RegexCalc — Clean, Modern Regular Expression Calculator & Tester

Regular Expression JAVASCRIPT REGEX

/ / gm

Recipes:

g i m s

Test String

0 characters

Extracted Matches

Instant client-side evaluation

Embed this tool on your site

Paste this where you want the tool to appear. It renders RegexCalc inside your own page — no ads in the embed, no input stored, and it links back to regexcalc.net so visitors can reach the full page.

<iframe src="https://regexcalc.net/embed/" title="RegexCalc — Regex Engine" width="100%" height="560" loading="lazy" style="border:1px solid #d0d7de;border-radius:10px;max-width:100%;"></iframe>

Frequently Asked Questions

Is my pattern or test data sent to a server?

No. Matching runs in your browser using the JavaScript engine itself, so neither your pattern nor your test string is transmitted.

Why does my pattern work here but not in grep?

This tester uses the JavaScript regex dialect: no delimiters, ECMAScript syntax, and whichever of the g, i, m, s flags you tick. That means lookahead, named capture groups, and — on modern engines — lookbehind all work, but \A, \z, \h, possessive quantifiers, and atomic groups do not. grep -E uses POSIX ERE and grep -P uses PCRE, so the same pattern can behave differently in each.

What is catastrophic backtracking?

A pattern with nested quantifiers, such as (a+)+, can require exponential time against a non-matching input and freeze the page. The tester evaluates in your browser, so it can hang that tab rather than a server. Rewrite the pattern with atomic-style rewrites or character classes that cannot overlap.

JavaScript is not your regex engine

The tester compiles a plain new RegExp(pattern, flags) literal — no leading/trailing /.../ slashes. Paste a /^a./gi style regex and those slashes become part of the pattern and match nothing.

No named groups and no backreferences? Wrong direction: named groups ((?<year>\d{4})) work in any current browser, so prefer them over counting group numbers in your head.

. never matches \n unless you add the s flag. A pattern that passes your single-line tests can still silently swallow a multi-line input in production.

^ and $ mean start/end of input with no m flag. Add m and they become line anchors — the single most common cause of "it matched locally but not on the site".

Gotchas that bite in production

\d means [0-9], not "any digit". Unicode digits like ٣ (Arabic-Indic) fail \d and pass \p{Nd} — and \p{...} needs the v (or u) flag.

Unescaped . matches any character, so a.com also accepts axcom. Escape literal dots in domains and IP patterns.

Greedy .+ will eat as much as the input allows; .? is lazy but backtracking-heavy. On user-supplied strings that is the shape that enables catastrophic backtracking — prefer explicit character classes over ..

Alternation is ordered and first-match-wins: cat|category never reaches category. Put longer alternatives first, or wrap with a boundary.

In a replacement string, $& is the whole match and $' is the remainder of the string, not a literal quote. Use $$ for a literal dollar sign.

String.replace(str, re) replaces once; replaceAll needs a global regex or a string pattern. Silently replacing only the first hit is a classic data-corruption bug.

test() on a global (g) regex advances lastIndex between calls, so the same regex object alternates true/false over equal inputs. Reset lastIndex or drop the g flag.

Frequently Asked Questions

🧭 Pro Tips & Gotchas

JavaScript is not your regex engine

Gotchas that bite in production