regex: benchmark and optimizations

1 files changed, 22 insertions, 0 deletions

diff --git a/benches/regex.rs b/benches/regex.rs
new file mode 100644
index 0000000..7f20606
--- /dev/null
+++ b/benches/regex.rs
@@ -0,0 +1,22 @@
+use criterion::{Criterion, criterion_group, criterion_main};
+use pish::parse::{
+    Parse,
+    regex::{Pattern, bc::BytecodeCompiledRegex},
+};
+
+/// https://mattmahoney.net/dc/enwik8.zip
+pub fn regex_throughput(c: &mut Criterion) {
+    let mut content = std::fs::read("enwik8").unwrap();
+    content.truncate(1_000_000);
+    let pat = Pattern::parse_from_bytes(b".*GNU.plus.Linux.*").unwrap();
+    let compiled = BytecodeCompiledRegex::try_from(pat).unwrap();
+
+    c.bench_function("enwik8 match-all", |b| {
+        b.iter(|| {
+            let _ = compiled.re_match(&content);
+        })
+    });
+}
+
+criterion_group!(benches, regex_throughput);
+criterion_main!(benches);