24 lines
551 B
Markdown
24 lines
551 B
Markdown
|
```typescript title="WASM"
|
||
|
|
import init, { extractFile } from "kreuzberg-wasm";
|
||
|
|
|
||
|
|
await init();
|
||
|
|
|
||
|
|
const config = {
|
||
|
|
chunking: {
|
||
|
|
maxChars: 500,
|
||
|
|
chunkOverlap: 50,
|
||
|
|
},
|
||
|
|
};
|
||
|
|
|
||
|
|
const result = await extractFile("research_paper.pdf", undefined, config);
|
||
|
|
|
||
|
|
if (result.chunks) {
|
||
|
|
for (const chunk of result.chunks) {
|
||
|
|
const meta = chunk.metadata;
|
||
|
|
console.log(`Chunk ${meta.chunkIndex + 1}/${meta.totalChunks}`);
|
||
|
|
console.log(`Position: ${meta.byteStart}-${meta.byteEnd}`);
|
||
|
|
console.log(`Content: ${chunk.content.slice(0, 100)}...`);
|
||
|
|
}
|
||
|
|
}
|
||
|
|
```
|