From 6b273165862a2ac227646108d0cfe1b51deab65d Mon Sep 17 00:00:00 2001
From: Ethiraric <ethiraric@gmail.com>
Date: Fri, 15 Mar 2024 17:34:39 +0100
Subject: [PATCH] Prepare for benchmarks.

---
 saphyr/Cargo.toml          |  4 +++
 saphyr/tools/run_bench.rs  | 71 ++++++++++++++++++++++++++++++++++++++
 saphyr/tools/time_parse.rs |  7 +++-
 3 files changed, 81 insertions(+), 1 deletion(-)
 create mode 100644 saphyr/tools/run_bench.rs

diff --git a/saphyr/Cargo.toml b/saphyr/Cargo.toml
index 15425df..a87b239 100644
--- a/saphyr/Cargo.toml
+++ b/saphyr/Cargo.toml
@@ -35,3 +35,7 @@ path = "tools/dump_events.rs"
 [[bin]]
 name = "time_parse"
 path = "tools/time_parse.rs"
+
+[[bin]]
+name = "run_bench"
+path = "tools/run_bench.rs"
diff --git a/saphyr/tools/run_bench.rs b/saphyr/tools/run_bench.rs
new file mode 100644
index 0000000..cda9db2
--- /dev/null
+++ b/saphyr/tools/run_bench.rs
@@ -0,0 +1,71 @@
+#![allow(clippy::cast_possible_truncation, clippy::cast_precision_loss)]
+
+use std::{env, fs::File, io::prelude::*};
+use yaml_rust2::{
+    parser::{MarkedEventReceiver, Parser},
+    scanner::Marker,
+    Event,
+};
+
+/// A sink which discards any event sent.
+struct NullSink {}
+
+impl MarkedEventReceiver for NullSink {
+    fn on_event(&mut self, _: Event, _: Marker) {}
+}
+
+/// Parse the given input, returning elapsed time in nanoseconds.
+fn do_parse(input: &str) -> u64 {
+    let mut sink = NullSink {};
+    let mut parser = Parser::new(input.chars());
+    let begin = std::time::Instant::now();
+    parser.load(&mut sink, true).unwrap();
+    let end = std::time::Instant::now();
+    (end - begin).as_nanos() as u64
+}
+
+fn main() {
+    let args: Vec<_> = env::args().collect();
+    let iterations: u64 = args[2].parse().unwrap();
+    let output_yaml = args.len() == 4 && args[3] == "--output-yaml";
+    let mut f = File::open(&args[1]).unwrap();
+    let mut s = String::new();
+    f.read_to_string(&mut s).unwrap();
+
+    // Warmup
+    do_parse(&s);
+    do_parse(&s);
+    do_parse(&s);
+
+    // Bench
+    let times: Vec<_> = (0..iterations).map(|_| do_parse(&s)).collect();
+
+    let mut sorted_times = times.clone();
+    sorted_times.sort_unstable();
+
+    // Compute relevant metrics.
+    let sum: u64 = times.iter().sum();
+    let avg = sum / iterations;
+    let min = sorted_times[0];
+    let max = sorted_times[(iterations - 1) as usize];
+    let percentile95 = sorted_times[((95 * iterations) / 100) as usize];
+
+    if output_yaml {
+        println!("parser: yaml-rust2");
+        println!("input: {}", args[1]);
+        println!("average: {avg}");
+        println!("min: {min}");
+        println!("max: {max}");
+        println!("percentile95: {percentile95}");
+        println!("iterations: {iterations}");
+        println!("times:");
+        for time in &times {
+            println!("  - {time}");
+        }
+    } else {
+        println!("Average: {}s", (avg as f64) / 1_000_000_000.0);
+        println!("Min: {}s", (min as f64) / 1_000_000_000.0);
+        println!("Max: {}s", (max as f64) / 1_000_000_000.0);
+        println!("95%: {}s", (percentile95 as f64) / 1_000_000_000.0);
+    }
+}
diff --git a/saphyr/tools/time_parse.rs b/saphyr/tools/time_parse.rs
index 9b551e1..014fcfc 100644
--- a/saphyr/tools/time_parse.rs
+++ b/saphyr/tools/time_parse.rs
@@ -27,5 +27,10 @@ fn main() {
     let begin = std::time::Instant::now();
     parser.load(&mut sink, true).unwrap();
     let end = std::time::Instant::now();
-    println!("Loaded {}MiB in {:?}", s.len() / 1024 / 1024, end - begin);
+
+    if args.len() == 3 && args[2] == "--short" {
+        println!("{}", (end - begin).as_nanos());
+    } else {
+        println!("Loaded {}MiB in {:?}", s.len() / 1024 / 1024, end - begin);
+    }
 }