Framework born of a WP devs Rails-envy
Switch branches/tags
Nothing to show
Pull request Compare This branch is 100 commits behind Streetwise-Media:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


swpMVC is a lightweight MVC framework built to bring some of the experience of other rapid application development frameworks to WordPress. Inspired largely by Rails, Express and FuelPHP, it aims to make routing, modeling, and rendering easy, giving you more control over your code structure than WordPress gives out of the box, without adding too much extra work.

The simplest way to cover some of the initial concepts and get started is by examining (and using) the starter plugin found in the starter_plugin directory.

##Starter Plugin

###Why a singleton?

I know a singleton is an upgraded global variable, but your plugins need to add WordPress filters and actions. Using a singleton gives you easy access to the plugin class, and also makes sure you don't end up running the methods you hang on your filters and actions more than once each.


This is called in the constructor, and this is where you should include your models and controllers. By creating the plugin instance on the swp_mvc_init action, we ensure that the swpMVC core is loaded, and the base classes which your models and controllers must extend will be available.


Also called in the constructor, this is used to hook the add_routes method to the swp_mvc_routes filter, where you can add your plugin routes using the syntax described in the router section. By placing this after the require_dependencies call in the constructor, you can be sure your controllers are loaded when you add the routes to the system.


swpMVC follows a routing structure that more closely resembles Sinatra or Express than WordPress' rewrite rules. See the next section for syntax.


Adding routes

swpMVC routes are stored as an array of arrays, with each array stored representing one route using the following structure


    $route = array('controller' => 'ControllerClass',
                            'method' => 'ControllerMethod',
                            'route' => '/url/of/route/:p/:p'

There is no "automagic" routing, everything must be declared. This is done so that your routing structure is exactly as you want, with no additional steps required to turn off magic routes.

Routing parameters

Parameters in your route are represented with the token ":p"

They will be passed to your controller method in the same order they appear in the route. Skipping named parameters allows the framework to use only one additional querystring variable

Auto-flush rewrite rules

The core framework will monitor whether swpMVC routes have been added, modified or removed, and flush the rewrite rules as needed, so there is no need to do this manually.


Here's a full example adding routes from your swpMVC plugin based off the example plugin included in the example directory

    public function add_routes($routes)
        $r[] = array('controller' => 'swpMVC_Example_Controller',
                    'method' => 'wp_style',
                    'route' => '/recent_thumbs/wp_style');
        $r[] = array('controller' => 'swpMVC_Example_Controller',
                        'method' => 'swpmvc_style',
                        'route' => '/recent_thumbs/swpmvc_style');
        $r[] = array('controller' => 'swpMVC_Example_Controller',
                        'method' => 'render_post_form',
                        'route' => '/post_form/:p');
        $s =  array_merge($routes, $r);
        return $s;


Models must extend the swpMVCBaseModel class. This class itself extends ActiveRecord\Model, from the PHP ActiveRecord library. For query syntax, CRUD operations, basic model definitions, and overloading, refer to the ActiveRecord docs. The copy included in swpMVC includes several modifications to make it more WordPress friendly, (the diff is backwards, sorry) , and the swpMVCBaseModel class includes several convenience methods, all of which are covered in this section.

###public static function tablename

Instead of declaring the model table with a static variable, we use a static method. This allows us to do the following

    public static function tablename()
        global $wpdb;
        return $wpdb->prefix.'posts';

This model would now be multisite aware. The advantage to using a method over a variable is that we can now dynamically define the table property for our model.

###public static function conditions

This defines any conditions that should apply to every finder query that is generated by the model. For example if I wanted to model draft posts only:

    public static function conditions()
        return array("post_status = ?", "draft");

###public static function joins

This defines any joins that should apply to every finder query that is generated by the model. In general for related eager loading, I favor include, using the joins method only when my conditions method relies on data in another table. An example of how this can be used to model categories:


class Category extends swpMVCBaseModel
    public static function tablename()
        global $wpdb;
        return $wpdb->prefix.'terms';

    public static function conditions()
        global $wpdb;
        $tt = $wpdb->prefix.'term_taxonomy';
        return array("$tt.taxonomy = ?", 'category');
    public static function joins()
        global $wpdb;
        $t = self::tablename();
        $tt = $wpdb->prefix.'term_taxonomy';
        return "LEFT JOIN $tt ON $t.term_id = $tt.term_id";

Now any finder queries generated by the Category class will include a left join on the term_taxonomy table, and filter results to include only those where the term_taxonomy.taxonomy field has a value of "category." Filtering subsets with models is particularly relevant in WordPress, where different "types" of data are frequently lumped together in single tables.

One catch to using the joins method, is that calls to the models finder methods will need to use table prefixes for any columns that are present in both the main and joined tables. For this I recommend your Model::tablename() methods.

###Automatic stripslashes

Model properties in string format are automatically run through stripslashes when accessed directly. To override this, call the properties method on an object, and access the properties from the resulting array.


swpMVCBaseModel comes with an instance method 'render,' which accepts as an argument a Stamp view object (see Views section for details,) and autopopulates the Stamp using the model properties.

###public function render_{{property_name}}

These methods act as overrides for your properties when called by the render method. For example, if a Stamp view object passed to the render method contains a tag called post_name, and your Post model has a method called render_post_name, the return value of that method will be used to populate the Stamp object in favor of the value of the instance property post_name.

###public function needs_template_cleanup

This method accepts two parameters, property_name and property_value. You can overload this method in your models to determine which values for a property will cause a Stamp tag of the format {{property_name}}_block to be stripped from the template when rendered. Here's an example that will strip _block tags for any falsy value, or a custom value if the property is post_title:

    public function needs_template_cleanup($property_name, $property_value)
        if ($property_name === 'post_title')
            return $property_value === false or
                trim($property_value) === '' or
                $property_value ===
                    'This post title strips the post_title_block tag';
        return $property_value === false or
            empty($property_value) or
            trim($property_value) === '';

As shown in the example above, the {{property_name}}_block tag is stripped when $model->needs_template_cleanup() returns true. The default behavior is to check for falsy values.

###public function renderers

This method allows you to define additional renderers (not limited to model properties) that the model should be able to handle. It must return an array where the keys are the Stamp tags to be replaced, and the value is an array where the first element is the name of an instance method on the model which will return the value to be used in view population, and the second element is an array of arguments, (an empty array if no arguments to be passed.)

For example:


    public function renderers()
        return array(
            'additional_render_tag' => array('additional_render_method', array())

This would replace a Stamp tag labeled 'additional_render_tag' with the result of calling $model->additional_render_method() with no arguments.

###public static function controls

This method allows you to define form controls that should be used to interact with model properties, which will then be used by the render method to populate any Stamp tags of the format control_{{property name}} and additionally control_label_{{property name}} if the label attribute is defined for the corresponding element in the returned array. The return value from this method is again an associative array of arrays which must follow the structure below:


    public static function controls()
        return array(
            'property_name' => array(
                    'type' => 'input',
                    'label' => 'Property Name',
                    'input_type' => 'button'

The following must be true of any element in the controls array for it to be valid:

  • The key must match the model property that the control corresponds to.
  • type must be either input, select, or textarea
  • input_type is optional and only applies when type is set to input. Default is text.
  • label is optional. control_label_{{property name}} tags will not be replaced if no label property is defined
  • If type is select, an additional element is required with key 'options'. The value of this element must be an associative array of options for the dropdown, where key is the text for the option, and value is the value when that option is selected.

An example of a valid select control:


    public function controls
        return array(
            'property_name' => array(
                    'type' => 'select',
                    'label' => 'Property Name',
                    'options' => array(
                        'Option 1 Text' => 'option_1_value',
                        'Option 2 Text' => 'option_2_value'

When called by the $model->render() method, the generated controls will have their values set according to the values of the model instance on which the render method was invoked, and the form will appear populated.

###Form prefixes

Each model rendering a form adds a prefix to its form elements. By default when invoked via $model->render(), the form prefix will be the class name of the model instance invoking the render method. To override this, set the '_prefix' property on the model instances form helper before calling render, as follows:

    $model->form_helper()->_prefix = 'my_custom_form_prefix';


This method can be called statically on a model class to render an empty form for the class properties. The method accepts two parameters. The first parameter is required and must be a valid Stamp view object to be populated, the second is an optional form prefix (defaults to the class name.)

###A note about "through" relationships

Personally I've not had much success declaring through relationships with my activerecord models, so I have stuck to nesting includes during my finder calls manually. While the through relationship does improve efficiency, I've decided it's not worth the trouble, as even without the generated joins you can typically get all the data you will need for less than 10 queries with simple nested eager loading. Here's an example of getting 10 posts with their related post tag and category data using simple nested eager loading:

    $posts = Post::all(array('limit' => 10,
            'include' => array('postterms' =>
                array('termtaxonomy' =>

If you do have some success modeling with the through relationships, please reach out via Github issues and I'll be happy to update the docs, repo as needed.

###Included models

  •   Post
  • PostMeta
  • TermRelationship
  • TermTaxonomy
  • Term
  • Comment
  • User
  • UserMeta

All models can be found in models/wordpress_models.php. There's not alot of code, and the best way to learn model definition, as well as see what added methods are available on each model is to view the source.

##Model meta

the swpMVCBaseModel class includes some methods for working with meta, a popular WordPress data structure. Any table with columns foreign_key, meta_key, meta_value will work with these methods, once you've defined a $has_many relationship to a model with the name meta.


This method will return an empty array if there is no $has_many relationship named meta defined for the model on which it is called.

If the meta relationship is defined, it must point to a Model of a table with columns foreign_key, meta_key and meta_value. By calling $model->meta() when these circumstances are met, the return value will be an associative array where keys are equal to meta_key, and values are equal to the meta_value. In the case of duplicate meta_key rows for one model instance, the meta_value will be an indexed array of all values found.

It is recommended to eager load the meta when querying for your models if you intend to use this method, to avoid the n+1 query problem

The method accepts two parameters, $key and $raw. Passing in $key returns the meta value where meta_key matches the provided key. Passing in true for the $raw parameter will return an array of actual meta objects, as opposed to the hydrated meta_values.

###public function hydrate_meta

When working with ActiveRecord Models, WordPress will not serialize and unserialize data automatically for you. This method accepts two parameters, meta_key and meta_value, and gives you the opportunity to modify meta values as necessary when retreiving them using the $model->meta() method. (This does not apply when the $raw parameter is passed as true.)

Here's an example that will unserialize meta when the key is equal to 'serialized_meta':

    public function hydrate_meta($meta_key, $meta_value)
        return ($meta_key === 'serialized_meta') ?
            unserialize($meta_value) : $meta_value;

###public function dehydrate_meta

This method serves the opposite purpose of hydrate_meta, and will be invoked on any values assigned to the 'meta_value' property of a meta object. For a class with dehydrate_meta defined as follows:


    public function dehydrate_meta($meta_key, $meta_value)
        return ($meta_key === 'serialized_meta') ?
            serialize($meta_value) : $meta_value;

Calling the below on a model instance of the class would make the following boolean statment true:


    $model->meta_value = array('one', 'two', 'three');
    $model->meta_value === 'a:3:{i:0;s:3:"one";i:1;s:3:"two";i:2;s:5:"three";}';


swpMVC Views are based off of an older version of Gabor DeMooij's Stamp library, using extremely simple principles. Your views will contain no logic at all, and in most cases will be completely valid HTML on their own.

###Stamp tags

Stamp tags take the form of html comments, with an opening and closing comment representing one replaceable block. For example:

<a href="<!-- url --><!-- /url -->">Link to somewhere</a>

gives you a region labeled url that can be manipulated from the stamp object.


If you place the above template code in a file called template.tpl, and call the below code from your controller:

    echo $this->template('template')->replace('url', '');

the result would be:

<a href="">Link to somewhere</a>


This method allows you to copy defined template regions. Given the below template in file template.tpl:

<p>Here's some stuff</p>
<!-- more_stuff -->
    <p>And here's some more stuff</p>
<!-- /more_stuff -->

The below code would yield true at the boolean in the last statement:


    $more_stuff = $this->template('template')->copy('more_stuff');
    $more_stuff === "<p>And here's some more stuff</p>";

This is useful when populating one view with multiple models. For example, given the below template in file post.tpl:

<h1><!-- post_title --><!-- /post_title --></h1>
<!-- author_data -->
    by <!-- display_name --><!-- /display_name -->
<!-- /author_data -->

The below controller code would replace post_title with the title of the post object, and display_name with the display name of the post author:

    $post = Post::first(array('include' => 'user');
    echo $post->render($this->template('post'))

###Populating views with $model->render()

When using the $model->render() method, your model will automatically replace tags named according to the following conventions:

<!-- attribute_name --><!-- /attribute_name -->

The above gets replaced with a model property named attribute_name, or the return value of model instance method render_attribute_name if such method exists.

<!-- attribute_name_block -->
    <!-- attribute_name --><!-- /attribute_name -->
<!-- /attribute_name_block -->

The above gets replaced with a model property named attribute_name, or the return value of model instance method render_attribute_name if such method exists. If the value returns true when passed through $model->needs_template_cleanup(), the entire attribute_name_block section is stripped from the template when rendered.

<!-- control_attribute_name --><!-- /control_attribute_name -->

The above gets replaced with the 'attribute_name' element of the array returned by static class method controls

<!-- control_label_attribute_name --><!-- /control_label_attribute_name -->

The above gets replaced with the value of the 'label' key under the 'attribute_name' key of the array returned by static class method controls


With well defined models, controllers can be relatively sparse. Your controller classes must extend swpMVCBaseController, which equips them with the following functionality.


Set this property in your controller method to what you want the title attribute on the generated page to be. This needs to be set before you call get_header().


Set this property to the directory where your views for the current controller are stored, including a trailing slash. Best practice is to define this in the constructor, in which case you'll want to make sure to call the parent constructor as well:

    public function __construct()
        $this->_templatedir = dirname(__FILE__).'/../views/';


Set this property to an array where each element is an array of arguments to be passed to wp_enqueue_script. Define the property before calling get_header() to have your scripts automatically enqueued on that page.


Same as $this->_scripts, except each element of this array should be an array of arguments for wp_enqueue_style.


Same as _styles and _scripts, except each element of this array should be an array of arguments for wp_localize_script.


Requires $this->_templatedir to be defined. Accepts the filename of a template (minus the file extension, which must be .tpl,) and returns a Stamp view object for population and rendering. Here's an example of using the template method to pass a view to a models render method:

    $post = Post::first();

In the above example, we assume that the controllers templatedir property points to a directory that contains a file called show_post.tpl, which contains the correct Stamp tags to be populated by the Post model.


This method generates a WordPress 404 page using the currently selected WordPress theme. It must be called before any output is generated. Here's an example of using this method within a controller method:


    public function show_post($slug=false)
        if (!$slug) return $this->set404();
        $post = Post::first(array('conditions' => array('post_name = ?', $slug)));
        if (!$post) return $this->set404();

In this method, if no slug was passed to the controller method, we return a 404. If a slug was passed, we attempt to find a post using the provided slug. If we cannot, we return a 404. Only at that point if we have found a post using the provided slug do we begin to generate output from the controller method.


This method accepts three arguments, a controller class name, a method name, and an optional array of parameters. It will then return the corresponding url for that controller method. For example, if I've defined the following route in my plugin:


    $route = array('controller' => 'ControllerClass',
                        'method' => 'ControllerMethod',
                        'route' => '/url/of/route/:p/:p'

Then the below statement would be true:


                array('arg1', 'arg2')
            ) === get_bloginfo('url').'/url/of/route/arg1/arg2';

The link method is preferable to hard coding any fragment of a url into your views or controllers, since this will automatically update any references if you change the route definitions for your plugin.


This method accepts a string as a parameter, and will write that string as an E_USER_WARNING level error to your PHP log, as well as an error notice to the pQp Console if you are running in the development environment.

###public function before()

This method will run before any controller method is executed.

###public function after()

This method will run after any controller method is executed


Several features have been incorporated into the framework to further improve the development experience.

###PHP Quick Profiler

The PHP Quick Profiler is included with the framework. To activate, add the following code to your wp-config file:

    define('SW_WP_ENVIRONMENT', 'development');

The pQp will show on the bottom of every page, and by default will log load time, memory usage, file includes, and all SQL queries.

The following additional methods are available to write debug info to the console:


Accepts two parameters. First is a variable to be dumped, second is an optional name. This writes a general information entry to the console.


Accepts same parameters as Console::log(). This writes an error entry to the console.

While best practice is to clean up any debug code you write after you don't need it anymore, leaving calls to Console::log() or Console::error() in your code will not cause any issues if you run that code without your SW_WP_ENVIRONMENT set to development.


In cases where you want to debug queries that cannot be seen by the pQp (typically queries run on ajax calls,) the following code will cause all queries to be written to your PHP log at level E_USER_WARNING:


    define('SW_LOG_QUERIES', true);

You can then set this to false again to only log a specific set of queries within your code.

###Underscore PHP

The Underscore PHP library is an immensely useful set of tools for eliminating bulky boilerplate code used for common data manipulations in favor of more elegant, concise alternatives, largely taking advantage of the closure support introduced in PHP > 5.3. To avoid naming clashes with WordPress' __ function, the included copy has been modified and can be accessed only via the class methods, using a single underscore to reference the class. For example:


    echo _::reduce(array('that\'s', 'all', 'for', 'now'),
        function($memo, $a) { return $memo.' '.$a; }, '');


The following open source works have been instrumental in the creation of swpMVC, either as starting points, or having been included in their entirety within the framework: