From 2804638d8440f1d8983077124b82458b0c503268 Mon Sep 17 00:00:00 2001
From: Camden Dixie O'Brien <cdo@wip.sh>
Date: Sat, 26 Oct 2024 23:32:04 +0100
Subject: [PATCH] Support sequences in construct()

---
 lib/construct.c         | 42 +++++++++++++++++++++++++++++++-
 tests/construct_tests.c | 53 +++++++++++++++++++++++++++++++++++++++--
 2 files changed, 92 insertions(+), 3 deletions(-)

diff --git a/lib/construct.c b/lib/construct.c
index e40956a..e54bcc6 100644
--- a/lib/construct.c
+++ b/lib/construct.c
@@ -7,6 +7,7 @@
 
 #include <assert.h>
 #include <stdlib.h>
+#include <string.h>
 
 static void construct_literal(char literal, fsa_t *out)
 {
@@ -34,9 +35,48 @@ static void construct_term(const regex_term_t *term, fsa_t *out)
 	}
 }
 
+static void concat_fsas(fsa_t *base, const fsa_t *other)
+{
+	const int new_count = base->count + other->count - 1;
+	if (base->capacity < new_count) {
+		do
+			base->capacity *= 2;
+		while (base->capacity < new_count);
+		base->states = realloc(base->states, base->capacity);
+		assert(base->states);
+	}
+
+	const int copy_size = (other->count - 1) * sizeof(fsa_state_t);
+	memcpy(&base->states[base->count], &other->states[1], copy_size);
+
+	const int id_offset = base->count - 1;
+	for (int i = base->count; i < new_count; ++i) {
+		fsa_state_t *state = &base->states[i];
+		for (int j = 0; j < state->count; ++j) {
+			if (0 == state->rules[j].next)
+				state->rules[j].next = base->initial;
+			else
+				state->rules[j].next += id_offset;
+		}
+	}
+
+	base->count = new_count;
+	base->initial = other->initial + id_offset;
+
+	free(other->states[0].rules);
+	free(other->states);
+}
+
 static void construct_sequence(const regex_sequence_t *seq, fsa_t *out)
 {
-	construct_term(&seq->contents[0], out);
+	assert(seq->count > 0);
+
+	fsa_t term_fsa;
+	construct_term(&seq->contents[seq->count - 1], out);
+	for (int i = seq->count - 2; i >= 0; --i) {
+		construct_term(&seq->contents[i], &term_fsa);
+		concat_fsas(out, &term_fsa);
+	}
 }
 
 void construct(const regex_t *regex, fsa_t *out)
diff --git a/tests/construct_tests.c b/tests/construct_tests.c
index 461ce76..83199a0 100644
--- a/tests/construct_tests.c
+++ b/tests/construct_tests.c
@@ -47,8 +47,56 @@ static void test_literal_expression(void)
 	ASSERT_EQ(2, fsa.count);
 	ASSERT_EQ(1, initial->count);
 	ASSERT_EQ('a', initial->rules[0].input);
-	ASSERT_TRUE(fsa.states[initial->rules[0].next].final);
-	ASSERT_EQ(0, fsa.states[initial->rules[0].next].count);
+
+	const int next = initial->rules[0].next;
+	ASSERT_TRUE(fsa.states[next].final);
+	ASSERT_EQ(0, fsa.states[next].count);
+
+	regex_free(&regex);
+	fsa_free(&fsa);
+}
+
+static void test_sequence(void)
+{
+	regex_term_t *terms = malloc(3 * sizeof(regex_term_t));
+	terms[0].type = REGEX_TERM_LITERAL;
+	terms[0].literal = 'a';
+	terms[1].type = REGEX_TERM_LITERAL;
+	terms[1].literal = 'b';
+	terms[2].type = REGEX_TERM_LITERAL;
+	terms[2].literal = 'c';
+	regex_sequence_t *alternatives = malloc(1 * sizeof(regex_sequence_t));
+	alternatives[0].count = alternatives[0].capacity = 3;
+	alternatives[0].contents = terms;
+	regex_t regex = { .count = 1, .capacity = 1, .contents = alternatives };
+
+	fsa_t fsa;
+	construct(&regex, &fsa);
+
+	int next = fsa.initial;
+	const fsa_state_t *state;
+
+	state = &fsa.states[next];
+	ASSERT_FALSE(state->final);
+	ASSERT_EQ(1, state->count);
+	ASSERT_EQ('a', state->rules[0].input);
+	next = state->rules[0].next;
+
+	state = &fsa.states[next];
+	ASSERT_FALSE(state->final);
+	ASSERT_EQ(1, state->count);
+	ASSERT_EQ('b', state->rules[0].input);
+	next = state->rules[0].next;
+
+	state = &fsa.states[next];
+	ASSERT_FALSE(state->final);
+	ASSERT_EQ(1, state->count);
+	ASSERT_EQ('c', state->rules[0].input);
+	next = state->rules[0].next;
+
+	state = &fsa.states[next];
+	ASSERT_TRUE(state->final);
+	ASSERT_EQ(0, state->count);
 
 	regex_free(&regex);
 	fsa_free(&fsa);
@@ -59,5 +107,6 @@ int main(void)
 	TESTING_BEGIN();
 	test_empty_expression();
 	test_literal_expression();
+	test_sequence();
 	return TESTING_END();
 }