From 2ba32a2d4845de1e245a5d54ab1144a652047798 Mon Sep 17 00:00:00 2001
From: Gary Pendergast <pento@git.wordpress.org>
Date: Mon, 17 Oct 2016 23:53:20 +0000
Subject: [PATCH] Charset: Allow `_canonical_charset()` to handle mixed-case
 strings.

Add improved unit tests, and collect existing unit tests together.

Props pbearne.
Fixes #38337.



git-svn-id: https://develop.svn.wordpress.org/trunk@38809 602fd350-edb4-49c9-b593-d223f7449a82
---
 src/wp-includes/functions.php                 | 14 +--
 .../tests/functions/canonical-charset.php     | 88 +++++++++++++++++++
 2 files changed, 96 insertions(+), 6 deletions(-)
 create mode 100755 tests/phpunit/tests/functions/canonical-charset.php

diff --git a/src/wp-includes/functions.php b/src/wp-includes/functions.php
index a434ea5622..989772e765 100644
--- a/src/wp-includes/functions.php
+++ b/src/wp-includes/functions.php
@@ -555,7 +555,7 @@ function do_enclose( $content, $post_ID ) {
 	global $wpdb;
 
 	//TODO: Tidy this ghetto code up and make the debug code optional
-	include_once( ABSPATH . WPINC . '/class-IXR.php' ); 
+	include_once( ABSPATH . WPINC . '/class-IXR.php' );
 
 	$post_links = array();
 
@@ -5250,13 +5250,15 @@ function get_tag_regex( $tag ) {
  * @return string The canonical form of the charset.
  */
 function _canonical_charset( $charset ) {
-	if ( 'UTF-8' === $charset || 'utf-8' === $charset || 'utf8' === $charset ||
-		'UTF8' === $charset )
-		return 'UTF-8';
+	if ( 'utf-8' === strtolower( $charset ) || 'utf8' === strtolower( $charset) ) {
+
+		return 'UTF-8';
+	}
+
+	if ( 'iso-8859-1' === strtolower( $charset ) || 'iso8859-1' === strtolower( $charset ) ) {
 
-	if ( 'ISO-8859-1' === $charset || 'iso-8859-1' === $charset ||
-		'iso8859-1' === $charset || 'ISO8859-1' === $charset )
 		return 'ISO-8859-1';
+	}
 
 	return $charset;
 }
diff --git a/tests/phpunit/tests/functions/canonical-charset.php b/tests/phpunit/tests/functions/canonical-charset.php
new file mode 100755
index 0000000000..cc6008a6ad
--- /dev/null
+++ b/tests/phpunit/tests/functions/canonical-charset.php
@@ -0,0 +1,88 @@
+<?php
+
+/*
+ * Validate that's badly named charsets always return the correct format for UTF-8 and ISO-8859-1
+ *
+ *  @since 4.8.0
+ */
+
+class Tests_Functions_canonical_charset extends WP_UnitTestCase {
+
+	public function test_utf_8_lower() {
+		$this->assertEquals( 'UTF-8', _canonical_charset( 'utf-8' ) );
+	}
+
+	public function test_utf_8_upper() {
+		$this->assertEquals( 'UTF-8', _canonical_charset( 'UTF-8' ) );
+	}
+
+	public function test_utf_8_mixxed() {
+		$this->assertEquals( 'UTF-8', _canonical_charset( 'Utf-8' ) );
+	}
+
+	public function test_utf_8() {
+		$this->assertEquals( 'UTF-8', _canonical_charset( 'UTF8' ) );
+	}
+
+	public function test_iso_lower() {
+		$this->assertEquals( 'ISO-8859-1', _canonical_charset( 'iso-8859-1' ) );
+	}
+
+	public function test_iso_upper() {
+		$this->assertEquals( 'ISO-8859-1', _canonical_charset( 'ISO-8859-1' ) );
+	}
+
+	public function test_iso_mixxed() {
+		$this->assertEquals( 'ISO-8859-1', _canonical_charset( 'Iso8859-1' ) );
+	}
+
+	public function test_iso() {
+		$this->assertEquals( 'ISO-8859-1', _canonical_charset( 'ISO8859-1' ) );
+	}
+
+	public function test_random() {
+		$this->assertEquals( 'random', _canonical_charset( 'random' ) );
+	}
+
+	public function test_empty() {
+		$this->assertEquals( '', _canonical_charset( '' ) );
+	}
+
+	/**
+	 * @ticket 23688
+	 */
+	function test_update_option_blog_charset() {
+		$orig_blog_charset = get_option( 'blog_charset' );
+
+		update_option( 'blog_charset', 'utf8' );
+		$this->assertEquals( 'UTF-8', get_option( 'blog_charset') );
+
+		update_option( 'blog_charset', 'utf-8' );
+		$this->assertEquals( 'UTF-8', get_option( 'blog_charset') );
+
+		update_option( 'blog_charset', 'UTF8' );
+		$this->assertEquals( 'UTF-8', get_option( 'blog_charset') );
+
+		update_option( 'blog_charset', 'UTF-8' );
+		$this->assertEquals( 'UTF-8', get_option( 'blog_charset') );
+
+		update_option( 'blog_charset', 'ISO-8859-1' );
+		$this->assertEquals( 'ISO-8859-1', get_option( 'blog_charset') );
+
+		update_option( 'blog_charset', 'ISO8859-1' );
+		$this->assertEquals( 'ISO-8859-1', get_option( 'blog_charset') );
+
+		update_option( 'blog_charset', 'iso8859-1' );
+		$this->assertEquals( 'ISO-8859-1', get_option( 'blog_charset') );
+
+		update_option( 'blog_charset', 'iso-8859-1' );
+		$this->assertEquals( 'ISO-8859-1', get_option( 'blog_charset') );
+
+		// Arbitrary strings are passed through.
+		update_option( 'blog_charset', 'foobarbaz' );
+		$this->assertEquals( 'foobarbaz', get_option( 'blog_charset') );
+
+		update_option( 'blog_charset', $orig_blog_charset );
+	}
+
+}