MDL-38466 filters: ReDoS protection for text to URL conversion.

filter/urltolink/filter.php

@@ -134,10 +134,9 @@ class filter_urltolink extends moodle_text_filter {
 
         // Lookbehind assertions.
         // Is not HTML attribute or CSS URL property. Unfortunately legit text like "url(http://...)" will not be a link.
-        $lookbehindstart = "(?<!=[\"']|\burl\([\"' ]|\burl\()";
         $lookbehindend = "(?<![]),.;])";
 
-        $regex = "$lookbehindstart$urlstart((?:$domainsegment\.)+$domainsegment|$numericip)" .
+        $regex = "$urlstart((?:$domainsegment\.)+$domainsegment|$numericip)" .
                 "($port?$path$querystring?$fragment?)$lookbehindend";
         if ($unicoderegexp) {
             $regex = '#' . $regex . '#ui';
@@ -145,7 +144,39 @@ class filter_urltolink extends moodle_text_filter {
             $regex = '#' . preg_replace(array('\pLl', '\PL'), 'a-z', $regex) . '#i';
         }
 
-        $text = preg_replace($regex, '<a href="http$1://$2$3$4" class="_blanktarget">$0</a>', $text);
+        // Locate any HTML tags.
+        $matches = preg_split('/(<[^>]*>)/i', $text, -1, PREG_SPLIT_NO_EMPTY | PREG_SPLIT_DELIM_CAPTURE);
+        $lookingforendstyle = false;
+
+        // Iterate through the tokenized text to handle chunks (html and content).
+        foreach ($matches as $idx => $chunk) {
+            // Nothing to do. We skip completely any html chunk.
+            if (strpos($chunk, '<') !== false) {
+                continue;
+            }
+
+            // Nothing to do. We skip any content chunk having any of these attributes.
+            if (preg_match('#(background=")|(action=")|(style="background)|(href=")|(src=")|(url [(])#', $chunk)) {
+                continue;
+            }
+
+            // Arrived here, we want to process every word in this chunk.
+            $text = $chunk;
+            $words = explode(' ', $text);
+
+            foreach ($words as $idx2 => $word) {
+                // ReDoS protection. Stop processing if a word is too large.
+                if (strlen($word) < 4096) {
+                    $words[$idx2] = preg_replace($regex, '<a href="http$1://$2$3$4" class="_blanktarget">$0</a>', $word);
+                }
+            }
+            $text = implode(' ', $words);
+
+            // Copy the result back to the array.
+            $matches[$idx] = $text;
+        }
+
+        $text = implode('', $matches);
 
         if (!empty($ignoretags)) {
             $ignoretags = array_reverse($ignoretags); /// Reversed so "progressive" str_replace() will solve some nesting problems.

filter/urltolink/tests/filter_test.php

@@ -32,6 +32,10 @@ require_once($CFG->dirroot . '/filter/urltolink/filter.php'); // Include the cod
 class filter_urltolink_testcase extends basic_testcase {
 
     function get_convert_urls_into_links_test_cases() {
+        // Create a 4095 and 4096 long URLs.
+        $superlong4095 = str_pad('http://www.superlong4095.com?this=something', 4095, 'a');
+        $superlong4096 = str_pad('http://www.superlong4096.com?this=something', 4096, 'a');
+
         $texts = array (
             //just a url
             'http://moodle.org - URL' => '<a href="http://moodle.org" class="_blanktarget">http://moodle.org</a> - URL',
@@ -130,6 +134,7 @@ class filter_urltolink_testcase extends basic_testcase {
             '<td background="http://moodle.org">&nbsp;</td>' => '<td background="http://moodle.org">&nbsp;</td>',
             '<td background="www.moodle.org">&nbsp;</td>' => '<td background="www.moodle.org">&nbsp;</td>',
             '<form name="input" action="http://moodle.org/submit.asp" method="get">'=>'<form name="input" action="http://moodle.org/submit.asp" method="get">',
+            '<input type="submit" value="Go to http://moodle.org">' => '<input type="submit" value="Go to http://moodle.org">',
             '<td background="https://www.moodle.org">&nbsp;</td>' => '<td background="https://www.moodle.org">&nbsp;</td>',
             // CSS URLs.
             '<table style="background-image: url(\'http://moodle.org/pic.jpg\');">' => '<table style="background-image: url(\'http://moodle.org/pic.jpg\');">',
@@ -148,6 +153,12 @@ class filter_urltolink_testcase extends basic_testcase {
             //Encoded URLs in the query
             'URL: http://127.0.0.1/path/to?param=value_with%28parenthesis%29&param2=1' => 'URL: <a href="http://127.0.0.1/path/to?param=value_with%28parenthesis%29&param2=1" class="_blanktarget">http://127.0.0.1/path/to?param=value_with%28parenthesis%29&param2=1</a>',
             'URL: www.localhost.com/path/to?param=value_with%28parenthesis%29&param2=1' => 'URL: <a href="http://www.localhost.com/path/to?param=value_with%28parenthesis%29&param2=1" class="_blanktarget">www.localhost.com/path/to?param=value_with%28parenthesis%29&param2=1</a>',
+            // Test URL less than 4096 characters in size is converted to link.
+            'URL: ' . $superlong4095 => 'URL: <a href="' . $superlong4095 . '" class="_blanktarget">' . $superlong4095 . '</a>',
+            // Test URL equal to or greater than 4096 characters in size is not converted to link.
+            'URL: ' . $superlong4096 => 'URL: ' . $superlong4096,
+            // Testing URL within a span tag.
+            'URL: <span style="kasd"> my link to http://google.com </span>' => 'URL: <span style="kasd"> my link to <a href="http://google.com" class="_blanktarget">http://google.com</a> </span>',
             //URLs in Javascript. Commented out as part of MDL-21183
             //'var url="http://moodle.org";'=>'var url="http://moodle.org";',
             //'var url = "http://moodle.org";'=>'var url = "http://moodle.org";',