dan/oxc
1
0
Fork 0
mirror of https://github.com/danbulant/oxc synced 2026-05-24 12:21:58 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

refactor(parser): small efficiencies in byte_search macro usage (#2554)

Browse source 
A few small efficiencies in usage of `byte_search` macro for lexing comments.
This commit is contained in:
overlookmotel 2024-03-01 13:23:34 +00:00 committed by GitHub
parent 3d354d44a3
commit c579620701
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 17 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

28
crates/oxc_parser/src/lexer/comment.rs
View file

@ -38,7 +38,7 @@ impl<'a> Lexer<'a> {
self.trivia_builder self.trivia_builder
.add_single_line_comment(self.token.start, self.source.offset_of(pos)); .add_single_line_comment(self.token.start, self.source.offset_of(pos));
// SAFETY: Safe to consume `\r` or `\n` as both are ASCII // SAFETY: Safe to consume `\r` or `\n` as both are ASCII
unsafe { pos = pos.add(1) }; pos = unsafe { pos.add(1) };
// We've found the end. Do not continue searching. // We've found the end. Do not continue searching.
false false
} else { } else {
@ -48,14 +48,14 @@ impl<'a> Lexer<'a> {
// SAFETY: Next byte is `0xE2` which is always 1st byte of a 3-byte UTF-8 char. // SAFETY: Next byte is `0xE2` which is always 1st byte of a 3-byte UTF-8 char.
// So safe to advance `pos` by 1 and read 2 bytes. // So safe to advance `pos` by 1 and read 2 bytes.
let next2 = unsafe { pos.add(1).read2() }; let next2 = unsafe { pos.add(1).read2() };
if next2 == LS_BYTES_2_AND_3 || next2 == PS_BYTES_2_AND_3 { if matches!(next2, LS_BYTES_2_AND_3 | PS_BYTES_2_AND_3) {
// Irregular line break // Irregular line break
self.trivia_builder self.trivia_builder
.add_single_line_comment(self.token.start, self.source.offset_of(pos)); .add_single_line_comment(self.token.start, self.source.offset_of(pos));
// Advance `pos` to after this char. // Advance `pos` to after this char.
// SAFETY: `0xE2` is always 1st byte of a 3-byte UTF-8 char, // SAFETY: `0xE2` is always 1st byte of a 3-byte UTF-8 char,
// so consuming 3 bytes will place `pos` on next UTF-8 char boundary. // so consuming 3 bytes will place `pos` on next UTF-8 char boundary.
unsafe { pos = pos.add(3) }; pos = unsafe { pos.add(3) };
// We've found the end. Do not continue searching. // We've found the end. Do not continue searching.
false false
} else { } else {
@ -63,7 +63,7 @@ impl<'a> Lexer<'a> {
// Skip 3 bytes (macro skips 1 already, so skip 2 here), and continue searching. // Skip 3 bytes (macro skips 1 already, so skip 2 here), and continue searching.
// SAFETY: `0xE2` is always 1st byte of a 3-byte UTF-8 char, // SAFETY: `0xE2` is always 1st byte of a 3-byte UTF-8 char,
// so consuming 3 bytes will place `pos` on next UTF-8 char boundary. // so consuming 3 bytes will place `pos` on next UTF-8 char boundary.
unsafe { pos = pos.add(2) }; pos = unsafe { pos.add(2) };
true true
} }
}) })
@ -93,13 +93,15 @@ impl<'a> Lexer<'a> {
continue_if: |next_byte, pos| { continue_if: |next_byte, pos| {
// Match found. Decide whether to continue searching. // Match found. Decide whether to continue searching.
if next_byte == b'*' { if next_byte == b'*' {
if pos.addr() < self.source.end_addr() - 1 { // SAFETY: Next byte is `*` (ASCII) so after it is UTF-8 char boundary
let after_star = unsafe { pos.add(1) };
if after_star.addr() < self.source.end_addr() {
// If next byte isn't `/`, continue // If next byte isn't `/`, continue
// SAFETY: Have checked there's at least 1 further byte to read // SAFETY: Have checked there's at least 1 further byte to read
if unsafe { pos.add(1).read() } == b'/' { if unsafe { after_star.read() } == b'/' {
// Consume `*/` // Consume `*/`
// SAFETY: Consuming `*/` leaves `pos` on a UTF-8 char boundary // SAFETY: Consuming `*/` leaves `pos` on a UTF-8 char boundary
unsafe { pos = pos.add(2) }; pos = unsafe { pos.add(2) };
false false
} else { } else {
true true
@ -115,8 +117,11 @@ impl<'a> Lexer<'a> {
cold_branch(|| { cold_branch(|| {
// SAFETY: Next byte is `0xE2` which is always 1st byte of a 3-byte UTF-8 char. // SAFETY: Next byte is `0xE2` which is always 1st byte of a 3-byte UTF-8 char.
// So safe to advance `pos` by 1 and read 2 bytes. // So safe to advance `pos` by 1 and read 2 bytes.
let next2 = unsafe { pos.add(1).read2() }; let next2 = unsafe {
if next2 == LS_BYTES_2_AND_3 || next2 == PS_BYTES_2_AND_3 { pos = pos.add(1);
pos.read2()
};
if matches!(next2, LS_BYTES_2_AND_3 | PS_BYTES_2_AND_3) {
// Irregular line break // Irregular line break
self.token.is_on_new_line = true; self.token.is_on_new_line = true;
// Ideally we'd go on to `skip_multi_line_comment_after_line_break` here // Ideally we'd go on to `skip_multi_line_comment_after_line_break` here
@ -124,10 +129,11 @@ impl<'a> Lexer<'a> {
// But irregular line breaks are rare anyway. // But irregular line breaks are rare anyway.
} }
// Either way, continue searching. // Either way, continue searching.
// Skip 3 bytes (macro skips 1 already, so skip 2 here), and continue searching. // Skip 3 bytes (skipped 1 byte above, macro skips 1 more, so skip 1 more here
// to make 3), and continue searching.
// SAFETY: `0xE2` is always 1st byte of a 3-byte UTF-8 char, // SAFETY: `0xE2` is always 1st byte of a 3-byte UTF-8 char,
// so consuming 3 bytes will place `pos` on next UTF-8 char boundary. // so consuming 3 bytes will place `pos` on next UTF-8 char boundary.
unsafe { pos = pos.add(2) }; pos = unsafe { pos.add(1) };
true true
}) })
} else { } else {