arcayr/saphyr-serde
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Improve buffer handling in scan_plain_scalar.

Browse source
This commit is contained in:
Ethiraric 2024-07-14 16:49:40 +02:00
parent 7275141203
commit 4a5241e0bb
2 changed files with 43 additions and 24 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
parser/justfile
View file

@ -17,11 +17,13 @@ ethi_bench:
cd ../libfyaml/build && ninja cd ../libfyaml/build && ninja
cargo bench_compare run_bench cargo bench_compare run_bench
ethi_build_dump: ethi_build_cg_dump:
(cargo test 2>&1 >/dev/null || (cargo test && false)) (cargo test 2>&1 >/dev/null || (cargo test && false))
CARGO_PROFILE_RELEASE_DEBUG=true cargo build --release --bin time_parse CARGO_PROFILE_RELEASE_DEBUG=true cargo build --release --bin time_parse
valgrind --tool=callgrind --dump-instr=yes --collect-jumps=yes ./target/release/time_parse ~/Projects/yaml-rust2/bench_yaml/strings_array.yaml valgrind --tool=callgrind --dump-instr=yes --collect-jumps=yes ./target/release/time_parse ~/Projects/yaml-rust2/bench_yaml/small_objects.yaml
ethi_build_dump: ethi_build_cg_dump
cg_file=`\ls -1t callgrind.out.* | head -n1` && callgrind_annotate $cg_file --auto=no --threshold=99.99 > cg/WORK && rm $cg_file
ethi_compare: ethi_build_dump ethi_compare: ethi_build_dump
cg_file=`\ls -1t callgrind.out.* | head -n1` && callgrind_annotate $cg_file --auto=no --threshold=99.99 > cg/WORK && rm $cg_file
callgrind_differ `\ls cg/0*` cg/WORK --show percentagediff,ircount --sort-by=-first-ir -a callgrind_differ `\ls cg/0*` cg/WORK --show percentagediff,ircount --sort-by=-first-ir -a

59
parser/src/scanner.rs
View file

@ -422,6 +422,9 @@ pub struct Scanner<T> {
/// [`Possible`]: ImplicitMappingState::Possible /// [`Possible`]: ImplicitMappingState::Possible
/// [`Inside`]: ImplicitMappingState::Inside /// [`Inside`]: ImplicitMappingState::Inside
implicit_flow_mapping_states: Vec<ImplicitMappingState>, implicit_flow_mapping_states: Vec<ImplicitMappingState>,
buf_leading_break: String,
buf_trailing_breaks: String,
buf_whitespaces: String,
} }
impl<T: Input> Iterator for Scanner<T> { impl<T: Input> Iterator for Scanner<T> {
@ -473,6 +476,10 @@ impl<T: Input> Scanner<T> {
leading_whitespace: true, leading_whitespace: true,
flow_mapping_started: false, flow_mapping_started: false,
implicit_flow_mapping_states: vec![], implicit_flow_mapping_states: vec![],
buf_leading_break: String::new(),
buf_trailing_breaks: String::new(),
buf_whitespaces: String::new(),
} }
} }
@ -564,6 +571,16 @@ impl<T: Input> Scanner<T> {
// If the next characters do not correspond to a line break. // If the next characters do not correspond to a line break.
#[inline] #[inline]
fn read_break(&mut self, s: &mut String) { fn read_break(&mut self, s: &mut String) {
self.skip_break();
s.push('\n');
}
// Read and consume a line break (either `\r`, `\n` or `\r\n`).
//
// # Panics (in debug)
// If the next characters do not correspond to a line break.
#[inline]
fn skip_break(&mut self) {
let c = self.input.peek(); let c = self.input.peek();
let nc = self.input.peek_nth(1); let nc = self.input.peek_nth(1);
debug_assert!(is_break(c)); debug_assert!(is_break(c));
@ -571,8 +588,6 @@ impl<T: Input> Scanner<T> {
self.skip_blank(); self.skip_blank();
} }
self.skip_nl(); self.skip_nl();
s.push('\n');
} }
/// Insert a token at the given position. /// Insert a token at the given position.
@ -2102,9 +2117,9 @@ impl<T: Input> Scanner<T> {
} }
let mut string = String::with_capacity(32); let mut string = String::with_capacity(32);
let mut leading_break = String::with_capacity(32); self.buf_whitespaces.clear();
let mut trailing_breaks = String::with_capacity(32); self.buf_leading_break.clear();
let mut whitespaces = String::with_capacity(32); self.buf_trailing_breaks.clear();
loop { loop {
self.input.lookahead(4); self.input.lookahead(4);
@ -2123,24 +2138,24 @@ impl<T: Input> Scanner<T> {
&& self.input.next_can_be_plain_scalar(self.flow_level > 0) && self.input.next_can_be_plain_scalar(self.flow_level > 0)
{ {
if self.leading_whitespace { if self.leading_whitespace {
if leading_break.is_empty() { if self.buf_leading_break.is_empty() {
string.push_str(&leading_break); string.push_str(&self.buf_leading_break);
string.push_str(&trailing_breaks); string.push_str(&self.buf_trailing_breaks);
trailing_breaks.clear(); self.buf_trailing_breaks.clear();
leading_break.clear(); self.buf_leading_break.clear();
} else { } else {
if trailing_breaks.is_empty() { if self.buf_trailing_breaks.is_empty() {
string.push(' '); string.push(' ');
} else { } else {
string.push_str(&trailing_breaks); string.push_str(&self.buf_trailing_breaks);
trailing_breaks.clear(); self.buf_trailing_breaks.clear();
} }
leading_break.clear(); self.buf_leading_break.clear();
} }
self.leading_whitespace = false; self.leading_whitespace = false;
} else if !whitespaces.is_empty() { } else if !self.buf_whitespaces.is_empty() {
string.push_str(&whitespaces); string.push_str(&self.buf_whitespaces);
whitespaces.clear(); self.buf_whitespaces.clear();
} }
// We can unroll the first iteration of the loop. // We can unroll the first iteration of the loop.
@ -2182,7 +2197,7 @@ impl<T: Input> Scanner<T> {
while self.input.next_is_blank_or_break() { while self.input.next_is_blank_or_break() {
if self.input.next_is_blank() { if self.input.next_is_blank() {
if !self.leading_whitespace { if !self.leading_whitespace {
whitespaces.push(self.input.peek()); self.buf_whitespaces.push(self.input.peek());
self.skip_blank(); self.skip_blank();
} else if (self.mark.col as isize) < indent && self.input.peek() == '\t' { } else if (self.mark.col as isize) < indent && self.input.peek() == '\t' {
// Tabs in an indentation columns are allowed if and only if the line is // Tabs in an indentation columns are allowed if and only if the line is
@ -2200,10 +2215,12 @@ impl<T: Input> Scanner<T> {
} else { } else {
// Check if it is a first line break // Check if it is a first line break
if self.leading_whitespace { if self.leading_whitespace {
self.read_break(&mut trailing_breaks); self.skip_break();
self.buf_trailing_breaks.push('\n');
} else { } else {
whitespaces.clear(); self.buf_whitespaces.clear();
self.read_break(&mut leading_break); self.skip_break();
self.buf_leading_break.push('\n');
self.leading_whitespace = true; self.leading_whitespace = true;
} }
} }