• Installation
• Configuration
• Usage
  • Search Modes
    • Submit Mode (Default) - Search on Enter key
    • Live Mode - Search as you type
    • Manual Mode Configuration
  • When to Use Each Mode
• How it works
• Examples
• Universal Language Support 🌍
  • Multi-Language Examples
  • How It Works
  • Mixed Language Queries
• Configuration Options
  • Environment Variables
• Requirements
• License

A simple Filament filter that converts natural language text into database queries using AI.

#Installation

composer require hayderhatem/filament-natural-language-filter

#Configuration

1. Publish the config file:

php artisan vendor:publish --tag="filament-natural-language-filter-config"

2. Add your OpenAI API key to your .env file:

OPENAI_API_KEY=your-openai-api-key-here

#Usage

Add the filter to your Filament table:

use HayderHatem\FilamentNaturalLanguageFilter\Filters\NaturalLanguageFilter;
 
public function table(Table $table): Table
{
    return $table
        ->columns([
            // your columns
        ])
        ->filters([
            NaturalLanguageFilter::make()
                ->availableColumns([
                    'id',
                    'name',
                    'email',
                    'status',
                    'created_at',
                    'updated_at'
                ])
        ]);
}

#Search Modes

You can configure how the filter triggers searches:

#Submit Mode (Default) - Search on Enter key

NaturalLanguageFilter::make()
    ->availableColumns(['name', 'email', 'status'])
    ->submitSearch() // Users press Enter to search

#Live Mode - Search as you type

NaturalLanguageFilter::make()
    ->availableColumns(['name', 'email', 'status'])
    ->liveSearch() // Search happens automatically as user types

#Manual Mode Configuration

NaturalLanguageFilter::make()
    ->availableColumns(['name', 'email', 'status'])
    ->searchMode('live') // or 'submit'

#When to Use Each Mode

Submit Mode (Default) - Best for:

• Large datasets where live search might be slow
• Complex queries that users want to perfect before searching
• Reducing API calls to OpenAI (only search when user is ready)

Live Mode - Best for:

• Instant feedback and better user experience
• Smaller datasets where performance isn't a concern
• Users who prefer immediate results as they type

#How it works

1. User enters natural language: "show users named john created after 2023"
2. AI processes the text: Converts it to structured filters based on your available columns
3. Database query is built: WHERE name LIKE '%john%' AND created_at > '2023-01-01'
4. Results are filtered: Table shows matching records

#Examples

• "users named john" → WHERE name LIKE '%john%'
• "active users" → WHERE status = 'active'
• "created after 2023" → WHERE created_at > '2023-01-01'
• "email contains gmail" → WHERE email LIKE '%gmail%'

#Universal Language Support 🌍

The filter supports ANY language with automatic AI translation and understanding:

#Multi-Language Examples

English:

• "show users named john" → WHERE name LIKE '%john%'
• "created after 2023" → WHERE created_at > '2023-01-01'

Arabic (العربية):

• "الاسم يحتوي على أحمد" → WHERE name LIKE '%أحمد%'
• "أنشئ بعد 2023" → WHERE created_at > '2023-01-01'

Spanish (Español):

• "usuarios con nombre juan" → WHERE name LIKE '%juan%'
• "creado después de 2023" → WHERE created_at > '2023-01-01'

French (Français):

• "nom contient marie" → WHERE name LIKE '%marie%'
• "créé après 2023" → WHERE created_at > '2023-01-01'

German (Deutsch):

• "benutzer mit namen hans" → WHERE name LIKE '%hans%'
• "erstellt nach 2023" → WHERE created_at > '2023-01-01'

Chinese (中文):

• "姓名包含张三" → WHERE name LIKE '%张三%'
• "2023年后创建" → WHERE created_at > '2023-01-01'

Japanese (日本語):

• "田中という名前のユーザー" → WHERE name LIKE '%田中%'
• "2023年以降に作成" → WHERE created_at > '2023-01-01'

#How It Works

1. AI Language Detection: Automatically detects the input language
2. Natural Understanding: Maps language-specific keywords to operators
3. Value Preservation: Keeps original values in their native language/script
4. Mixed Language: Handles mixed-language queries seamlessly

#Mixed Language Queries

The AI can handle mixed-language queries naturally:

• "name يحتوي على john" ✅
• "usuario con email gmail.com" ✅
• "姓名 contains 张三" ✅

#Configuration Options

// config/filament-natural-language-filter.php
return [
    'model' => 'gpt-3.5-turbo', // OpenAI model
    'openai' => [
        'api_key' => env('OPENAI_API_KEY'),
        'max_tokens' => 500,
        'temperature' => 0.1,
    ],
    'cache' => [
        'enabled' => true,
        'ttl' => 3600, // 1 hour
    ],
    'languages' => [
        'universal_support' => true,
        'auto_detect_direction' => true,
        'preserve_original_values' => true,
    ],
];

#Environment Variables

OPENAI_API_KEY=your-openai-api-key-here
FILAMENT_NL_FILTER_UNIVERSAL_SUPPORT=true
FILAMENT_NL_FILTER_AUTO_DETECT_DIRECTION=true
FILAMENT_NL_FILTER_PRESERVE_ORIGINAL_VALUES=true

#Requirements

• PHP 8.1+
• Laravel 10+
• Filament 3+
• OpenAI API key

#License

MIT