Index and search Laravel models on Amazon's CloudSearch. To get started, you should have a basic knowledge of how CloudSearch works.
From the command line run:
$ composer require torann/laravel-cloudsearch
Once installed you need to register the service provider with the application. Open up config/app.php
and find the providers
key.
'providers' => [
LaravelCloudSearch\LaravelCloudSearchServiceProvider::class,
]
For Lumen register the service provider in bootstrap/app.php
.
$app->register(LaravelCloudSearch\LaravelCloudSearchServiceProvider::class);
Run this on the command line from the root of your project:
$ php artisan vendor:publish --provider="LaravelCloudSearch\LaravelCloudSearchServiceProvider" --tag=config
A configuration file will be publish to config/cloud-search.php
.
The package uses a batch queue system for updating the documents on AWS. This is done to help reduce the number of calls made to the API (will save money in the long run).
php artisan vendor:publish --provider="LaravelCloudSearch\LaravelCloudSearchServiceProvider" --tag=migrations
Run this on the command line from the root of your project to generate the table for storing currencies:
$ php artisan migrate
The better help manage fields, the package ships with a simple field management command. This is completely optional, as you can manage them in the AWS console.
NOTE: If you choose not to use this command to manage or setup your fields, you will still need to add the field
searchable_type
as aliteral
. This is used to store the model type.
They can be found in the config/cloud-search.php
file under the fields
property:
'fields' => [
'title' => 'text',
'status' => 'literal',
],
search:fields
Initialize an Eloquent model map.
search:index <model>
Name or comma separated names of the model(s) to index.
Arguments:
model Name or comma separated names of the model(s) to index
search:flush <model>
Flush all of the model documents from the index.
Arguments:
model Name or comma separated names of the model(s) to index
search:queue
Reduces the number of calls made to the CloudSearch server by queueing the updates and deletes.
Once you have added the LaravelCloudSearch\Eloquent\Searchable
trait to a model, all you need to do is save a model instance and it will automatically be added to your index when the search:queue
command is ran.
$post = new App\Post;
// ...
$post->save();
Note: if the model document has already been indexed, then it will simply be updated. If it does not exist, it will be added.
To update an index model, you only need to update the model instance's properties and `save`` the model to your database. The package will automatically persist the changes to your search index:
$post = App\Post::find(1);
// Update the post...
$post->save();
To remove a document from your index, simply delete
the model from the database. This form of removal is even compatible with soft deleted models:
$post = App\Post::find(1);
$post->delete();
You may begin searching a model using the search
method. The search method accepts a single string that will be used to search your models. You should then chain the get
method onto the search query to retrieve the Eloquent models that match the given search query:
$posts = App\Post::search('Kitten fluff')->get();
Since package searches return a collection of Eloquent models, you may even return the results directly from a route or controller and they will automatically be converted to JSON:
use Illuminate\Http\Request;
Route::get('/search', function (Request $request) {
return App\Post::search($request->search)->get();
});
In addition to retrieving a collection of models, you may paginate your search results using the paginate
method. This method will return a Paginator
instance just as if you had paginated a traditional Eloquent query:
$posts = App\Post::search('Kitten fluff')->paginate();
You may specify how many models to retrieve per page by passing the amount as the first argument to the paginate
method:
$posts = App\Post::search('Kitten fluff')->paginate(15);
Once you have retrieved the results, you may display the results and render the page links using Blade just as if you had paginated a traditional Eloquent query:
<div class="container">
@foreach ($posts as $post)
{{ $post->title }}
@endforeach
</div>
{{ $posts->links() }}
Initialize a builder instance:
$query = app(\LaravelCloudSearch\CloudSearcher::class)->newQuery();
You can chain query methods like so:
$query->phrase('ford')
->term('National Equipment', 'seller')
->range('year', '2010');
use the get()
or paginate()
methods to submit query and retrieve results from AWS.
$results = $query->get();
In the example above we did not set the search type, so this means the results that are returned will match any document on CloudSearch domain. To refine you search to certain model, either use the model like shown in the example previously or use the searchableType()
method to set the class name of the model (this is done automatically in the model instance call):
$query = app(\LaravelCloudSearch\CloudSearcher::class)->newQuery();
$results = $query->searchableType(\App\LawnMower::class)
->term('honda', 'name')
->get();
You can use the and
, or
, and not
operators to build compound and nested queries. The corresponding and()
, or()
, and not()
methods expect a closure as their argument. You can chain all available methods as well nest more sub-queries inside of closures.
$query->or(function($builder) {
$builder->phrase('ford')
->phrase('truck');
});
The help reduce the number of bulk requests made to the CloudSearch endpoint (because they cost) a queue system is used. This can be set in Laravel Task Scheduling. You can decide how often it is ran using the scheduled task frequency options. Please note this uses the DB to function.
Example of the task added to /app/Console/Kernel.php
:
/**
* Define the application's command schedule.
*
* @param \Illuminate\Console\Scheduling\Schedule $schedule
*
* @return void
*/
protected function schedule(Schedule $schedule)
{
$schedule->command('search:queue')->everyTenMinutes();
}
This feature is experimental
Laravel CloudSearch can support multiple languages by appending the language code to the index type, so when the system performs a search it will only look for data that is on in the current system locale suffixed index type. For this to work the model needs to use the LaravelCloudSearch\Eloquent\Localized
trait or something similar to it.