Merge pull request #29470 from nextcloud/fix/translit-php8

Avoid use of iconv to get rid of unicode

Author: blizzz

apps/user_ldap/lib/Access.php

@@ -1433,12 +1433,15 @@ public function sanitizeUsername($name) {
 			return $name;
 		}
 
-		// Transliteration to ASCII
-		$transliterated = @iconv('UTF-8', 'ASCII//TRANSLIT', $name);
-		if ($transliterated !== false) {
-			// depending on system config iconv can work or not
-			$name = $transliterated;
-		}
+		// Use htmlentities to get rid of accents
+		$name = htmlentities($name, ENT_NOQUOTES, 'UTF-8');
+
+		// Remove accents
+		$name = preg_replace('#&([A-Za-z])(?:acute|cedil|caron|circ|grave|orn|ring|slash|th|tilde|uml);#', '\1', $name);
+		// Remove ligatures
+		$name = preg_replace('#&([A-Za-z]{2})(?:lig);#', '\1', $name);
+		// Remove unknown leftover entities
+		$name = preg_replace('#&[^;]+;#', '', $name);
 
 		// Replacements
 		$name = str_replace(' ', '_', $name);

apps/user_ldap/tests/AccessTest.php

@@ -688,16 +688,14 @@ public function testFetchListOfGroupsKnown() {
 	}
 
 	public function intUsernameProvider() {
-		// system dependent :-/
-		$translitExpected = @iconv('UTF-8', 'ASCII//TRANSLIT', 'fränk') ? 'frank' : 'frnk';
-
 		return [
 			['alice', 'alice'],
 			['b/ob', 'bob'],
 			['charly🐬', 'charly'],
 			['debo rah', 'debo_rah'],
 			['epost@poste.test', 'epost@poste.test'],
-			['fränk', $translitExpected],
+			['fränk', 'frank'],
+			[' UPPÉR Case/[\]^`', 'UPPER_Case'],
 			[' gerda ', 'gerda'],
 			['🕱🐵🐘🐑', null],
 			[
@@ -731,9 +729,6 @@ public function groupIDCandidateProvider() {
 	 * @param $expected
 	 */
 	public function testSanitizeUsername($name, $expected) {
-		if ($name === 'fränk' && PHP_MAJOR_VERSION > 7) {
-			$this->markTestSkipped('Special chars do boom still on CI in php8');
-		}
 		if ($expected === null) {
 			$this->expectException(\InvalidArgumentException::class);
 		}