Skip to content

Extend macrobenchmark suite to cover namespaces, buffered vs. nocopy #451

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
merged 1 commit into from
Aug 1, 2022
Merged

Extend macrobenchmark suite to cover namespaces, buffered vs. nocopy #451

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
230 changes: 180 additions & 50 deletions benches/macrobenches.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,27 +1,78 @@
use criterion::{self, criterion_group, criterion_main, Criterion, Throughput};
use quick_xml::events::Event;
use quick_xml::Reader;
use quick_xml::Result as XmlResult;
use quick_xml::{NsReader, Reader};

static RPM_PRIMARY: &str = include_str!("../tests/documents/rpm_primary.xml");
static RPM_PRIMARY2: &str = include_str!("../tests/documents/rpm_primary2.xml");
static RPM_FILELISTS: &str = include_str!("../tests/documents/rpm_filelists.xml");
static RPM_OTHER: &str = include_str!("../tests/documents/rpm_other.xml");
static LIBREOFFICE_DOCUMENT: &str = include_str!("../tests/documents/libreoffice_document.fodt");
static DOCUMENT: &str = include_str!("../tests/documents/document.xml");
static TEST_WRITER_INDENT: &str = include_str!("../tests/documents/test_writer_indent.xml");
static SAMPLE_1: &str = include_str!("../tests/documents/sample_1.xml");
static LINESCORE: &str = include_str!("../tests/documents/linescore.xml");
static SAMPLE_RSS: &str = include_str!("../tests/documents/sample_rss.xml");
static SAMPLE_NS: &str = include_str!("../tests/documents/sample_ns.xml");
static PLAYERS: &str = include_str!("../tests/documents/players.xml");

static INPUTS: &[(&str, &str)] = &[
Copy link
Collaborator Author

@dralley dralley Aug 1, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We do probably want to start asking the question - which benchmarks are strictly necessary to get coverage of the things we want to test. Because otherwise it will take a very long time to run them all.

Or at least document what the most representative subset is so that it can be specified on the command line.

Copy link
Collaborator

@Mingun Mingun Aug 1, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think, that documentation would be enough

// long, mix of attributes and text, not much escaping, mix of attribute lengths, some namespaces
("rpm_primary.xml", RPM_PRIMARY),
// long, mix of attributes and text, not much escaping, mix of attribute lengths, some namespaces
("rpm_primary2.xml", RPM_PRIMARY2),
// long, mostly medium-length text elements, not much escaping
("rpm_filelists.xml", RPM_FILELISTS),
// long, mix of attributes and text, lots of escaping (both entity and char literal), long attributes
("rpm_other.xml", RPM_OTHER),
// long, mix of attributes and text, not much escaping, lots of non-ascii characters, lots of namespaces
("libreoffice_document.fodt", LIBREOFFICE_DOCUMENT),
// medium length, mostly empty tags, a few short attributes per element, no escaping
("document.xml", DOCUMENT),
// medium length, lots of namespaces, no escaping
("test_writer_ident.xml", TEST_WRITER_INDENT),
// short, mix of attributes and text, lots of escapes
("sample_1.xml", SAMPLE_1),
// medium length, lots of attributes, short attributes, few escapes
("linescore.xml", LINESCORE),
// short, lots of namespaces, no escapes
("sample_ns.xml", SAMPLE_NS),
// long, few attributes, mix of attribute lengths, escapes in text content
("sample_rss.xml", SAMPLE_RSS),
// long, lots of attributes, short attributes, no text, no escapes
("players.xml", PLAYERS),
];

// TODO: use fully normalized attribute values
fn parse_document_from_str(doc: &str) -> XmlResult<()> {
let mut r = Reader::from_str(doc);
loop {
match criterion::black_box(r.read_event()?) {
Event::Start(e) | Event::Empty(e) => {
for attr in e.attributes() {
criterion::black_box(attr?.decode_and_unescape_value(&r)?);
}
}
Event::Text(e) => {
criterion::black_box(e.unescape()?);
}
Event::CData(e) => {
criterion::black_box(e.into_inner());
}
Event::End(_) => (),
Event::Eof => break,
_ => (),
}
}
Ok(())
}

static RPM_PRIMARY: &[u8] = include_bytes!("../tests/documents/rpm_primary.xml");
static RPM_PRIMARY2: &[u8] = include_bytes!("../tests/documents/rpm_primary2.xml");
static RPM_FILELISTS: &[u8] = include_bytes!("../tests/documents/rpm_filelists.xml");
static RPM_OTHER: &[u8] = include_bytes!("../tests/documents/rpm_other.xml");
static LIBREOFFICE_DOCUMENT: &[u8] = include_bytes!("../tests/documents/libreoffice_document.fodt");
static DOCUMENT: &[u8] = include_bytes!("../tests/documents/document.xml");
static TEST_WRITER_INDENT: &[u8] = include_bytes!("../tests/documents/test_writer_indent.xml");
static SAMPLE_1: &[u8] = include_bytes!("../tests/documents/sample_1.xml");
static LINESCORE: &[u8] = include_bytes!("../tests/documents/linescore.xml");
static SAMPLE_RSS: &[u8] = include_bytes!("../tests/documents/sample_rss.xml");
static SAMPLE_NS: &[u8] = include_bytes!("../tests/documents/sample_ns.xml");
static PLAYERS: &[u8] = include_bytes!("../tests/documents/players.xml");

// TODO: read the namespaces too
// TODO: use fully normalized attribute values
fn parse_document(doc: &[u8]) -> XmlResult<()> {
fn parse_document_from_bytes(doc: &[u8]) -> XmlResult<()> {
let mut r = Reader::from_reader(doc);
let mut buf = Vec::new();
loop {
match r.read_event()? {
match criterion::black_box(r.read_event_into(&mut buf)?) {
Event::Start(e) | Event::Empty(e) => {
for attr in e.attributes() {
criterion::black_box(attr?.decode_and_unescape_value(&r)?);
Expand All @@ -37,49 +88,128 @@ fn parse_document(doc: &[u8]) -> XmlResult<()> {
Event::Eof => break,
_ => (),
}
buf.clear();
}
Ok(())
}

pub fn bench_fully_parse_document(c: &mut Criterion) {
let mut group = c.benchmark_group("fully_parse_document");

let inputs = [
// long, mix of attributes and text, not much escaping, mix of attribute lengths, some namespaces
("rpm_primary.xml", RPM_PRIMARY),
// long, mix of attributes and text, not much escaping, mix of attribute lengths, some namespaces
("rpm_primary2.xml", RPM_PRIMARY2),
// long, mostly medium-length text elements, not much escaping
("rpm_filelists.xml", RPM_FILELISTS),
// long, mix of attributes and text, lots of escaping (both entity and char literal), long attributes
("rpm_other.xml", RPM_OTHER),
// long, mix of attributes and text, not much escaping, lots of non-ascii characters, lots of namespaces
("libreoffice_document.fodt", LIBREOFFICE_DOCUMENT),
// medium length, mostly empty tags, a few short attributes per element, no escaping
("document.xml", DOCUMENT),
// medium length, lots of namespaces, no escaping
("test_writer_ident.xml", TEST_WRITER_INDENT),
// short, mix of attributes and text, lots of escapes
("sample_1.xml", SAMPLE_1),
// medium length, lots of attributes, short attributes, few escapes
("linescore.xml", LINESCORE),
// short, lots of namespaces, no escapes
("sample_ns.xml", SAMPLE_NS),
// long, few attributes, mix of attribute lengths, escapes in text content
("sample_rss.xml", SAMPLE_RSS),
// long, lots of attributes, short attributes, no text, no escapes
("players.xml", PLAYERS),
];

for (id, data) in inputs.iter() {
// TODO: use fully normalized attribute values
fn parse_document_from_str_with_namespaces(doc: &str) -> XmlResult<()> {
let mut r = NsReader::from_str(doc);
loop {
match criterion::black_box(r.read_resolved_event()?) {
(resolved_ns, Event::Start(e) | Event::Empty(e)) => {
criterion::black_box(resolved_ns);
for attr in e.attributes() {
criterion::black_box(attr?.decode_and_unescape_value(&r)?);
}
}
(resolved_ns, Event::Text(e)) => {
criterion::black_box(e.unescape()?);
criterion::black_box(resolved_ns);
}
(resolved_ns, Event::CData(e)) => {
criterion::black_box(e.into_inner());
criterion::black_box(resolved_ns);
}
(_, Event::End(_)) => (),
(_, Event::Eof) => break,
_ => (),
}
}
Ok(())
}

// TODO: use fully normalized attribute values
fn parse_document_from_bytes_with_namespaces(doc: &[u8]) -> XmlResult<()> {
let mut r = NsReader::from_reader(doc);
let mut buf = Vec::new();
loop {
match criterion::black_box(r.read_resolved_event_into(&mut buf)?) {
(resolved_ns, Event::Start(e) | Event::Empty(e)) => {
criterion::black_box(resolved_ns);
for attr in e.attributes() {
criterion::black_box(attr?.decode_and_unescape_value(&r)?);
}
}
(resolved_ns, Event::Text(e)) => {
criterion::black_box(e.unescape()?);
criterion::black_box(resolved_ns);
}
(resolved_ns, Event::CData(e)) => {
criterion::black_box(e.into_inner());
criterion::black_box(resolved_ns);
}
(_, Event::End(_)) => (),
(_, Event::Eof) => break,
_ => (),
}
buf.clear();
}
Ok(())
}

/// Just parse - no decoding overhead
pub fn bench_parse_document_nocopy(c: &mut Criterion) {
let mut group = c.benchmark_group("parse_document_nocopy");

for (id, data) in INPUTS.iter() {
group.throughput(Throughput::Bytes(data.len() as u64));
group.bench_with_input(*id, *data, |b, input| {
b.iter(|| parse_document_from_str(input).unwrap())
});
}

group.finish();
}

/// Decode into a buffer, then parse
Copy link
Collaborator Author

@dralley dralley Aug 1, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm aware that this is a premature description

pub fn bench_decode_and_parse_document(c: &mut Criterion) {
let mut group = c.benchmark_group("decode_and_parse_document");

for (id, data) in INPUTS.iter() {
group.throughput(Throughput::Bytes(data.len() as u64));
group.bench_with_input(*id, *data, |b, input| {
b.iter(|| parse_document_from_bytes(input.as_bytes()).unwrap())
});
}

group.finish();
}

/// Just parse - no decoding overhead - including namespaces
pub fn bench_parse_document_nocopy_with_namespaces(c: &mut Criterion) {
let mut group = c.benchmark_group("parse_document_nocopy_with_namespaces");

for (id, data) in INPUTS.iter() {
group.throughput(Throughput::Bytes(data.len() as u64));
group.bench_with_input(*id, *data, |b, input| {
b.iter(|| parse_document_from_str_with_namespaces(input).unwrap())
});
}

group.finish();
}

/// Decode into a buffer, then parse - including namespaces
pub fn bench_decode_and_parse_document_with_namespaces(c: &mut Criterion) {
let mut group = c.benchmark_group("decode_and_parse_document_with_namespaces");

for (id, data) in INPUTS.iter() {
group.throughput(Throughput::Bytes(data.len() as u64));
group.bench_with_input(*id, *data, |b, input| {
b.iter(|| parse_document(input).unwrap())
b.iter(|| parse_document_from_bytes_with_namespaces(input.as_bytes()).unwrap())
});
}

group.finish();
}

criterion_group!(benches, bench_fully_parse_document,);
criterion_group!(
benches,
bench_parse_document_nocopy,
bench_decode_and_parse_document,
bench_parse_document_nocopy_with_namespaces,
bench_decode_and_parse_document_with_namespaces,
);
criterion_main!(benches);