Dakota

Your best friend for hauling data in and out of MySQL.

A lightweight Active Record implementation for PHP5. Built on top of Idiorm. Forked from Paris.

Tested on PHP 5.2.0+ - may work on earlier versions with PDO and the correct database drivers.

Released under a BSD license.

Features

Extremely simple configuration.
Exposes the full power of Idiorm's fluent query API.
Supports associations.
Simple mechanism to encapsulate common queries in filter methods.
Built on top of PDO.
Uses prepared statements throughout to protect against SQL injection attacks.
Database agnostic. Currently supports SQLite and MySQL. May support others, please give it a try!

Philosophy

Dakota is built with the same less is more philosophy as Idiorm.

What's different from Paris?

Model instantiation is more like Kohana's ORM
- the new operator works as expected, e.g. $user = new User;
- a model instance represents a new row unless find_one() or find_many() is used
- therefore the create() method is unnecessary and has been removed
- find_one() won't return FALSE if it doesn't load a row; use the new loaded() method to check for this
Filters are simpler because model class methods can call query builder functions on $this
Transaction functions are made available in Model::

Let's See Some Code

Setup

Dakota requires Idiorm. Install Dakota with Idiorm somewhere in your project directory, and require both. The latest version of Idiorm that Dakota is known to work with is included in the test directory of this repository, but it's possible that the bleeding edge version at j4mie's repository will still work.

require_once 'your/path/to/idiorm.php';
require_once 'your/path/to/dakota.php';

Then, you need to tell Idiorm how to connect to your database. For full details of how to do this, see Idiorm's documentation.

Briefly, you need to pass a Data Source Name connection string to the configure method of the ORM class.

ORM::configure('sqlite:./example.db');

You may also need to pass a username and password to your database driver, using the username and password configuration options. For example, if you are using MySQL:

ORM::configure('mysql:host=localhost;dbname=my_database');
ORM::configure('username', 'database_user');
ORM::configure('password', 'top_secret');

Model Classes

You should create a model class for each entity in your application. For example, if you are building an application that requires users, you should create a User class. Your model classes should extend the base Model class:

class User extends Model {
}

Dakota takes care of creating instances of your model classes, and populating them with data from the database. You can then add behaviour to this class in the form of public methods which implement your application logic. This combination of data and behaviour is the essence of the Active Record pattern.

Database Tables

Your User class should have a corresponding user table in your database to store its data.

By default, Dakota assumes your class names are in CapWords style, and your table names are in lowercase_with_underscores style. It will convert between the two automatically. For example, if your class is called CarTyre, Dakota will look for a table named car_tyre.

To override this default behaviour, add a public static property to your class called $_table:

class User extends Model {
    public static $_table = 'my_user_table';
}

ID Column

Dakota requires that your database tables have a unique primary key column. By default, Dakota will use a column called id. To override this default behaviour, add a public static property to your class called $_id_column:

class User extends Model {
    public static $_id_column = 'my_id_column';
}

Note - Dakota has its own default ID column name mechanism, and does not respect column names specified in Idiorm's configuration.

Dakota's Active Record pattern

Each instance of your class represents a row in the database table. For example, to create an object representing a new row in the user table,

$user = new User;

Properties on model objects represent columns from the database table. You can retrieve, set, and unset them just like any other PHP variable. However, like Idiorm, any changes made are not saved to the database until you call the save method.

$user->name = "Frank";
echo $user->name;
// prints "Frank"
$user->save();
// assuming id is the primary key column,
echo $user->id;
// prints the id assigned to the newly inserted row

To operate on rows already in the database table the find_one and find_many methods from Idiorm are available, but slightly enhanced. Instead of returning an instance of the ORM class find_one loads data from the database into the current instance of your model. As in Idiorm, it can accept a primary key ID as its argument. For example, to load the user with id $id:

$user = new User;
$user->find_one($id);

As an aside, since Dakota uses nearly the same fluent interface as Idiorm, it relies on a lot of chaining. For this reason, a convenience factory method called factory on the base Model class is provided. It takes a single argument: the name of the model class you wish to use, and immediately returns a new empty instance of this model to facilitate chaining.

$user = Model::factory('User')->find_one($id);
// equivalent to the previous example.

Since the $user object now represents an existing row in the database, when it is modified and saved, it performs an update on the database instead of inserting a new row.

$user->name = "Fred";
$user->email = "fred@example.com";
$user->pageviews++;
$user->save();
echo $user->id;
// should still be the same as $id.

To delete the user, the delete method can be applied to $user. This takes effect immediately and does not require a subsequent call to save.

$user->delete();

Querying

Querying allows you to select particular rows from your database to populate instances of your model class. Query methods operate on instances of your class, chain as in Idiorm, and then execute when a call to find_one or find_many is performed. Dakota gives you full access to Idiorm's fluent query API. See Idiorm's documentation for details of this API.

For example:

$users = Model::factory('User')
    ->where('name', 'Fred')
    ->where_gte('age', 20)
    ->find_many();

This sets $users to an array of User objects that can be manipulated, or an empty array if no records were found. To find and immediately instantiate a User from one row only, use query methods and then find_one without an argument:

$user = Model::factory('User')->where('email', 'fred@example.com')->find_one();

To check if a row was found and loaded, use the loaded method.

if (!$user->loaded())
    echo "User not found!";

You may also want to retrieve a count of the number of rows returned by your query. This method behaves exactly like Idiorm's count method:

$count = Model::factory('User')->where_lt('age', 20)->count();

This returns an integer and leaves the model instance unaffected.

We should summarize differences between using Idiorm and using Dakota for querying:

You do not need to call the for_table method to specify the database table to use. Dakota will supply this automatically based on the class name (or the $_table static property, if present).
The find_one and find_many methods will return instances of your model subclass, instead of the base ORM class. Like Idiorm, find_one will return a single instance if a row matches, while find_many will return an array of instances, which may be empty if no rows matched.

Unlike Idiorm, find_one does not return false if no rows matched, it returns an unloaded instance of the model. You should always use the loaded method to check if a row was loaded. This is because Dakota model instances represent potentially new rows until an existing row is loaded into them via the find_one or find_many methods. As an example:

 $user = Model::factory('User')->where('name', 'Fred')->find_one();
 if (!$user->loaded()) {
     // Fred doesn't yet exist in the database.
     $user->name = "Fred";
     $user->save();
     // Now he does.
 }
 // Further operations can take place with the $user object at this point.
 $comment = Model::factory('Comment');
 $comment->user_id = $user->id;
 $comment->text = $posted_comment;
 $comment->save();

Associations

Dakota provides a simple API for one-to-one, one-to-many and many-to-many relationships (associations) between models. It takes a different approach to many other ORMs, which use associative arrays to add configuration metadata about relationships to model classes. These arrays can often be deeply nested and complex, and are therefore quite error-prone.

Instead, Dakota treats the act of querying across a relationship as a behaviour, and supplies a family of helper methods to help generate such queries. These helper methods should be called from within methods on your model classes which are named to describe the relationship. These methods return ORM instances (rather than actual Model instances) and so, if necessary, the relationship query can be modified and added to before it is run.

Summary

The following list summarises the associations provided by Dakota, and explains which helper method supports each type of association:

One-to-one

Use has_one in the base, and belongs_to in the associated model.

One-to-many

Use has_many in the base, and belongs_to in the associated model.

Many-to-many

Use has_many_through in both the base and associated models.

Below, each association helper method is discussed in detail.

Has-one

One-to-one relationships are implemented using the has_one method. For example, say we have a User model. Each user has a single Profile, and so the user table should be associated with the profile table. To be able to find the profile for a particular user, we should add a method called profile to the User class (note that the method name here is arbitrary, but should describe the relationship). This method calls the protected has_one method provided by Dakota, passing in the class name of the related object. The profile method should return an ORM instance ready for (optional) further filtering.

class Profile extends Model {
}

class User extends Model {
    public function profile() {
        return $this->has_one('Profile');
    }
}

The API for this method works as follows:

// Select a particular user from the database
$user = Model::factory('User')->find_one($user_id);

// Find the profile associated with the user
$profile = $user->profile()->find_one();

By default, Dakota assumes that the foreign key column on the related table has the same name as the current (base) table, with _id appended. In the example above, Dakota will look for a foreign key column called user_id on the table used by the Profile class. To override this behaviour, add a second argument to your has_one call, passing the name of the column to use.

Has many

One-to-many relationships are implemented using the has_many method. For example, say we have a User model. Each user has several Post objects. The user table should be associated with the post table. To be able to find the posts for a particular user, we should add a method called posts to the User class (note that the method name here is arbitrary, but should describe the relationship). This method calls the protected has_many method provided by Dakota, passing in the class name of the related objects. Pass the model class name literally, not a pluralised version. The posts method should return an ORM instance ready for (optional) further filtering.

class Post extends Model {
}

class User extends Model {
    public function posts() {
        return $this->has_many('Post'); // Note we use the model name literally - not a pluralised version
    }
}

The API for this method works as follows:

// Select a particular user from the database
$user = Model::factory('User')->find_one($user_id);

// Find the posts associated with the user
$posts = $user->posts()->find_many();

By default, Dakota assumes that the foreign key column on the related table has the same name as the current (base) table, with _id appended. In the example above, Dakota will look for a foreign key column called user_id on the table used by the Post class. To override this behaviour, add a second argument to your has_many call, passing the name of the column to use.

Belongs to

The 'other side' of has_one and has_many is belongs_to. This method call takes identical parameters as these methods, but assumes the foreign key is on the current (base) table, not the related table.

class Profile extends Model {
    public function user() {
        return $this->belongs_to('User');
    }
}

class User extends Model {
}

The API for this method works as follows:

// Select a particular profile from the database
$profile = Model::factory('Profile')->find_one($profile_id);

// Find the user associated with the profile
$user = $profile->user()->find_one();

Again, Dakota makes an assumption that the foreign key on the current (base) table has the same name as the related table with _id appended. In the example above, Dakota will look for a column named user_id. To override this behaviour, pass a second argument to the belongs_to method, specifying the name of the column on the current (base) table to use.

Has many through

Many-to-many relationships are implemented using the has_many_through method. This method has only one required argument: the name of the related model. Supplying further arguments allows us to override default behaviour of the method.

For example, say we have a Book model. Each Book may have several Author objects, and each Author may have written several Books. To be able to find the authors for a particular book, we should first create an intermediary model. The name for this model should be constructed by concatenating the names of the two related classes, in alphabetical order. In this case, our classes are called Author and Book, so the intermediate model should be called AuthorBook.

We should then add a method called authors to the Book class (note that the method name here is arbitrary, but should describe the relationship). This method calls the protected has_many_through method provided by Dakota, passing in the class name of the related objects. Pass the model class name literally, not a pluralised version. The authors method should return an ORM instance ready for (optional) further filtering.

class Author extends Model {
    public function books() {
        return $this->has_many_through('Book');
    }
}

class Book extends Model {
    public function authors() {
        return $this->has_many_through('Author');
    }
}

class AuthorBook extends Model {
}

The API for this method works as follows:

// Select a particular book from the database
$book = Model::factory('Book')->find_one($book_id);

// Find the authors associated with the book
$authors = $book->authors()->find_many();

// Get the first author
$first_author = $authors[0];

// Find all the books written by this author
$first_author_books = $first_author->books()->find_many();

Overriding defaults

The has_many_through method takes up to four arguments, which allow us to progressively override default assumptions made by the method.

First argument: associated model name - this is mandatory and should be the name of the model we wish to select across the association.

Second argument: intermediate model name - this is optional and defaults to the names of the two associated models, sorted alphabetically and concatenated.

Third argument: custom key to base table on intermediate table - this is optional, and defaults to the name of the base table with _id appended.

Fourth argument: custom key to associated table on intermediate table - this is optional, and defaults to the name of the associated table with _id appended.

Filters

It is often desirable to create reusable queries that can be used to extract particular subsets of data without repeating large sections of code. Dakota allows you to do this by writing functions that extend your model class, within which you run Idiorm query builder methods on the $this object. By returning $this from your function, you will allow your method to be chainable.

It is easiest to illustrate this with an example. Imagine an application in which users can be assigned a role, which controls their access to certain pieces of functionality. In this situation, you may often wish to retrieve a list of users with the role 'admin'. To do this, add a method called (for example) admins to your Model class:

class User extends Model {
    public function admins() {
        return $this->where('role', 'admin');
    }
}

You can then use this filter in your queries:

$admin_users = Model::factory('User')->admins()->find_many();

You can also chain it with other methods as normal:

$young_admins = Model::factory('User')
                    ->admins()
                    ->where_lt('age', 18)
                    ->find_many();

Here is an example utilizing a filter method with an argument.

class User extends Model {
    public function has_role($role) {
        return $this->where('role', $role);
    }
}

$admin_users = Model::factory('User')->has_role('admin')->find_many();
$guest_users = Model::factory('User')->has_role('guest')->find_many();

These examples may seem simple (has_role('admin') could just as easily be achieved using where('role', 'admin')), but remember that filters can contain arbitrarily complex code - adding raw_where clauses or even complete raw_query calls to perform joins, etc. Filters provide a powerful mechanism to hide complexity in your model's query API.

Transactions

Dakota provides three shortcuts for initiating, committing and rolling back transactions as public static methods on the Model class. Your database must support transactions: e.g., if you are using MySQL, your tables must be using the InnoDB and not the MyISAM engine.

Transactions are started with Model::start_transaction like this:

Model::start_transaction();
$user = Model::factory('User')->find_one($user_id);
$user->balance -= $price;
$user->save();

At this point changes have not actually been saved to the database, since the transaction is still open. The reason this sort of update is best encapsulated in a transaction is that other processes that attempt to simultaneously update the user table will not interfere with this balance calculation. For example, a simultaneous credit operation that updates this user's balance between the queries normally executed by find_one and save would have been silently overwritten unless a transaction was in progress, as it is here.

At any point a transaction can be canceled with Model::rollback, and the database will be left in the state it was before the transaction began.

if ($user->balance < 0) {
  echo "Sorry, you have insufficient funds.";
  Model::rollback();
  return;
}

When the changes contained in the transaction are ready to be applied, use Model::commit.

Model::commit();
echo "Your new balance is: {$user->balance}";

Transactions are not necessary for everyday use of Dakota. If a transaction has not been started with Model::start_transaction, every save and delete call is implicitly committed to the database immediately following the query's execution. For more information, see PDO's documentation on transactions.

Examples: Getting, updating and inserting data

The model instances returned by your queries behave similarly to instances of Idiorm's raw ORM class. To distinguish whether data was successfully loaded into your model instance with find_one. always use the loaded method.

Accessing data:

$user = Model::factory('User')->find_one($id);
echo $user->loaded() ? $user->name : 'User not found';

Update data and save the instance:

$user = Model::factory('User')->find_one($id);
if ($user->loaded()) {
    $user->name = 'Dakota';
    $user->save();
}

Of course, because these objects are instances of your base model classes, you can also call methods that you have defined on them:

class User extends Model {
    public function full_name() {
        return $this->first_name . ' ' . $this->last_name;
    }
}

$user = Model::factory('User')->find_one($id);
echo $user->loaded() ? $user->full_name() : 'User not found';

To delete the database row associated with an instance of your model, call its delete method:

$user = Model::factory('User')->find_one($id);
if ($user->loaded())
    $user->delete();

To insert a database row, create a new instance of your model and save it.

$user = Model::factory('User');
// equivalently, $user = new User;
$user->name = "Frank";
$user->save();

You can also get the all the data wrapped by a model subclass instance using the as_array method. This will return an associative array mapping column names (keys) to their values.

The as_array method takes column names as optional arguments. If one or more of these arguments is supplied, only matching column names will be returned.

class Person extends Model {
}

$person = Model::factory('Person');

$person->first_name = 'Fred';
$person->surname = 'Bloggs';
$person->age = 50;

// Returns array('first_name' => 'Fred', 'surname' => 'Bloggs', 'age' => 50)
$data = $person->as_array();

// Returns array('first_name' => 'Fred', 'age' => 50)
$data = $person->as_array('first_name', 'age');

A word on validation

It's generally considered a good idea to centralise your data validation in a single place, and a good place to do this is inside your model classes. This is preferable to handling validation alongside form handling code, for example. Placing validation code inside models means that if you extend your application in the future to update your model via an alternative route (say a REST API rather than a form) you can re-use the same validation code.

Despite this, Dakota doesn't provide any built-in support for validation. This is because validation is potentially quite complex, and often very application-specific. Dakota is deliberately quite ignorant about your actual data - it simply executes queries, and gives you the responsibility of making sure the data inside your models is valid and correct. Adding a full validation framework to Dakota would probably require more code than Dakota itself!

However, there are several simple ways that you could add validation to your models without any help from Dakota. You could override the save() method, check the data is valid, and return false on failure, or return parent::save() on success. You could create your own subclass of the Model base class and add your own generic validation methods. Or you could write your own external validation framework which you pass model instances to for checking. Choose whichever approach is most suitable for your own requirements.

Configuration

The only configuration options provided by Dakota itself are the $_table and $_id_column static properties on model classes. To configure the database connection, you should use Idiorm's configuration system via the ORM::configure method. See Idiorm's documentation for full details.

Query logging

Idiorm can log all queries it executes. To enable query logging, set the logging option to true (it is false by default).

ORM::configure('logging', true);

When query logging is enabled, you can use two static methods to access the log. ORM::get_last_query() returns the most recent query executed. ORM::get_query_log() returns an array of all queries executed.

powerpak/dakota

Dakota

Features

Philosophy

What's different from Paris?

Let's See Some Code

Setup

Model Classes

Database Tables

ID Column

Dakota's Active Record pattern

Querying

Associations

Summary

One-to-one

One-to-many

Many-to-many

Has-one

Has many

Belongs to

Has many through

Overriding defaults

Filters

Transactions

Examples: Getting, updating and inserting data

A word on validation

Configuration

Query logging