Regex Engine Design

The Trieste regex engine is a header-only C++20 NFA-based matcher in include/trieste/regex_engine.h. It compiles UTF-8 regexes with a shunting-yard + Thompson pipeline and simulates them with epoch-based state deduplication.

The engine started as an iregexp-focused implementation, but now includes compatibility extensions used by Trieste parsers and rewrite helpers.

Architecture Overview

The engine processes a regex in four phases:

regex string ──► postfix runes ──► NFA ──► finalization ──► simulation
               (shunting-yard)   (Thompson)  (closures+bitmaps) (parallel)