Database Schema Definition
With Propel, a database schema is defined in an XML-file. Each module carries it’s own part of the big schema that is collected and merged.
Propel’s documentation on defining the database schema: Database schema
Example Schema XML File
This file can be copied into your moduleinto (Module)/Persistence/Propel/Schema/. Don’t forget to adopt it to your needs.
<?xml version="1.0"?> <database name="zed" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="http://static.spryker.com/schema-01.xsd" namespace="Orm\Zed\Customer\Persistence" package="src.Orm.Zed.Customer.Persistence"> <table name="spy_customer" idMethod="native"> <column name="id_customer" required="true" type="INTEGER" autoIncrement="true" primaryKey="true" /> <column name="email" required="true" size="255" type="VARCHAR" /> <column name="first_name" size="100" type="VARCHAR" /> <column name="last_name" size="100" type="VARCHAR" /> <unique> <unique-column name="email" /> </unique> </table> </database>
XML File Prefix
The name of the XML file that contains DB schema definition must be prefixed with the spy_ string
When you need to use a composed name for a field, you define in your table, you need to use small cases and separate the words using underscores.
To apply a change in the database schema, you need to run a command line call.
The schema migration is part of the setup:install call.
The workflow can be described like this:
- Collect the schema-xml files from all bundles
- Merge schemas and copy them into one directory ( src/Generated/Propel/DE/Schema )
- Generation of entities and query objects
- Diff of new and existing schema
- Run migration
Schema File Merge
Sometimes it is useful to add columns to a table which belongs to another module. Possible use cases:
- A core module wants to inject a foreign key into a table from another core module. This happens when the dependency direction is in contrast to the direction of the relation.
- A project module wants to add a column to a table from a core module.
When you add a column to a core table, this column must not be mandatory, because in this case the core classes, not knowing the mandatory fields, become unusable and thus have to be overwritten, which increases the difficulty to update. In most cases you can and should avoid adding a column to a core table. This can cause compatibility problems in the future. What happens, when the next release adds a column with the same name but another meaning? To avoid this problem it is a good practice to add a new table and use a one-to-one relationship.
The merging works like this:
- Find all schema xml files from core and project level
- Group them by filename
- Check number of files per filename
- If one filename has one path: copy to src/Orm/Propel/DE/Schema
- If one filename has more than one path: merge them
- Check if the database attributes name, package and namespace are consistent, otherwise throw exception
- Merge xml by the name attribute on all levels
- Copy merged xml to src/Orm/Propel/DE/Schema
<table name="spy_customer" idMethod="native"> <column name="first_name" size="100" type="VARCHAR" /> </table>
<table name="spy_customer"> <column name="last_name" size="100" type="VARCHAR" /> </table>
Merged file :
<table name="spy_customer" idMethod="native"> <column name="first_name" size="100" type="VARCHAR" /> <column name="last_name" size="100" type="VARCHAR" /> </table>
Change attribute values
It is possible to change the value of an attribute e.g. type="VARCHAR" to type="LONGVARCHAR" or size="100" to size="200". Be aware that this can have strong side effects and should be avoided when possible.
To avoid accidental change of attribute values, you can run vendor/bin/console propel:schema:validate after you added the PropelSchemaValidatorConsole class to your ConsoleDependencyProvider. This will show you when a merge of your schema files would lead to a change of an attribute value.
To give you more control about the validation of the attribute values, we added a whitelist so you can run the validator also when you have changed an attribute value on purpose. Please see PropelConfig::getWhitelistForAllowedAttributeValueChanges() for more details.
Case Insensitive Fields
MySQL provides case insensitive text type and this paragraph is mostly related to PostgreSql. But you still could have some benefits from using case insensitive text fields (e.g handling lower/upper case comparison on DB side).
Sometime a table requires a case insensitive field. A good example is email field. A customer should not be able to register twice with the same email but different character case.
To enable this option just set column attribute caseInsensitive with true in a schema XML file of a target table.
Last review date: Dec. 12th, 2017