Add performance numbers

[zhiltsov-max]

• Added performance test
• Added performance numbers in Readme

[Copilot]

Pull request overview

Adds a Criterion benchmark and publishes initial benchmark results in the README to document string_distance::levenshtein() performance.

Changes:

• Added a Criterion benchmark (cargo bench) to measure levenshtein() over multiple string lengths/charsets.
• Documented benchmark results and benchmarking instructions in README.md.
• Added benchmark-related dev dependencies and bench configuration in Cargo.toml (and updated Cargo.lock).

Reviewed changes

Copilot reviewed 4 out of 5 changed files in this pull request and generated 5 comments.

File	Description
benches/my_benchmark.rs	New Criterion benchmark for levenshtein() over ASCII and UTF-8 inputs.
README.md	Adds a Performance section with reported timings and a Benchmarking section.
Cargo.toml	Adds criterion + rand dev-dependencies and declares the benchmark target.
Cargo.lock	Locks new dev-dependency transitive graph for benchmarking.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The generic bound ?std::marker::Sized is invalid Rust syntax and will not compile. Use ?Sized (or just omit it, since R: Rng is Sized by default) in the sample signature.

Suggested change

	fn sample<R: rand::Rng + ?std::marker::Sized>(&self, rng: &mut R) -> char {
	fn sample<R: rand::Rng>(&self, rng: &mut R) -> char {

[Copilot]

Pull request overview

Copilot reviewed 4 out of 5 changed files in this pull request and generated 4 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

use rand; and use string_distance; are unused here (crate names are already in the extern prelude, and you reference them via rand::... / string_distance::...). This will trigger unused-import warnings when running cargo bench. Remove these imports or import the specific items you need (e.g., rand::Rng).

Suggested change

	use rand;
	use std::hint::black_box;
	use string_distance;
	use std::hint::black_box;

[Copilot]

This uses iter_batched with random string generation in the setup closure; Criterion measures the whole loop, so the reported time includes string generation/allocation as well as levenshtein(). If the goal is to benchmark the distance algorithm, pre-generate inputs per length (or use a pool and clone/select) and benchmark only the levenshtein call.

Suggested change

	let mut rng = rand::rng();
	for length in lengths {
	c.bench_function(
	&format!(
	"levenshtein for {charset} strings of length {length}",
	charset = "ASCII",
	length = length
	),
	|b| {
	b.iter_batched(
	|| {
	return (
	generate_random_string(length, &mut rng, &ascii_chars),
	generate_random_string(length, &mut rng, &ascii_chars),
	);
	},
	|v| string_distance::levenshtein(black_box(&v.0), black_box(&v.1)),
	BatchSize::SmallInput,
	)
	},
	);
	}
	for length in lengths {
	c.bench_function(
	&format!(
	"levenshtein for {charset} strings of length {length}",
	charset = "UTF8",
	length = length
	),
	|b| {
	b.iter_batched(
	|| {
	return (
	generate_random_string(length, &mut rng, &utf8_chars),
	generate_random_string(length, &mut rng, &utf8_chars),
	);
	},
	|v| string_distance::levenshtein(black_box(&v.0), black_box(&v.1)),
	BatchSize::SmallInput,
	)
	// Benchmark ASCII inputs
	for length in lengths {
	// Pre-generate a pool of input pairs so the benchmark measures only levenshtein()
	let mut rng = rand::rng();
	let sample_size = 1024usize;
	let mut samples: Vec<(String, String)> = Vec::with_capacity(sample_size);
	for _ in 0..sample_size {
	samples.push((
	generate_random_string(length, &mut rng, &ascii_chars),
	generate_random_string(length, &mut rng, &ascii_chars),
	));
	}
	c.bench_function(
	&format!(
	"levenshtein for {charset} strings of length {length}",
	charset = "ASCII",
	length = length
	),
	move |b| {
	let mut idx = 0usize;
	b.iter(|| {
	let pair = &samples[idx];
	idx += 1;
	if idx == samples.len() {
	idx = 0;
	}
	string_distance::levenshtein(black_box(&pair.0), black_box(&pair.1))
	})
	},
	);
	}
	// Benchmark UTF-8 inputs
	for length in lengths {
	// Pre-generate a pool of input pairs so the benchmark measures only levenshtein()
	let mut rng = rand::rng();
	let sample_size = 1024usize;
	let mut samples: Vec<(String, String)> = Vec::with_capacity(sample_size);
	for _ in 0..sample_size {
	samples.push((
	generate_random_string(length, &mut rng, &utf8_chars),
	generate_random_string(length, &mut rng, &utf8_chars),
	));
	}
	c.bench_function(
	&format!(
	"levenshtein for {charset} strings of length {length}",
	charset = "UTF8",
	length = length
	),
	move |b| {
	let mut idx = 0usize;
	b.iter(|| {
	let pair = &samples[idx];
	idx += 1;
	if idx == samples.len() {
	idx = 0;
	}
	string_distance::levenshtein(black_box(&pair.0), black_box(&pair.1))
	})

[Copilot]

For the "UTF8" benchmarks, length is the number of char scalar values generated, but the library’s UTF path operates on Unicode grapheme clusters (via unicode-segmentation). That means the label "strings of length {length}" may not reflect the actual units processed. Either constrain the generated characters to single-codepoint graphemes or adjust the label/documentation to clarify what "length" means.

Suggested change

	"levenshtein for {charset} strings of length {length}",
	"levenshtein for {charset} strings with {length} Unicode scalar values",