Consilidate methods for preparing drupal titles.

CHANGELOG.md

-migration_tools 7.x-1.x **-**-****
+migration_tools 7.x-1.x  ** - ** - ****
 -----------------------------------------------
+
+* Consolidate methods for cleaning node titles.
 * Add CHANGELOG.md
 * Moved URL and redirect related methods into UrlTools.inc
 

examples/classes/MTBaseMigration.inc

@@ -148,11 +148,6 @@ abstract class MTBaseMigration extends Migration {
   public function prepareRow($row) {
     parent::prepareRow($row);
 
-    // Prepare title property by trimming to 255 chars and fixing encoding.
-    if (!empty($row->title)) {
-      $row->title = mb_strimwidth($row->title, 0, 255, "...");
-      $row->title = html_entity_decode($row->title, ENT_QUOTES, 'UTF-8');
-    }
   }
 
   /**

includes/StringCleanUp.inc

@@ -393,6 +393,21 @@ class StringCleanUp {
     return $string;
   }
 
+  /**
+   * Prepares a string to be a Drupal node title.
+   *
+   * @param string $string
+   *   The string to be cleaned.
+   *
+   * @return string
+   *   The string html decoded, capped and trimmed.
+   */
+  public static function cleanTitle($string) {
+    $string = ObtainTitle::cleanString($string);
+    $string = ObtainTitle::truncateString($string);
+    return $string;
+  }
+
   /**
    * Callback helper.
    */

includes/obtainers/ObtainTitle.php

@@ -26,8 +26,11 @@ class ObtainTitle extends ObtainHtml {
   protected function truncateString($string) {
     $split = $this->truncateThisWithoutHTML($string, 255, 2);
 
-    // @todo Add debugging to display $split['remaining'].
-    // $this->setTextDiscarded($split['remaining']);
+    // If something got trimmed off, message it.
+    if (!empty($split['remaining'])) {
+      $message = "The title was shortened and lost: @remainder";
+      MigrationMessage::makeMessage($message, array('@remainder' => $split['remaining']), WATCHDOG_ERROR, 2);
+    }
 
     return $split['truncated'];
   }
@@ -51,7 +54,7 @@ class ObtainTitle extends ObtainHtml {
     return $this->findSubBannerAttr('title');
   }
 
-/**
+  /**
    * Grab method to find the content sub-banner attribute.
    * @return string
    *   The text found.
@@ -103,7 +106,7 @@ class ObtainTitle extends ObtainHtml {
     $text = str_ireplace($break_tags, ' ', $text);
     $text = strip_tags($text);
     // Titles can not have html entities.
-    $text = html_entity_decode($text, ENT_COMPAT, 'UTF-8');
+    $text = html_entity_decode($text, ENT_QUOTES, 'UTF-8');
 
     // There are also numeric html special chars, let's change those.
     module_load_include('inc', 'migration_tools', 'includes/migration_tools');