sheikhheera / requent
A GraphQL like interface to map a request to eloquent query with data transformation.
Requires
- php: >=5.6.4
- illuminate/config: >=5.3
- illuminate/database: >=5.3
- illuminate/http: >=5.3
- illuminate/pagination: >=5.3
Requires (Dev)
- fzaninotto/faker: ^1.6
- phpunit/phpunit: 5.7.*
This package is auto-updated.
Last update: 2024-12-20 12:28:31 UTC
README
An elegant, light-weight GQL (Graph Query Language) like interface for Eloquent with zero configuration. It maps a request to eloquent query and transforms the result based on query parameters. It also supports to transform the query result explicitly using user defined transformers which provides a more secured way to exchange data from a public API with minimal effort.
- Installation
- How It Works
- Basic Example
- Resource
- Methods
- Resource Key By
- Data Filtering Using Transformers
- Get Raw Result
- Query Modifier Clause
- Customizations
Installation
You can simply run the following command from your terminal to install the package:
composer require sheikhheera/requent
Or add the following line in "composer.json" file within "require" section and run composer install
from terminal:
"sheikhheera/requent": "1.0.*"
If you are using version 5.5 or greater, then you are all done, no need to add service provider or alias in
config\app
. So you can simply skip the following steps because of Laravel's package auto discovery feature.
This will install the package. Now add the following entry in your config/app.php
file inside the providers
section:
Requent\RequentServiceProvider::class
Also add the following entry in aliases
section of your config/app.php
file:
'Requent' => Requent\Facade\Requent::class,
If you've done everything right then you can start using it without any configuration but you may customize it.
How It Works
This package will allow us to query resources through the request query string parameter. For example, if we've a User
model and the User
model has many posts (Post
model) and each post has many comments (Comment
model) then we can query the users with their posts and comments of each posts by sending a request like the followig: http://example.com/users?fields=posts{comments}
. This is the most basic use case but it offers more. We can also select properties of each model through query string, for example, if we want to select only the emal field from the User
model and title from Post
and body from the Comment
model then we can just do it by sending a request using the following URL
:
http://example.com/users?fields=email,posts.orderByDesc(id){title,comments{body}}
It'll be translated into something similar to following (Not literally):
User::select('email') ->with(['posts' => function($query) { $query ->orderByDesc('id') ->select('title') ->with(['comments' => function($query) { $query->select('body'); }]) }]);
Basic Example
To use this package, we need to create some resources (Eloquent Models). For this demonstration, we'll use the same idea using User, Post and Comment models for an imaginary blog. The User model has a hasMany
relation for posts and the Post model has a hasMany
relation for comments. So, we need a route, which could be a resourceful route but we'll use an explicite route declaration here:
Route::get('users', 'UserController@index');
Now, we need a controller which is just a simple controller, for example:
<?php namespace App\Http\Controllers; use Requent; use App\User; use App\Http\Controllers\Controller; class UserController extends Controller { public function index() { return Requent::resource(User::class)->get(); } }
Now, we can make a request using: http://example.com/users?fields=email,posts{title,comments{body}}
. This will give us the expected result, which would be an array of users (only email column from User
) with all the related posts (only title column from Post
) and all the comments of each post (only body column from Comment
).
If we want to load any resource with relations without selecting any properties then we can just do it using the following request: http://example.com/users?fields=posts{comments}
. This was the most basic example but let's explore it's features.
Resource
Actually, a resource is just an eloquent model, the first method we should call on the Requent
class is resource
which sets the primary resource we want to query on. So we can set the resource using couple of ways, for example:
$resource = Requent::resource(User::class);
Also, we can use an object, for example:
$resource = Requent::resource(new User);
We can also pass a Query Builder
for example:
$resource = Requent::resource(app(User::class)->where('role', 'admin'));
So, we can call any scope methods as well, which just returns a Query Builder
instance. The resource
method returns the Requent
object so we can chain methods, for example, we can call any query executing method (including other available methods in Requent
), for example:
$result = Requent::resource( app(User::class)->where('role', 'admin') ) ->transformBy(UserTransformer::class) ->keyBy('users') ->get();
We'll walk-through all the available methods and features that Requent
offers. Let's continue.
Methods
Get
We've seen get
method earlier which just returns an array of users which is:
return Requent::resource(User::class)->get();
Paginated Result
At this point, we'll get an array but we can retrieve paginated result using same get
method and in this case we, we only need to provide a query string parameter in our URL
like the following example:
http://example.com/users?fields=posts{comments}&paginate
Also, we can set the paginator, for example:
http://example.com/users?fields=posts{comments}&paginate=simple
This will return the paginated result using SimplePaginator
but by default it'll use LengthAwarePaginator
.
Per Page
We can also tell how many pages we want to get for per page and it's just another parameter, for example:
http://example.com/users?fields=posts{comments}&paginate=simple&per_page=5
If we provide per_page=n
then we don't need to provide &paginate
parameter unless we want to use the simple paginator instead of default
. We can also customize these parameters, we'll check later on.
Paginate
Also, we can call the paginate
method on the Requent
directly, for example:
return Requent::resource(User::class)->paginate(); // or paginate(10)
Simple Paginate
The simplePaginate
will return paginated result using Simple Paginator
. Check Laravel Documentation.
return Requent::resource(User::class)->simplePaginate(); // or simplePaginate(10)
Find
If we want to retrieve a single user
then we can use find
and first
method, for example:
return Requent::resource(User::class)->find($id);
First
For the first item we can call the first
method:
return Requent::resource(User::class)->first();
Fetch
These are all the available methods for executing query but there is one more method which is fetch
. This method can return any kind of result, an collection (array), paginated result or a singlr resource. Let's see an example:
// In Controller public function fetch($id = null) { return Requent::resource(User::class)->fetch($id); }
To use this method we need a route like: Route::get('users/{id?}', 'UserController@fetch')
and then we can use this single route to get all kind of results, for example:
Get a collection of users (Array)
http://example.com/users?fields=posts{comments}
Get paginated result
http://example.com/users?fields=posts{comments}&paginate=simple&per_page=5
Get a single user (Array)
http://example.com/users/1?fields=posts{comments}
The fetch
method will be useful if we declare explicit route other than RESTfull routes for Resource Controllers. Check Laravel Documentation.
When selecting properties of a model/resource using query string, i.e:
fields=name,posts{title}
, we can select a dynamic property (getter/accessor), defined using agetPropertyAttribute
method. Check the documentation for Defining An Accessor.
Resource Key By
The query results for a collection is simply an array with a zero based index but if we want then we can wrap our collection in a key using keyBy
method, for example:
return Requent::resource(User::class)->keyBy('users')->get();
This will return a collection of users (Array) as a key value pair where the key will be users
and the result will be the valuse of that key. We can also use a key for a single user for example:
return Requent::resource(User::class)->keyBy('user')->find(1);
In case of fetch
we can use something like the following:
public function fetch($id = null) { return Requent::resource(User::class)->keyBy($id ? 'user' : 'users')->fetch($id); }
The paginated result will remain the same, by default Laravel
wraps the collection using the data
as key.
Data Filtering Using Transformers
The idea of transformers is taken from Fractal Transformer package. This looks like re-inventing the wheel but actually it's not. The main intention for building the
Requent
package was to allow an easy to use interface for fetching resource/data form a web application (non-publicAPI
), which allows to read data from server using anyjavaScript
framework/library even without defining any transformers. Also, theEloquent
query is built dynamically on the run-time to load everything eagerly, whileFractal
uses lazy loading. So, theRequent
couldn't utilize the data transforming feature thatFractal
offers. So, to provide the data filtering layer (for publicAPI
), theRequent
needed it's own data filtering mechanism but theFractal
package is great and I've used it exclusively on my projects.
So far we've seen the default data transformation, which means that, a user can get any property or available relations of the resource just by asking it through the query string parameter fields
(we can use something else other than fields
), but there is no way to keep some data private if you are using this for a public API
. Here, the transformer
comes into play.
By default, the Requent
uses a DefaultTransformer
class to return only selected properties/relations, for example, if you send a request using a URL
like following: http://example.com/users?fields=email,posts{title,comments}
then it'll return only selected properties/relations. In this case, it'll return what you ask for it but you may need to define explicitly what properties/relations a user can get from a request through query parameter. For this, you can create a custom transformer where you can tell what to return. To create a trunsformer, you just need to create transformer classes by extending the Requent\Transformer\Transformer
class. For example:
<?php namespace App\Http\Transformers; use Requent\Transformer\Transformer; class UserTransformer extends Transformer { public function transform($model) { return [ 'id' => $model->id, 'name' => $model->name, 'email' => $model->email, ]; } }
To use your custom transformer, all you need to pass the transformer class to the resource
method, i.e:
return Requent::resource(User::class, UserTransformer::class)->fetch($id);
Also you can pass the class instance:
return Requent::resource(User::class, new UserTransformer)->fetch($id);
Also, you can set the transformer using transformBy
method:
return Requent::resource(User::class)->transformBy(UserTransformer::class)->fetch($id);
Also, the transformBy
method accepts a transformer object instance:
return Requent::resource(User::class)->transformBy(new UserTransformer)->fetch($id);
This will transform the resource using by calling the transform
method defined in the transformer class you created. In this case, the transform mathod will called to transform the User
model but right now, it'll not load any relations. Which means, if the URL
is something like: http://example.com/users?fields=posts{comments}
then only the User
model will be transformed and the result would be something like the following:
[
{
id: 1,
name: "Aurelio Graham",
email: "hharvey@example.org"
},
{
id: 2,
name: "Adolfo Weissnat",
email: "serena78@example.com",
}
// ...
]
To load any relations from the root transformer (UserTransformer
in this case), we also need to explicitly declare a method using the same name the relation is defined in the model, so for example to load the related posts with each User
model we need to declare a posts
method in our UserTransformer
class. For example:
class UserTransformer extends Transformer { public function transform($model) { return [ 'id' => $model->id, 'name' => $model->name, 'email' => $model->email, ]; } // To allow inclussion of posts public function posts($model) { return $this->items($model, PostTransformer::class); } }
In this example, we've added a filtering for Post
model and that's why a user can select the related posts with users from the URL
, for example: http://example.com/users?fields=posts
. without the posts
method in UserTransformer
a user can't read/fetch the posts relation. At this point, we are not done yet. As you can assume that, we are transforming the posts (Collection) using items
method available in UserTransformer
(extended from abstract Transform class) and passing another transformer (PostTransformer) to transform the collection of posts. So, we need to implement the PostTransformer
and have to implement the transform
method where we'll explicitly return the transformed array for each Post
model, for example:
namespace App\Http\Transformers; use Requent\Transformer\Transformer; class PostTransformer extends Transformer { public function transform($model) { return [ 'post_id' => $model->id, 'post_title' => $model->title, 'post_body' => $model->body, ]; } // User can select related user for each Post model public function user($model) { return $this->item($model, new UserTransformer); } // User can select related comments for each Post model public function comments($collection) { return $this->items($collection, new CommentTransformer); } }
In this example, we've implemented the transform
method for the Post
model for response filtering so only the id
, title
and body
column will be available for the Post
model in the response and the related posts will be included only if the user selects the posts through the query string parameter in the URL
.
In the exmple given above, we've also defined two additional methods, user
and comments
. Those methods are also relations of Post
model. The user
method is defined as a belongsTo
relationship which simply maps the related user who published the post and the comments method loads the related comments published under the post. The Post
model looks something like the followin:
namespace App; use App\User; use App\Comment; use Illuminate\Database\Eloquent\Model; class Post extends Model { public function user() { return $this->belongsTo(User::class); } public function comments() { return $this->hasMany(Comment::class); } }
According to the setup given above, we can make a request using the following URL
to get all the users with posts and it's related comments and user: http://example.com/users?fields=posts{user,comments}
.
In the PostTransformer
class we've used item
method inside user
method, which actually resieves a single Eloquent
model in $model
parameter and so we've called the item
method from the transformer. In the comments method, we've used the items
method because the $collection
parameter in comments method recieves a collection of Comment
models.
So, it's obvious that, to allow the inclussion of any relation from a resource we've to declare a method for that relation using the same name we've used to declare the relation in the Eloquent
model and relations will be included only if the user selects/includes it within the fields
parameter. If user selects a relation from a resource that is not exposed throught the transformer using a method, then it'll not be available in the response.
The user defined transformers will be used to transform the data only a transformer class is passed as the second parameter in the
resource
method or by calling thetransformBy
method, otherwise, everything will be included in the result/response the user asked for (if those fields/relations are available in the corresponding model).
Get Raw Result
Requent has a raw
method which could be useful if someone doesn't want to apply any transformation, because after the transformation, the returned data is an array. So if you want to execute the query but you want to ommit the data transformation by default (selection of columns through the query string) then you can use raw
method, for example:
$result = Requent::resource(User::class)->raw()->fetch($id);
In this case, when you don't provide a custom transformer to transform data then the requent will transform the data using the defalt transformer. So, if you make a request using http://example.com/users?fields=email,posts{title}
, then it should return only email
from the User
model and title
from the Post
model.
In this case, because of raw
, the requent will execute the query to load the resource with mentioned relations but it'll not filter the result so the original result returned by the Eloquent
(could be a collection, paginated data or a modeld) will be returned as the result of Requent
query.
Transform Raw Result
use Requent; use App\User; use App\Http\Controllers\Controller; use Requent\Transformer\TransformerHelper; use App\Http\Transformers\UserTransformer; class HomeController extends Controller { use TransformerHelper; public function index() { $result = Requent::resource(User::class)->raw()->get(); return $this->transform($result, UserTransformer::class, 'users'); } }
In this example, the third parameter users
passed to transform method is optional, which would be used as the resource key.
The
Requent
doesn't use lazy loading, which means that, everything (relations) is eager loaded and when it comes to transformer (transform method), the quwey has been already executed and everything is loaded. So, at this point, thetransform
method will recieve a model and and if we want to load other relations that was not loaded through the query, then we can simply call theload
method on the model before transforming it, for example:
public function transform($model) { if(!$model->relationLoaded('comments')) { $model->load('comments'); } return [ 'post_id' => $model->id, 'post_title' => $model->title, 'post_body' => $model->body, ]; }
Query Modifier Clause
When we make a request, we can add some query modifier clauses for example, orderBy
, orderByDesc
e.t.c. There are several clauses that Requent
offers to use in the URL
, those are given below:
orderBy
http://blog54.dev/1?fields=posts{user,comments.orderBy(id){user}}
orderByDesc
http://example.com/1?fields=posts{user,comments.orderByDesc(id){user}}
skip & take
http://example.com/1?fields=posts{user,comments.skip(2).take(1){user}}
offset & limit
http://example.com/1?fields=posts{user,comments.offset(2).limit(1){user}}
Multiple Clauses
http://example.com/1?fields=posts.orderBy(title).limit(3){user,comments.orderByDesc(id).skip(2).take(1){user}}
When using limit/take on a relationship and loading a collection of that relation, you may get wrong result because the limit is applied only once on the query by Laravel. An old issue here.
Customizations
Requent uses some base settings from a config file. By default, it'll work as it's configured but if you need to modify any of the settings then you can publish the config file from vendor to your local app config directory. To publish the config, just execute the following command from your terminal:
php artisan vendor:publish --provider="Requent\RequentServiceProvider" --tag="config"
Once you publish the config file to your local /config
directory then you can modify any settings to customize Requent
for your need. Follwing code is taken from the config file which is documented itself.
/*
|--------------------------------------------------------------------------
| Query Identifier
|--------------------------------------------------------------------------
|
| Here you may define the parameter name to pass your query string
| to select fields/columns. By default, "fields" is set but you may
| override it if you wish.
|
*/
'query_identifier' => 'fields',
/*
|--------------------------------------------------------------------------
| Paginator Identifier
|--------------------------------------------------------------------------
|
| Here you may define the parameter name to get paginated data.
| By default, "paginate" is set but you may override it if you wish.
|
*/
'paginator_identifier' => 'paginate',
/*
|--------------------------------------------------------------------------
| Default Paginator
|--------------------------------------------------------------------------
|
| Here you may define the default paginator to be used when geting paginated
| result. By default, the length aware paginator will be used unless you override
| it here or pass the pagination type in the query string.
|
| Available Options: "simple", "default"
|
*/
'default_paginator' => 'default',
/*
|--------------------------------------------------------------------------
| Per Page Identifier
|--------------------------------------------------------------------------
|
| Here you may define the name of the query string to pass the number
| of pages you want to retrieve from paginated result. By default, the
| package will use "per_page" so you can pass ?per_page=10 to get 10 items
| per page unless you override it here.
*/
'per_page_identifier' => 'per_page',
/*
|--------------------------------------------------------------------------
| Default Attributes Selection
|--------------------------------------------------------------------------
|
| Here you may define whether you would like to load all properties/attributes
| from a model if no property was explicitly selected using the query string. If
| you just select relations of a model, the package will load all the attributes
| by default unless you override it here by setting the value to false.
*/
'select_default_attributes' => true,