cmb_pdf_convert.js

const fs = require('fs');
const pdfParse = require('pdf-parse');

let content = []

const readPdf = async (uri) => {
    const buffer = fs.readFileSync(uri);
    try {
        const data = await pdfParse(buffer);

        content = data.text.split('\n')
        console.log(readLine(content).join('\n'))

    }catch(err){
        throw new Error(err);
    }
}

const readLine = (content) => {
    let currentLine = 0
    let total = content.length
    let lines = []
    let lastLineValid = false
    while (currentLine < total) {
        const regex = /(\d{4}-\d{2}-\d{2})(CNY)/g;
        const valid = regex.test(content[currentLine])
        if (valid) {
            lines.push(content[currentLine])
            lastLineValid = true
        } else {
            if (lastLineValid) {
                if (content[currentLine]) {
                    const l = lines.length
                    lines[l - 1] = lines[l - 1] + content[currentLine]
                } else {
                    lastLineValid = false
                }
            }
        }
        currentLine += 1
    }

    return lines.map((line) => {
        const r = /(\d{4}-\d{2}-\d{2})(CNY)(-?[0-9,]*.\d{2})(-?[0-9,]*.\d{2})(\S*)/g
        const part = r.exec(line)
        if (part) {
            const [, date, , amount, balance, payee] = part
            return `${date},${amount},${balance},${payee}`
        }
        return []
    })
}

// Testing
const DUMMY_PDF = './cmb.pdf';
readPdf(DUMMY_PDF);