Onigmo in pure Go — Ruby's regexp engine, with the features RE2 leaves out, no cgo.
🌐 Website · 📚 Documentation
go-ruby-regexp is a pure-Go (no cgo) reimplementation of Onigmo,
the regular-expression engine used by Ruby. Go's standard regexp is RE2 —
linear-time, but deliberately without backreferences or lookaround, and
with different (leftmost-longest) match semantics. A byte-compatible Ruby regexp
therefore needs a backtracking engine, and that is what go-ruby-regexp is: a
faithful backtracking VM supporting backreferences, lookahead/lookbehind,
possessive quantifiers, atomic groups, named groups, subexpression calls, and
Ruby's leftmost-first semantics — hardened against catastrophic backtracking
(ReDoS) with memoization and a deterministic time/step budget (as Ruby ≥3.2).
It is standalone and reusable by any Go program, and is the regexp backend for the sibling org go-embedded-ruby.
| Repo | What it is |
|---|---|
| regexp | the engine: syntax/compile/vm/charset/encoding + the public regexp API |
| docs | MkDocs Material documentation, versioned with mike, served at /docs/ |
| go-ruby-regexp.github.io | the Hugo landing page |
| brand | logos and brand assets |
- Pure Go, zero cgo. Cross-compiles and embeds anywhere; a static binary by default.
- Backtracking VM, not RE2. Backreferences, lookaround and leftmost-first semantics require it — that is precisely what makes go-ruby-regexp Ruby-compatible where the standard library cannot be.
- ReDoS-hardened. Memoization plus a deterministic timeout / backtrack-step budget that always terminates — never a host watchdog alone.
- Differentially tested against Onigmo/MRI; behaviour is byte-for-byte.
- Standalone & reusable. No dependency on the Ruby runtime.
- 100% test coverage is the target, enforced as a CI gate.
Engine roadmap (Phases 0–4) complete. A backtracking VM with leftmost-first
semantics covering literals/escapes, ., character classes (incl. POSIX
[[:alpha:]] and multibyte members [é] / [à-ï]), anchors (\A \z \Z \G ^ $),
every quantifier mode (greedy, lazy *? +? ??, possessive *+ ++ ?+, atomic
groups (?>…)), capturing/non-capturing/named groups, alternation,
backreferences \1 / \k<name>, lookahead and fixed/bounded-width
lookbehind, recursive subexpression calls \g<…>, inline flags (?imx),
\p{…} Unicode properties, rune-level /i case folding, \h / \H, \R,
and UTF-8 / ASCII-8BIT multi-encoding with a char-advancing .. It is
ReDoS-hardened (memoization + step budget + recursion cap + WithTimeout) and
carries a transparent start-position / interior-literal optimizer (up to
~210× faster) plus a lazy-NFA + cached-DFA fast path, with a benchmark suite —
differential-tested against MRI, 100% coverage, CI green across 6 arches. Phase 5
(the Ruby Regexp/MatchData
surface and replacement DSL) is downstream in the go-embedded-ruby adapter that
consumes this engine, not part of this module. The architecture, the documented
out-of-scope boundaries, and the six-phase roadmap live in
regexp/docs/plan-regexp.md
and on the documentation site.
BSD-3-Clause.