From a721ae0ae6106b49c3a33b08a8b6a24bdcbd8801 Mon Sep 17 00:00:00 2001
From: Xavier Ripoll <xavier@wiris.com>
Date: Wed, 17 Oct 2018 17:19:54 +0200
Subject: [PATCH] MOODLE-556: Normalization of whitespace in MathML token
 elements

---
 library/HTMLPurifier.includes.php             |  1 +
 library/HTMLPurifier.safe-includes.php        |  1 +
 library/HTMLPurifier/HTMLModule/Math.php      |  3 ++
 .../Injector/MathSpaceNormalize.php           | 40 +++++++++++++++++++
 4 files changed, 45 insertions(+)
 create mode 100644 library/HTMLPurifier/Injector/MathSpaceNormalize.php

diff --git a/library/HTMLPurifier.includes.php b/library/HTMLPurifier.includes.php
index 75144193..28202e62 100644
--- a/library/HTMLPurifier.includes.php
+++ b/library/HTMLPurifier.includes.php
@@ -192,6 +192,7 @@ require 'HTMLPurifier/HTMLModule/Tidy/XHTML.php';
 require 'HTMLPurifier/Injector/AutoParagraph.php';
 require 'HTMLPurifier/Injector/DisplayLinkURI.php';
 require 'HTMLPurifier/Injector/Linkify.php';
+require 'HTMLPurifier/Injector/MathSpaceNormalize.php';
 require 'HTMLPurifier/Injector/PurifierLinkify.php';
 require 'HTMLPurifier/Injector/RemoveEmpty.php';
 require 'HTMLPurifier/Injector/RemoveSpansWithoutAttributes.php';
diff --git a/library/HTMLPurifier.safe-includes.php b/library/HTMLPurifier.safe-includes.php
index 0da1e6ed..73251972 100644
--- a/library/HTMLPurifier.safe-includes.php
+++ b/library/HTMLPurifier.safe-includes.php
@@ -186,6 +186,7 @@ require_once $__dir . '/HTMLPurifier/HTMLModule/Tidy/XHTML.php';
 require_once $__dir . '/HTMLPurifier/Injector/AutoParagraph.php';
 require_once $__dir . '/HTMLPurifier/Injector/DisplayLinkURI.php';
 require_once $__dir . '/HTMLPurifier/Injector/Linkify.php';
+require_once $__dir . '/HTMLPurifier/Injector/MathSpaceNormalize.php';
 require_once $__dir . '/HTMLPurifier/Injector/PurifierLinkify.php';
 require_once $__dir . '/HTMLPurifier/Injector/RemoveEmpty.php';
 require_once $__dir . '/HTMLPurifier/Injector/RemoveSpansWithoutAttributes.php';
diff --git a/library/HTMLPurifier/HTMLModule/Math.php b/library/HTMLPurifier/HTMLModule/Math.php
index 827e6338..6f535ae2 100644
--- a/library/HTMLPurifier/HTMLModule/Math.php
+++ b/library/HTMLPurifier/HTMLModule/Math.php
@@ -19,6 +19,9 @@ class HTMLPurifier_HTMLModule_Math extends HTMLPurifier_HTMLModule
     public function setup($config)
     {
 
+        // Normalize whitespace inside text elements as per MathML spec 2.1.7
+        $this->info_injector[] = new HTMLPurifier_Injector_MathSpaceNormalize();
+
         /*****************************************************************
          * Meta variables
          * Used in this file to simplify code and help adapt the DTD
diff --git a/library/HTMLPurifier/Injector/MathSpaceNormalize.php b/library/HTMLPurifier/Injector/MathSpaceNormalize.php
new file mode 100644
index 00000000..1228f934
--- /dev/null
+++ b/library/HTMLPurifier/Injector/MathSpaceNormalize.php
@@ -0,0 +1,40 @@
+<?php
+
+class HTMLPurifier_Injector_MathSpaceNormalize extends HTMLPurifier_Injector
+{
+
+    /**
+     * Elements to apply handleText to.
+     * These are those that accept #PCDATA except <cs> and <cbytes>.
+     * @type array
+     */
+    protected $tags = array('mi', 'mn', 'mo', 'ms', 'mtext', 'ci', 'cn', 'csymbol', 'annotation');
+
+    /**
+     * @param HTMLPurifier_Token $token
+     */
+    public function handleText(&$token)
+    {
+
+        // No parent tag => return to avoid error on following line
+        if (count($this->currentNesting) == 0) {
+            return;
+        }
+
+        // Get the parent tag
+        $parent_token = $this->currentNesting[count($this->currentNesting) - 1];
+
+        // If we're not in a "token element" (specified in $tags above), return
+        if ($parent_token === null || !in_array($parent_token->name, $this->tags)) {
+            return;
+        }
+
+        // Replace as per the MathML specification, section 2.1.7
+        $token->data = preg_replace(
+            '/[ \t\n\r]+/',
+            ' ',
+            trim($token->data) // Using trim($token->data, ' \t\n\r') trims t,n,r
+        );
+    }
+
+}