GitHub - mnurzia/bbre: Compact, non-backtracking, unicode aware regular expressions in C89

bbre

Regular expression engine written in ANSI C.

Features:

• Non-backtracking (linear-time execution guarantee)
• One .h file and one .c file
• Supports matching many distinct patterns at once (regexp sets)
• Syntax parity with re2
• ~6000 lines of code (only ~1800 semicolons)
• Extensively tested
• Pluggable allocator support

Usage

#include <assert.h>
#include <string.h>

#include "bbre.h"

int main(void)
{
  const char *text = "Hello WorLd!";
  bbre *reg = bbre_init_pattern("Hel*o (?i)[w]orld!");
  assert(bbre_is_match(reg, text, strlen(text)) == 1);
  bbre_destroy(reg);
  return 0;
}

FAQ

Why write this?

I like regular expressions and C89.

You should have written this in Rust.

That would have been too much fun.

What does the bb stand for?

"blueberry", because I ate a lot of frozen blueberries while I wrote this.

Caveats

• Written in C89.
• Not optimized for memory fragmentation: uses lots of variable-size buffers.
• Assumes width of integer types in a way that's not completely compliant with C89/99. This works on 99% of platforms out there, but of course part of the fun in C is the esoteric 1%.
• Probably won't work on EBCDIC systems.