Create tables using utf8mb4 by default for MySQL

Signed-off-by: Jon Stovell <jonstovell@gmail.com>
SimpleMachines · Jan 30, 2025 · d7c25ff · d7c25ff
1 parent 9cce008
commit d7c25ff
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 11 deletions.
diff --git a/Sources/Db/APIs/MySQL.php b/Sources/Db/APIs/MySQL.php
@@ -1691,8 +1691,12 @@ public function create_table(string $table_name, array $columns, array $indexes
 
 		$table_query .= ') ENGINE=' . $parameters['engine'];
 
-		if (!empty($this->character_set) && $this->character_set == 'utf8') {
-			$table_query .= ' DEFAULT CHARSET=utf8 COLLATE=utf8_general_ci';
+		if (!empty($this->character_set) && str_starts_with($this->character_set, 'utf8')) {
+			if ($this->mb4) {
+				$table_query .= ' DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci';
+			} else {
+				$table_query .= ' DEFAULT CHARSET=utf8mb3 COLLATE=utf8mb3_general_ci';
+			}
 		}
 
 		// Create the table!

diff --git a/other/install.php b/other/install.php
@@ -935,10 +935,12 @@ function DatabaseSettings()
 		if (Db::$db->name != '' && !$needsDB) {
 			Db::$db->query(
 				'',
-				'CREATE DATABASE IF NOT EXISTS `' . Db::$db->name . '`',
+				'CREATE DATABASE IF NOT EXISTS {identifier:db_name} {raw:extra}',
 				[
 					'security_override' => true,
 					'db_error_skip' => true,
+					'db_name' => Db::$db->name,
+					'extra' => Config::$db_type === 'mysql' ? ' CHARACTER SET utf8mb4 COLLATE utf8mb4_0900_ai_ci' : '',
 				],
 				Db::$db->connection,
 			);
@@ -947,10 +949,12 @@ function DatabaseSettings()
 			if (!Db::$db->select(Db::$db->name, Db::$db->connection) && Db::$db->name != '') {
 				Db::$db->query(
 					'',
-					'CREATE DATABASE IF NOT EXISTS `' . Db::$db->prefix . Db::$db->name . '`',
+					'CREATE DATABASE IF NOT EXISTS {identifier:db_name} {raw:extra}',
 					[
 						'security_override' => true,
 						'db_error_skip' => true,
+						'db_name' => Db::$db->prefix . Db::$db->name,
+						'extra' => Config::$db_type === 'mysql' ? ' CHARACTER SET utf8mb4 COLLATE utf8mb4_0900_ai_ci' : '',
 					],
 					Db::$db->connection,
 				);
@@ -1235,8 +1239,8 @@ function DatabasePopulation()
 		$replaces['{$memory}'] = (!$has_innodb && in_array('MEMORY', $engines)) ? 'MEMORY' : $replaces['{$engine}'];
 
 		// UTF-8 is required.
-		$replaces['{$engine}'] .= ' DEFAULT CHARSET=utf8 COLLATE=utf8_general_ci';
-		$replaces['{$memory}'] .= ' DEFAULT CHARSET=utf8 COLLATE=utf8_general_ci';
+		$replaces['{$engine}'] .= ' DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci';
+		$replaces['{$memory}'] .= ' DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci';
 
 		// One last thing - if we don't have InnoDB, we can't do transactions...
 		if (!$has_innodb) {

diff --git a/other/upgrade.php b/other/upgrade.php
@@ -2392,10 +2392,10 @@ function ($errno, $errstr, $errfile, $errline) use ($support_js) {
 		},
 	);
 
-	// If we're on MySQL, set {db_collation}; this approach is used throughout upgrade_2-0_mysql.php to set new tables to utf8
+	// If we're on MySQL, set {db_collation}; this approach is used throughout upgrade_2-0_mysql.php to set new tables to utf8mb4
 	// Note it is expected to be in the format: ENGINE=InnoDB{$db_collation};
 	if (Config::$db_type == 'mysql') {
-		$db_collation = ' DEFAULT CHARSET=utf8 COLLATE=utf8_general_ci';
+		$db_collation = ' DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci';
 	} else {
 		$db_collation = '';
 	}
@@ -2410,7 +2410,7 @@ function ($errno, $errstr, $errfile, $errline) use ($support_js) {
 	$last_step = '';
 
 	// Make sure all newly created tables will have the proper characters set; this approach is used throughout upgrade_2-1_mysql.php
-	$lines = preg_replace('/\) ENGINE=(InnoDB|MyISAM);/', ') ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_general_ci;', $lines);
+	$lines = preg_replace('/\) ENGINE=(InnoDB|MyISAM);/', ') ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci;', $lines);
 
 	// Count the total number of steps within this file - for progress.
 	$file_steps = substr_count(implode('', $lines), '---#');
@@ -3301,8 +3301,7 @@ function ConvertUtf8()
 		'',
 		'SHOW INDEX
 		FROM {db_prefix}messages',
-		[
-		],
+		[],
 	);
 
 	$upcontext['dropping_index'] = false;