Merge ab899e0 into 47c5ee2

CHANGELOG.md

@@ -1,3 +1,11 @@
+## [Unreleased]
+
+### Changed
+
+* Language codes: more strict normalization is added
+  to ensure that language codes more closely conform to
+  [RFC5646](https://tools.ietf.org/html/rfc5646).
+
 ## v2.1.0 (2019-11-24)
 
 ### Changed

src/Intuition.php

@@ -356,11 +356,14 @@ protected function normalizeDomain( string $domain ) : string {
 	}
 
 	/**
+	 * Normalize a language string (which may be user-supplied and unsafe) to be an all-lowercase
+	 * hyphen-separated string that approximately conforms to IETF's 'Tags for Identifying Languages'.
+	 * @link https://tools.ietf.org/html/rfc5646
 	 * @param string $lang
 	 * @return string
 	 */
 	protected function normalizeLang( $lang ) {
-		$lang = strtolower( str_replace( '_', '-', $lang ) );
+		$lang = trim( preg_replace( '/[^a-z0-9-]+/', '-', strtolower( $lang ) ), '-' );
 		if ( isset( $this->deprecatedLangCodes[$lang] ) ) {
 			return $this->deprecatedLangCodes[$lang];
 		}

tests/phpunit/IntuitionTest.php

@@ -73,7 +73,13 @@ public function testSetLang() {
 			$this->i18n->msg( 'test-value', 'test-domain' ),
 			'Change default lang'
 		);
+		// Language must be a non-empty string.
 		$this->assertFalse( $this->i18n->setLang( 42 ), 'Bad value' );
+		// Language is normalized to lowercase and hyphen-separated.
+		$this->i18n->setLang( 'en_AU' );
+		$this->assertEquals( 'en-au', $this->i18n->getLang() );
+		$this->i18n->setLang( '"bad" lang <string>' );
+		$this->assertEquals( 'bad-lang-string', $this->i18n->getLang() );
 	}
 
 	/**