Write some matching benchmarks

CMakeLists.txt

@@ -19,3 +19,4 @@ endfunction()
 add_subdirectory(lib)
 add_subdirectory(tests)
 add_subdirectory(demo)
+add_subdirectory(benchmarks)

benchmarks/CMakeLists.txt

@@ -0,0 +1,20 @@
+add_library(benchmarking benchmarking.c)
+set_default_target_options(benchmarking)
+target_include_directories(benchmarking PUBLIC include)
+
+function(add_benchmark_suite source)
+  string(REGEX REPLACE ".c$" "" name ${source})
+  add_executable(${name} ${source})
+  set_default_target_options(${name})
+  target_link_libraries(${name} PRIVATE lib benchmarking)
+endfunction()
+
+function(add_benchmark_suites)
+  foreach(source ${ARGN})
+	add_benchmark_suite(${source})
+  endforeach()
+endfunction()
+
+add_benchmark_suites(
+  matching_benchmarks.c
+)

benchmarks/benchmarking.c

@@ -0,0 +1,85 @@
+/*
+ * Copyright (c) Camden Dixie O'Brien
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+#include "benchmarking.h"
+
+#include <assert.h>
+#include <math.h>
+#include <stdio.h>
+
+#define SWAP(x, y) \
+	do { \
+		const double tmp = x; \
+		x = y; \
+		y = tmp; \
+	} while (0)
+
+clock_t benchmark_start, benchmark_end;
+
+static void sort(double *xs, int n)
+{
+	if (n <= 0)
+		return;
+
+	const double pivot = xs[(n - 1) / 2];
+	int lt = 0;
+	int eq = 0;
+	int gt = n - 1;
+	while (eq <= gt) {
+		if (xs[eq] < pivot) {
+			SWAP(xs[eq], xs[lt]);
+			++lt;
+			++eq;
+		} else if (xs[eq] > pivot) {
+			SWAP(xs[eq], xs[gt]);
+			--gt;
+		} else {
+			++eq;
+		}
+	}
+
+	sort(xs, lt);
+	sort(xs + gt + 1, n - (gt + 1));
+}
+
+void benchmark_summarise(double *res, int reps, benchmark_summary_t *out)
+{
+	assert(reps > 0);
+
+	sort(res, reps);
+	const double median = res[reps / 2];
+
+	double sum = 0;
+	for (int i = 0; i < reps; ++i)
+		sum += res[i];
+	const double mean = sum / reps;
+
+	double diff_sum = 0;
+	for (int i = 0; i < reps; ++i)
+		diff_sum += pow(res[i] - mean, 2);
+	const double variance = diff_sum / (reps - 1);
+
+	out->reps = reps;
+	out->total = sum;
+	out->median = median;
+	out->mean = mean;
+	out->min = res[0];
+	out->max = res[reps - 1];
+	out->stddev = sqrt(variance);
+}
+
+void benchmark_print_header(void)
+{
+	printf(
+	    "%-12s %13s %13s %13s %13s %12s\n", "benchmark", "median (µs)",
+	    "mean (µs)", "min (µs)", "max (µs)", "stddev");
+}
+
+void benchmark_print(const char *name, const benchmark_summary_t *s)
+{
+	printf(
+	    "%-12s %12.2f %12.2f %12.2f %12.2f %12.2f\n", name, s->median,
+	    s->mean, s->min, s->max, s->stddev);
+}

benchmarks/include/benchmarking.h

@@ -0,0 +1,50 @@
+/*
+ * Copyright (c) Camden Dixie O'Brien
+ * SPDX-License-Identifier: AGPL-3.0-only
+ */
+
+#ifndef BENCHMARKING_H
+#define BENCHMARKING_H
+
+#include <time.h>
+
+typedef struct {
+	int reps;
+	double total, median, mean, min, max, stddev;
+} benchmark_summary_t;
+
+#define CLOCK_MICROS(c) (1000000 * (double)c / CLOCKS_PER_SEC)
+
+#define BENCHMARKING_BEGIN() benchmark_print_header()
+#define BENCHMARKING_END() 0
+
+#define START_CLOCK() \
+	do { \
+		benchmark_start = clock(); \
+	} while (0)
+
+#define STOP_CLOCK() \
+	do { \
+		benchmark_end = clock(); \
+	} while (0)
+
+#define RUN_BENCHMARK(reps, name, fn, ...) \
+	do { \
+		double res[reps]; \
+		for (int i = 0; i < reps; ++i) { \
+			fn(__VA_ARGS__); \
+			res[i] = CLOCK_MICROS(benchmark_end) \
+			    - CLOCK_MICROS(benchmark_start); \
+		} \
+		benchmark_summary_t summary; \
+		benchmark_summarise(res, reps, &summary); \
+		benchmark_print(name, &summary); \
+	} while (0)
+
+extern clock_t benchmark_start, benchmark_end;
+
+void benchmark_summarise(double *res, int reps, benchmark_summary_t *out);
+void benchmark_print_header(void);
+void benchmark_print(const char *name, const benchmark_summary_t *summary);
+
+#endif

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();
+}

scripts/entr.sh

@@ -1,4 +1,4 @@
 #!/bin/sh
 cd "$(git rev-parse --show-toplevel)"
 find . -not \( -path './.git' -prune \) -not \( -path './build' -prune \) \
-	| entr -s 'clear && scripts/build.sh && scripts/test.sh'
+	| entr -cs 'scripts/build.sh && scripts/test.sh'