From bbecd12c91cc45bcf044f4d4efb90cdf8ac6d49f Mon Sep 17 00:00:00 2001
From: Camden Dixie O'Brien <cdo@wip.sh>
Date: Sun, 27 Oct 2024 13:24:29 +0000
Subject: [PATCH] Implement star construction

---
 lib/construct.c         | 61 ++++++++++++++++++++++++++++++++++++++++-
 tests/construct_tests.c | 25 +++++++++++++++++
 2 files changed, 85 insertions(+), 1 deletion(-)

diff --git a/lib/construct.c b/lib/construct.c
index 7f79b15..f66e764 100644
--- a/lib/construct.c
+++ b/lib/construct.c
@@ -17,6 +17,53 @@ static void construct_literal(char literal, fsa_t *out)
 	out->initial = id;
 }
 
+static void star_fsa(fsa_t *fsa)
+{
+	// If the initial state is already the final state then nothing
+	// needs to be done.
+	if (0 == fsa->initial)
+		return;
+
+	// Copy inital state's rules to final state.
+	fsa_state_t *final = &fsa->states[0];
+	const fsa_state_t *initial = &fsa->states[fsa->initial];
+	if (final->capacity < final->count + initial->count) {
+		do
+			final->capacity *= 2;
+		while (final->capacity < final->count + initial->count);
+		final->rules
+		    = realloc(final->rules, final->capacity * sizeof(fsa_rule_t));
+		assert(final->rules);
+	}
+	const int copy_size = initial->count * sizeof(fsa_rule_t);
+	memcpy(&final->rules[final->count], initial->rules, copy_size);
+	final->count += initial->count;
+
+	// Move states that come after initial state if there are any.
+	if (fsa->count - 1 > fsa->initial) {
+		const int count = fsa->count - fsa->initial - 1;
+		fsa_state_t *start = &fsa->states[fsa->initial];
+		memmove(start, start + 1, count * sizeof(fsa_state_t));
+	}
+
+	// Retarget all states' rules.
+	for (int i = 0; i < fsa->count - 1; ++i) {
+		for (int j = 0; j < fsa->states[i].count; ++j) {
+			if (fsa->states[i].rules[j].next == fsa->initial)
+				fsa->states[i].rules[j].next = 0;
+			else if (fsa->states[i].rules[j].next > fsa->initial)
+				// All states after the initial state have been moved
+				// down by one position.
+				--fsa->states[i].rules[j].next;
+		}
+	}
+
+	--fsa->count;
+	fsa->initial = 0;
+
+	free(initial->rules);
+}
+
 static void construct_term(const regex_term_t *term, fsa_t *out)
 {
 	switch (term->type) {
@@ -28,10 +75,22 @@ static void construct_term(const regex_term_t *term, fsa_t *out)
 		break;
 	case REGEX_TERM_SUBEXPR:
 		return;
-
 	case REGEX_TERM_WILDCARD:
 	case REGEX_TERM_CLASS:
 		assert(false);
+		break;
+	}
+
+	switch (term->quantifier) {
+	case REGEX_QUANTIFIER_NONE:
+		break;
+	case REGEX_QUANTIFIER_STAR:
+		star_fsa(out);
+		break;
+	case REGEX_QUANTIFIER_PLUS:
+	case REGEX_QUANTIFIER_QMARK:
+		assert(false);
+		break;
 	}
 
 	assert(out->states[0].final);
diff --git a/tests/construct_tests.c b/tests/construct_tests.c
index 304cbc9..d81891b 100644
--- a/tests/construct_tests.c
+++ b/tests/construct_tests.c
@@ -132,6 +132,30 @@ static void test_union(void)
 	fsa_free(&fsa);
 }
 
+static void test_star(void)
+{
+	regex_term_t *terms = malloc(1 * sizeof(regex_term_t));
+	terms[0].quantifier = REGEX_QUANTIFIER_STAR;
+	terms[0].type = REGEX_TERM_LITERAL;
+	terms[0].literal = 'a';
+	regex_sequence_t *alternatives = malloc(1 * sizeof(regex_sequence_t));
+	alternatives[0].count = alternatives[0].capacity = 1;
+	alternatives[0].contents = terms;
+	regex_t regex = { .count = 1, .capacity = 1, .contents = alternatives };
+
+	fsa_t fsa;
+	construct(&regex, &fsa);
+
+	const fsa_state_t *initial = &fsa.states[fsa.initial];
+	ASSERT_TRUE(initial->final);
+	ASSERT_EQ(1, initial->count);
+	ASSERT_EQ('a', initial->rules[0].input);
+	ASSERT_EQ(fsa.initial, initial->rules[0].next);
+
+	regex_free(&regex);
+	fsa_free(&fsa);
+}
+
 int main(void)
 {
 	TESTING_BEGIN();
@@ -139,5 +163,6 @@ int main(void)
 	test_literal_expression();
 	test_sequence();
 	test_union();
+	test_star();
 	return TESTING_END();
 }