28 lines
730 B
Markdown
28 lines
730 B
Markdown
|
```dart title="Dart"
|
||
|
|
import 'package:kreuzberg/kreuzberg.dart';
|
||
|
|
|
||
|
|
Future<void> main() async {
|
||
|
|
final config = ExtractionConfig(
|
||
|
|
useCache: true,
|
||
|
|
enableQualityProcessing: true,
|
||
|
|
forceOcr: false,
|
||
|
|
disableOcr: false,
|
||
|
|
ocr: const OcrConfig(
|
||
|
|
enabled: true,
|
||
|
|
backend: 'tesseract',
|
||
|
|
language: 'eng',
|
||
|
|
autoRotate: false,
|
||
|
|
),
|
||
|
|
resultFormat: ResultFormat.unified,
|
||
|
|
outputFormat: OutputFormat.plain(),
|
||
|
|
includeDocumentStructure: false,
|
||
|
|
maxArchiveDepth: 3,
|
||
|
|
useLayoutForMarkdown: false,
|
||
|
|
);
|
||
|
|
|
||
|
|
final result = await KreuzbergBridge.extractFile('scanned.pdf', null, config);
|
||
|
|
print('Content length: ${result.content.length}');
|
||
|
|
print('Tables detected: ${result.tables.length}');
|
||
|
|
}
|
||
|
|
```
|