Add config option to update charset of mysql to utf8mb4

* fully optional
* requires additional options set in the database

9 files changed, 66 insertions, 9 deletions

diff --git a/config/config.sample.php b/config/config.sample.php
index 3aa0f353c59..df1e2d16fc0 100644
--- a/config/config.sample.php
+++ b/config/config.sample.php
@@ -129,6 +129,7 @@ $CONFIG = array(
  */
 'dbtableprefix' => '',
 
+
 /**
  * Indicates whether the Nextcloud instance was installed successfully; ``true``
  * indicates a successful installation, and ``false`` indicates an unsuccessful
@@ -1080,6 +1081,34 @@ $CONFIG = array(
 'sqlite.journal_mode' => 'DELETE',
 
 /**
+ * If this setting is set to true MySQL can handle 4 byte characters instead of
+ * 3 byte characters
+ *
+ * MySQL requires a special setup for longer indexes (> 767 bytes) which are
+ * needed:
+ *
+ * [mysqld]
+ * innodb_large_prefix=true
+ * innodb_file_format=barracuda
+ * innodb_file_per_table=true
+ *
+ * Tables will be created with
+ *  * character set: utf8mb4
+ *  * collation:     utf8mb4_bin
+ *  * row_format:    compressed
+ *
+ * See:
+ * https://dev.mysql.com/doc/refman/5.7/en/charset-unicode-utf8mb4.html
+ * https://dev.mysql.com/doc/refman/5.7/en/innodb-parameters.html#sysvar_innodb_large_prefix
+ * https://mariadb.com/kb/en/mariadb/xtradbinnodb-server-system-variables/#innodb_large_prefix
+ * http://www.tocker.ca/2013/10/31/benchmarking-innodb-page-compression-performance.html
+ * http://mechanics.flite.com/blog/2014/07/29/using-innodb-large-prefix-to-avoid-error-1071/
+ *
+ * WARNING: EXPERIMENTAL
+ */
+'mysql.utf8mb4' => false,
+
+/**
  * Database types that are supported for installation.
  *
  * Available:
diff --git a/core/register_command.php b/core/register_command.php
index a6da3cbd899..fa85aea8956 100644
--- a/core/register_command.php
+++ b/core/register_command.php
@@ -83,7 +83,7 @@ if (\OC::$server->getConfig()->getSystemValue('installed', false)) {
 	$application->add(new OC\Core\Command\Config\System\SetConfig(\OC::$server->getSystemConfig()));
 
 	$application->add(new OC\Core\Command\Db\GenerateChangeScript());
-	$application->add(new OC\Core\Command\Db\ConvertType(\OC::$server->getConfig(), new \OC\DB\ConnectionFactory()));
+	$application->add(new OC\Core\Command\Db\ConvertType(\OC::$server->getConfig(), new \OC\DB\ConnectionFactory(\OC::$server->getSystemConfig())));
 
 	$application->add(new OC\Core\Command\Encryption\Disable(\OC::$server->getConfig()));
 	$application->add(new OC\Core\Command\Encryption\Enable(\OC::$server->getConfig(), \OC::$server->getEncryptionManager()));
diff --git a/lib/private/DB/AdapterMySQL.php b/lib/private/DB/AdapterMySQL.php
index 3e2fceda8db..0c0c6b31021 100644
--- a/lib/private/DB/AdapterMySQL.php
+++ b/lib/private/DB/AdapterMySQL.php
@@ -39,7 +39,8 @@ class AdapterMySQL extends Adapter {
 	}
 
 	public function fixupStatement($statement) {
-		$statement = str_replace(' ILIKE ', ' COLLATE utf8_general_ci LIKE ', $statement);
+		$characterSet = \OC::$server->getConfig()->getSystemValue('mysql.utf8mb4', false) ? 'utf8mb4' : 'utf8';
+		$statement = str_replace(' ILIKE ', ' COLLATE ' . $characterSet . '_general_ci LIKE ', $statement);
 		return $statement;
 	}
 }
diff --git a/lib/private/DB/ConnectionFactory.php b/lib/private/DB/ConnectionFactory.php
index b2c356edef7..a7aae32f670 100644
--- a/lib/private/DB/ConnectionFactory.php
+++ b/lib/private/DB/ConnectionFactory.php
@@ -28,6 +28,7 @@ namespace OC\DB;
 use Doctrine\DBAL\Event\Listeners\OracleSessionInit;
 use Doctrine\DBAL\Event\Listeners\SQLSessionInit;
 use Doctrine\DBAL\Event\Listeners\MysqlSessionInit;
+use OC\SystemConfig;
 
 /**
 * Takes care of creating and configuring Doctrine connections.
@@ -64,6 +65,12 @@ class ConnectionFactory {
 		),
 	);
 
+	public function __construct(SystemConfig $systemConfig) {
+		if($systemConfig->getValue('mysql.utf8mb4', false)) {
+			$defaultConnectionParams['mysql']['charset'] = 'utf8mb4';
+		}
+	}
+
 	/**
 	* @brief Get default connection parameters for a given DBMS.
 	* @param string $type DBMS type
diff --git a/lib/private/DB/MDB2SchemaReader.php b/lib/private/DB/MDB2SchemaReader.php
index 3f183c9723a..c198bb31e00 100644
--- a/lib/private/DB/MDB2SchemaReader.php
+++ b/lib/private/DB/MDB2SchemaReader.php
@@ -33,6 +33,7 @@ namespace OC\DB;
 
 use Doctrine\DBAL\Platforms\AbstractPlatform;
 use Doctrine\DBAL\Schema\SchemaConfig;
+use Doctrine\DBAL\Platforms\MySqlPlatform;
 use OCP\IConfig;
 
 class MDB2SchemaReader {
@@ -54,12 +55,16 @@ class MDB2SchemaReader {
 	/** @var \Doctrine\DBAL\Schema\SchemaConfig $schemaConfig */
 	protected $schemaConfig;
 
+	/** @var IConfig */
+	protected $config;
+
 	/**
 	 * @param \OCP\IConfig $config
 	 * @param \Doctrine\DBAL\Platforms\AbstractPlatform $platform
 	 */
 	public function __construct(IConfig $config, AbstractPlatform $platform) {
 		$this->platform = $platform;
+		$this->config = $config;
 		$this->DBNAME = $config->getSystemValue('dbname', 'owncloud');
 		$this->DBTABLEPREFIX = $config->getSystemValue('dbtableprefix', 'oc_');
 
@@ -118,8 +123,15 @@ class MDB2SchemaReader {
 					$name = str_replace('*dbprefix*', $this->DBTABLEPREFIX, $name);
 					$name = $this->platform->quoteIdentifier($name);
 					$table = $schema->createTable($name);
-					$table->addOption('collate', 'utf8_bin');
 					$table->setSchemaConfig($this->schemaConfig);
+
+					if($this->platform instanceof MySqlPlatform && $this->config->getSystemValue('mysql.utf8mb4', false)) {
+						$table->addOption('charset', 'utf8mb4');
+						$table->addOption('collate', 'utf8mb4_bin');
+						$table->addOption('row_format', 'compressed');
+					} else {
+						$table->addOption('collate', 'utf8_bin');
+					}
 					break;
 				case 'create':
 				case 'overwrite':
diff --git a/lib/private/DB/MDB2SchemaWriter.php b/lib/private/DB/MDB2SchemaWriter.php
index 26e32b036fe..7664b4359ab 100644
--- a/lib/private/DB/MDB2SchemaWriter.php
+++ b/lib/private/DB/MDB2SchemaWriter.php
@@ -43,7 +43,11 @@ class MDB2SchemaWriter {
 		$xml->addChild('name', $config->getSystemValue('dbname', 'owncloud'));
 		$xml->addChild('create', 'true');
 		$xml->addChild('overwrite', 'false');
-		$xml->addChild('charset', 'utf8');
+		if($config->getSystemValue('dbtype', 'sqlite') === 'mysql' && $config->getSystemValue('mysql.utf8mb4', false)) {
+			$xml->addChild('charset', 'utf8mb4');
+		} else {
+			$xml->addChild('charset', 'utf8');
+		}
 
 		// FIX ME: bloody work around
 		if ($config->getSystemValue('dbtype', 'sqlite') === 'oci') {
diff --git a/lib/private/Repair/Collation.php b/lib/private/Repair/Collation.php
index 74c4ca2e6ac..c19b8eea5ec 100644
--- a/lib/private/Repair/Collation.php
+++ b/lib/private/Repair/Collation.php
@@ -61,10 +61,12 @@ class Collation implements IRepairStep {
 			return;
 		}
 
+		$characterSet = $this->config->getSystemValue('mysql.utf8mb4', false) ? 'utf8mb4' : 'utf8';
+
 		$tables = $this->getAllNonUTF8BinTables($this->connection);
 		foreach ($tables as $table) {
 			$output->info("Change collation for $table ...");
-			$query = $this->connection->prepare('ALTER TABLE `' . $table . '` CONVERT TO CHARACTER SET utf8 COLLATE utf8_bin;');
+			$query = $this->connection->prepare('ALTER TABLE `' . $table . '` CONVERT TO CHARACTER SET ' . $characterSet . ' COLLATE ' . $characterSet . '_bin;');
 			$query->execute();
 		}
 	}
@@ -75,11 +77,12 @@ class Collation implements IRepairStep {
 	 */
 	protected function getAllNonUTF8BinTables($connection) {
 		$dbName = $this->config->getSystemValue("dbname");
+		$characterSet = $this->config->getSystemValue('mysql.utf8mb4', false) ? 'utf8mb4' : 'utf8';
 		$rows = $connection->fetchAll(
 			"SELECT DISTINCT(TABLE_NAME) AS `table`" .
 			"	FROM INFORMATION_SCHEMA . COLUMNS" .
 			"	WHERE TABLE_SCHEMA = ?" .
-			"	AND (COLLATION_NAME <> 'utf8_bin' OR CHARACTER_SET_NAME <> 'utf8')" .
+			"	AND (COLLATION_NAME <> '" . $characterSet . "_bin' OR CHARACTER_SET_NAME <> '" . $characterSet . "')" .
 			"	AND TABLE_NAME LIKE \"*PREFIX*%\"",
 			array($dbName)
 		);
diff --git a/lib/private/Server.php b/lib/private/Server.php
index 291714b23d1..063ff4a3d3b 100644
--- a/lib/private/Server.php
+++ b/lib/private/Server.php
@@ -407,8 +407,8 @@ class Server extends ServerContainer implements IServerContainer {
 			return new CredentialsManager($c->getCrypto(), $c->getDatabaseConnection());
 		});
 		$this->registerService('DatabaseConnection', function (Server $c) {
-			$factory = new \OC\DB\ConnectionFactory();
 			$systemConfig = $c->getSystemConfig();
+			$factory = new \OC\DB\ConnectionFactory($systemConfig);
 			$type = $systemConfig->getValue('dbtype', 'sqlite');
 			if (!$factory->isValidType($type)) {
 				throw new \OC\DatabaseException('Invalid database type');
diff --git a/lib/private/Setup/MySQL.php b/lib/private/Setup/MySQL.php
index 4ad6926c2d7..c022616d8b3 100644
--- a/lib/private/Setup/MySQL.php
+++ b/lib/private/Setup/MySQL.php
@@ -58,8 +58,9 @@ class MySQL extends AbstractDatabase {
 		try{
 			$name = $this->dbName;
 			$user = $this->dbUser;
-			//we can't use OC_BD functions here because we need to connect as the administrative user.
-			$query = "CREATE DATABASE IF NOT EXISTS `$name` CHARACTER SET utf8 COLLATE utf8_bin;";
+			//we can't use OC_DB functions here because we need to connect as the administrative user.
+			$characterSet = \OC::$server->getSystemConfig()->getValue('mysql.utf8mb4', false) ? 'utf8mb4' : 'utf8';
+			$query = "CREATE DATABASE IF NOT EXISTS `$name` CHARACTER SET $characterSet COLLATE ${characterSet}_bin;";
 			$connection->executeUpdate($query);
 		} catch (\Exception $ex) {
 			$this->logger->error('Database creation failed: {error}', [