From 1fd91de82c0be74ae2305f1412b1cd1b2948dfbc Mon Sep 17 00:00:00 2001
From: Zach Copley <zach@status.net>
Date: Mon, 1 Mar 2010 14:58:06 -0800
Subject: [PATCH] Upgrade XML output scrubbing to better deal with newline and
 a few other chars

---
 lib/util.php | 24 ++++++++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

diff --git a/lib/util.php b/lib/util.php
index 8381bc63c0..f793cc10e9 100644
--- a/lib/util.php
+++ b/lib/util.php
@@ -770,8 +770,28 @@ function common_shorten_links($text)
 
 function common_xml_safe_str($str)
 {
-    // Neutralize control codes and surrogates
-	return preg_replace('/[\p{Cc}\p{Cs}]/u', '*', $str);
+    // Replace common eol and extra whitespace input chars
+    $unWelcome = array(
+        "\t",  // tab
+        "\n",  // newline
+        "\r",  // cr
+        "\0",  // null byte eos
+        "\x0B" // vertical tab
+    );
+
+    $replacement = array(
+        ' ', // single space
+        ' ',
+        '',  // nothing
+        '',
+        ' '
+    );
+
+    $str = str_replace($unWelcome, $replacement, $str);
+
+    // Neutralize any additional control codes and UTF-16 surrogates
+    // (Twitter uses '*')
+    return preg_replace('/[\p{Cc}\p{Cs}]/u', '*', $str);
 }
 
 function common_tag_link($tag)