How to Split a String by Regex in TypeScript

You split a string by regex in TypeScript by passing a RegExp to split() instead of a string. Every match becomes a cut point, and you get back a string[]:

const parts: string[] = "react,  node;typescript".split(/\s*[,;]\s*/);
// ["react", "node", "typescript"]

Each snippet here was compiled in strict mode with TypeScript 7.0.2 and Node.js v22.22.2. I pasted the console output straight from those runs.

How split() works with a regular expression

The pattern describes the separator, so one regex can eat messy spacing around commas and semicolons. Tabs and runs of spaces collapse with \s+:

const tags: string = "react,  node;typescript ,  vite";
const parts: string[] = tags.split(/\s*[,;]\s*/);
console.log(parts);

const address = "1600   Main St\tSuite 210";
console.log(address.split(/\s+/));
console.log(",Austin,,Denver,".split(/,/));

Output:

[ 'react', 'node', 'typescript', 'vite' ]
[ '1600', 'Main', 'St', 'Suite', '210' ]
[ '', 'Austin', '', 'Denver', '' ]

Leading, trailing and doubled separators leave empty strings, and the regex doesn’t fix that on its own. Clean them up with removing empty strings from an array.

I keep the two everyday cases in their own guides: splitting a string with multiple separators and splitting a string by new line.

Do regex flags work with split()?

Most do, but g doesn’t matter. split() hands the work to the regex’s [Symbol.split] method, which copies the pattern and finds every match anyway:

const re = /and/g;
const items = "Salt and Pepper AND Garlic and Thyme";

console.log("g:   ", items.split(re));
console.log("lastIndex after split:", re.lastIndex);
console.log("g+i: ", items.split(/\s*and\s*/gi));
console.log("i:   ", items.split(/\s*and\s*/i));

Output:

g:    [ 'Salt ', ' Pepper AND Garlic ', ' Thyme' ]
lastIndex after split: 0
g+i:  [ 'Salt', 'Pepper', 'Garlic', 'Thyme' ]
i:    [ 'Salt', 'Pepper', 'Garlic', 'Thyme' ]

Because it works on a copy, lastIndex stays at 0, unlike exec() and test() with a shared global regex. For case-insensitive checks without a regex, see comparing strings in TypeScript.

Keep the separators with capture groups

Parentheses in the pattern change the output: whatever a group captures is spliced into the array between the pieces. Use (?:...) for plain grouping:

const formula = "120+35.5-18*2";

console.log("no group:   ", formula.split(/[-+*]/));
console.log("with group: ", JSON.stringify(formula.split(/([-+*])/)));
console.log("non-capture:", formula.split(/(?:[-+*])/));

// Two alternatives, two groups: the group that didn't match gives undefined
const mixed: string[] = "a,b:c".split(/(,)|(:)/);
console.log(mixed);

// limit truncates AFTER splitting, and captured separators count
console.log("limit 3:    ", formula.split(/([-+*])/, 3));

Output:

no group:    [ '120', '35.5', '18', '2' ]
with group:  ["120","+","35.5","-","18","*","2"]
non-capture: [ '120', '35.5', '18', '2' ]
[ 'a', ',', undefined, 'b', undefined, ':', 'c' ]
limit 3:     [ '120', '+', '35.5' ]

Watch that undefined. TypeScript still types the result as string[], so a later .trim() on it compiles but crashes at runtime. One group with a character class, ([,:]), avoids it.

The second argument, limit, doesn’t stop the search early. It trims the finished array, so /([-+*])/ with 3 returns two numbers and one operator. For the final item, see getting the last element after a split.

Split camelCase and before capital letters

Lookahead (?=...) and lookbehind (?<=...) match a position, not characters. Nothing is removed, so you can cut before a capital letter or after a period and keep both:

console.log("camelCase:".padEnd(14), "shippingAddressLine".split(/(?=[A-Z])/));

// Keep acronyms together: split between lower->Upper and before the last capital of a run
const acronym = /(?<=[a-z0-9])(?=[A-Z])|(?<=[A-Z])(?=[A-Z][a-z])/;
console.log("acronyms:".padEnd(14), "parseHTTPResponse2026Code".split(acronym));

const words = "XMLHttpRequest".split(acronym).map((w) => w.toLowerCase()).join("-");
console.log("kebab-case:".padEnd(14), words);

// Lookbehind: split after sentence punctuation, keep the punctuation
const note = "Order shipped. Arrives Oct 21! Need help? Call us.";
console.log("sentences:".padEnd(14), note.split(/(?<=[.!?])\s+/));

Output:

camelCase:     [ 'shipping', 'Address', 'Line' ]
acronyms:      [ 'parse', 'HTTP', 'Response2026', 'Code' ]
kebab-case:    xml-http-request
sentences:     [ 'Order shipped.', 'Arrives Oct 21!', 'Need help?', 'Call us.' ]

Lookbehind is supported in every current browser and Node.js release. If you only need the text after one marker, getting the string after a character is simpler.

How do I split Unicode and emoji safely?

split("") cuts UTF-16 code units, which breaks emoji in half. An empty regex with the u flag splits by code point instead, and Intl.Segmenter splits by what readers see as one character:

const reaction = "ok👍🏽";

console.log("split(''):    ", reaction.split("").length, "pieces");
console.log("split(/(?:)/u):", reaction.split(/(?:)/u));

const family = "hi👨‍👩‍👧";
const seg = new Intl.Segmenter("en-US", { granularity: "grapheme" });
console.log("graphemes:    ", Array.from(seg.segment(family), (s) => s.segment));

// \p{...} classes need the u flag: split on any punctuation or symbol
console.log("\\p{P}:        ", "Café—Seattle/Boston·2026".split(/[\p{P}\p{S}]+/u));

Output:

split(''):     6 pieces
split(/(?:)/u): [ 'o', 'k', '👍', '🏽' ]
graphemes:     [ 'h', 'i', '👨‍👩‍👧' ]
\p{P}:         [ 'Café', 'Seattle', 'Boston', '2026' ]

Skin tones and family emoji are several code points glued together, so only the segmenter keeps them whole. The same code-unit counting explains surprises in a TypeScript string length check.

Write a custom splitter with Symbol.split

split() accepts any object with a [Symbol.split] method, and TypeScript’s types allow it. It’s a tidy home for a reusable rule:

// Any object with a [Symbol.split] method can be passed to split()
const csvLine = {
  [Symbol.split](value: string): string[] {
    // split on commas that are NOT inside double quotes
    return value.split(/,(?=(?:[^"]*"[^"]*")*[^"]*$)/).map((f) => f.replace(/^"|"$/g, ""));
  },
};

const row = '1042,"Portland, OR",249.99,2026-10-21';
console.log(row.split(csvLine));

Output:

[ '1042', 'Portland, OR', '249.99', '2026-10-21' ]

It suits simple CSV lines; files with escaped quotes need a real parser. Stripping quotes is covered in removing quotes from a string.

Catch broken regex patterns at compile time

TypeScript 5.5 and later check regex literal syntax, so a bad pattern fails the build instead of throwing in production:

const zip = "Dallas TX 75201";

const a = zip.split(/\s+{2}/);
const b = zip.split(/(?<state[A-Z]{2})/);
const c = zip.split(/\s/x);

Output:

regex_broken.ts(3,25): error TS1507: There is nothing available for repetition.
regex_broken.ts(4,30): error TS1005: '>' expected.
regex_broken.ts(5,25): error TS1499: Unknown regular expression flag.

Patterns built with new RegExp() aren’t checked, so they still throw a SyntaxError at runtime. Wrap dynamic ones in try:

const pattern: string = "\\s+{2}";
try {
  "Dallas TX 75201".split(new RegExp(pattern));
} catch (err) {
  console.log((err as Error).name + ":", (err as Error).message);
}

Output:

SyntaxError: Invalid regular expression: /\s+{2}/: Nothing to repeat

Want more string tutorials? These pick up where this one stops:

• Convert a string to an array in TypeScript
• Replace all occurrences of a substring
• Remove spaces from a string in TypeScript
• Convert split pieces to numbers in TypeScript

Frequently asked questions