Wrap in a Try catch

Islandora · Apr 15, 2019 · c979e03 · c979e03
1 parent efd3185
commit c979e03
Show file tree

Hide file tree

Showing 4 changed files with 31 additions and 6 deletions.
diff --git a/includes/coins.inc b/includes/coins.inc
@@ -51,7 +51,16 @@ class Coins {
       drupal_set_message(t('No MODS datastream found!'), 'warning');
       return " ";
     }
-    $xml = new SimpleXMLElement($xmlstr);
+    try {
+      $xml = new SimpleXMLElement(islandora_sanitize_input_for_valid_xml($xmlstr));
+    }
+    catch (Exception $e) {
+      watchdog('islandora_scholar', 'The "@dsid" of "@pid" could not be parsed as XML.', array(
+        '@dsid' => 'MODS',
+        '@pid' => $object->id,
+      ), WATCHDOG_WARNING, l(t("manage object"), "islandora/object/{$object->id}/manage/datastreams"));
+      return " ";
+    }
     $xml->registerXPathNamespace('mods', 'http://www.loc.gov/mods/v3');
 
     $this->mods = $xml;

diff --git a/includes/upload.tab.inc b/includes/upload.tab.inc
@@ -43,7 +43,16 @@ function islandora_scholar_modify_form(array $form, array &$form_state, Abstract
 
   $mods_doc = new DOMDocument();
   $mods_doc->loadXML($mods_str);
-  $mods_xpath = new DOMXPath($mods_doc);
+  try {
+    $mods_xpath = new DOMXPath(islandora_sanitize_input_for_valid_xml($mods_doc));
+  }
+  catch (Exception $e) {
+    watchdog('islandora_scholar', 'The upload of "@dsid" of "@pid" could not be parsed as XML.', array(
+      '@dsid' => 'MODS',
+      '@pid' => $object->id,
+    ), WATCHDOG_WARNING, l(t("manage object"), "islandora/object/{$object->id}/manage/datastreams"));
+    return " ";
+  }
   $mods_xpath->registerNamespace('m', 'http://www.loc.gov/mods/v3');
   $usage = t('No usage statement');
   $mods_usages = $mods_xpath->query('//m:mods/m:accessCondition[@type="use and reproduction"]');

diff --git a/modules/citeproc/includes/converter.inc b/modules/citeproc/includes/converter.inc
@@ -106,7 +106,7 @@ function convert_mods_to_citeproc_jsons($mods_in) {
   }
   else {
     try {
-      $mods = simplexml_load_string($mods_in);
+      $mods = simplexml_load_string(islandora_sanitize_input_for_valid_xml($mods_in));
     }
     catch (Exception $e) {
       watchdog('citeproc', 'Got exception while parsing.  Message: !msg Errors: !error', array(

diff --git a/modules/islandora_google_scholar/islandora_google_scholar.module b/modules/islandora_google_scholar/islandora_google_scholar.module
@@ -24,10 +24,17 @@ function islandora_google_scholar_create_meta_tags($object) {
   else {
     $tags = array();
     $tags['citation_author'] = array();
-    // remove breaking control characters.
-    $object['MODS']->content = preg_replace('/[\x00-\x09\x0B\x0C\x0E-\x1F\x7F]/', '', $object['MODS']->content);
     $mods = $object['MODS']->content;
-    $mods_xml = new SimpleXMLElement($mods);
+    try {
+      $mods_xml = new SimpleXMLElement(islandora_sanitize_input_for_valid_xml($mods));
+    }
+    catch (Exception $e) {
+      watchdog('islandora_google_scholar', 'The "@dsid" of "@pid" could not be parsed as XML.', array(
+        '@dsid' => 'MODS',
+        '@pid' => $object->id,
+      ), WATCHDOG_WARNING, l(t("manage object"), "islandora/object/{$object->id}/manage/datastreams"));
+      return FALSE;
+    }
     $mods_xml->registerXPathNamespace('mods', 'http://www.loc.gov/mods/v3');
 
     $title_results = $mods_xml->xpath(variable_get('islandora_scholar_xpaths_title', '//mods:mods[1]/mods:titleInfo/mods:title'));