113 lines
2.0 KiB
C
113 lines
2.0 KiB
C
/*
|
|
* Copyright (c) Camden Dixie O'Brien
|
|
* SPDX-License-Identifier: AGPL-3.0-only
|
|
*/
|
|
|
|
#include "parser.h"
|
|
|
|
#include <stdbool.h>
|
|
#include <stdlib.h>
|
|
|
|
#define SEQUENCE_START_CAPACITY 8
|
|
|
|
static bool is_special(char c)
|
|
{
|
|
switch (c) {
|
|
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_term(const char *input, int rem, term_t *out)
|
|
{
|
|
int result, used = 0;
|
|
|
|
if (used < rem && '.' == input[0]) {
|
|
out->type = TERM_TYPE_WILDCARD;
|
|
++used;
|
|
} else {
|
|
result = parse_literal(input + used, rem - used, &out->literal);
|
|
if (result < 0)
|
|
return -1;
|
|
out->type = TERM_TYPE_LITERAL;
|
|
used += result;
|
|
}
|
|
|
|
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 = calloc(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) {
|
|
if (input[used] != '|')
|
|
return -1;
|
|
++used;
|
|
|
|
out->alternative = calloc(1, 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;
|
|
}
|
|
|
|
return used;
|
|
}
|