Regex Parser Tester

Yes. You can toggle global, ignore case, multiline, dotAll, unicode, and sticky flags in the tester.

The parser reads your pattern token by token and generates plain-language explanations for anchors, classes, groups, and quantifiers.

Yes. The sample generator creates a representative string using your current pattern so you can quickly sanity-check intent.

A greedy quantifier like .* consumes as much as possible, then backtracks; its lazy twin .*? stops at the first match. On the text say "hi" bye, the pattern "..*" captures the whole line, while ".*?" captures only hi. Greedy is the default - append ? to make any quantifier lazy.

Parentheses capture: (\d{4})-(\d{2}) against 2026-09-19 gives group 1 = 2026 and group 2 = 09, and in a replace pattern $1 reuses the captured text. Named groups like (?<year>\d{4}) read better in complex patterns. JavaScript numbers groups by their opening parenthesis, starting at 1 - group 0 is always the whole match.

Add u whenever the pattern or the text involves non-ASCII characters: it makes \w and \d follow Unicode rules, enables property escapes like \p{L} and \p{Script=Arabic}, and treats emoji as single characters. Without u, a UTF-16 surrogate pair can be split in half by a dot match.

By default ^ and $ anchor the whole string and . skips line breaks. The m (multiline) flag makes ^ and $ match at every line boundary instead, and the s (dotAll) flag lets . cross newlines. A common recipe for log parsing is both flags together: /error.*timeout/gms.

It runs the JavaScript engine in your browser, so it follows ECMAScript (ECMA-262) regex semantics - the same engine String.prototype.match uses in production JS. Note the differences from PCRE or Python: JavaScript has no \K or recursion, and lookbehind arrived only in ES2018. Patterns written for Perl or PCRE may need small adjustments.