zeptech / dbup
Database update helper
Requires
- php: >=5.4
- psr/log: 1.0.0
- zeptech/database: dev-master
- zeptech/utility: dev-master
Requires (Dev)
- hamcrest/hamcrest: 1.1.0
- mockery/mockery: dev-master@dev
- phpunit/phpunit: 4.1.*
This package is not auto-updated.
Last update: 2024-11-09 12:57:47 UTC
README
Database version manager written in PHP
Phases
The update algorithm supports 3 phases for each schema revision, pre-alter, alter and post-alter. Pre-alter and post-alter scripts are PHP scripts. The alter phase is an SQL script. Each phase can return data which will be passed to subsequent phases including from PHP to SQL and vice versa (more on this later).
WIP PHP Script execution methods
DbUp supports a couple of different ways of structuring/encapsulating PHP pre/post alter scripts. The default simply requires the PHP files containing the alter logic. These files can be structured whichever way best suits the logic being performed. The return values from the pre-alter script is passed into the subsequent alter SQL script.
File system structure
Database schema revisions should all be stored in the same directory. Each revision will consist of a set of 1 to 3 files, one for phase where all phases are optional. The files need to named according to the following convention:
- Pre-alter: pre-alter-000001.php
- Alter: alter-000001.sql
- Post-alter: post-alter-000001.php
SQL Enhancements
The .sql scripts of the alter phase support a superset of SQL syntax. The additional syntax makes it eaiser to write portable scripts, as well as to perform some common tasks that would normally require a pre/post alter script.
INSERT ID variables
When performing INSERTs into tables that contain a serial (auto increment) field it is difficult to determine the generated ID for subsequent use. DbUp supports special syntax to store these IDs in variables and to use them in subsequent INSERT statements. To store the insert ID use the following syntax:
my_insert_id := INSERT INTO my_table (key, value)
VALUES ('aKey', 'aValue');
To then use the saved insert ID in a subsequent statement, reference the variable using the following syntax:
INSERT INTO my_table_ref (my_table_id, refd)
VALUES (:my_insert_id, 'refVal');
In order for this to work with PostgreSQL the id field for the table needs to be
backed by a sequence named <table-name>_id_seq
.
WIP Normalization of SERIAL fields
Most databases support a field type that is an auto incrementing integer most commonly used for primary keys, however they all seem to have a slightly different syntax. When writing your alter scripts you can write these statements for whichever database you are comfortable with and dbUp will handle it correctly for whichever database your script is being run against.
This feature is currently a WIP and only supports a limitted number of use cases (i.e the ones I've need personally). If you find a various that's not working create an issue, it should be fairly easy to add support.
Current support includes:
- MySQL
integer [NOT NULL] AUTO_INCREMENT
field can be executed against a PostgreSQL connection. Field can be specified as not null. - That's it.
WIP Normalization of ALTER TABLE
Some support for normalizing the various ALTER TABLE
commands
- MySQL
ALTER TABLE <table> MODIFY [COLUMN] <column> <new-type>
and PgSQLALTER TABLE <table> ALTER COLUMN <column> [SET DATA] TYPE <new-type>
statements are normalized.