Meilisearch Integration for Website

Meilisearch Integration for Website Search

Meilisearch — full-text search engine written in Rust. Response time under 50ms on most corpora, built-in typo support, fuzzy search, faceted filtering. Installs as separate process beside main application.

When Needed

Standard LIKE '%query%' in PostgreSQL stops working around 50,000 records — query time grows and database load increases. Meilisearch solves differently: builds inverted index separately, queries bypass the database.

Suitable for:

• Product catalogs with attribute filters
• Blogs and knowledge bases (10,000+ articles)
• User, address, product directories in B2B cabinets
• Documentation search

Integration Architecture

Browser → Backend API → Meilisearch HTTP API
                    ↓
              PostgreSQL (data source)
              Indexer (Queued Job / Cron)

Meilisearch isn't main database replacement. Data lives in PostgreSQL, only search-needed data goes to Meilisearch. Sync via job queues on record changes or periodic index rebuild.

Installation and Configuration

Docker Compose:

services:
  meilisearch:
    image: getmeili/meilisearch:v1.7
    environment:
      MEILI_MASTER_KEY: "${MEILI_MASTER_KEY}"
      MEILI_ENV: production
    volumes:
      - meili_data:/meili_data
    ports:
      - "7700:7700"

Index attribute configuration:

{
  "searchableAttributes": ["name", "description", "sku", "brand"],
  "filterableAttributes": ["category_id", "brand", "in_stock", "price"],
  "sortableAttributes": ["price", "created_at", "rating"],
  "rankingRules": ["words", "typo", "proximity", "attribute", "sort", "exactness"],
  "typoTolerance": {
    "enabled": true,
    "minWordSizeForTypos": { "oneTypo": 5, "twoTypos": 9 }
  }
}

Data Synchronization

Laravel Scout + official driver:

// composer require laravel/scout meilisearch/meilisearch-php

use Laravel\Scout\Searchable;

class Product extends Model
{
    use Searchable;

    public function toSearchableArray(): array
    {
        return [
            'id'          => $this->id,
            'name'        => $this->name,
            'description' => strip_tags($this->description),
            'brand'       => $this->brand->name,
            'category_id' => $this->category_id,
            'price'       => $this->price,
            'in_stock'    => $this->stock > 0,
        ];
    }
}

Initial indexing:

php artisan scout:import "App\Models\Product"

Search with filters:

$results = Product::search($query)
    ->where('in_stock', true)
    ->where('category_id', $categoryId)
    ->orderBy('price')
    ->paginate(20);

Faceted Filtering

Meilisearch returns aggregations by facets in single request:

$client = new \Meilisearch\Client('http://localhost:7700', $apiKey);

$response = $client->index('products')->search($query, [
    'filter'         => ['category_id = 5', 'price 100 TO 5000'],
    'facets'         => ['brand', 'color', 'in_stock'],
    'hitsPerPage'    => 20,
    'page'           => 1,
]);

// $response->getFacetDistribution() — array with counts per facet

Browser Direct Search

Direct JavaScript requests to Meilisearch via Search-only API Key — key with limited rights (search only on specific indexes). Lowers latency, removes extra hop.

import { MeiliSearch } from 'meilisearch'

const client = new MeiliSearch({
  host: 'https://search.example.com',
  apiKey: SEARCH_ONLY_KEY,
})

const results = await client.index('products').search(query, {
  filter: 'in_stock = true',
  limit: 10,
})

Multilingual Search

Separate index per language (products_ru, products_en) or single with language filter. Meilisearch supports Russian stemming via dictionaries.

Timelines and Work Stages

Total: 7–9 working days for typical catalog.

Monitoring

Meilisearch exposes metrics at /metrics in Prometheus format (if enabled). Key indicators: index size, indexing task time, queries per second. Indexing task status available at /tasks.