bench

#!/usr/bin/env bash
set -euo pipefail

source "$HOME/dotfiles/bin/src/capture.bash"

root_dir="$(pwd)"

par2_cli="${PAR2_CLI_BIN:-$root_dir/zig-out/bin/par2z-cli}"
other_cli="${PAR2_OTHER_BIN:-par2}"
other_name="${PAR2_OTHER_NAME:-par2cmdline}"
bench_log="${PAR2_BENCH_LOG:-bench-results.tsv}"
prng_gen="${PAR2_PRNG_GEN:-$root_dir/zig-out/par2z/bin/prng-gen}"
bench_optimize="${PAR2_BENCH_OPTIMIZE:-ReleaseFast}"
bench_build="${PAR2_BENCH_BUILD:-1}"

bench_size="${PAR2_BENCH_SIZE:-33554432}"
block_size="${PAR2_BENCH_BLOCK_SIZE:-4096}"
redundancy="${PAR2_BENCH_REDUNDANCY:-10}"
iters="${PAR2_BENCH_ITERS:-2}"
corrupt_bytes="${PAR2_BENCH_CORRUPT_BYTES:-4096}"
bench_seed="${PAR2_BENCH_SEED:-1}"
bench_seq="${PAR2_BENCH_SEQ:-1}"

out=""
err=""
rc=0

if command -v mktmp >/dev/null 2>&1; then
	tmpdir="$(mktmp --tmpdir)"
else
	tmpdir="$(mktemp -d "${TMPDIR:-/tmp}/par2-bench.XXXXXX")"
	echo "warning: mktmp not found; using mktemp" >&2
fi
cleanup() {
	if [ -n "${tmpdir-}" ] && [ -d "$tmpdir" ]; then
		rm -rf "$tmpdir"
	fi
}
trap cleanup EXIT

now_ms() {
	perl -MTime::HiRes -e 'printf "%.0f\n", Time::HiRes::time()*1000'
}

run_timed() {
	local tool="$1"
	local phase="$2"
	shift 2
	local start end elapsed rate
	start="$(now_ms)"
	capture -- "$@"
	end="$(now_ms)"
	elapsed=$((end - start))
	if [ "$rc" -ne 0 ]; then
		echo "$tool $phase failed: $err" >&2
		exit 1
	fi
	rate="$(awk -v size="$bench_size" -v ms="$elapsed" 'BEGIN { if (ms == 0) { print "inf"; } else { printf "%.2f", (size/1048576)/(ms/1000) } }')"
	printf '%-24s %8s ms %8s MiB/s\n' "$tool $phase" "$elapsed" "$rate"
	log_result "$tool" "$phase" "$elapsed" "$rate"
}

gen_data() {
	local path="$1"
	if [ -x "$prng_gen" ]; then
		capture "$prng_gen" "$path" "$bench_size" "$bench_seed" "$bench_seq"
		if [ "$rc" -ne 0 ]; then
			echo "prng-gen failed: $err" >&2
			exit 1
		fi
	else
		capture dd if=/dev/urandom of="$path" bs=1 count="$bench_size" status=none
		if [ "$rc" -ne 0 ]; then
			echo "dd failed: $err" >&2
			exit 1
		fi
	fi
}

corrupt_file() {
	local path="$1"
	capture dd if=/dev/urandom of="$path" bs=1 count="$corrupt_bytes" seek=4096 conv=notrunc status=none
	if [ "$rc" -ne 0 ]; then
		echo "dd corrupt failed: $err" >&2
		exit 1
	fi
}

bench_tool() {
	local name="$1"
	local kind="$2"
	local dir="$3"
	local data="$4"
	mkdir -p "$dir"
	cp "$data" "$dir/data.bin"
	cp "$data" "$dir/orig.bin"

	if [ "$kind" = "par2" ]; then
		local par2_path="$dir/data.par2"
		run_timed "$name" "create" "$other_cli" create "-s$block_size" "-r$redundancy" "$par2_path" "$dir/data.bin"
		run_timed "$name" "verify" "$other_cli" verify "$par2_path" "$dir/data.bin"
		corrupt_file "$dir/data.bin"
		run_timed "$name" "repair" "$other_cli" repair -q "$par2_path" "$dir/data.bin"
	else
		local par2_path="$dir/data.par2"
		PAR2_MUTE_DEFAULTS=1 run_timed "$name" "create" "$par2_cli" create --block-size "$block_size" --redundancy-percent "$redundancy" "$par2_path" "$dir/data.bin"
		run_timed "$name" "verify" "$par2_cli" verify "$par2_path" "$dir/data.bin"
		corrupt_file "$dir/data.bin"
		run_timed "$name" "recover" "$par2_cli" recover -o "$dir/out" "$par2_path" "$dir/data.bin"
	fi
}

if [ "$bench_build" -ne 0 ]; then
	capture zig build -Doptimize="$bench_optimize"
	if [ "$rc" -ne 0 ]; then
		echo "zig build failed: $err" >&2
		exit 1
	fi
	par2_cli="$root_dir/zig-out/bin/par2z-cli"
fi
capture "$par2_cli" -h
if [ "$rc" -ne 0 ]; then
	echo "par2z-cli not runnable at $par2_cli" >&2
	exit 1
fi
capture "$other_cli" -V
if [ "$rc" -ne 0 ]; then
	echo "other par2 not runnable at $other_cli" >&2
	exit 1
fi

bench_head="$(git rev-parse HEAD 2>/dev/null || true)"
if [ -z "$bench_head" ]; then
	bench_head="$(jj log --ignore-working-copy -n1 --template 'commit_id' 2>/dev/null || true)"
fi
if [ -z "$bench_head" ]; then
	bench_head="unknown"
fi

if [ ! -f "$bench_log" ]; then
	printf "timestamp\tcommit\ttool\tphase\telapsed_ms\trate_mib_s\tbench_size\tblock_size\tredundancy\titers\tother_cli\tpar2_cli\n" >> "$bench_log"
fi

log_result() {
	local tool="$1"
	local phase="$2"
	local elapsed="$3"
	local rate="$4"
	local ts
	ts="$(date -u +"%Y-%m-%dT%H:%M:%SZ")"
	printf "%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\n" \
		"$ts" "$bench_head" "$tool" "$phase" "$elapsed" "$rate" \
		"$bench_size" "$block_size" "$redundancy" "$iters" \
		"$other_cli" "$par2_cli" >> "$bench_log"
}

echo "bench size=$bench_size block_size=$block_size redundancy=$redundancy iters=$iters"
data="$tmpdir/data.bin"
gen_data "$data"

iter=1
while [ "$iter" -le "$iters" ]; do
	echo "iteration $iter/$iters"
	bench_tool "$other_name" "par2" "$tmpdir/par2-$iter" "$data"
	bench_tool "par2z-cli" "ours" "$tmpdir/ours-$iter" "$data"
	iter=$((iter + 1))
done