abduns/laravel-ocr

Laravel-first wrapper around Tesseract 5 OCR.

dev-main 2026-05-26 14:21 UTC

Requires

• php: ^8.2
• illuminate/bus: ^10.0 || ^11.0 || ^12.0 || ^13.0
• illuminate/console: ^10.0 || ^11.0 || ^12.0 || ^13.0
• illuminate/filesystem: ^10.0 || ^11.0 || ^12.0 || ^13.0
• illuminate/queue: ^10.0 || ^11.0 || ^12.0 || ^13.0
• illuminate/support: ^10.0 || ^11.0 || ^12.0 || ^13.0
• symfony/process: ^6.0 || ^7.0 || ^8.0

Requires (Dev)

• laravel/pint: ^1.13
• orchestra/testbench: ^8.0 || ^9.0 || ^10.0 || ^11.0
• phpstan/phpstan: ^2.0
• phpunit/phpunit: ^10.0 || ^11.0 || ^12.0

MIT

• Abdun Syakuur

• image
• pdf
• Tesseract
• OCR
• laravel

README

Laravel-first wrapper around Tesseract 5 OCR for images and PDFs.

Features

• Modern PHP support
• Fluent Laravel API for image and PDF OCR
• Queueable OCR jobs
• Lifecycle events for started, completed, and failed runs
• Filesystem disk support, including remote source disks
• Publishable config file
• Artisan command for direct scans
• Setup check command for Tesseract, language data, temp storage, and PDF tooling
• Lazy validation so apps can boot without a local OCR binary
• AI agent instructions included for common coding assistants
• Laravel Boost guidelines and skill included for AI-assisted Laravel development

Requirements

• PHP 8.2+
• Laravel 10, 11, 12, or 13
• Tesseract 5.0+
• For PDF input: Imagick PHP extension or Ghostscript

Install system dependencies first:

# Ubuntu/Debian
sudo apt-get install tesseract-ocr tesseract-ocr-eng ghostscript

# macOS with Homebrew
brew install tesseract ghostscript

Installation

composer require abduns/laravel-ocr

Publish the config file when you need to customize paths, defaults, queues, or PDF rasterization:

php artisan vendor:publish --tag=ocr-config

The package is auto-discovered by Laravel. The facade alias is Ocr.

The Composer package and GitHub repository are abduns/laravel-ocr. The PHP namespace intentionally remains Dunn\LaravelOcr.

Quick Start

use Dunn\LaravelOcr\Facades\Ocr;

$text = Ocr::image(storage_path('app/invoices/invoice.png'))
    ->language('eng')
    ->run();

PDF output is keyed by page number:

$pages = Ocr::pdf(storage_path('app/contracts/contract.pdf'))
    ->language('eng')
    ->runAll();

// [1 => '...', 2 => '...', 3 => '...']

Why This Package?

• OCR packages often hide the local binary details that matter in production
• PDF handling usually needs predictable temporary files and cleanup
• Queue and event integration should feel native in Laravel applications

This package focuses on a small, headless Laravel surface around the local Tesseract CLI. It ships no routes, views, frontend assets, cloud OCR SDKs, or language-pack downloaders.

Usage

Image OCR

use Dunn\LaravelOcr\Facades\Ocr;

$text = Ocr::image(storage_path('app/invoices/invoice.png'))
    ->language('eng')
    ->psm(6)
    ->timeout(30)
    ->run();

Multiple languages are supported and passed to Tesseract in order:

$text = Ocr::image($path)
    ->languages(['eng', 'ind'])
    ->run();

PDF OCR

$pages = Ocr::pdf(storage_path('app/contracts/contract.pdf'))
    ->language('eng')
    ->dpi(300)
    ->pages([1, 2, 3])
    ->runAll();

If pages() is omitted, every page is OCRed in ascending order. Temporary rasterized page images are deleted after success or failure.

Storage Disks

Use disk() when the source lives on a configured Laravel filesystem disk:

$text = Ocr::image('uploads/receipt.png')
    ->disk('s3')
    ->language('eng')
    ->run();

Non-local source disks are streamed to the configured local temp disk before OCR.

Queues

Dispatch a queued OCR job with onQueue():

Ocr::pdf('documents/report.pdf')
    ->disk('s3')
    ->language('eng')
    ->onQueue('ocr')
    ->dispatch();

The queue connection comes from ocr.queue.connection. The queue name comes from the explicit onQueue('name') value, then ocr.queue.name, then Laravel's default queue.

Events

Each OCR invocation dispatches lifecycle events:

• Dunn\LaravelOcr\Events\OcrStarted
• Dunn\LaravelOcr\Events\OcrCompleted
• Dunn\LaravelOcr\Events\OcrFailed

PDF OCR emits one event sequence per page.

use Dunn\LaravelOcr\Events\OcrCompleted;
use Illuminate\Support\Facades\Event;

Event::listen(OcrCompleted::class, function (OcrCompleted $event) {
    logger()->info('OCR completed', [
        'job_id' => $event->jobId,
        'length' => strlen($event->text),
    ]);
});

Artisan

php artisan ocr:check
php artisan ocr:check --lang=eng,ind
php artisan ocr:check --skip-pdf
php artisan ocr:scan storage/app/invoice.png --lang=eng --psm=6
php artisan ocr:scan storage/app/report.pdf --lang=eng --dpi=300

Use ocr:check before wiring OCR into production or queue workers. It verifies the configured Tesseract binary, Tesseract 5 version, requested language data, local temp disk, and PDF backend. Use --skip-pdf when the app only needs image OCR.

For PDFs, page output is separated by a form-feed character.

Advanced Usage

Configuration

Published config lives at config/ocr.php.

return [
    'binary' => env('TESSERACT_BIN', '/usr/bin/tesseract'),
    'default_language' => env('OCR_LANG', 'eng'),
    'default_psm' => 3,
    'default_oem' => 3,
    'tessdata_path' => env('TESSDATA_PREFIX'),
    'temp_disk' => 'local',
    'temp_path' => 'ocr/tmp',
    'timeout' => 120,
    'pdf' => [
        'driver' => env('OCR_PDF_DRIVER', 'auto'),
        'default_dpi' => 300,
    ],
    'queue' => [
        'connection' => env('OCR_QUEUE_CONNECTION'),
        'name' => env('OCR_QUEUE_NAME', 'ocr'),
    ],
];

temp_disk must be a local filesystem disk because Tesseract and the PDF rasterizers need real local file paths.

Tesseract Options

Method	Description
language('eng')	Use one Tesseract language
languages(['eng', 'ind'])	Use multiple Tesseract languages in order
psm(6)	Set page segmentation mode, 0..13
oem(3)	Set OCR engine mode, 0..3
timeout(30)	Override process timeout in seconds
tessdataPath('/path/to/tessdata')	Override traineddata lookup path
dpi(300)	Set PDF rasterization DPI
pages([1, 2, 3])	Limit PDF OCR to selected pages

Error Handling

The package exposes typed runtime exceptions:

• Dunn\LaravelOcr\Exceptions\TesseractNotFoundException
• Dunn\LaravelOcr\Exceptions\UnsupportedLanguageException
• Dunn\LaravelOcr\Exceptions\OcrProcessingException

Configuration validation is lazy. The service provider can boot even if Tesseract is missing; failures surface when OCR is executed.

AI Agent Instructions

This package includes AI coding-agent instructions for common tools:

• resources/boost/guidelines/core.blade.php for Laravel Boost package guidelines
• resources/boost/skills/laravel-ocr-development/SKILL.md for the Laravel Boost package skill
• AGENTS.md for generic repository-wide agent guidance
• .github/copilot-instructions.md for GitHub Copilot
• .cursor/rules/laravel-ocr.mdc for Cursor
• .windsurfrules for Windsurf
• GEMINI.md for Gemini
• CLAUDE.md and .claude/skills/laravel-ocr for Claude Code

Use these when an AI agent needs to add OCR to a Laravel app, configure local Tesseract, wire image or PDF OCR, dispatch queued OCR jobs, listen for lifecycle events, or debug system dependency failures.

In apps using Laravel Boost, the package guidelines and laravel-ocr-development skill can be installed by running:

php artisan boost:install

If the package was added after Boost resources were already installed, run:

php artisan boost:update --discover

Standards / Specifications

References:

• https://tesseract-ocr.github.io
• https://github.com/tesseract-ocr/tesseract

Supported Features

Feature	Support
Image OCR	Yes
PDF OCR	Yes
Multiple Languages	Yes
Laravel Filesystem Disks	Yes
Queued Jobs	Yes
Lifecycle Events	Yes
Artisan Scans	Yes
Setup Diagnostics	Yes

Compatibility

Platform	Supported
PHP 8.2+	Yes
Laravel 10.0+	Yes
Tesseract 5.0+	Yes
Symfony Process 6.0+	Yes

Design Goals

• Laravel-native API
• Predictable local-process execution
• Explicit configuration
• Safe temporary file handling
• Minimal runtime dependencies
• Strong validation
• Testable internals

Architecture

• Facade-backed OcrManager
• Immutable builder state for image and PDF workflows
• Symfony Process execution through the local Tesseract binary
• PDF rasterization through Ghostscript or Imagick
• Laravel queue payload rebuild for deferred OCR jobs
• Events emitted around each OCR operation

Performance

Operation	Constraint
Image OCR	Bound by Tesseract and source image size
PDF OCR	Bound by page count, rasterization DPI, and Tesseract
Remote disk source	Streamed once to a local temp file
Cleanup	Temporary files removed after success or failure

Testing

vendor/bin/phpunit
vendor/bin/phpstan analyse --configuration=phpstan.neon.dist
vendor/bin/pint --test

Real Tesseract integration tests are opt-in:

OCR_INTEGRATION=1 TESSERACT_BIN=/usr/bin/tesseract vendor/bin/phpunit --testsuite Integration

PDF integration also needs Ghostscript or a working Imagick PDF policy.

Roadmap

• Add optional cloud OCR driver integrations while keeping local Tesseract as the default engine
• Add more CLI output format options
• Add per-page PDF progress callbacks
• Add richer queue result handling examples

Contributing

Contributions, issues, and discussions are welcome.

Security

If you discover security issues, please report them responsibly.

License

The MIT License (MIT). Please see LICENSE for more information.