The DatabaseSchema component provides a unified data structure to describe relational databases. DatabaseSchema can read and write the structure from and to several databases (such as MySQL, SQLite and Oracle) and also read and write in two web language files (XML and PHP).
The main intention of the component is to extract an already existing schema from a database, and compare it with either a predefined schema to check for any changes. Although the component does allow you to create the schema manually through the API, it is often easier to create the schema yourself in a database, and then extract the data from it.
DatabaseSchema can also generate PersistentObject definition files. For further information, please refer to the API documentation of ezcPersistentObjectDefinition and ezcPersistentObjectSchemaGenerator.
This section gives you an overview of the main classes of the DatabaseSchema component.
Databases can be generated from two different types of sources. The first source is the file, where there are multiple format handlers available. There is a format for reading and writing XML files, and one for reading and writing files that store the ezcDbSchema structure in a PHP array.
This example shows how to create a database schema from the two different sources. With the first parameter in ezcDbSchema::createFromFile(), you define which type to use ('xml' or 'array'). The second parameter is the file that you are reading the schema definition from.
In both cases, the methods will return an object of the class ezcDbSchema. The documentation for ezcDbSchemaHandlerManager lists the supported types of readers and writers. The ezcDbSchema->getSchema() method returns the schema as an array of ezcDbSchemaTable objects. The ezcDbSchemaTable object then contains, in two member variables, an array of ezcDbSchemaField objects to describe the fields and an array of ezcDbSchemaIndex objects to describe the indices.
Once a ezcDbSchema object has been created, its tables and fields can be manipulated in different ways. Tables can be created, edited and deleted. Fields can be added to tables, edited and deleted. See the example below:
Once a ezcDbSchema object has been created, it can be manipulated in any way. Then, there are different possible ways to save the database schema. The ezcDbSchema->writeToFile() method stores an ezcDbSchema object to a file on disk. As stated, you can use the XML and PHP array formats. This is shown in the first part of this example:
The second and third methods both deal directly with a database. The second method in the example uses the database connection $db for creating a schema. This method only saves to the tables in the schema definition. The schemas that are defined will be overwritten.
The third method merely uses the database connection to determine which SQL DLL dialect to use for the query to create the schema in the database.
The ezcDbSchemaComparator class provides the ezcDbSchemaComparator::compareSchemas() method to compare two different ezcDbSchema objects. The result of this operation is an ezcDbSchemaDiff object. The next example illustrates this:
First, we create the two different schemas: one from the database, and one from a file. In line 10, we utilize the ezcDbSchemaComparator::compareSchemas() function to compare the two schemas. This process returns the differences as the $diffSchema variable.
You can perform multiple actions with the generated differences. These three methods are explained in the inline comments in the above example.
The last feature that the DatabaseSchema component offers is the validation of schemas. The ezcDbSchemaValidator::validate() method accepts an ezcDbSchema object and returns an array of strings describing potential problems with the schema. This is shown in the following example:
To keep the compability between different relational database management systems, there are restrictions you have to cope with:
Lowercase column names
As some RDBMS always do not store column names case sensitive, but return and store them completely as upper case letters, ezcDatabase converts all column names to lower case to offer a uniform interface.
You are limited to an amount of field types common to all RDBMS, or for which at least some kind of mapping exists for all RDBMS.
@TODO: Mapping table (issue #10975).