Implement first iteration of parser and write test scripts

lib/parser.c

@@ -0,0 +1,84 @@
+/*
+ * Copyright (c) Camden Dixie O'Brien
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+#include "parser.h"
+
+#include <stdbool.h>
+#include <stdlib.h>
+
+static bool is_special(char c)
+{
+	switch (c) {
+	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 {
+		return -1;
+	}
+}
+
+static int parse_term(const char *input, int rem, term_t *out)
+{
+	int result, used = 0;
+
+	result = parse_literal(input + used, rem - used, &out->literal);
+	if (result < 0)
+		return -1;
+	out->quantifier = QUANTIFIER_NONE;
+	out->type = TERM_TYPE_LITERAL;
+	used += result;
+
+	return used;
+}
+
+static int parse_sequence(const char *input, int rem, sequence_t *out)
+{
+	int result, used = 0;
+
+	out->contents = calloc(1, sizeof(term_t));
+	out->len = 0;
+	out->capacity = 1;
+
+	result = parse_term(input + used, rem - used, &out->contents[0]);
+	if (result < 0)
+		return -1;
+	++out->len;
+	used += result;
+
+	return used;
+}
+
+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));
+		result = parse_regex(input + used, rem - used, out->alternative);
+		if (result < 0)
+			return -1;
+		used += result;
+	}
+
+	return used;
+}

lib/parser.h

@@ -0,0 +1,13 @@
+/*
+ * Copyright (c) Camden Dixie O'Brien
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+#ifndef PARSER_H
+#define PARSER_H
+
+#include "regex.h"
+
+int parse_regex(const char *input, int rem, regex_t *out);
+
+#endif