From 2d5f6833fb5054937752c64f0c5bf64ddb91c2c1 Mon Sep 17 00:00:00 2001
From: Sergey Biryukov <sergeybiryukov@git.wordpress.org>
Date: Mon, 28 Feb 2022 15:40:15 +0000
Subject: [PATCH] I18N: Add a `$locale` parameter for `remove_accents()`.

This highlights the fact that `remove_accents()` is locale-aware and makes it easier to utilize the function with different locales without having to use `switch_to_locale()` or the `locale` filter.

Additionally, this commit relaxes the check for character replacements in German locales to include formal and informal variants of any `de_*` locale, even if WordPress does not have a native translation for some of them yet.

Props malthert, johnbillion, knutsp, ocean90, SergeyBiryukov.
Fixes #54415.

git-svn-id: https://develop.svn.wordpress.org/trunk@52809 602fd350-edb4-49c9-b593-d223f7449a82
---
 src/wp-includes/formatting.php                | 19 +++++++--
 .../tests/formatting/removeAccents.php        | 42 ++-----------------
 2 files changed, 19 insertions(+), 42 deletions(-)

diff --git a/src/wp-includes/formatting.php b/src/wp-includes/formatting.php
index 41c6afbec2..ff90ebd022 100644
--- a/src/wp-includes/formatting.php
+++ b/src/wp-includes/formatting.php
@@ -1583,11 +1583,15 @@ function utf8_uri_encode( $utf8_string, $length = 0, $encode_ascii_characters =
  * @since 4.7.0 Added locale support for `sr_RS`.
  * @since 4.8.0 Added locale support for `bs_BA`.
  * @since 5.7.0 Added locale support for `de_AT`.
+ * @since 6.0.0 Added the `$locale` parameter.
  *
- * @param string $string Text that might have accent characters
+ * @param string $string Text that might have accent characters.
+ * @param string $locale Optional. The locale to use for accent removal. Some character
+ *                       replacements depend on the locale being used (e.g. 'de_DE').
+ *                       Defaults to the current locale.
  * @return string Filtered string with replaced "nice" characters.
  */
-function remove_accents( $string ) {
+function remove_accents( $string, $locale = '' ) {
 	if ( ! preg_match( '/[\x80-\xff]/', $string ) ) {
 		return $string;
 	}
@@ -1923,9 +1927,16 @@ function remove_accents( $string ) {
 		);
 
 		// Used for locale-specific rules.
-		$locale = get_locale();
+		if ( empty( $locale ) ) {
+			$locale = get_locale();
+		}
 
-		if ( in_array( $locale, array( 'de_DE', 'de_DE_formal', 'de_CH', 'de_CH_informal', 'de_AT' ), true ) ) {
+		/*
+		 * German has various locales (de_DE, de_CH, de_AT, ...) with formal and informal variants.
+		 * There is no 3-letter locale like 'def', so checking for 'de' instead of 'de_' is safe,
+		 * since 'de' itself would be a valid locale too).
+		 */
+		if ( str_starts_with( $locale, 'de' ) ) {
 			$chars['Ä'] = 'Ae';
 			$chars['ä'] = 'ae';
 			$chars['Ö'] = 'Oe';
diff --git a/tests/phpunit/tests/formatting/removeAccents.php b/tests/phpunit/tests/formatting/removeAccents.php
index 19fd50586b..2ebea564a5 100644
--- a/tests/phpunit/tests/formatting/removeAccents.php
+++ b/tests/phpunit/tests/formatting/removeAccents.php
@@ -80,67 +80,33 @@ class Tests_Formatting_RemoveAccents extends WP_UnitTestCase {
 		$this->assertSame( 'aaeiouuAEIOUU', remove_accents( 'aɑeiouüAEIOUÜ' ) );
 	}
 
-	public function remove_accents_germanic_umlauts_cb() {
-		return 'de_DE';
-	}
-
 	/**
 	 * @ticket 3782
 	 */
 	public function test_remove_accents_germanic_umlauts() {
-		add_filter( 'locale', array( $this, 'remove_accents_germanic_umlauts_cb' ) );
-
-		$this->assertSame( 'AeOeUeaeoeuess', remove_accents( 'ÄÖÜäöüß' ) );
-
-		remove_filter( 'locale', array( $this, 'remove_accents_germanic_umlauts_cb' ) );
-	}
-
-	public function set_locale_to_danish() {
-		return 'da_DK';
+		$this->assertSame( 'AeOeUeaeoeuess', remove_accents( 'ÄÖÜäöüß', 'de_DE' ) );
 	}
 
 	/**
 	 * @ticket 23907
 	 */
 	public function test_remove_danish_accents() {
-		add_filter( 'locale', array( $this, 'set_locale_to_danish' ) );
-
-		$this->assertSame( 'AeOeAaaeoeaa', remove_accents( 'ÆØÅæøå' ) );
-
-		remove_filter( 'locale', array( $this, 'set_locale_to_danish' ) );
-	}
-
-	public function set_locale_to_catalan() {
-		return 'ca';
+		$this->assertSame( 'AeOeAaaeoeaa', remove_accents( 'ÆØÅæøå', 'da_DK' ) );
 	}
 
 	/**
 	 * @ticket 37086
 	 */
 	public function test_remove_catalan_middot() {
-		add_filter( 'locale', array( $this, 'set_locale_to_catalan' ) );
-
-		$this->assertSame( 'allallalla', remove_accents( 'al·lallaŀla' ) );
-
-		remove_filter( 'locale', array( $this, 'set_locale_to_catalan' ) );
-
+		$this->assertSame( 'allallalla', remove_accents( 'al·lallaŀla', 'ca' ) );
 		$this->assertSame( 'al·lallalla', remove_accents( 'al·lallaŀla' ) );
 	}
 
-	public function set_locale_to_serbian() {
-		return 'sr_RS';
-	}
-
 	/**
 	 * @ticket 38078
 	 */
 	public function test_transcribe_serbian_crossed_d() {
-		add_filter( 'locale', array( $this, 'set_locale_to_serbian' ) );
-
-		$this->assertSame( 'DJdj', remove_accents( 'Đđ' ) );
-
-		remove_filter( 'locale', array( $this, 'set_locale_to_serbian' ) );
-
+		$this->assertSame( 'DJdj', remove_accents( 'Đđ', 'sr_RS' ) );
 		$this->assertSame( 'Dd', remove_accents( 'Đđ' ) );
 	}
 }