24 lines
537 B
Markdown
24 lines
537 B
Markdown
|
```php title="PHP"
|
||
|
|
<?php
|
||
|
|
declare(strict_types=1);
|
||
|
|
|
||
|
|
require_once __DIR__ . '/vendor/autoload.php';
|
||
|
|
|
||
|
|
use Kreuzberg\Kreuzberg;
|
||
|
|
use Kreuzberg\Config\ExtractionConfig;
|
||
|
|
use Kreuzberg\Config\OcrConfig;
|
||
|
|
|
||
|
|
$config = new ExtractionConfig(
|
||
|
|
ocr: new OcrConfig(
|
||
|
|
backend: 'paddle-ocr',
|
||
|
|
language: 'en',
|
||
|
|
// paddleOcrConfig: new PaddleOcrConfig(modelTier: 'server') // for max accuracy
|
||
|
|
)
|
||
|
|
);
|
||
|
|
|
||
|
|
$kreuzberg = new Kreuzberg($config);
|
||
|
|
$result = $kreuzberg->extractFile('scanned_document.pdf');
|
||
|
|
|
||
|
|
echo $result->content . "\n";
|
||
|
|
```
|