aymdev / fregata
Database migration framework allowing data migration between different DBMS or database structures.
Installs: 158
Dependents: 2
Suggesters: 0
Security: 0
Stars: 23
Watchers: 3
Forks: 2
Open Issues: 3
Type:framework
Requires
- php: >=7.4
- doctrine/dbal: ^2.10
- hanneskod/classtools: ^1.2
- marcj/topsort: ^2.0
- symfony/config: ^4.4||^5.0||^6.0
- symfony/console: ^4.4||^5.0||^6.0
- symfony/dependency-injection: ^4.4||^5.0||^6.0
- symfony/string: ^5.0||^6.0
- symfony/yaml: ^4.4||^5.0||^6.0
Requires (Dev)
- dealerdirect/phpcodesniffer-composer-installer: ^0.7.1
- mikey179/vfsstream: ^1.6
- phpcompatibility/php-compatibility: ^9.3
- phpstan/phpstan: ^1.2
- phpstan/phpstan-phpunit: ^1.0
- phpstan/phpstan-strict-rules: ^1.1
- phpunit/phpunit: ^9.1
- sebastian/phpcpd: ^6.0
- squizlabs/php_codesniffer: ^3.6
README
Fregata is a data migration framework. You can use it to migrate any kind of data, but it has features to help you migrate between different DBMS or database structures.
Documentation:
Introduction
Fregata is a data migration framework. It can probably be compared to an ETL (Extract Transform - Load) tool.
You can use it to migrate data from files, databases, or anything you want, it is completely agnostic on this part (some of its test migrate data between PHP arrays). But note that it was initially targeting databases, providing a way to migrate data between different DBMS, even with a different structure. Some included features are specifically built for databases.
Why creating a framework for data migration ?
While database migrations might not be your everyday task, I encountered it multiple times on different projects. That's why I created Fregata to have a migration workflow I could reuse.
What are the use cases ?
Here are some example use cases (from experience):
- when you want to change from a DBMS to another
- when you want to sync your staging database with the production one (useful for CMS-based projects)
Setup
Installation
Install with Composer:
composer require aymdev/fregata
Configuration
Fregata expects you to have a config
and a cache
directory at your project root by default.
Kernel and service container
If you need to use a different directory structure than the default one, you can extend the
Fregata\Configuration\AbstractFregataKernel
class.
Then you will have to implement methods to specify your configuration and cache directory.
Important: your kernel full qualified class name must be
App\FregataKernel
.
The kernel holds a service container, built from Symfony's DependencyInjection component.
This means you can define your own services as you would do it in a Symfony application, in a
services.yaml
file in your configuration directory.
Here's a recommended minimal services.yaml to start your project:
services: _defaults: autowire: true App\: resource: '../src/'
YAML configuration
To configure Fregata itself, you will need a fregata.yaml
file in your configuration directory.
Example configuration file:
fregata: migrations: # define any name for your migration main_migration: # define custom options for your migrations options: custom_opt: 'opt_value' special_cfg: foo: bar # load migrators from a directory # use the %fregata.root_dir% parameter to define a relative path from the project root migrators_directory: '%fregata.root_dir%/src/MainMigration' # load individual migrators # can be combined with migrators_directory migrators: - App\MainMigration\FirstMigrator # load tasks to execute before or after the migrators tasks: before: - App\MainMigration\BeforeTask after: - App\MainMigration\AfterTask other_migration: # extend an other migration to inherit its options, tasks and migrators parent: main_migration # overwrite a part of the options options: custom_opt: 'another_value' # load additional migrators or tasks migrators: - App\OtherMigration\Migrator
Components
Migration Registry
The migration registry contains every defined migrations. You shouldn't have to interact with it.
Migration
A migration project holds the steps of a migration. For example, data migration from your production database to staging one. Each migration is created and saved into the registry based on your configuration. You don't need to instantiate migration objects by yourself.
Migrations contain tasks and migrators. When a migration is run, components are executed in the following order:
- before tasks
- migrators
- after tasks
Options
You may need to set specific configuration to your migration project, which could be used by tasks
or migrators.
With the options
key you can define your migration specific configuration, they will be accessible to
the components from the migration context.
Parent migration
When having multiple migrations for different environments, you probably want to avoid duplicating your
whole configuration.
You can extend a migration with the parent
key. The "child" migration will inherit the parent's
options, tasks and migrators. You can still add more tasks and migrators, and overwrite options.
Task
A task can be executed before or after migrators. They can be useful to bootstrap your migration (before tasks) or to clean temporary data at the end (after tasks):
use Fregata\Migration\TaskInterface; class MyTask implements TaskInterface { public function execute() : ?string { // perform some verifications, delete temporary data, ... return 'Optional result message'; } }
Migrator
The migrators are the main components of the framework. A single migrator holds 3 components:
- a puller
- a pusher
- an executor
It must return its components from getter methods by implementing
Fregata\Migration\Migrator\MigratorInterface
.
A migrator represents the migration of a data from a source to a target. For example, migrating data
from a MySQL table to a PostgreSQL one.
Puller
A puller is a migrator component responsible for pulling data from a source. It returns data and optionally the number of items to migrate:
use Doctrine\DBAL\Connection; use Fregata\Migration\Migrator\Component\PullerInterface; class Puller implements PullerInterface { private Connection $connection; public function __construct(Connection $connection) { $this->connection = $connection; } public function pull() { return $this->connection ->executeQuery('SELECT * FROM my_table') ->fetchAllAssociative(); } public function count() : ?int { return $this->connection ->executeQuery('SELECT COUNT(*) FROM my_table') ->fetchColumn(); } }
Pusher
A pusher gets item fetched by the puller 1 by 1 and has to push the data to a target:
use Doctrine\DBAL\Connection; use Fregata\Migration\Migrator\Component\PusherInterface; class Pusher implements PusherInterface { private Connection $connection; public function __construct(Connection $connection) { $this->connection = $connection; } /** * @return int number of items inserted */ public function push($data): int { return $this->connection->executeStatement( 'INSERT INTO my_table VALUES (:foo, :bar, :baz)', [ 'foo' => $data['foo'], 'bar' => some_function($data['bar']), 'baz' => 'default value', ] ); } }
Here $data
is a single item from the example puller returned value. The push()
method is called
multiple times.
The separation of pullers and pushers allow you to migrate between different sources: pull from
a file and push to a database, etc.
Executor
The executor is the component which plugs a puller with a pusher. A default one is provided
and should work for most cases: Fregata\Migration\Migrator\Component\Executor
.
Extend the default executor if you need a specific behaviour.
Tools
Migration Context
You can get some informations about the current migration by injecting the
Fregata\Migration\MigrationContext
service in a task or migration.
It provides:
- current migration object
- current migration name
- migration options
- parent migration name if applicable
Features
Dependent migrators
If your migrators need to be executed in a specific order you can define dependencies, and they will be sorted automatically:
use Fregata\Migration\Migrator\DependentMigratorInterface; class DependentMigrator implements DependentMigratorInterface { public function getDependencies() : array { return [ DependencyMigrator::class, ]; } // other migrator methods ... }
Here, DependencyMigrator
will be executed before DependentMigrator
.
Batch pulling
When a puller works with very large datasets you might want to pull the data by chunks:
use Doctrine\DBAL\Connection; use Fregata\Migration\Migrator\Component\BatchPullerInterface; class BatchPulling implements BatchPullerInterface { private Connection $connection; private ?int $count = null; public function __construct(Connection $connection) { $this->connection = $connection; } public function pull(): \Generator { $limit = 50; $offset = 0; while ($offset < $this->count()) { yield $this->connection ->executeQuery(sprintf('SELECT * FROM my_table LIMIT %d, %d', $offset, $limit)) ->fetchAllAssociative(); $offset += $limit; } } public function count() : ?int { if (null === $this->count) { $this->count = $this->connection ->executeQuery('SELECT COUNT(*) FROM my_table') ->fetchColumn(); } return $this->count; } }
Foreign Key migrations
One of the most complex parts of a database migration is about foreign keys. There are multiple steps to follow to perform a valid foreign key migration. This is done using Doctrine DBAL.
You must add 2 tasks to your migration:
- before task:
Fregata\Adapter\Doctrine\DBAL\ForeignKey\Task\ForeignKeyBeforeTask
- after task:
Fregata\Adapter\Doctrine\DBAL\ForeignKey\Task\ForeignKeyAfterTask
The before task will create temporary columns in your target database to keep the original referenced and
referencing columns. It may also change referencing columns to allow NULL
(only if you specify it).
The after task will set the real values in your original referencing columns and then drop the temporary
columns.
Then the migrators must provide the database connection and the list of foreign keys:
use Doctrine\DBAL\Connection; use Doctrine\DBAL\Schema\ForeignKeyConstraint; use Fregata\Adapter\Doctrine\DBAL\ForeignKey\ForeignKey; use Fregata\Adapter\Doctrine\DBAL\ForeignKey\Migrator\HasForeignKeysInterface; class ReferencingMigrator implements HasForeignKeysInterface { private Connection $connection; public function __construct(Connection $connection) { $this->connection = $connection; } public function getConnection() : Connection { return $this->connection; } /** * List the foreign keys constraints to keep * @return ForeignKey[] */ public function getForeignKeys() : array { $constraints = $this->connection->getSchemaManager()->listTableForeignKeys('my_table'); return array_map( function (ForeignKeyConstraint $constraint) { return new ForeignKey( $constraint, // DBAL constraint object 'target_referencing', // name of the referencing table ['fk'] // columns to change to allow NULL (will be set back to NOT NULL in the after task) ); }, $constraints ); } // other migrator methods ... }
The migrators are responsible for the data migration, this means you need to fill the temporary columns
with original primary/foreign key from the source database.
To get the name of a temporary column, require the CopyColumnHelper
service in your pusher:
use Doctrine\DBAL\Connection; use Fregata\Adapter\Doctrine\DBAL\ForeignKey\CopyColumnHelper; use Fregata\Migration\Migrator\Component\PusherInterface; class ReferencingForeignKeyPusher implements PusherInterface { private Connection $connection; private CopyColumnHelper $columnHelper; public function __construct(Connection $connection, CopyColumnHelper $columnHelper) { $this->connection = $connection; $this->columnHelper = $columnHelper; } /** * @return int number of items inserted */ public function push($data): int { return $this->connection->executeStatement( sprintf( 'INSERT INTO my_table (column, %s) VALUES (:value, :old_fk)', $this->columnHelper->localColumn('my_table', 'fk_column') ), [ 'value' => $data['value'], 'old_fk' => $data['fk_column'], ] ); } }
This example show the local (or referencing) side but this need to be done for the foreign (or
referenced) side too, using CopyColumnHelper::foreignColumn()
.
CLI usage
Fregata provides a simple program to run the migrations, you can launch it with:
php vendor/bin/fregata
List migrations
To list the migrations of your installation, run the migration:list
command:
> php vendor/bin/fregata migration:list
Registered migrations: 2
========================
main_migration
other_migration
Get details of a migration
To get information about a single migration, run the migration:show
command:
> php vendor/bin/fregata migration:show main_migration main_migration : 1 migrators ============================ --- --------------------------------- # Migrator Name --- --------------------------------- 0 App\MainMigration\FirstMigrator --- ---------------------------------
Execute a migration
And the most important one to run a migration: migration:execute
.
> php vendor/bin/fregata migration:execute main_migration Confirm execution of the "main_migration" migration ? (yes/no) [no]: > yes [OK] Starting "main_migration" migration: 1 migrators Before tasks: 1 =============== App\MainMigration\BeforeTask : OK Migrators: 1 ============ 0 - Executing "App\MainMigration\FirstMigrator" [3 items] : =========================================================== 3/3 [▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓] 100% After tasks: 1 ============== App\MainMigration\AfterTask : OK [OK] Migrated successfully !
Contributing
A Docker setup is available, providing a MySQL 5.7 service.
If you want to test the implementation of the framework (using a Composer
path repository), install it in a _implementation
directory at the root of the project, it is ignored by Git by default and will ensure you are using your
implementation autoloader.