Regex Tester

Write a pattern, toggle the flags, and every match lights up in the test string as you type. Capture groups are listed per match and a replacement preview runs alongside, so you can confirm both what the pattern finds and what it would produce. The engine is the browser's own JavaScript RegExp, which means what you see is exactly what your code will do — and nothing leaves the page, which matters when the sample you are debugging is real data.

Being the real JavaScript engine also means inheriting its specific personality. \d matches only the ASCII digits 0 through 9, so a string of Arabic-Indic numerals fails a pattern that looks like it should accept any digits; \p{Nd} with the u flag is the one that does. The dot skips newlines unless you set the s flag. There is no \A or \z — anchoring to the whole string means ^ and $ without m. Lookbehind works in current browsers but arrived late, so a pattern using it may fail on an older device even though it passes here.

Portability is worth checking before you copy a pattern into another language. Go's standard library uses RE2, which has no backreferences and no lookaround at all, by design — a pattern that works perfectly here will simply be rejected there. Python and PCRE accept more than JavaScript does, including atomic groups and possessive quantifiers, so a pattern copied the other way may silently behave differently rather than erroring.

The failure mode to actually fear is catastrophic backtracking. A pattern with nested quantifiers over overlapping alternatives can take exponential time on input that nearly matches: /^(a+)+b$/ tested against twenty-eight a characters followed by an x takes seconds in this engine, and each extra character roughly doubles it. On a server that is a denial of service from a single short request, which is why the pattern class has its own name. If your regex has a quantifier inside a quantifier, try it against a long near-miss before shipping it — this page is a good place to find out.