From 5d109f5f06d4e812c6eac70319cf5714bb0f9af6 Mon Sep 17 00:00:00 2001
From: Camden Dixie O'Brien <cdo@wip.sh>
Date: Sun, 10 Nov 2024 14:08:58 +0000
Subject: [PATCH] Write some matching benchmarks

---
 benchmarks/CMakeLists.txt        |  4 +++
 benchmarks/matching_benchmarks.c | 52 ++++++++++++++++++++++++++++++++
 2 files changed, 56 insertions(+)
 create mode 100644 benchmarks/matching_benchmarks.c

diff --git a/benchmarks/CMakeLists.txt b/benchmarks/CMakeLists.txt
index ea18878..29f82f5 100644
--- a/benchmarks/CMakeLists.txt
+++ b/benchmarks/CMakeLists.txt
@@ -14,3 +14,7 @@ function(add_benchmark_suites)
 	add_benchmark_suite(${source})
   endforeach()
 endfunction()
+
+add_benchmark_suites(
+  matching_benchmarks.c
+)
diff --git a/benchmarks/matching_benchmarks.c b/benchmarks/matching_benchmarks.c
new file mode 100644
index 0000000..c98cc12
--- /dev/null
+++ b/benchmarks/matching_benchmarks.c
@@ -0,0 +1,52 @@
+/*
+ * Copyright (c) Camden Dixie O'Brien
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+#include "benchmarking.h"
+#include "compile.h"
+
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/time.h>
+
+#define LEN 100
+#define RANGE_FIRST 'a'
+#define RANGE_LAST 'z'
+
+#define CLAMP_CHAR(x) (RANGE_FIRST + x % (RANGE_LAST - RANGE_FIRST + 1))
+
+#define RUN_MATCHING_BENCHMARK(reps, name, regex) \
+	do { \
+		fsa_t fsa; \
+		compile(regex, strlen(regex), &fsa); \
+		RUN_BENCHMARK(reps, name, matching_benchmark, &fsa); \
+		fsa_free(&fsa); \
+	} while (0)
+
+static void matching_benchmark(const fsa_t *fsa)
+{
+	char s[LEN];
+	for (int j = 0; j < LEN; ++j)
+		s[j] = CLAMP_CHAR(rand());
+
+	START_CLOCK();
+	fsa_accepts(fsa, s, LEN);
+	STOP_CLOCK();
+}
+
+int main(void)
+{
+	struct timeval tv;
+	gettimeofday(&tv, NULL);
+	srand(tv.tv_usec);
+
+	BENCHMARKING_BEGIN();
+
+	RUN_MATCHING_BENCHMARK(10000, "foo or bar", ".*(foo|bar).*");
+	RUN_MATCHING_BENCHMARK(10000, "regex #1", ".*(abc!?)*|dd+.*");
+	RUN_MATCHING_BENCHMARK(10000, "regex #2", ".*(l|wh)?[aeiou]+.*");
+
+	return BENCHMARKING_END();
+}