regex-engine/lib/construct.c

/*
 * Copyright (c) Camden Dixie O'Brien
 * SPDX-License-Identifier: AGPL-3.0-only
 */

#include "construct.h"

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

static void construct_literal(char literal, fsa_t *out)
{
	fsa_init(out);
	const int id = fsa_add_state(out);
	fsa_add_rule(out, id, out->initial, literal);
	out->initial = id;
}

static void construct_term(const regex_term_t *term, fsa_t *out)
{
	switch (term->type) {
	case REGEX_TERM_EMPTY:
		fsa_init(out);
		break;
	case REGEX_TERM_LITERAL:
		construct_literal(term->literal, out);
		break;
	case REGEX_TERM_SUBEXPR:
		return;

	case REGEX_TERM_WILDCARD:
	case REGEX_TERM_CLASS:
		assert(false);
	}

	assert(out->states[0].final);
}

static void concat_fsas(fsa_t *base, const fsa_t *other)
{
	const int new_count = base->count + other->count - 1;
	if (base->capacity < new_count) {
		do
			base->capacity *= 2;
		while (base->capacity < new_count);
		base->states = realloc(base->states, base->capacity);
		assert(base->states);
	}

	const int copy_size = (other->count - 1) * sizeof(fsa_state_t);
	memcpy(&base->states[base->count], &other->states[1], copy_size);

	const int id_offset = base->count - 1;
	for (int i = base->count; i < new_count; ++i) {
		fsa_state_t *state = &base->states[i];
		for (int j = 0; j < state->count; ++j) {
			if (0 == state->rules[j].next)
				state->rules[j].next = base->initial;
			else
				state->rules[j].next += id_offset;
		}
	}

	base->count = new_count;
	base->initial = other->initial + id_offset;

	free(other->states[0].rules);
	free(other->states);

	assert(base->states[0].final);
}

static void construct_sequence(const regex_sequence_t *seq, fsa_t *out)
{
	assert(seq->count > 0);

	fsa_t term_fsa;
	construct_term(&seq->contents[seq->count - 1], out);
	for (int i = seq->count - 2; i >= 0; --i) {
		construct_term(&seq->contents[i], &term_fsa);
		concat_fsas(out, &term_fsa);
	}

	assert(out->states[0].final);
}

void construct(const regex_t *regex, fsa_t *out)
{
	construct_sequence(&regex->contents[0], out);
}