1 files changed, 43 insertions, 0 deletions
diff --git a/www/wiki/extensions/SemanticMediaWiki/src/Utils/CharArmor.php b/www/wiki/extensions/SemanticMediaWiki/src/Utils/CharArmor.php
new file mode 100644
index 00000000..9a1e20c0
--- /dev/null
+++ b/www/wiki/extensions/SemanticMediaWiki/src/Utils/CharArmor.php
@@ -0,0 +1,43 @@
+<?php
+
+namespace SMW\Utils;
+
+/**
+ * @license GNU GPL v2+
+ * @since 3.0
+ *
+ * @author mwjames
+ */
+class CharArmor {
+
+	/**
+	 * Remove invisible control characters and unused code points (using a
+	 * negated character class to avoid removing spaces)
+	 *
+	 * @see http://www.regular-expressions.info/unicode.html#category
+	 * @since 3.0
+	 *
+	 * @param string $text
+	 *
+	 * @return text
+	 */
+	public static function removeControlChars( $text ) {
+		return preg_replace('/[^\PC\s]/u', '', $text );
+	}
+
+	/**
+	 * @since 3.0
+	 *
+	 * @param string $text
+	 *
+	 * @return text
+	 */
+	public static function removeSpecialChars( $text ) {
+		return str_replace(
+			[ '&shy;', '&lrm;', " ", " ", " " ],
+			[ '', '', ' ', ' ', ' ' ],
+			$text
+		);
+	}
+
+}