Skip to content

perf: reduce scans and allocations in version and range parsing #86

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ryuapp merged 1 commit into from
Aug 24, 2026
Merged

perf: reduce scans and allocations in version and range parsing #86

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
25 changes: 20 additions & 5 deletions benches/parse.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,10 +8,25 @@ use criterion2::{Criterion, Throughput, black_box, criterion_group, criterion_ma
use js_semver::{Range, Version};

fn bench_version_parse(c: &mut Criterion) {
let version = "4.5.3";
let mut group = c.benchmark_group("version_parse");
bench_version_case(c, "version_parse", "4.5.3");
bench_version_case(c, "version_parse_prefixed", "v4.5.3");
bench_version_case(c, "version_parse_whitespace", " 4.5.3 ");
bench_version_case(c, "version_parse_prerelease", "4.1.0-rc.1");
bench_version_case(c, "version_parse_build", "4.1.0+build.42");
bench_version_case(c, "version_parse_prerelease_build", "4.1.0-rc.1+build.42");
bench_version_case(
c,
"version_parse_long_metadata",
"19.3.0-canary-044d56f3-20260330+sha.abcdef0123456789",
);
bench_version_case(c, "version_parse_invalid_core", "04.1.0");
bench_version_case(c, "version_parse_invalid_metadata", "4.1.0-alpha..1");
}

fn bench_version_case(c: &mut Criterion, name: &str, version: &str) {
let mut group = c.benchmark_group(name);
group.throughput(Throughput::Bytes(u64::try_from(version.len()).unwrap_or(0)));
group.bench_function("version", |b| {
group.bench_function("js-semver", |b| {
b.iter(|| {
if let Ok(parsed) = Version::parse(black_box(version)) {
black_box(parsed);
Expand All @@ -25,7 +40,7 @@ fn bench_range_parse(c: &mut Criterion) {
let range = "^4.2.0";
let mut group = c.benchmark_group("range_parse");
group.throughput(Throughput::Bytes(u64::try_from(range.len()).unwrap_or(0)));
group.bench_function("range", |b| {
group.bench_function("js-semver", |b| {
b.iter(|| {
if let Ok(parsed) = Range::parse(black_box(range)) {
black_box(parsed);
Expand All @@ -42,7 +57,7 @@ fn bench_parse_and_satisfies(c: &mut Criterion) {
let bytes = u64::try_from(range.len() + version.len()).unwrap_or(0);

group.throughput(Throughput::Bytes(bytes));
group.bench_function("full", |b| {
group.bench_function("js-semver", |b| {
b.iter(|| {
if let (Ok(parsed_range), Ok(parsed_version)) = (
Range::parse(black_box(range)),
Expand Down
203 changes: 88 additions & 115 deletions src/identifier.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,7 +36,7 @@ impl PreRelease {
if s.is_empty() {
return Err(SemverErrorKind::Empty.into());
}
validate_identifiers(s, parse_prerelease_identifier)?;
validate_prerelease(s)?;
Ok(Self(Box::from(s)))
}

Expand Down Expand Up @@ -114,7 +114,7 @@ impl BuildMetadata {
if s.is_empty() {
return Err(SemverErrorKind::Empty.into());
}
validate_identifiers(s, parse_build_metadata_identifier)?;
validate_build_metadata(s)?;
Ok(Self(Box::from(s)))
}

Expand Down Expand Up @@ -182,54 +182,27 @@ impl<'a> Identifier<'a> {
}
}

fn parse_prerelease_identifier(raw: &str) -> Result<Identifier<'_>, SemverError> {
fn parse_prerelease_identifier(raw: &str) -> Identifier<'_> {
let bytes = raw.as_bytes();
if bytes.is_empty() {
return Err(SemverErrorKind::EmptySegment.into());
}

let mut all_digits = true;
for &byte in bytes {
if byte.is_ascii_digit() {
} else if byte.is_ascii_alphabetic() || byte == b'-' {
all_digits = false;
} else {
return Err(SemverErrorKind::UnexpectedCharacter(char::from(byte)).into());
}
}

if all_digits && bytes.len() > 1 && bytes[0] == b'0' {
return Err(SemverErrorKind::LeadingZero.into());
}

Ok(Identifier::new(
Identifier::new(
raw,
if all_digits {
if bytes.iter().all(u8::is_ascii_digit) {
IdentifierKind::Numeric
} else {
IdentifierKind::AlphaNumeric
},
))
)
}

fn parse_build_metadata_identifier(raw: &str) -> Result<Identifier<'_>, SemverError> {
if raw.is_empty() {
return Err(SemverErrorKind::EmptySegment.into());
}
if let Some(byte) = raw
.bytes()
.find(|byte| !byte.is_ascii_alphanumeric() && *byte != b'-')
{
return Err(SemverErrorKind::UnexpectedCharacter(char::from(byte)).into());
}
Ok(Identifier::new(
fn parse_build_metadata_identifier(raw: &str) -> Identifier<'_> {
Identifier::new(
raw,
if raw.bytes().all(|byte| byte.is_ascii_digit()) {
IdentifierKind::Numeric
} else {
IdentifierKind::AlphaNumeric
},
))
)
}

impl PartialOrd for Identifier<'_> {
Expand All @@ -251,26 +224,73 @@ impl Ord for Identifier<'_> {
}
}

fn validate_identifiers<'a>(
s: &'a str,
parser: fn(&'a str) -> Result<Identifier<'a>, SemverError>,
fn validate_prerelease(s: &str) -> Result<(), SemverError> {
let bytes = s.as_bytes();
let mut segment_start = 0;
let mut all_digits = true;

for (pos, &byte) in bytes.iter().enumerate() {
match byte {
b'.' => {
validate_prerelease_segment(bytes, segment_start, pos, all_digits)?;
segment_start = pos + 1;
all_digits = true;
}
b'0'..=b'9' => {}
b'A'..=b'Z' | b'a'..=b'z' | b'-' => all_digits = false,
_ => {
return Err(SemverErrorKind::UnexpectedCharacter(char::from(byte)).into());
}
}
}

validate_prerelease_segment(bytes, segment_start, bytes.len(), all_digits)
}

fn validate_prerelease_segment(
bytes: &[u8],
start: usize,
end: usize,
all_digits: bool,
) -> Result<(), SemverError> {
let mut start = 0;
while start <= s.len() {
let end = next_separator(s, start);
parser(&s[start..end])?;
if end == s.len() {
break;
if start == end {
return Err(SemverErrorKind::EmptySegment.into());
}
if all_digits && end - start > 1 && bytes[start] == b'0' {
return Err(SemverErrorKind::LeadingZero.into());
}
Ok(())
}

fn validate_build_metadata(s: &str) -> Result<(), SemverError> {
let bytes = s.as_bytes();
let mut segment_start = 0;

for (pos, &byte) in bytes.iter().enumerate() {
match byte {
b'.' => {
if pos == segment_start {
return Err(SemverErrorKind::EmptySegment.into());
}
segment_start = pos + 1;
}
b'0'..=b'9' | b'A'..=b'Z' | b'a'..=b'z' | b'-' => {}
_ => {
return Err(SemverErrorKind::UnexpectedCharacter(char::from(byte)).into());
}
}
start = end + 1;
}

if segment_start == bytes.len() {
return Err(SemverErrorKind::EmptySegment.into());
}
Ok(())
}

fn cmp_dot_separated<'a>(
left: &'a str,
right: &'a str,
parser: fn(&'a str) -> Result<Identifier<'a>, SemverError>,
parser: fn(&'a str) -> Identifier<'a>,
) -> Ordering {
let mut left_start = 0;
let mut right_start = 0;
Expand All @@ -279,25 +299,24 @@ fn cmp_dot_separated<'a>(
let right_end = next_separator(right, right_start);
let left_part = &left[left_start..left_end];
let right_part = &right[right_start..right_end];
let Ok(left_id) = parser(left_part) else {
return Ordering::Equal;
};
let Ok(right_id) = parser(right_part) else {
return Ordering::Equal;
};
let left_id = parser(left_part);
let right_id = parser(right_part);
match left_id.cmp(&right_id) {
Ordering::Equal => {}
ord @ (Ordering::Less | Ordering::Greater) => return ord,
}
let left_done = left_end == left.len();
let right_done = right_end == right.len();
if left_done || right_done {
return match (left_done, right_done) {
(true, false) => Ordering::Less,
(false, true) => Ordering::Greater,
(true, true) | (false, false) => Ordering::Equal,
if left_done {
return if right_done {
Ordering::Equal
} else {
Ordering::Less
};
}
if right_done {
return Ordering::Greater;
}
left_start = left_end + 1;
right_start = right_end + 1;
}
Expand Down Expand Up @@ -326,61 +345,39 @@ mod tests {

use super::*;

fn fail_on_bad_left(raw: &str) -> Result<Identifier<'_>, SemverError> {
if raw == "bad-left" {
return Err(SemverErrorKind::Empty.into());
}
parse_prerelease_identifier(raw)
}

fn fail_on_bad_right(raw: &str) -> Result<Identifier<'_>, SemverError> {
if raw == "bad-right" {
return Err(SemverErrorKind::Empty.into());
}
parse_prerelease_identifier(raw)
}

#[test]
fn identifier_ordering() {
assert_eq!(
parse_build_metadata_identifier("1")
.unwrap()
.cmp(&parse_build_metadata_identifier("2").unwrap()),
parse_build_metadata_identifier("1").cmp(&parse_build_metadata_identifier("2")),
Ordering::Less
);
assert_eq!(
parse_build_metadata_identifier("18446744073709551615")
.unwrap()
.cmp(&parse_build_metadata_identifier("18446744073709551616").unwrap()),
.cmp(&parse_build_metadata_identifier("18446744073709551616")),
Ordering::Less
);
assert_eq!(
parse_build_metadata_identifier("18446744073709551616")
.unwrap()
.cmp(&parse_build_metadata_identifier("18446744073709551617").unwrap()),
.cmp(&parse_build_metadata_identifier("18446744073709551617")),
Ordering::Less
);
assert_eq!(
parse_prerelease_identifier("1")
.unwrap()
.cmp(&parse_prerelease_identifier("alpha").unwrap()),
parse_prerelease_identifier("1").cmp(&parse_prerelease_identifier("alpha")),
Ordering::Less
);
assert_eq!(
parse_prerelease_identifier("beta")
.unwrap()
.cmp(&parse_prerelease_identifier("1").unwrap()),
parse_prerelease_identifier("beta").cmp(&parse_prerelease_identifier("1")),
Ordering::Greater
);
}

#[test]
fn prerelease_identifier_validation() {
assert!(parse_prerelease_identifier("").is_err());
assert!(parse_prerelease_identifier("01").is_err());
assert!(parse_prerelease_identifier("a!b").is_err());
assert!(PreRelease::new("").is_err());
assert!(PreRelease::new("01").is_err());
assert!(PreRelease::new("a!b").is_err());
assert_eq!(
parse_prerelease_identifier("alpha-1").unwrap().kind,
parse_prerelease_identifier("alpha-1").kind,
IdentifierKind::AlphaNumeric
);
}
Expand Down Expand Up @@ -452,9 +449,7 @@ mod tests {
#[test]
fn identifier_partial_cmp_and_prefix_order() {
assert_eq!(
parse_prerelease_identifier("alpha")
.unwrap()
.partial_cmp(&parse_prerelease_identifier("alpha").unwrap()),
parse_prerelease_identifier("alpha").partial_cmp(&parse_prerelease_identifier("alpha")),
Some(Ordering::Equal)
);
assert_eq!(
Expand All @@ -476,9 +471,7 @@ mod tests {
Some(Ordering::Less)
);
assert_eq!(
parse_build_metadata_identifier("1")
.unwrap()
.partial_cmp(&parse_build_metadata_identifier("2").unwrap()),
parse_build_metadata_identifier("1").partial_cmp(&parse_build_metadata_identifier("2")),
Some(Ordering::Less)
);
assert_eq!("rc.1".parse::<PreRelease>().unwrap().to_string(), "rc.1");
Expand All @@ -488,26 +481,6 @@ mod tests {
);
}

#[test]
fn cmp_dot_separated_parser_failures_fallback_to_equal() {
assert_eq!(
fail_on_bad_left("alpha").unwrap(),
parse_prerelease_identifier("alpha").unwrap()
);
assert_eq!(
fail_on_bad_right("alpha").unwrap(),
parse_prerelease_identifier("alpha").unwrap()
);
assert_eq!(
cmp_dot_separated("bad-left", "alpha", fail_on_bad_left),
Ordering::Equal
);
assert_eq!(
cmp_dot_separated("alpha", "bad-right", fail_on_bad_right),
Ordering::Equal
);
}

#[test]
fn prerelease_cmp_identifiers_covers_empty_and_equal_cases() {
assert_eq!(
Expand Down
Loading