spqr/vendor/robmorgan/phinx/docs/en/configuration.rst

535 lines
17 KiB
ReStructuredText
Raw Permalink Normal View History

2024-11-05 12:10:06 +08:00
.. index::
single: Configuration
Configuration
=============
When you initialize your project using the :doc:`Init Command<commands>`, Phinx
creates a default file in the root of your project directory. By default, this
file uses the YAML data serialization format, but you can use the ``--format``
command line option to specify either ``yaml``, ``yml``, ``json``, or ``php``.
If a ``--configuration`` command line option is given, Phinx will load the
specified file. Otherwise, it will attempt to find ``phinx.php``, ``phinx.json``,
``phinx.yml``, or ``phinx.yaml`` and load the first file found. See the
:doc:`Commands <commands>` chapter for more information.
.. warning::
Remember to store the configuration file outside of a publicly accessible
directory on your webserver. This file contains your database credentials
and may be accidentally served as plain text.
Note that while JSON and YAML files are *parsed*, the PHP file is *included*.
This means that:
* It must `return` an array of configuration items.
* The variable scope is local, i.e. you would need to explicitly declare
any global variables your initialization file reads or modifies.
* Its standard output is suppressed.
* Unlike with JSON and YAML, it is possible to omit environment connection details
and instead specify ``connection`` which must contain an initialized PDO instance.
This is useful when you want your migrations to interact with your application
and/or share the same connection. However remember to also pass the database name
as Phinx cannot infer this from the PDO connection.
.. code-block:: php
$app = require 'app/phinx.php';
$pdo = $app->getDatabase()->getPdo();
return [
'environments' => [
'default_environment' => 'development',
'development' => [
'name' => 'devdb',
'connection' => $pdo
]
]
];
Migration Paths
---------------
The first option specifies the path to your migration directory. Phinx uses
``%%PHINX_CONFIG_DIR%%/db/migrations`` by default.
.. note::
``%%PHINX_CONFIG_DIR%%`` is a special token and is automatically replaced
with the root directory where your phinx configuration file is stored.
In order to overwrite the default ``%%PHINX_CONFIG_DIR%%/db/migrations``, you
need to add the following to the configuration.
.. code-block:: yaml
paths:
migrations: /your/full/path
You can also provide multiple migration paths by using an array in your configuration:
.. code-block:: yaml
paths:
migrations:
- application/module1/migrations
- application/module2/migrations
You can also use the ``%%PHINX_CONFIG_DIR%%`` token in your path.
.. code-block:: yaml
paths:
migrations: '%%PHINX_CONFIG_DIR%%/your/relative/path'
Migrations are captured with ``glob``, so you can define a pattern for multiple
directories.
.. code-block:: yaml
paths:
migrations: '%%PHINX_CONFIG_DIR%%/module/*/{data,scripts}/migrations'
Custom Migration Base
---------------------
By default all migrations will extend from Phinx's `AbstractMigration` class.
This can be set to a custom class that extends from `AbstractMigration` by
setting ``migration_base_class`` in your config:
.. code-block:: yaml
migration_base_class: MyMagicalMigration
Seed Paths
----------
The second option specifies the path to your seed directory. Phinx uses
``%%PHINX_CONFIG_DIR%%/db/seeds`` by default.
.. note::
``%%PHINX_CONFIG_DIR%%`` is a special token and is automatically replaced
with the root directory where your configuration file is stored.
In order to overwrite the default ``%%PHINX_CONFIG_DIR%%/db/seeds``, you
need to add the following to the yaml configuration.
.. code-block:: yaml
paths:
seeds: /your/full/path
You can also provide multiple seed paths by using an array in your configuration:
.. code-block:: yaml
paths:
seeds:
- /your/full/path1
- /your/full/path2
You can also use the ``%%PHINX_CONFIG_DIR%%`` token in your path.
.. code-block:: yaml
paths:
seeds: '%%PHINX_CONFIG_DIR%%/your/relative/path'
Custom Seeder Base
---------------------
By default all seeders will extend from Phinx's `AbstractSeed` class.
This can be set to a custom class that extends from `AbstractSeed` by
setting ``seed_base_class`` in your config:
.. code-block:: yaml
seed_base_class: MyMagicalSeeder
Custom Migration Template
-------------------------
Custom template for Migrations could be used either by defining template file path
in configuration file:
.. code-block:: yaml
templates:
file: src/templates/customMigrationTemplate.php
Custom Seeder Template
----------------------
Custom Seeder template could be used either by defining template file path
in configuration file:
.. code-block:: yaml
templates:
seedFile: src/templates/customSeederTemplate.php
Environments
------------
One of the key features of Phinx is support for multiple database environments.
You can use Phinx to create migrations on your development environment, then
run the same migrations on your production environment. Environments are
specified under the ``environments`` nested collection. For example:
.. code-block:: yaml
environments:
default_migration_table: phinxlog
default_environment: development
production:
adapter: mysql
host: localhost
name: production_db
user: root
pass: ''
port: 3306
charset: utf8mb4
collation: utf8mb4_unicode_ci
would define a new environment called ``production``.
In a situation when multiple developers work on the same project and each has
a different environment (e.g. a convention such as ``<environment
type>-<developer name>-<machine name>``), or when you need to have separate
environments for separate purposes (branches, testing, etc) use environment
variable `PHINX_ENVIRONMENT` to override the default environment in the yaml
file:
.. code-block:: bash
export PHINX_ENVIRONMENT=dev-`whoami`-`hostname`
Migration Table
---------------
To keep track of the migration statuses for an environment, phinx creates
a table to store this information. You can customize where this table
is created by configuring ``default_migration_table`` to be used as default
for all environments:
.. code-block:: yaml
environment:
default_migration_table: phinxlog
If this field is omitted, then it will default to ``phinxlog``. For
databases that support it, e.g. Postgres, the schema name can be prefixed
with a period separator (``.``). For example, ``phinx.log`` will create
the table ``log`` in the ``phinx`` schema instead of ``phinxlog`` in the
``public`` (default) schema.
You may also specify the ``migration_table`` on a per environment basis.
Any environment that does not have a ``migration_table`` specified will
fallback to using the ``default_migration_table`` that is defined at the
top level. An example of how you might use this is as follows:
.. code-block:: yaml
environment:
default_migration_table: phinxlog
development:
migration_table: phinxlog_dev
# rest of the development settings
production:
# rest of the production settings
In the above example, ``development`` will look to the ``phinxlog_dev``
table for migration statues while ``production`` will use ``phinxlog``.
Table Prefix and Suffix
-----------------------
You can define a table prefix and table suffix:
.. code-block:: yaml
environments:
development:
....
table_prefix: dev_
table_suffix: _v1
testing:
....
table_prefix: test_
table_suffix: _v2
Socket Connections
------------------
When using the MySQL adapter, it is also possible to use sockets instead of
network connections. The socket path is configured with ``unix_socket``:
.. code-block:: yaml
environments:
default_migration_table: phinxlog
default_environment: development
production:
adapter: mysql
name: production_db
user: root
pass: ''
unix_socket: /var/run/mysql/mysql.sock
charset: utf8
External Variables
------------------
Phinx will automatically grab any environment variable prefixed with ``PHINX_``
and make it available as a token in the config file. The token will have
exactly the same name as the variable but you must access it by wrapping two
``%%`` symbols on either side. e.g: ``'%%PHINX_DBUSER%%'``. This is especially
useful if you wish to store your secret database credentials directly on the
server and not in a version control system. This feature can be easily
demonstrated by the following example:
.. code-block:: yaml
environments:
default_migration_table: phinxlog
default_environment: development
production:
adapter: mysql
host: '%%PHINX_DBHOST%%'
name: '%%PHINX_DBNAME%%'
user: '%%PHINX_DBUSER%%'
pass: '%%PHINX_DBPASS%%'
port: 3306
charset: utf8
Data Source Names
-----------------
Phinx supports the use of data source names (DSN) to specify the connection
options, which can be useful if you use a single environment variable to hold
the database credentials. PDO has a different DSN formats depending on the
underlying driver, so Phinx uses a database-agnostic DSN format used by other
projects (Doctrine, Rails, AMQP, PaaS, etc).
.. code-block:: text
<adapter>://[<user>[:<pass>]@]<host>[:<port>]/<name>[?<additionalOptions>]
* A DSN requires at least ``adapter``, ``host`` and ``name``.
* You cannot specify a password without a username.
* ``port`` must be a positive integer.
* ``additionalOptions`` takes the form of a query string, and will be passed to
the adapter in the options array.
.. code-block:: yaml
environments:
default_migration_table: phinxlog
default_environment: development
production:
# Example data source name
dsn: mysql://root@localhost:3306/mydb?charset=utf8
Once a DSN is parsed, it's values are merged with the already existing
connection options. Values in specified in a DSN will never override any value
specified directly as connection options.
.. code-block:: yaml
environments:
default_migration_table: phinxlog
default_environment: development
development:
dsn: '%%DATABASE_URL%%'
production:
dsn: '%%DATABASE_URL%%'
name: production_database
If the supplied DSN is invalid, then it is completely ignored.
Supported Adapters
------------------
Phinx currently supports the following database adapters natively:
* `MySQL <https://www.mysql.com/>`_: specify the ``mysql`` adapter.
* `PostgreSQL <https://www.postgresql.org/>`_: specify the ``pgsql`` adapter.
* `SQLite <https://www.sqlite.org/>`_: specify the ``sqlite`` adapter.
* `SQL Server <https://www.microsoft.com/sqlserver>`_: specify the ``sqlsrv`` adapter.
For each adapter, you may configure the behavior of the underlying PDO object by setting in your
config object the lowercase version of the constant name. This works for both PDO options
(e.g. ``\PDO::ATTR_CASE`` would be ``attr_case``) and adapter specific options (e.g. for MySQL
you may set ``\PDO::MYSQL_ATTR_IGNORE_SPACE`` as ``mysql_attr_ignore_space``). Please consult
the `PDO documentation <https://www.php.net/manual/en/book.pdo.php>`_ for the allowed attributes
and their values.
For example, to set the above example options:
.. code-block:: php
$config = [
"environments" => [
"development" => [
"adapter" => "mysql",
# other adapter settings
"attr_case" => \PDO::ATTR_CASE,
"mysql_attr_ignore_space" => 1,
],
],
];
By default, the only attribute that Phinx sets is ``\PDO::ATTR_ERRMODE`` to ``PDO::ERRMODE_EXCEPTION``. It is
not recommended to override this.
MySQL
`````````````````
The MySQL adapter has an unfortunate limitation in that it certain actions causes an
`implicit commit <https://dev.mysql.com/doc/refman/8.0/en/implicit-commit.html>`_ regardless of transaction
state. Notably this list includes ``CREATE TABLE``, ``ALTER TABLE``, and ``DROP TABLE``, which are the most
common operations that Phinx will run. This means that unlike other adapters which will attempt to gracefully
rollback a transaction on a failed migration, if a migration fails for MySQL, it may leave your DB in a partially
migrated state.
SQLite
`````````````````
Declaring an SQLite database uses a simplified structure:
.. code-block:: yaml
environments:
development:
adapter: sqlite
name: ./data/derby
suffix: ".db" # Defaults to ".sqlite3"
testing:
adapter: sqlite
memory: true # Setting memory to *any* value overrides name
Starting with PHP 8.1 the SQlite adapter supports ``cache`` and ``mode``
query parameters by using the `URI scheme <https://www.sqlite.org/uri.html>`_ as long as ``open_basedir`` is unset.
.. code-block:: yaml
environments:
testing:
adapter: sqlite
name: my_app
mode: memory # Determines if the new database is opened read-only, read-write, read-write and created if it does not exist, or that the database is a pure in-memory database that never interacts with disk, respectively.
cache: shared # Determines if the new database is opened using shared cache mode or with a private cache.
SQL Server
`````````````````
When using the ``sqlsrv`` adapter and connecting to a named instance you should
omit the ``port`` setting as SQL Server will negotiate the port automatically.
Additionally, omit the ``charset: utf8`` or change to ``charset: 65001`` which
corresponds to UTF8 for SQL Server.
Custom Adapters
`````````````````
You can provide a custom adapter by registering an implementation of the `Phinx\\Db\\Adapter\\AdapterInterface`
with `AdapterFactory`:
.. code-block:: php
$name = 'fizz';
$class = 'Acme\Adapter\FizzAdapter';
AdapterFactory::instance()->registerAdapter($name, $class);
Adapters can be registered any time before `$app->run()` is called, which normally
called by `bin/phinx`.
Templates
---------
You may override how phinx generates the template used with in a handful of ways:
* file - path to an alternative file to use.
* class - class to use for the template, must implement the ``Phinx\Migration\CreationInterface`` interface.
* style - style to use for template, either ``change`` or ``up_down``, defaults to ``change`` if not set.
You should only use one of these options. These can be overridden by passing command line options to the
:doc:`Create Command <commands`. Example usage within the config file is:
.. code-block:: yaml
templates:
style: up_down
Aliases
-------
Template creation class names can be aliased and used with the ``--class`` command line option for the :doc:`Create Command <commands>`.
The aliased classes will still be required to implement the ``Phinx\Migration\CreationInterface`` interface.
.. code-block:: yaml
aliases:
permission: \Namespace\Migrations\PermissionMigrationTemplateGenerator
view: \Namespace\Migrations\ViewMigrationTemplateGenerator
Version Order
-------------
When rolling back or printing the status of migrations, Phinx orders the executed migrations according to the
``version_order`` option, which can have the following values:
* ``creation`` (the default): migrations are ordered by their creation time, which is also part of their filename.
* ``execution``: migrations are ordered by their execution time, also known as start time.
Bootstrap Path
---------------
You can provide a path to a `bootstrap` php file that will included before any commands phinx commands are run. Note that
setting External Variables to modify the config will not work because the config has already been parsed by this point.
.. code-block:: yaml
paths:
bootstrap: 'phinx-bootstrap.php'
Within the bootstrap script, the following variables will be available:
.. code-block:: php
/**
* @var string $filename The file name as provided by the configuration
* @var string $filePath The absolute, real path to the file
* @var \Symfony\Component\Console\Input\InputInterface $input The executing command's input object
* @var \Symfony\Component\Console\Output\OutputInterface $output The executing command's output object
* @var \Phinx\Console\Command\AbstractCommand $context the executing command object
*/
Feature Flags
-------------
For some breaking changes, Phinx offers a way to opt-out of new behavior. The following flags are available:
* ``unsigned_primary_keys``: Should Phinx create primary keys as unsigned integers? (default: ``true``)
* ``column_null_default``: Should Phinx create columns as null by default? (default: ``true``)
.. code-block:: yaml
feature_flags:
unsigned_primary_keys: false
These values can also be set by modifying class fields on the ```Phinx\Config\FeatureFlags``` class, converting
the flag name to ``camelCase``, for example:
.. code-block:: php
Phinx\Config\FeatureFlags::$unsignedPrimaryKeys = false;