za-z0-9._\-]+ ([0-9.]{5,9})/', function ($matches) {
+            return $matches[1] . 'hostname ' .$matches[2];
+        }, $entry['sysDescr']);
+
+        // wipe serial numbers, preserve the format
+        $entry['sysDescr'] = preg_replace_callback('/(SN[:=]? ?)([A-Za-z0-9.\-]{4,16})/', function ($matches) {
+            $patterns = array(
+                '/[A-Z]/',
+                '/[a-z]/',
+                '/[0-9]/'
+            );
+            $replacements = array(
+                'A',
+                'a',
+                '0'
+            );
+            return $matches[1] . preg_replace($patterns, $replacements, $matches[2]);
+        }, $entry['sysDescr']);
         return $entry;
     }, $device_info);