regex-engine/lib/parser.c

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

#include "parser.h"

#include <stdbool.h>
#include <stdlib.h>

#define CLASS_START_CAPACITY 4
#define SEQUENCE_START_CAPACITY 8

static bool is_special(char c)
{
	switch (c) {
	case '|':
	case '.':
	case '\\':
	case '(':
	case ')':
	case '*':
	case '+':
	case '?':
	case '[':
	case ']':
	case '^':
		return true;
	default:
		return false;
	}
}

static int parse_literal(const char *input, int rem, char *out)
{
	if (rem > 0 && !is_special(input[0])) {
		*out = input[0];
		return 1;
	} else if (rem > 1 && '\\' == input[0]) {
		*out = input[1];
		return 2;
	} else {
		return -1;
	}
}

static int parse_class(const char *input, int rem, class_t *out)
{
	int result, used = 0;

	if (used >= rem || '[' != input[used])
		return -1;
	++used;

	if (used < rem && '^' == input[used]) {
		out->negated = true;
		++used;
	} else {
		out->negated = false;
	}

	out->count = 0;
	out->capacity = CLASS_START_CAPACITY;
	out->contents = malloc(out->capacity);
	if (NULL == out->contents)
		return -1;

	while (used < rem) {
		if (out->count >= out->capacity) {
			out->capacity *= 2;
			out->contents = realloc(out->contents, out->capacity);
			if (NULL == out->contents)
				return -1;
		}

		result = parse_literal(
		    input + used, rem - used, &out->contents[out->count]);
		if (result < 0)
			break;
		used += result;
		++out->count;
	}

	if (used >= rem || ']' != input[used])
		return -1;
	++used;

	return out->count > 0 ? used : -1;
}

static int parse_term(const char *input, int rem, term_t *out)
{
	int result, used = 0;

	if (1 > rem)
		return -1;

	if ('.' == input[0]) {
		out->type = TERM_TYPE_WILDCARD;
		++used;
	} else if ('(' == input[0]) {
		++used;

		result = parse_regex(input + used, rem - used, &out->regex);
		if (result < 0)
			return -1;
		out->type = TERM_TYPE_REGEX;
		used += result;

		if (')' != input[used])
			return -1;
		++used;
	} else if ('[' == input[0]) {
		result = parse_class(input + used, rem - used, &out->class);
		if (result < 0)
			return -1;
		out->type = TERM_TYPE_CLASS;
		used += result;
	} else {
		result = parse_literal(input + used, rem - used, &out->literal);
		if (result < 0)
			return -1;
		out->type = TERM_TYPE_LITERAL;
		used += result;
	}

	if (used < rem) {
		switch (input[used]) {
		case '*':
			out->quantifier = QUANTIFIER_ZERO_OR_MORE;
			++used;
			break;
		case '+':
			out->quantifier = QUANTIFIER_ONE_OR_MORE;
			++used;
			break;
		case '?':
			out->quantifier = QUANTIFIER_ZERO_OR_ONE;
			++used;
			break;
		default:
			out->quantifier = QUANTIFIER_NONE;
		}
	} else {
		out->quantifier = QUANTIFIER_NONE;
	}

	return used;
}

static int parse_sequence(const char *input, int rem, sequence_t *out)
{
	int result, used = 0;

	out->len = 0;
	out->capacity = SEQUENCE_START_CAPACITY;
	out->contents = malloc(out->capacity * sizeof(term_t));
	if (NULL == out->contents)
		return -1;

	while (used < rem) {
		if (out->len >= out->capacity) {
			out->capacity *= 2;
			out->contents
			    = realloc(out->contents, out->capacity * sizeof(term_t));
			if (NULL == out->contents)
				return -1;
		}

		result
		    = parse_term(input + used, rem - used, &out->contents[out->len]);
		if (result < 0)
			break;
		++out->len;
		used += result;
	}

	return out->len > 0 ? used : -1;
}

int parse_regex(const char *input, int rem, regex_t *out)
{
	int result, used = 0;

	result = parse_sequence(input + used, rem - used, &out->sequence);
	if (result < 0)
		return -1;
	used += result;

	if (used < rem && '|' == input[used]) {
		++used;

		out->alternative = malloc(sizeof(regex_t));
		if (NULL == out->alternative)
			return -1;
		result = parse_regex(input + used, rem - used, out->alternative);
		if (result < 0)
			return -1;
		used += result;
	} else {
		out->alternative = NULL;
	}

	return used;
}