這個正則表達式效果更好
/^(?:[^"\]|\.|"(?:\.|[^"\])")$/g
這是通過讀取每一行來處理大文件的更復雜的工作腳本
import csv from 'csvtojson'
import fs from 'fs-extra'
import lineReader from 'line-reader'
import { __dirname } from '../../../utils.js'
const CSV2JSON = async(dumb, editDumb, headers, {
options = {
trim: true,
delimiter: '|',
quote: '"',
escape: '"',
fork: true,
headers: headers
}
} = {}) => {
try {
log(`\n\nStarting CSV2JSON - Current directory: ${__dirname()} - Please wait..`)
await new Promise((resolve, reject) => {
let firstLine, counter = 0
lineReader.eachLine(dumb, async(line, last) => {
counter++
// log(`line before convert: ${line}`)
let json = (
await csv(options).fromString(headers + '\n\r' + line)
.preFileLine((fileLineString, lineIdx) => {
// if it its not the first line
// eslint-disable-next-line max-len
if (counter !== 1 && !fileLineString.match(/^(?:[^"\\]|\\.|"(?:\\.|[^"\\])*")*$/g)) {
// eslint-disable-next-line max-len
console.log(`Line #${lineIdx + 1} is invalid. It has unescaped quotes. We will skip this line.. Invalid Line: ${fileLineString}`)
fileLineString = ''
}
return fileLineString
})
.on('error', e => {
e = `Error while converting CSV to JSON.
Line before convert: ${line}
Error: ${e}`
throw new BaseError(e)
})
)[0]
// log(`line after convert: ${json}`)
if (json) {
json = JSON.stringify(json).replace(/\\"/g, '')
if (json.match(/^(?:[^"\\]|\\.|"(?:\\.|[^"\\])*")*$/g)) {
await fs.appendFile(editDumb, json)
}
}
if (last) {
resolve()
}
})
})
} catch (e) {
throw new BaseError(`Error while converting CSV to JSON - Error: ${e}`)
}
}
export { CSV2JSON }
