This plugin provides a SQL table schema builder for the PHP Fat-Free Framework. It might be useful for installation scripts, dynamic applications or CMS environments.
Currently drivers for MySQL, SQLite, PostgreSQL & SQL Server are supported and fully tested. Further support for Sybase, Oracle and DB2 drivers are already included, but experimental. Handle with care and test your application. No warranty at all.
This plugin was made for F3 version 3.x and requires PHP 5.4+
Just copy schema.php into F3's lib/db/sql
folder. Done.
If you use composer, you can add this package by running composer require ikkez/f3-schema-builder:dev-master
To work with the Schema builder, you need an active SQL Connection. Create one like this:
$db = new \DB\SQL('mysql:host=localhost;port=3306;dbname='.$DBname, $user, $pass);
Now create a Schema object to work on. Inject the DB object into its constructor:
$schema = new \DB\SQL\Schema( $db );
Creating new tables is super easy. Let's have a look at this example:
$table = $schema->createTable('products');
$table->addColumn('title')->type($schema::DT_VARCHAR128);
$table->addColumn('description')->type($schema::DT_TEXT);
$table->build();
The createTable()
method returns a new table object (instance of TableCreator) for creation purpose. You may add new columns, indexes and change the primary key with it. New tables will always contain an auto-incremented, primary-key field named id
, which is required for further SQL\Mapper usage. All actions on a table object that affects its schema, are collected first and needs an additional build()
command to be executed to really take effect on the database. If you're unsure of the result, you can run a simulation of that build method and have a look at the generated queries, the Schema Builder would have executed, with the following call:
$generated_queries = $table->build(false);
print_r($generated_queries);
Using the $table->addColumn()
method will create a new Column object and adds it to the table object. We can use fluent calls for configuring these columns.
$column = $table->addColumn('deleted');
$column->type($schema::DT_BOOL);
$column->nullable(false);
$column->defaults(0);
$column->after('id');
// or in a fluent way:
$table->addColumn('deleted')->type($schema::DT_BOOL)->nullable(false)->defaults(0)->after('id');
Here is a list of possible configuration methods:
Set datatype of this column. Usually a constant of type \DB\SQL\Schema::DT_{datatype}. Have a look at the Column Class API for more details about datatypes.
When $force
is TRUE, the $datatype
string is used as a raw value as it is and passed through the creation query (useful for custom data types).
Set this column as NULL or NOT NULL. Default is true / nullable.
Adds a default value for records.
Trys to place the new column behind an existing one.
Add an index for that field. $unique
makes it a UNIQUE INDEX.
Altering existing tables is quite similar to creating them, but offers a bunch more possibilities. A basic example:
$table = $schema->alterTable('products');
$table->addColumn('prize')->type($schema::DT_DECIMAL);
$table->addColumn('stock')->type($schema::DT_INT);
$table->dropColumn('foo_bar');
$table->renameColumn('title','name');
$table->build();
As you can see, $schema->alterTable()
returns a new table object (instance of TableModifier) for altering purpose, which provides all methods of the TableCreator plus some more actions like removing or renaming columns. Here is a list of method you can use:
- renameColumn( string $name, string $new_name );
- updateColumn( string $name, string $datatype );
- dropColumn( string $name );
- addIndex( string | array $columns, [ bool $unique = false ]);
- dropIndex( string | array $columns );
- listIndex();
- getCols([ bool $types = false ]);
The SchemaBuilder will quote all your table and column identifiers and should be resistant against preserved word errors.
The Schema class provides you the following simple methods for:
Returns an array of all databases available (except for SQLite). Can be useful for installation purpose, when you want the user to select a database to work on. Therefor just create your DB connection without selecting a database like:
$db = new \DB\SQL('mysql:host=localhost;port=3306;dbname=', $user, $password);
Some DB engine default setups also grants simple read operations, without setting a user / password.
Returns an array of all tables available within the current database.
Returns a new table object for creation purpose.
Returns a table object for altering operations on already existing tables.
Renames a table. If you set $exec
to FALSE
, it will return the generated query instead of executing it.
You can also use a short-cut on an altering table object, like $table->rename( string $new_name, [ bool $exec = true ]);
.
Clear the contents of a table. Set $exec
to FALSE
will return the generated query instead of executing it.
Deletes a table. Set $exec
to FALSE
will return the generated query instead of executing it.
You can also use a short-cut on an altering table object, like $table->drop([ bool $exec = true ]);
.
This is useful for reverse lookup. It checks if a data type is compatible with a given column definition,
I.e. $schema->isCompatible('BOOLEAN','tinyint(1)');
.
This class is meant for creating new tables. It can be created by using $schema->createTable($name)
.
This creates a new Column object and saves a reference to it. You can configure the Column for your needs using further fluent calls, setting its public parameters or directly via config array like this:
$table->addColumn('title',array(
'type'=>\DB\SQL\Schema::DT_INT4,
'nullable'=>false,
'default'=>'untitled new entry',
'after'=>'id',
'index'=>true,
'unique'=>true,
));
You can add an index to a column by configuring the Column object while adding the new column, or like this:
$table->addIndex('name');
For adding an combined index on multiple columns, just use an array as parameter:
$table->addIndex(array('name','email'));
If you like to change the default id
named primary-key right on the creation of a new table, you can use this one:
$table->primary('uid');
This will rename the id
field to uid
. If you like to set a primary key on multiple columns (a composite key), use an array:
$table->primary(array('uid','version'));
The first element of this pkey array will always be treated as an auto-incremented field.
example:
$table = $schema->createTable('news');
$table->addColumn('title')->type($schema::DT_VARCHAR128);
$table->addColumn('bodytext')->type($schema::DT_TEXT);
$table->addColumn('version')->type($schema::DT_INT8)->nullable(false)->defaults(1);
$table->primary(array('id', 'version'));
$table->build();
Now your primary key is build upon 2 columns, to use records like id=1, version=1
and id=1, version=2
.
This method will set a custom charset and default collation to a new table.
In example, this will set an utf8mb4
charset and a utf8mb4_unicode_ci
collation as default for the new table:
$table = $schema->createTable('comments');
$table->setCharset('utf8mb4');
// ...
NB: currently only effects MySQL. 1-4 Multibyte UTF8 chars work out of the box in Postgre, SQlite. No workaround for SQL Server yet.
This will start the table generation process and executes all queries if $exec
is TRUE
, otherwise it will just return all queries as array.
This class is ment for creating new tables. It can be created by using $schema->alterTable($name)
.
Adds a new column
This is used to rename an existing column.
This is used to modify / update the column's datatype.
Tries to removes a column from the table, if it exists.
Creates a index or unique index for one or multiple columns on the table.
Drops an index.
Returns an associative array with index name as key and array('unique'=>$value)
as value.
Creates a new primary or composite key on the table.
Returns an array of existing table columns. If $types
is set to TRUE
, it will return an associative array with column name as key and the schema array as value.
This generates the queries needed for the table alteration and executes them when $exec
is true, otherwise it returns them as array.
This will instantly rename the table. Notice: Instead of being executed on calling build()
the execution is controlled by $exec
.
This will instantly drop the table. Notice: Instead of being executed on calling build()
the execution is controlled by $exec
.
This is useful for reverse lookup. It checks if a data type is compatible with an existing column type,
I.e. $table->isCompatible('BOOLEAN','hidden');
.
This method will set a custom charset and collation and will convert existing tables upon build()
.
In this example we will convert a table to an utf8mb4
charset and a utf8mb4_general_ci
collation:
$table = $schema->alterTable('comments');
$table->setCharset('utf8mb4','general');
// ...
The method $table->addColumn($columnName);
adds a further column field to the selected table and creates and returns a new Column object, that can be configured in different ways, before finally building it.
Set datatype of this column. The $force
argument will disable the datatype check with the included mappings and uses your raw string as type definition.
You can use these available mapped types as constants in \DB\SQL\Schema:
Type | Description | Storage size | Save Range |
---|---|---|---|
DT_BOOL DT_BOOLEAN |
resolves in a numeric | at least 1 byte | 0,1 |
DT_INT1 DT_TINYINT |
exact integer | at least 1 byte | lower: 0, upper; 255 |
DT_INT2 DT_SMALLINT |
exact integer | at least 2 bytes | ±32,768 |
DT_INT4 DT_INT |
exact integer | 4 bytes | ±2,147,483,648 |
DT_INT8 DT_BIGINT |
exact integer | at most 8 bytes | ±2^63 |
DT_FLOAT | approximate numeric | 4 bytes | ±1.79E + 308 |
DT_DECIMAL DT_DOUBLE |
exact numeric | at least 5 bytes | ±10^38+1 |
DT_VARCHAR128 | character string | 128 bytes | 128 chars |
DT_VARCHAR256 | character string | 256 bytes | 256 chars |
DT_VARCHAR512 | character string | 512 bytes | 512 chars |
DT_TEXT | character string | max length 2,147,483,647 | |
DT_LONGTEXT | character string | max length 4,294,967,295 | |
DT_DATE | Y-m-d | 3 bytes | |
DT_DATETIME | Y-m-d H:i:s | 8 bytes | |
DT_TIMESTAMP | Y-m-d H:i:s | 8 bytes | |
DT_BLOB DT_BINARY |
bytes |
usage:
$table = $schema->alterTable('news');
$table->addColumn('author')->type(\DB\SQL\Schema::DT_VARCHAR128);
// or
$table->addColumn('bodytext')->type($schema::DT_TEXT);
// or the shorthand
$table->addColumn('bodytext')->type_text();
// save changes to database
$table->build();
there are also a bunch of shorthand methods available, you can use instead of type()
:
- type_tinyint()
- type_smallint()
- type_int()
- type_bigint()
- type_float()
- type_decimal()
- type_text()
- type_longtext()
- type_varchar([ $length = 255 ])
- type_date()
- type_datetime()
- type_timestamp([ $asDefault = FALSE ])
- type_blob()
- type_bool()
When pass-through is enabled, the datatype value is treated as a raw value, which makes it possible to set any other custom data type that is not covered by the included aliases.
This is equivalent to type()
with $force = TRUE
.
Set this column as NULL or NOT NULL. Default is TRUE
/ nullable.
You can set defaults to nullable fields as well.
Adds a default value for records. Usually a string or integer value or NULL
.
CURRENT_TIMESTAMP as dynamic default value
But if you like to add a timestamp of the current time to new inserted records, you can use a TIMESTAMP field with a special default value to achieve this.
$table->addColumn('creation_date')->type($schema::DT_TIMESTAMP)->defaults($schema::DF_CURRENT_TIMESTAMP);
// a shorthand would be:
$table->addColumn('creation_date')->type_timestamp(TRUE);
Try to place the new column behind an existing one. (only works for SQLite and MySQL)
Add an index for that field. $unique
makes it a UNIQUE INDEX.
Feed column from array or hive key.
Returns an array of the current column configuration
Returns the resolved column data type.
GPLv3
Like this Plugin?