From 27b4b28fba5250c96782bc51c7fe3c409e36a129 Mon Sep 17 00:00:00 2001
From: Camden Dixie O'Brien <cdo@wip.sh>
Date: Fri, 25 Oct 2024 13:30:33 +0100
Subject: [PATCH] Define regex_t data structure and write build script

---
 lib/regex.c      | 43 ++++++++++++++++++++++++++++++++++++++
 lib/regex.h      | 54 ++++++++++++++++++++++++++++++++++++++++++++++++
 scripts/build.sh | 11 ++++++++++
 3 files changed, 108 insertions(+)
 create mode 100644 lib/regex.c
 create mode 100644 lib/regex.h
 create mode 100644 scripts/build.sh

diff --git a/lib/regex.c b/lib/regex.c
new file mode 100644
index 0000000..2100db9
--- /dev/null
+++ b/lib/regex.c
@@ -0,0 +1,43 @@
+/*
+ * Copyright (c) Camden Dixie O'Brien
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+#include "regex.h"
+
+#include <stdlib.h>
+
+static void class_free(class_t *c)
+{
+	if (NULL != c->contents)
+		free(c->contents);
+}
+
+static void sequence_free(sequence_t *s)
+{
+	if (NULL != s->contents) {
+		for (int i = 0; i < s->len; ++i) {
+			switch (s->contents[i].type) {
+			case TERM_TYPE_CLASS:
+				class_free(&s->contents[i].class);
+				break;
+			case TERM_TYPE_REGEX:
+				regex_free_children(&s->contents[i].regex);
+				break;
+			case TERM_TYPE_WILDCARD:
+			case TERM_TYPE_LITERAL:
+				break;
+			}
+		}
+		free(s->contents);
+	}
+}
+
+void regex_free_children(regex_t *r)
+{
+	sequence_free(&r->sequence);
+	if (NULL != r->alternative) {
+		regex_free_children(r->alternative);
+		free(r->alternative);
+	}
+}
diff --git a/lib/regex.h b/lib/regex.h
new file mode 100644
index 0000000..5d9efe2
--- /dev/null
+++ b/lib/regex.h
@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) Camden Dixie O'Brien
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+#ifndef REGEX_H
+#define REGEX_H
+
+#include <stdbool.h>
+
+typedef struct {
+	bool negated;
+	int count, capacity;
+	char *contents;
+} class_t;
+
+typedef enum {
+	QUANTIFIER_NONE,
+	QUANTIFIER_ZERO_OR_MORE,
+	QUANTIFIER_ONE_OR_MORE,
+	QUANTIFIER_ZERO_OR_ONE,
+} quantifier_t;
+
+typedef enum {
+	TERM_TYPE_WILDCARD,
+	TERM_TYPE_CLASS,
+	TERM_TYPE_LITERAL,
+	TERM_TYPE_REGEX,
+} term_type_t;
+
+struct _term;
+typedef struct {
+	int len, capacity;
+	struct _term *contents;
+} sequence_t;
+
+typedef struct _regex {
+	sequence_t sequence;
+	struct _regex *alternative;
+} regex_t;
+
+typedef struct _term {
+	quantifier_t quantifier;
+	term_type_t type;
+	union {
+		class_t class;
+		char literal;
+		regex_t regex;
+	};
+} term_t;
+
+void regex_free_children(regex_t *r);
+
+#endif
diff --git a/scripts/build.sh b/scripts/build.sh
new file mode 100644
index 0000000..2dd4a01
--- /dev/null
+++ b/scripts/build.sh
@@ -0,0 +1,11 @@
+cd "$(git rev-parse --show-toplevel)"
+
+CFLAGS="$CFLAGS -std=c11 -pedantic -Wall -Wextra"
+CFLAGS="$CFLAGS -fsanitize=address,undefined"
+CFLAGS="$CFLAGS -O0 -ggdb"
+
+mkdir -p build
+
+# Build library
+clang $CFLAGS -Ilib -c lib/regex.c -o build/regex.o
+ar -crs build/lib.a build/regex.o