regex — Overview

A complete, SIMD-accelerated regex engine written in pure Vex. The engine compiles patterns into Thompson NFA bytecode and employs multiple execution strategies for optimal performance.

Execution Strategy Pipeline

When you call regex.test(text), the engine routes through the fastest available path:

Pattern Compile → Thompson NFA → Bytecode Program
                                       ↓
                  ┌────────────────────┼────────────────────┐
                  ↓                    ↓                    ↓
            One-Pass DFA         SIMD Prefix        Lazy DFA
           (small patterns)    (multi-byte lit)   (complex patterns)
                  ↓                    ↓                    ↓
              Direct match      Fast candidate       On-demand
              (no backtrack)     pre-filter          state cache

Quick Start

import { Regex, Match } from "regex";

let re = Regex.new("\\d{3}-\\d{4}");

// Simple test
if re.test("Call 555-1234 now") {
    println("Found phone number!");
}

// Extract match
let! m = Match.new();
if re.exec("Call 555-1234 now", &m) {
    println("Match: {m.value()}");       // "555-1234"
    println("Start: {m.start()}");       // position
}

Module Architecture

File	Lines	Purpose
parser.vx	18,837	Regex syntax parser → AST
ast.vx	6,472	Pattern/Sequence/Span types
thompson.vx	14,511	NFA compiler (AST → bytecode)
prog.vx	4,296	Bytecode program + metadata
onepass.vx	14,227	One-pass DFA builder & executor
lazy_dfa.vx	15,444	Lazy DFA with on-demand states
thompson_fast.vx	22,240	SIMD-accelerated NFA execution
thompson_exec.vx	7,729	Basic Thompson NFA executor
matcher.vx	17,197	High-level search + capture groups
regex.vx	6,179	Public API (Regex, Match)