General: prevent antispambot() from breaking UTF8 strings.

Author: arnt

src/wp-includes/formatting.php

@@ -3027,7 +3027,9 @@ function antispambot( $email_address, $hex_encoding = 0 ) {
 	for ( $i = 0, $len = strlen( $email_address ); $i < $len; $i++ ) {
 		$j = rand( 0, 1 + $hex_encoding );
 
-		if ( 0 === $j ) {
+		if ( ord( $email_address[ $i ] ) > 127 ) {
+			$email_no_spam_address .= $email_address[ $i ];
+		} elseif ( 0 === $j ) {
 			$email_no_spam_address .= '&#' . ord( $email_address[ $i ] ) . ';';
 		} elseif ( 1 === $j ) {
 			$email_no_spam_address .= $email_address[ $i ];

tests/phpunit/tests/formatting/antispambot.php

@@ -0,0 +1,26 @@
+<?php
+
+/**
+ * @group formatting
+ *
+ * @covers ::antispambot
+ */
+class Tests_Formatting_antispambot extends WP_UnitTestCase {
+	public function test_returns_valid_utf8() {
+		$data = array(
+			'[email protected]',
+			'[email protected]',
+			'[email protected]',
+			'[email protected]',
+			'[email protected]',
+			'[email protected]',
+			'info@grå.org',
+			'grå@grå.org',
+			"gr\u{0061}\u{030a}blå@grå.org",
+			'[email protected]',
+		);
+		foreach ( $data as $datum ) {
+			$this->assertTrue( seems_utf8( antispambot( $datum ) ) );
+		}
+	}
+}