Files
firecrawl/apps/api/src/lib/extract/build-document.ts
T

16 lines
586 B
TypeScript
Raw Normal View History

2024-11-20 12:48:10 -08:00
import { Document } from "../../controllers/v1/types";
export function buildDocument(document: Document): string {
const metadata = document.metadata;
const markdown = document.markdown;
2024-11-20 12:51:16 -08:00
// for each key in the metadata allow up to 250 characters
const metadataString = Object.entries(metadata).map(([key, value]) => {
return `${key}: ${value?.toString().slice(0, 250)}`;
}).join('\n');
2024-11-20 12:48:10 -08:00
2024-11-20 12:51:16 -08:00
const documentMetadataString = `\n- - - - - Page metadata - - - - -\n${metadataString}`;
2024-11-20 12:48:10 -08:00
const documentString = `${markdown}${documentMetadataString}`;
2024-11-20 12:51:16 -08:00
return documentString;
2024-11-20 12:48:10 -08:00
}