28 lines
770 B
Markdown
28 lines
770 B
Markdown
|
```kotlin title="Kotlin"
|
||
|
|
import dev.kreuzberg.*
|
||
|
|
import java.nio.file.Paths
|
||
|
|
import java.util.Optional
|
||
|
|
|
||
|
|
fun main() {
|
||
|
|
val embedding = EmbeddingConfig.builder()
|
||
|
|
.withModel(EmbeddingModelType.Preset("balanced"))
|
||
|
|
.withBatchSize(16L)
|
||
|
|
.withNormalize(true)
|
||
|
|
.withShowDownloadProgress(true)
|
||
|
|
.build()
|
||
|
|
|
||
|
|
val chunking = ChunkingConfig.builder()
|
||
|
|
.withMaxCharacters(1000L)
|
||
|
|
.withOverlap(200L)
|
||
|
|
.withEmbedding(Optional.of(embedding))
|
||
|
|
.build()
|
||
|
|
|
||
|
|
val config = ExtractionConfig.builder()
|
||
|
|
.withChunking(Optional.of(chunking))
|
||
|
|
.build()
|
||
|
|
|
||
|
|
val result = Kreuzberg.extractFileSync(Paths.get("document.pdf"), null, config)
|
||
|
|
println("Chunks with embeddings: ${result.chunks()?.size ?: 0}")
|
||
|
|
}
|
||
|
|
```
|