Regex for Splitting CSV Strings — Handling Quoted Fields Correctly

line.split(',') is the first thing every developer tries on CSV data, and it works right up until a field contains a comma inside quotes — then every column after it shifts by one. This guide shows the regex fix, its limits, and when to reach for a real CSV parsing library instead.

The Problem with a Plain Comma Split

const row = 'John,"Doe, Jr.",Mumbai,India';

row.split(',');
// ['John', '"Doe', ' Jr."', 'Mumbai', 'India']
// WRONG — the quoted "Doe, Jr." field got split into two columns

Splitting with a Lookahead-Aware Regex

The trick is to only split on a comma when it is followed by an even number of quote characters for the rest of the line — meaning it is outside any open quote:

const row = 'John,"Doe, Jr.",Mumbai,India';

const fields = row.split(/,(?=(?:(?:[^"]*"){2})*[^"]*$)/);
console.log(fields);
// ['John', '"Doe, Jr."', 'Mumbai', 'India']  — correct!

// Strip the surrounding quotes afterward
const cleaned = fields.map((f) => f.replace(/^"|"$/g, ''));
// ['John', 'Doe, Jr.', 'Mumbai', 'India']

The lookahead (?=(?:(?:[^"]*"){2})*[^"]*$) counts quote characters ahead of each comma in pairs — if there's an even number of quotes remaining, the comma is "outside" a quoted field and safe to split on.

Extracting Fields Directly with a Match Regex

An alternative approach matches each field directly instead of splitting, which is often easier to reason about:

function parseCsvLine(line) {
  const fieldRegex = /(?:^|,)("(?:[^"]|"")*"|[^,]*)/g;
  const fields = [];
  let match;

  while ((match = fieldRegex.exec(line)) !== null) {
    if (match[0] === '' && match.index === line.length) break;
    let value = match[1];
    if (value.startsWith('"') && value.endsWith('"')) {
      value = value.slice(1, -1).replace(/""/g, '"'); // unescape doubled quotes
    }
    fields.push(value);
  }
  return fields;
}

parseCsvLine('John,"Doe, Jr.",Mumbai,"Says ""hi""!"');
// ['John', 'Doe, Jr.', 'Mumbai', 'Says "hi"!']

Where Regex-Based CSV Parsing Still Falls Short

• Embedded newlines — a quoted field can legally contain a line break, which breaks any line-by-line regex approach that splits on \n first
• Different encodings — UTF-8 BOM markers, non-UTF-8 files, and mixed line endings (CRLF vs LF) need dedicated handling
• Alternate delimiters — semicolon-delimited "CSV" (common in European locales) or tab-delimited TSV need the pattern rewritten
• Performance on large files — regex-per-line parsing of a multi-million-row file is far slower than a streaming parser built for the format

For any production import/export feature, use a library:

// Node.js — using csv-parse
const { parse } = require('csv-parse/sync');

const records = parse(csvString, {
  columns: true,
  skip_empty_lines: true,
});

When the Regex Approach Is Fine

• You control the data source and know it never contains embedded newlines in fields
• You are doing a quick one-off script or a small admin tool, not a production import pipeline
• You need zero dependencies for a small serverless function with tight bundle size limits

Frequently Asked Questions