import PDFParser from "pdf2json"; export function withTimeout( func: (param: T) => Promise, timeoutSec: number, ) { return async (param: T): Promise => { return await Promise.race([ func(param), new Promise((_resolve, reject) => setTimeout( () => reject(new Error(`Timed-out after ${timeoutSec} secs`)), timeoutSec * 1000, ), ), ]); }; } export async function readPDFText(buffer: Buffer): Promise<{ text: string; metadata: Record; }> { return new Promise((resolve, reject) => { // Need raw text flag represents as number (1), reference : https://github.com/modesty/pdf2json/issues/76#issuecomment-236569265 const pdfParser = new PDFParser(null, 1); pdfParser.on("pdfParser_dataError", reject); pdfParser.on("pdfParser_dataReady", (pdfData) => { resolve({ // The type isn't set correctly, reference : https://github.com/modesty/pdf2json/issues/327 // eslint-disable-next-line text: (pdfParser as any).getRawTextContent(), metadata: pdfData.Meta, }); }); pdfParser.parseBuffer(buffer); }); }