Bug 10332 - Import of incorrectly formatted Moodle XML is not handle…

…d well

lang/en_utf8/error.php

@@ -41,6 +41,7 @@
 $string['errorreadingfile'] = 'Error reading file \"$a\"';
 $string['errorsavingrequest'] = 'An error occurred when trying to save your request.';
 $string['errorunzippingfiles'] = 'Error unzipping files';
+$string['errorxmlfile'] = 'Incorrect xml file (<strong>$a->errorstring</strong>, line <strong>$a->errorline</strong>, character <strong>$a->errorchar</strong>)';
 $string['fieldrequired'] = '\"$a\" is a required field';
 $string['filenotfound'] = 'Sorry, the requested file could not be found';
 $string['forumblockingtoomanyposts'] = 'You have exceeded the posting threshold set for this forum';

lib/xmlize.php

@@ -1,4 +1,20 @@
 <?php
+/**
+ * Exception indicating errors in xml file
+ *
+ * @package moodlecore
+ * @copyright 2010 The Open University
+ * @license http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later
+ */
+class xml_format_exception extends moodle_exception {
+    function __construct($errorstring, $errorline, $errorchar) {
+        $errorinfo = new stdClass;
+        $errorinfo->errorstring = $errorstring;
+        $errorinfo->errorline = $errorline;
+        $errorinfo->errorchar = $errorchar;
+        parent::__construct('errorxmlfile', 'error', '', $errorinfo);
+    }
+}
 
 /**
  * xmlize.php - xmlize() is by Hans Anderson, {@link http://www.hansanderson.com/contact/}
@@ -41,16 +57,28 @@
  * @param array $data The array to be converted
  * @param int $WHITE  If set to 1 allows the parser to skip "space" characters in xml document. Default is 1
  * @param string $encoding Specify an OUTPUT encoding. If not specified, it defaults to UTF-8.
+ * @param bool $error if set on true it checks whether xml-file is well-formed and throw an exception when xml is not well-formed,
  * @return array
  */
-function xmlize($data, $WHITE=1, $encoding='UTF-8') {
+function xmlize($data, $WHITE=1, $encoding='UTF-8', $error=false) {
 
     $data = trim($data);
     $vals = $index = $array = array();
     $parser = xml_parser_create($encoding);
     xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
     xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, $WHITE);
-    xml_parse_into_struct($parser, $data, $vals, $index);
+    xml_parse_into_struct($parser, $data, &$vals, &$index);
+
+    // Error handling when the xml file is not well-formed
+    if ($error) {
+        $errorcode = xml_get_error_code($parser);
+        if ($errorcode) {
+            $errorstring = xml_error_string($errorcode);
+            $errorline = xml_get_current_line_number($parser);
+            $errorchar = xml_get_current_column_number($parser);
+            throw new xml_format_exception($errorstring, $errorline, $errorchar);
+        }
+    }
     xml_parser_free($parser);
 
     $i = 0;

question/format.php

@@ -242,6 +242,9 @@ function importprocess() {
         }
 
         if (! $questions = $this->readquestions($lines)) {   // Extract all the questions
+            if (!is_array($questions)) {
+                return false;
+            }
             notify( get_string('noquestionsinfile','quiz') );
             return false;
         }

question/format/xml/format.php

@@ -652,8 +652,12 @@ function readquestions($lines) {
 
         // This converts xml to big nasty data structure
         // the 0 means keep white space as it is (important for markdown format)
-        $xml = xmlize($text, 0);
-
+        try {
+            $xml = xmlize($text, 0, 'UTF-8', true);
+        } catch (xml_format_exception $e){
+            $this->error ($e->getMessage(), '');
+            return false;
+        }
         // Set up array to hold all our questions
         $questions = array();