Add Micro Benchmark (#3466)

.cargo/config.toml

@@ -1,5 +1,6 @@
 [alias]
 dev = "run --package ruff_dev --bin ruff_dev"
+benchmark = "bench -p ruff_benchmark --"
 
 [target.'cfg(all())']
 rustflags = [

Cargo.lock

@@ -2024,6 +2024,18 @@ dependencies = [
  "toml",
 ]
 
+[[package]]
+name = "ruff_benchmark"
+version = "0.0.0"
+dependencies = [
+ "criterion",
+ "mimalloc",
+ "ruff",
+ "tikv-jemallocator",
+ "ureq",
+ "url",
+]
+
 [[package]]
 name = "ruff_cache"
 version = "0.0.0"

Cargo.toml

@@ -4,6 +4,10 @@ members = ["crates/*"]
 [workspace.package]
 edition = "2021"
 rust-version = "1.67"
+homepage = "https://beta.ruff.rs/docs/"
+documentation = "https://beta.ruff.rs/docs/"
+repository = "https://github.com/charliermarsh/ruff"
+authors = ["Charlie Marsh <charlie.r.marsh@gmail.com>"]
 
 [workspace.dependencies]
 anyhow = { version = "1.0.69" }
@@ -59,3 +63,9 @@ opt-level = 3
 # https://github.com/bytecodealliance/wasm-tools/blob/b5c3d98e40590512a3b12470ef358d5c7b983b15/crates/wasmparser/src/limits.rs#L29
 [profile.dev.package.rustpython-parser]
 opt-level = 1
+
+# Use the `--profile release-debug` flag to show symbols in release mode.
+# e.g. `cargo build --profile release-debug`
+[profile.release-debug]
+inherits = "release"
+debug = 1

crates/ruff/Cargo.toml

@@ -1,12 +1,12 @@
 [package]
 name = "ruff"
 version = "0.0.255"
-authors = ["Charlie Marsh <charlie.r.marsh@gmail.com>"]
-edition = { workspace = true }
-rust-version = { workspace = true }
-documentation = "https://github.com/charliermarsh/ruff"
-homepage = "https://github.com/charliermarsh/ruff"
-repository = "https://github.com/charliermarsh/ruff"
+authors.workspace = true
+edition.workspace = true
+rust-version.workspace = true
+documentation.workspace = true
+homepage.workspace = true
+repository.workspace = true
 readme = "README.md"
 license = "MIT"
 

crates/ruff_benchmark/Cargo.toml

@@ -0,0 +1,31 @@
+[package]
+name = "ruff_benchmark"
+version = "0.0.0"
+publish = false
+edition.workspace = true
+authors.workspace = true
+homepage.workspace = true
+documentation.workspace = true
+repository.workspace = true
+description = "Ruff Micro-benchmarks"
+
+[lib]
+bench = false
+
+[[bench]]
+name = "linter"
+harness = false
+
+[dependencies]
+ruff.path = "../ruff"
+url = "2.3.1"
+ureq = "2.6.2"
+
+[dev-dependencies]
+criterion = { version = "0.4.0"}
+
+[target.'cfg(target_os = "windows")'.dependencies]
+mimalloc = "0.1.34"
+
+[target.'cfg(all(not(target_os = "windows"), not(target_os = "openbsd"), any(target_arch = "x86_64", target_arch = "aarch64", target_arch = "powerpc64")))'.dependencies]
+tikv-jemallocator = "0.5.0"

crates/ruff_benchmark/README.md

@@ -0,0 +1,87 @@
+## Ruff Micro-benchmarks
+
+Benchmarks for the different Ruff-tools.
+
+### Run Benchmark
+
+You can run the benchmarks with
+
+```shell
+cargo benchmark
+```
+
+### Benchmark driven Development
+
+You can use `--save-baseline=<name>` to store an initial baseline benchmark (e.g. on `main`) and then use
+`--benchmark=<name>` to compare against that benchmark. Criterion will print a message telling you if the benchmark improved/regressed compared to that baseline.
+
+```shell
+# Run once on your "baseline" code
+cargo benchmark --save-baseline=main
+
+# Then iterate with
+cargo benchmark --baseline=main
+```
+
+### PR Summary
+You can use `--save-baseline` and `critcmp` to get a pretty comparison between two recordings.
+This is useful to illustrate the improvements of a PR.
+
+```shell
+# On main
+cargo benchmark --save-baseline=main
+
+# After applying your changes
+cargo benchmark --save-baseline=pr
+
+critcmp main pr
+```
+
+You must install [`critcmp`](https://github.com/BurntSushi/critcmp) for the comparison.
+
+```bash
+cargo install critcmp
+```
+
+### Tips
+
+* Use `cargo benchmark <filter>` to only run specific benchmarks. For example: `cargo benchmark linter/pydantic` to only run the pydantic tests.
+* Use `cargo benchmark --quiet` for a more cleaned up output (without statistical relevance)
+* Use `cargo benchmark --quick` to get faster results (more prone to noise)
+
+## Profiling
+
+### Linux
+
+Install `perf` and build `ruff_benchmark` with the `release-debug` profile and then run it with perf
+
+```shell
+cargo bench -p ruff_benchmark --no-run --profile=release-debug && perf record -g -F 9999 cargo bench -p ruff_benchmark --profile=release-debug -- --profile-time=1
+```
+
+Then convert the recorded profile
+
+```shell
+perf script -F +pid > /tmp/test.perf
+```
+
+You can now view the converted file with [firefox profiler](https://profiler.firefox.com/)
+
+You can find a more in-depth guide [here](https://profiler.firefox.com/docs/#/./guide-perf-profiling)
+
+### Mac
+
+Install [`cargo-instruments`](https://crates.io/crates/cargo-instruments):
+
+```shell
+cargo install cargo-instruments
+```
+
+Then run the profiler with
+
+```shell
+cargo instruments -t time --bench linter --profile release-debug -p ruff_benchmark -- --profile-time=1
+```
+
+* `-t`: Specifies what to profile. Useful options are `time` to profile the wall time and `alloc` for profiling the allocations.
+* You may want to pass an additional filter to run a single test file

crates/ruff_benchmark/benches/linter.rs

@@ -0,0 +1,71 @@
+use criterion::{black_box, criterion_group, criterion_main, BenchmarkId, Criterion, Throughput};
+use ruff::linter::lint_only;
+use ruff::settings::{flags, Settings};
+use ruff_benchmark::{TestCase, TestCaseSpeed, TestFile, TestFileDownloadError};
+use std::time::Duration;
+
+#[cfg(target_os = "windows")]
+#[global_allocator]
+static GLOBAL: mimalloc::MiMalloc = mimalloc::MiMalloc;
+
+#[cfg(all(
+    not(target_os = "windows"),
+    not(target_os = "openbsd"),
+    any(
+        target_arch = "x86_64",
+        target_arch = "aarch64",
+        target_arch = "powerpc64"
+    )
+))]
+#[global_allocator]
+static GLOBAL: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
+
+fn create_test_cases() -> Result<Vec<TestCase>, TestFileDownloadError> {
+    Ok(vec![
+        TestCase::fast(TestFile::try_download("numpy/globals.py", "https://github.com/numpy/numpy/blob/89d64415e349ca75a25250f22b874aa16e5c0973/numpy/_globals.py")?),
+        TestCase::normal(TestFile::try_download(
+            "pydantic/types.py",
+            "https://raw.githubusercontent.com/pydantic/pydantic/main/pydantic/types.py",
+        )?),
+        TestCase::normal(TestFile::try_download("numpy/ctypeslib.py", "https://github.com/numpy/numpy/blob/main/numpy/ctypeslib.py")?),
+        TestCase::slow(TestFile::try_download(
+            "large/dataset.py",
+            "https://raw.githubusercontent.com/DHI/mikeio/b7d26418f4db2909b0aa965253dbe83194d7bb5b/tests/test_dataset.py",
+        )?),
+    ])
+}
+
+fn benchmark_linter(criterion: &mut Criterion) {
+    let test_cases = create_test_cases().unwrap();
+    let mut group = criterion.benchmark_group("linter");
+
+    for case in test_cases {
+        group.throughput(Throughput::Bytes(case.code().len() as u64));
+        group.measurement_time(match case.speed() {
+            TestCaseSpeed::Fast => Duration::from_secs(10),
+            TestCaseSpeed::Normal => Duration::from_secs(20),
+            TestCaseSpeed::Slow => Duration::from_secs(30),
+        });
+        group.bench_with_input(
+            BenchmarkId::from_parameter(case.name()),
+            &case,
+            |b, case| {
+                b.iter(|| {
+                    lint_only(
+                        case.code(),
+                        &case.path(),
+                        None,
+                        &black_box(Settings::default()),
+                        flags::Noqa::Enabled,
+                        flags::Autofix::Enabled,
+                    )
+                });
+            },
+        );
+    }
+
+    group.finish();
+}
+
+criterion_group!(benches, benchmark_linter);
+criterion_main!(benches);

crates/ruff_benchmark/src/lib.rs

@@ -0,0 +1,144 @@
+use std::fmt::{Display, Formatter};
+use std::path::{Path, PathBuf};
+use url::Url;
+
+/// Relative size of a test case. Benchmarks can use it to configure the time for how long a benchmark should run to get stable results.
+#[derive(Copy, Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
+pub enum TestCaseSpeed {
+    /// A test case that is fast to run
+    Fast,
+
+    /// A normal test case
+    Normal,
+
+    /// A slow test case
+    Slow,
+}
+
+#[derive(Debug, Clone)]
+pub struct TestCase {
+    file: TestFile,
+    speed: TestCaseSpeed,
+}
+
+impl TestCase {
+    pub fn fast(file: TestFile) -> Self {
+        Self {
+            file,
+            speed: TestCaseSpeed::Fast,
+        }
+    }
+
+    pub fn normal(file: TestFile) -> Self {
+        Self {
+            file,
+            speed: TestCaseSpeed::Normal,
+        }
+    }
+
+    pub fn slow(file: TestFile) -> Self {
+        Self {
+            file,
+            speed: TestCaseSpeed::Slow,
+        }
+    }
+}
+
+impl TestCase {
+    pub fn code(&self) -> &str {
+        &self.file.code
+    }
+
+    pub fn name(&self) -> &str {
+        &self.file.name
+    }
+
+    pub fn speed(&self) -> TestCaseSpeed {
+        self.speed
+    }
+
+    pub fn path(&self) -> PathBuf {
+        Path::new("target").join(self.name())
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct TestFile {
+    name: String,
+    code: String,
+}
+
+impl TestFile {
+    pub fn new(name: String, code: String) -> Self {
+        Self { name, code }
+    }
+
+    #[allow(clippy::print_stderr)]
+    pub fn try_download(name: &str, url: &str) -> Result<TestFile, TestFileDownloadError> {
+        let url = Url::parse(url)?;
+
+        let cached_filename = Path::new("target").join(name);
+
+        if let Ok(content) = std::fs::read_to_string(&cached_filename) {
+            Ok(TestFile::new(name.to_string(), content))
+        } else {
+            // File not yet cached, download and cache it in the target directory
+            let response = ureq::get(url.as_str()).call()?;
+
+            let content = response.into_string()?;
+
+            // SAFETY: There's always the `target` directory
+            let parent = cached_filename.parent().unwrap();
+            if let Err(error) = std::fs::create_dir_all(parent) {
+                eprintln!("Failed to crate the directory for the test case {name}: {error}");
+            } else if let Err(error) = std::fs::write(cached_filename, &content) {
+                eprintln!("Failed to cache test case file downloaded from {url}: {error}");
+            }
+
+            Ok(TestFile::new(name.to_string(), content))
+        }
+    }
+}
+
+#[derive(Debug)]
+pub enum TestFileDownloadError {
+    UrlParse(url::ParseError),
+    Request(Box<ureq::Error>),
+    Download(std::io::Error),
+}
+
+impl From<url::ParseError> for TestFileDownloadError {
+    fn from(value: url::ParseError) -> Self {
+        Self::UrlParse(value)
+    }
+}
+
+impl From<ureq::Error> for TestFileDownloadError {
+    fn from(value: ureq::Error) -> Self {
+        Self::Request(Box::new(value))
+    }
+}
+
+impl From<std::io::Error> for TestFileDownloadError {
+    fn from(value: std::io::Error) -> Self {
+        Self::Download(value)
+    }
+}
+
+impl Display for TestFileDownloadError {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        match self {
+            TestFileDownloadError::UrlParse(inner) => {
+                write!(f, "Failed to parse url: {inner}")
+            }
+            TestFileDownloadError::Request(inner) => {
+                write!(f, "Failed to download file: {inner}")
+            }
+            TestFileDownloadError::Download(inner) => {
+                write!(f, "Failed to download file: {inner}")
+            }
+        }
+    }
+}
+
+impl std::error::Error for TestFileDownloadError {}