23 lines
855 B
Markdown
23 lines
855 B
Markdown
|
```java title="Java"
|
||
|
|
import dev.kreuzberg.Kreuzberg;
|
||
|
|
import dev.kreuzberg.ExtractionResult;
|
||
|
|
import dev.kreuzberg.ExtractionConfig;
|
||
|
|
import dev.kreuzberg.TokenReductionConfig;
|
||
|
|
|
||
|
|
ExtractionConfig config = ExtractionConfig.builder()
|
||
|
|
.tokenReduction(TokenReductionConfig.builder()
|
||
|
|
.mode("moderate")
|
||
|
|
.preserveImportantWords(true)
|
||
|
|
.build())
|
||
|
|
.build();
|
||
|
|
|
||
|
|
ExtractionResult result = Kreuzberg.extractFile("verbose_document.pdf", config);
|
||
|
|
|
||
|
|
Object originalTokens = result.getMetadata().get("original_token_count");
|
||
|
|
Object reducedTokens = result.getMetadata().get("token_count");
|
||
|
|
Object reductionRatio = result.getMetadata().get("token_reduction_ratio");
|
||
|
|
|
||
|
|
System.out.println("Reduced from " + originalTokens + " to " + reducedTokens + " tokens");
|
||
|
|
System.out.println("Reduction: " + ((Number)reductionRatio).doubleValue() * 100 + "%");
|
||
|
|
```
|