Regex Tester

The tester uses the native JavaScript RegExp object via new RegExp(pattern, flags). That means you are testing against the ECMAScript regex dialect specified in Annex B of the language standard — backtracking NFA, no atomic groups, no possessive quantifiers, no recursion. Patterns that work here will also work in Node 18+, Deno, Bun, and any modern browser.

If your target runtime is PCRE (PHP preg_*, nginx, many Linux utilities) or RE2 (Go, Cloudflare Workers, BigQuery), behavior diverges in edge cases: lookbehind support, possessive quantifiers, and catastrophic-backtracking protection differ. Test on the actual engine before shipping production patterns.

A surprisingly common bug is using \d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3} and calling it an IPv4 validator. That matches 999.999.999.999. The correct bounded version:

^(25[0-5]|2[0-4]\d|1\d{2}|[1-9]?\d)(\.(25[0-5]|2[0-4]\d|1\d{2}|[1-9]?\d)){3}$

192.168.1.1
255.255.255.255
256.1.1.1
10.0.0.0
01.02.03.04

✓ 192.168.1.1
✓ 255.255.255.255
✗ 256.1.1.1
✓ 10.0.0.0
✗ 01.02.03.04

• g — without it, .exec() and .test() return at most one match, and replace() only replaces the first occurrence. Almost every multi-match bug starts here.
• i — case-insensitive. Combine with \p{L} and the u flag for proper Unicode case folding (e.g. ß ↔ SS).
• m — changes ^ and $ to match line boundaries inside the string. Does not enable . to cross lines; that is the s flag.
• s — "dotall". Lets . match \n. Required for matching across multi-line YAML or JSON values.
• u — Unicode mode. Enables \p{} property escapes, surrogate-pair-aware character classes, and stricter escape validation. Use this whenever your input may contain emoji or non-Latin scripts.
• y — sticky. Anchors matching to lastIndex without requiring ^. Useful for tokenizers, rarely useful otherwise.

• You have a regex that works on three test cases and you need to confirm it fails on twenty edge cases before merging.
• You inherited a pattern with five nested groups and need to see exactly what each group captures on real data.
• You are migrating patterns from PCRE/RE2 and need to verify each one still works in V8 — paste both the pattern and a representative corpus.
• You are writing extraction logic and want to compare 4-5 candidate patterns side-by-side on the same input.

• It will not protect you from catastrophic backtracking. A pattern like (a+)+$ on "aaaaaaaaaaaaaaaaaaaaab" can freeze your tab. The engine is the same one your production code uses — if it hangs here, it will hang there.
• It will not translate between dialects. PCRE recursion (?R), .NET balanced groups, and Oniguruma named conditional groups are silently invalid in ECMAScript and will throw SyntaxError.
• It will not test performance. Browser RegExp does not expose step counts; for serious profiling use regex101 in PCRE mode or a real benchmark harness.