6k6g-v3 / src /lib /knowledge /parser /index.ts
fanyubo
feat(kb): HF Space KB Worker — 解析/分块/向量化/标签/质量审核
142ad12
Raw
History Blame Contribute Delete
586 Bytes
// hf-space/src/lib/knowledge/parser/index.ts
import { parsePDF } from './pdf';
import { parseDOCX } from './docx';
import { parseText } from './text';
export interface ParsedDocument {
content: string;
pages?: number;
}
export async function parseDocument(buffer: Buffer, fileType: string): Promise<ParsedDocument> {
switch (fileType) {
case 'pdf':
return parsePDF(buffer);
case 'docx':
return parseDOCX(buffer);
case 'md':
case 'txt':
return parseText(buffer);
default:
throw new Error(`Unsupported file type: ${fileType}`);
}
}