Configuration reference

Settings live in config/meilisearch-connect.php.

Root settings#

Key	Default	Purpose
`meiliHostUrl`	`null`	Meilisearch server URL.
`meiliAdminApiKey`	`null`	Key used for settings, indexing, and deletes.
`meiliSearchApiKey`	`null`	Key used for searches.
`maxDependencyRecursionLevel`	`4`	Maximum follow-up sync depth for dependencies.
`garbageCollectionAge`	`'1 day ago'`	Age of failed refresh indexes that Craft garbage collection can remove. Use a PHP relative date string or `null` to remove all of them.
`indices`	`[]`	Index definitions keyed by handle.

Index handles and IDs#

The array key is the index handle. Use it in commands and searches. indexId is the index name in Meilisearch itself. If not set, indexId defaults to the handle.

'indices' => [
    'pages' => IndexBuilder::create()
        ->withIndexId('website_pages')
        ->build(),
],

Use a different indexId in each environment when several environments share one Meilisearch server.

.env:

...
MEILISEARCH_PAGES_INDEX=pages_local
# MEILISEARCH_PAGES_INDEX=pages_production

config.php:

use craft\helpers\App;

'indices' => [
    'pages' => IndexBuilder::create()
        ->withIndexId(App::env('MEILISEARCH_PAGES_INDEX') ?? 'pages')
        ->build(),
],

Search-only configuration#

Use a search-only configuration when another service indexes data in Meilisearch, but Craft still needs to search it.

The most basic configuration is to simply specify the index handle. This plugin only needs to know where to look in Meilisearch to enable searching.

return [
    'meiliHostUrl' => 'http://localhost:7700',
    'meiliSearchApiKey' => '<Meilisearch Search Key>',
    'indices' => [
        'pages' => IndexBuilder::create()->build(),
    ],
];

Full configuration#

Use a full configuration when this plugin indexes data. A managed index has a fetch callback, which withElementQuery() sets up for you. You need an admin API key to sync its settings and documents.

Index builder#

Method	Purpose
`withIndexId(string $id)`	Set the Meilisearch index ID.
`withPageSize(int $size)`	Number of source records fetched at once. Default: `100`.
`withElementQuery($query, $transformer)`	Index Craft elements.
`withFetchFn(callable $fetch)`	Index custom data. Not needed if `withElementQuery` is used.
`withPagesFn(callable $pages)`	Provide progress for a custom fetch. Not needed if `withElementQuery` is used.
`withNameFn(callable $name)`	Set the source name shown in queue jobs. Not needed if `withElementQuery` is used.
`withAutoSync(bool $enabled)`	Enable or disable automatic element sync. Default: `true`.
`withActiveStatuses(array $statuses)`	Statuses that count as active during automatic sync. Default: `Element::STATUS_ENABLED` and `Entry::STATUS_LIVE`.

Use an ElementQueryInterface instance or a callable that returns one with withElementQuery(). Prefer a callable so that queries don’t execute during bootstrap. This is the case when using filters such as site($siteHandle) which does a lookup query to get the site ID.

Callable arguments#

`withElementQuery($query, $transformer)`

$query can be an ElementQueryInterface or a no-argument callable that returns an ElementQueryInterface.

The transformer receives:

Argument	Type	Purpose
`$element`	`ElementInterface`	The current result from the query. Use its concrete element type in your callback.
`$registerDependency`	`callable(ElementInterface): void`	Call this for an element whose data is included in the document.

Return one document array, a list of document arrays, null, or [].

withElementQuery() handles the setup of withFetchFn, withPagesFn, and withNameFn for element indexes. You only need to use those methods for custom data or when element queries are not sufficient.

`withFetchFn($fetch)`

The fetch callback receives:

Argument	Type	Purpose
`$index`	`Index`	The configured index.
`$sourceHandle`	`null	string
`$extra`	`mixed`	Extra data passed to the fetch operation.

Return one DocumentList, an array of DocumentList objects, or a generator that yields arrays of DocumentList objects.

`withPagesFn($pages)`

The pages callback receives one argument:

Argument	Type	Purpose
`$index`	`Index`	The configured index.

Return the number of expected fetch batches. This is used for queue progress.

`withNameFn($name)`

The name callback receives:

Argument	Type	Purpose
`$index`	`Index`	The configured index.
`$sourceHandle`	`string	int`
`$extra`	`mixed`	Extra data passed to the name operation.

Return a string for the queue job description.

Meilisearch settings#

Pass IndexSettingsBuilder::create() to IndexBuilder::fromSettings().

The builder has methods for the Meilisearch index settings: primary key, searchable, filterable, sortable, and displayed attributes; ranking rules; faceting; pagination; synonyms; typo tolerance; stop words; dictionary; tokens; and embedders.

IndexSettingsBuilder::create()
    ->withPrimaryKey('id')
    ->withSearchableAttributes(['title', 'body'])
    ->withFilterableAttributes(['section'])
    ->withSortableAttributes(['postDate'])
    ->withPagination(['maxTotalHits' => 1000])
    ->build()

Except for primaryKey, the plugin resets a setting left as null when sync/settings runs. Check the Meilisearch settings reference for valid values.

Meilisearch Connect