Integration of squirrelphp/queries into Symfony through service tags and bundle configuration.
composer require squirrelphp/queries-bundle
Enable the bundle in your project by adding Squirrel\QueriesBundle\SquirrelQueriesBundle
to the list of bundles (usually in config/bundles.php
).
Create a Symfony service for each of your Doctrine DBAL connections and tag it with squirrel.connection, for example like this:
services:
database_connection:
class: Doctrine\DBAL\Connection
factory: Doctrine\DBAL\DriverManager::getConnection
arguments:
$params:
driver: pdo_mysql
host: "%database_host%"
port: "%database_port%"
dbname: "%database_name%"
user: "%database_user%"
password: "%database_password%"
charset: UTF8
tags:
- { name: squirrel.connection, connectionName: somename, connectionType: mysql, isDefault: true }
You can use any DBAL connection settings, and the service name (database_connection
in this case) is irrelevant. For the tag, just make sure:
- to use one of the three supported database types as
connectionType
:mysql
for MySQL/MariaDB,pgsql
for PostgreSQL,sqlite
for SQLite - set a unique
connectionName
for each tag entry
If you set isDefault
to true, that connection will be registered as Squirrel\Queries\DBInterface
which you can then use as a type hint in your services. Only one connection can be the default!
If you have multiple connections and need to reference them in your service definitions, you can specifically inject them through the connectionName
- just prefix it with squirrel.connection.
to get the correct registered service name. So for a connectionName of mysql_remote
, the service in Symfony would be called squirrel.connection.mysql_remote
.
- For all connections,
PDO::ATTR_EMULATE_PREPARES
is set to false, so real query and values separation is enabled instead of emulating it via PDO. You should not notice this in any way, even in terms of performance: it was tested, and when script and database are running in the same network there is no measureable difference. Your script and database would need to be apart by some distance for any possible effect to manifest. - For MySQL,
PDO::MYSQL_ATTR_FOUND_ROWS
is set to true, meaning the "affected rows" reported for UPDATE queries are the found rows in the database, even if nothing changed by executing the UPDATE. By default with MySQL you get the "changed" rows, which is a behavior no other database has or even supports, so it is not a good behavior to rely on. - For MySQL,
PDO::MYSQL_ATTR_MULTI_STATEMENTS
is set to false, so multiple statements in one query are not possible. When using Squirrel Queries regularly this should make no difference whatsoever (as the library only does one query at a time), but if you do something custom with the Doctrine connection this makes sure you cannot shoot yourself in the foot, as multiple statements per query were a source of security exploits in the past and have little real world relevance.
By default, this bundle creates DBInterface services with an implementation layer and an error handling layer (see squirrelphp/queries for details).
If you want to add additional layers to decorate DBInterface, create a service for each additional layer and tag it with squirrel.layer
. Make sure the service implements Squirrel\Queries\DBRawInterface
and to add the trait Squirrel\Queries\DBPassToLowerLayerTrait
in the service. Define a priority
for the tag and set it to below zero if you want to inject it between the implementation and the error handler, or above zero if it should be above the error handler.
Example for the service definition of a logger which logs deadlocks / connection timeouts before the error handler automatically retries the query/transaction:
services:
Squirrel\QueriesBundle\Examples\SQLLogTemporaryFailuresListener:
tags:
- { name: squirrel.layer, priority: -250 }
Because the priority is below zero it is a layer beneath the error handler. You can find a possible implementation in the examples directory.
When using Symfony Profiler this library offers similar integration like the DoctrineBundle automatically - so you can check what queries were sent to the database and how long they took.