Merge pull request #35 from dof-dss/uilc

Ensuring body links aren't changed by the link checker

Author: soda736

unity_internal_link_checker/unity_internal_link_checker.module

@@ -39,7 +39,7 @@ function unity_internal_link_checker_entity_presave(EntityInterface $entity) {
                 preg_match('|https://(.*)|', $original_link, $matches2)) {
                 // This is an absolute URL does it match one of the domain names specified
                 // at /admin/config/unity_internal_link_checker/link_checker_form ?
-                $body = _convert_absolute_link($body, $original_link);
+                $body = unity_internal_link_checker__convert_absolute_link($body, $original_link);
               }
             }
           }
@@ -53,33 +53,32 @@ function unity_internal_link_checker_entity_presave(EntityInterface $entity) {
 /**
  * Utility function to convert internal link to relative.
  */
-function _convert_absolute_link(&$body_text, $original_link) {
+function unity_internal_link_checker__convert_absolute_link(&$body_text, $original_link) {
   // Get list of urls to replace
   // (these have been added at
   // /admin/config/unity_internal_link_checker/link_checker_form )
   $config = \Drupal::config('unity_internal_link_checker.linksettings');
-  $replace_url_list = _urls_to_replace() . $config->get('site_url_list');
+  $replace_url_list = unity_internal_link_checker__urls_to_replace() . $config->get('site_url_list');
   if (strlen($replace_url_list) == 0) {
     \Drupal::messenger()
       ->addMessage(t("Please set some URLs at /admin/config/unity_internal_link_checker/link_checker_form"), "error");
     return $body_text;
   }
   $replace_url_list = explode(PHP_EOL, $replace_url_list);
+  // Remove empty array items.
+  $replace_url_list = array_filter($replace_url_list);
   foreach ($replace_url_list as $replace_url) {
     // Make sure url is 'clean'.
     $replace_url = str_replace(["\n", "\t", "\r"], '', $replace_url);
     // Make sure that there is a trailing slash.
     if (!preg_match('|\/$|', $replace_url)) {
       $replace_url .= '/';
     }
-    // Parse the original link as we need the path from it to replace in the body text.
-    $original_link_parsed = parse_url($original_link);
     // Does the start of this link match one of the
     // urls that we are looking for ?
     if (str_contains($original_link, $replace_url)) {
-      // We have a match, change this absolute link
-      // to a relative link.
-      $body_text = preg_replace('~' . $original_link . '~', $original_link_parsed['path'], $body_text, '1');
+      // We have a match, change this absolute link to a relative link. Only target href values.
+      $body_text = preg_replace('~href\=[\'"]+' . $replace_url . '~', 'href="/', $body_text, 1);
     }
   }
   return $body_text;
@@ -88,7 +87,7 @@ function _convert_absolute_link(&$body_text, $original_link) {
 /**
  * Utility function to add all domains and extensions for the current site.
  */
-function _urls_to_replace(): string {
+function unity_internal_link_checker__urls_to_replace(): string {
   // Get the host name of the site and remove the domain extension.
   $host = \Drupal::request()->getHost();
   if (preg_match('/\S*?\.(uk|com|net|eu)/', $host, $matches)) {