-
Notifications
You must be signed in to change notification settings - Fork 14
Home
MTV is a plugin for WordPress that provides a new API for developing plugins and themes. Born out of frustration with the undocumented, inconsistent WordPress API, MTV provides a simple, familiar, consistent way to develop heavily customized WordPress sites.
MTV borrows a lot from existing MVC-style frameworks, namely Django and Backbone.js. If you're familiar with those frameworks, you should feel at home using MTV. If you're not familiar, go try some tutorials. It'll make you a better programmer.
This plugin hijacks and takes over how WordPress handles URLs and templates, and gives you new ORM-style tools for handling posts, users and blogs. This plugin does nothing by itself, and (hopefully) will not break stuff that you already have. It's just a set of tools for developers to use.
The way that WordPress handles requests and serves responses is not clear to anyone who is not well acquainted with the internals. This is fine for theme development and for simple plugin development. Great actually. Create template files in the correct spot with the right names; use the magical template tags in the right places in the right files, and you can quickly build a WordPress theme with the structure and style you want.
The problem with this, and the genesis of this project, is that once you try to step outside the lines of the standard theme and plugin development, the simplicity of WordPress unravels frighteningly quickly.
WordPress is magic. When you want to setup a quick blog, make it your own and start writing, WordPress is magic. But when you want to heavily customize it, adding post types, new URLs and rivers for those post types, new widgets, the list goes on and on, WordPress's magic goes awry.
We love WordPress, but we need it to be more serviceable. We need to be able to look under the hood and figure things out quickly. We need a code base that is simple, explicit, legible and self-documenting. That's what we're trying to do with MTV.
- Bare MTV
- My first theme
MTV is built around the concept of apps, an idea that comes from Django. An app is package of models, views, templates and other code that can mostly stand alone.
register_app \mtv\register_app( $name, $path_to_app )
This makes MTV aware of an app. This function doesn't load the app, it simply names the app for later reference.
run \mtv\run( $args )
run
makes MTV go. It takes an array containing three things: the current URL, an array of url patterns and their associated views, and an ordered array of apps to load.
The MTV wordpress plugin uses run
internally. Here's how it works:
- Checks to see if the currently active theme has a urls.php file. If the theme does not have a urls.php, but it has a parent theme, MTV looks for a urls.php file in the parent theme.
- Loads the
$url_patterns
variable from the urls.php file. - Looks for a global variable named
$apps
. - Calls
run
with$url_patterns
,$apps
, and$_REQUEST['path']
as the URL. -
run
loads all the code for the selected apps and configures the Twig template library. -
run
matches the url to one of the url patterns and executes the associated view function. - Internets!
If you're using the MTV plugin for WordPress development, you should never need to use this function directly. The more you know...
load \mtv\load( $apps )
If you ever need to use your models views or templates, but don't want to run MTV, use load
. You pass it an array of apps to load, and it includes the code and configures the twig template library, but doesn't execute any views.
The model layer is a collection of classes which handle the retrieval and manipulation data. Models are objects which represent a single item or row from a database.
In MTV, two types of classes comprise the model layer: Model
and ModelCollection
. A Model
represents a single item from the database; a post, user, category, etc., and contains all the logic for accessing data, saving to the database, validating changes, and retrieving related models or data. A ModelCollection
represents an array of a single type of Model
and contains all the logic needed to retrieve and filter one or more models from the database, as well as logic to work on multiple models at once.
Model classes are stored in the models.php
file of your
Constructor new Model( [$attributes] )
The constructor creates a real live model for you to use. It takes one optional parameter, an associative array of attributes to attach to the model.
$data = array(
"title" => "My blog post",
"published_on" => now(),
"author" => "Bobby Tables",
"content" => "I can blog!!"
);
$model = new Model($data);
print "<h1>" . $model->title . "</h1>";
print "by " . $model->author;
print "<p>" . $model->content . "<p>";
save $model->save()
Save this model to the database.
$model = new Model();
$model->title = "My blog post";
$model->save();
The base Model
class does not implement save
. It's up to you to extend the Model
class and write this function.
validate $model->validate()
Validate the attributes in this model. Call this before saving the model to verify that all of the attributes are correct.
$model = new Model();
$model->email_address = "bobby";
$model->validate();
// ERROR! "bobby" is not a valid email address
The base Model
class does not implement validate
. It's up to you to extend the Model
class and write this function.
initialize $model->initialize( $attributes )
Process the attributes when creating a new model. If you need to do something when a model object is created, you can implement this method instead of overriding the constructor. The base Model
class does not implement this.
clear $model->clear()
Delete all of the attributes on the model.
set $model->set( $attributes )
Takes an associative array of attributes and sets them on the model.
$data = array(
"title" => "My blog post",
"published_on" => now(),
"author" => "Bobby Tables",
"content" => "I can blog!!"
);
$model = new Model($data);
$model->set(array(
"title"=>"Better title",
"status"=>"draft"
))
print $model->status; // draft
print $model->author; // Bobby Tables
print $model->title; // Better title
fetch $model->fetch()
Populates the model's attributes from the database.
The base Model
class does not implement fetch
. It's up to you to extend the Model
class and write this function.
parse $model->parse( $data )
Parses the raw data from the database and returns an associative array of attributes for the model.
reload $model->reload( $data )
Takes a variable containing raw data from the database and loads it into the model's attributes. reload
calls parse
then set
.
to_json $model->to_json()
Returns an associative array containing a subset of the model's attributes that are appropriate to send over the wire as json. Does not return an actual JSON string.
to_json
uses an variable called json_fields
to determine which attributes it should return.
$data = array(
"name" => "Bobby Tables",
"birthday" => "02/01/00",
"secret_desire" => "space man"
);
$model = new Model($data);
$model->json_fields = array('name', 'birthday');
print json_encode($model->to_json()) // only shows name and birthday
from_json Model::from_json( $data )
set_from_json $model->set_from_json( $data )
The reverse of to_json
, takes an associative array of attributes, filters and sets them on the model. Does not accept an actual JSON string.
set_from_json
uses an variable called editable_json_fields
to determine which attributes it should set.
$json = '{
"name": "Bobby Tables",
"birthday": "02/01/00",
"secret_desire": "space man"
}'
$model = new Model();
$model->editable_json_fields = array("name", "birthday");
$model->set_from_json(json_decode($json));
print $model->name; // Bobby Tables
print $model->secret_desire; // Null
from_json
is a static function that you can use to create a new model object from an array
$json = '{
"name": "Bobby Tables",
"birthday": "02/01/00",
}'
$model = Model::from_json(json_decode($json));
print $model->name; // Bobby Tables
You can extend a Model
like you would any PHP class. When extending a model there are a few variables you might be interested in.
class MyModel extends Model {
// Default model data used when a new model is created
public $defaults = array();
// Attributes that are OK to send over the wire
public $json_fields = array();
// Attributes that can be updated over the wire
public $editable_json_fields = array();
}
A ModelCollection
represents a bunch of models. It has only a few functions but it acts as the gateway to retrieving models. You can treat a ModelCollection
object as an array.
Constructor $collection = new ModelCollection( [$attributes] )
add $collection->add( $model )
clear $collection->clear()
to_json $collection->to_json()
get $model = ModelCollection::get( $attributes )
get_by $model = ModelCollection::get_by( $attributes )
filter $collection = ModelCollection::filter( $attributes )
You can extend a ModelCollection
like you would any PHP class. There is only one thing you absolutely must put in your new class.
class MyModelCollection extends Collection {
public static $model = 'myapp\models\MyModel';
}
The static variable $model
must be the globally accessible name of the model class that this collection will be handling.
The Post
model represents a WordPress post in the database. It inherits and implements all of the features of the Model
class.
Here are the typical WordPress post fields you'll find on a Post
object:
id // ID of the post
post_author // ID of the post author
post_date // timestamp in local time
post_date_gmt // timestamp in gmt time
post_content // Full body of the post
post_title // title of the post
post_excerpt // excerpt field of the post, caption if attachment
post_status // post status: publish, new, pending, draft, auto-draft, future, private, inherit, trash
comment_status // comment status: open, closed
ping_status // ping/trackback status
post_password // password of the post
post_name // post slug, string to use in the URL
to_ping // to ping ??
pinged // pinged ??
post_modified // timestamp in local time
post_modified_gmt // timestatmp in gmt tim
post_content_filtered // filtered content ??
post_parent // id of the parent post. If attachment, id of the post that uses this image
guid // global unique id of the post
menu_order // menu order
post_type // type of post: post, page, attachment, or custom string
post_mime_type // mime type for attachment posts
comment_count // number of comments
filter // filter ??
The Post
model also adds these fields, not typical to WordPress:
post_meta // an array containing all of the meta for this post
post_format // the post_format of this post
blogid // id number of the blog this post lives on
url // attachments only, url of the original uploaded image or whatever
thumb_url // attachments only, url of the thumbnail image, if thumbnails are enabled
With the exception of the last three fields above (blogid, url, thumb_url), you can read, update and save any field using the Post
model:
$post = new Post(array('id'=>1));
$post->fetch();
$post->post_status = 'publish';
$post->post_title = 'Better headline';
$post->post_meta['my_special_meta'] = 'this post is special';
$post->post_meta['my_related_items'] = array(2,5,7);
$post->save();
The Post
model also has these extra special functions:
password_required $post->password_required()
is_sticky $post->is_sticky()
post_class $post->post_class( [$extra_classes] )
permalink $post->permalink()
categories $post->categories()
tags $post->tags()
the_time $post->the_time( [$format] )
the_date $post->the_date( [$format] )
make_excerpt $post->make_excerpt( [$more_text] )
Your gateway to the wonderful world of WordPress posts. PostCollection
only implements the basic features set out in the ModelCollection
above. With a couple twists. PostCollection
really is just a thin wrapper around WordPress's WP_Query object.
$query = array(
'post_type' => 'post',
'posts_per_page' => 10,
'order' => 'DESC'
);
$posts = PostCollection::filter( $query );
foreach ( $posts as $post )
print $post->post_title;
The PostCollection
object ($posts
in this example), holds on to the WP_Query
object for later reference. So if you need to see how many pages of posts you have:
$max_pages = $posts->wp_query->max_num_pages;
Not too shabby.
The view layer is the switching yard for all the requests your crazy WordPress/MTV website will receive. It's made up of two files: urls.php
and views.php
. Every MTV app can have a views.php and a urls.php
This file contains a single variable. Not a global variable, mind you, we're trying to keep those to a minimum. MTV includes your urls.php file only when it needs to read the url patterns.
$url_patterns = array(
"URL_REGEX" => "FUNCTION_CALLBACK"
)
That's all there is to it. $url_patterns
is an associative array where the key is a regular expression and the value is the name of the function to execute.
MTV's url resolver (the function that matches the current url to a url pattern) reads the $url_patterns
in order and uses the first view that matches.
The url resolver will also pass interesting bits of the matched url into your view function.
$url_patterns = array(
'/(?P<slug>[^\/]+)\/?$/' => 'myapp\views\page'
)
In the regex above, we capture part of the url using parentheses and name that captured part slug: (?P<slug>[^\/]+)
. When our url resolver executes our view function, it gives the view an array of the captured bits.
function page( $request ) {
print "I'm on page " . $request['slug'] . "!";
}
Fun fun.
If you have multiple apps, you will probably want to have a urls.php
in each app and separate stuff out. Maybe not, but we've found a need for it.
The MTV WordPress plugin will load the urls from your active theme, so what if you want to use urls from another app?
include_urls_for \mtv\http\include_urls_for( $app )
Include url patterns from another registered app.
$url_patterns = array(
\mtv\http\include_urls_for('myotherapp'),
'/(?P<slug>[^\/]+)\/?$/' => 'myapp\views\page'
)
Again, order is everything.
- urls.php
- Including urls from other places
- Writing a view
- Available wp template functions
- How to add a template functions, tags, filters
- Deep links to twig
- Setting up ajax urls
- Writing an ajax view
- Calling ajax