robin/php-literal-parser
1
0
Fork 0
mirror of https://codeberg.org/icewind/php-literal-parser.git synced 2026-06-03 10:34:08 +02:00
Code Issues Projects Releases Wiki Activity

also optimize single quoted strings

Browse source
This commit is contained in:
Robin Appelman 2020-12-08 21:36:08 +01:00
commit e9e13df444
2 changed files with 54 additions and 24 deletions
Download patch file Download diff file Expand all files Collapse all files

22
benches/parse.rs
View file

@ -15,7 +15,7 @@ fn perf_parse_int_basic(b: &mut Bencher) {
}
#[bench]
fn perf_str_basic(b: &mut Bencher) {
fn perf_str_double_basic(b: &mut Bencher) {
let input = r#""aut dolores excepturi rerum est velit ad natus eveniet quo tenetur et fugiat sit velit ipsam nesciunt sint et architecto""#;
b.iter(|| {
@ -24,10 +24,28 @@ fn perf_str_basic(b: &mut Bencher) {
}
#[bench]
fn perf_str_escape(b: &mut Bencher) {
fn perf_str_double_escape(b: &mut Bencher) {
let input = r#""aut dolores excepturi rerum est velit ad natus \"eveniet\" quo tenetur et fugiat sit velit ipsam nesciunt sint et architecto""#;
b.iter(|| {
assert!(parse(input).unwrap().is_string());
});
}
#[bench]
fn perf_str_single_basic(b: &mut Bencher) {
let input = r#"'aut dolores excepturi rerum est velit ad natus eveniet quo tenetur et fugiat sit velit ipsam nesciunt sint et architecto'"#;
b.iter(|| {
assert!(parse(input).unwrap().is_string());
});
}
#[bench]
fn perf_str_single_escape(b: &mut Bencher) {
let input = r#"'aut dolores excepturi rerum est velit ad natus \"eveniet\" quo tenetur et fugiat sit velit ipsam nesciunt sint et architecto'"#;
b.iter(|| {
assert!(parse(input).unwrap().is_string());
});
}

42
src/string.rs
View file

@ -71,34 +71,46 @@ fn parse_u32(
Ok(result)
}
/// Un-escape a string, following php single quote rules
pub fn unescape_single(s: &str) -> UnescapeResult<String> {
let mut state = UnescapeState::new();
let mut ins = s.chars();
while let Some(c) = ins.next() {
if c == '\\' {
fn handle_single_escape<'a>(
bytes: &'a [u8],
state: &mut UnescapeState,
) -> UnescapeResult<&'a [u8]> {
let mut ins = PeekableBytes::new(bytes);
debug_assert_eq!(ins.next(), Some(b'\\'));
match ins.next() {
None => {
return Err(UnescapeError);
}
Some(d) => match d {
'\\' | '\'' => state.push_char(d),
b'\\' | b'\'' => state.push_u8(d),
_ => {
state.push_char('\\');
state.push_char(d)
state.push_u8(b'\\');
state.push_u8(d)
}
},
}
} else {
state.push_char(c);
}
Ok(ins.as_slice())
}
/// Un-escape a string, following php single quote rules
pub fn unescape_single(s: &str) -> UnescapeResult<String> {
let mut state = UnescapeState::with_capacity(s.len());
let mut bytes = s.as_bytes();
while let Some(escape_index) = memchr::memchr(b'\\', bytes) {
state.push_slice(&bytes[0..escape_index]);
bytes = &bytes[escape_index..];
bytes = handle_single_escape(bytes, &mut state)?;
}
state.push_slice(&bytes[0..]);
Ok(state.finalize())
}
fn handle_escape<'a>(bytes: &'a [u8], state: &mut UnescapeState) -> UnescapeResult<&'a [u8]> {
fn handle_double_escape<'a>(
bytes: &'a [u8],
state: &mut UnescapeState,
) -> UnescapeResult<&'a [u8]> {
let mut ins = PeekableBytes::new(bytes);
debug_assert_eq!(ins.next(), Some(b'\\'));
match ins.next() {
@ -156,7 +168,7 @@ pub fn unescape_double(s: &str) -> UnescapeResult<String> {
while let Some(escape_index) = memchr::memchr(b'\\', bytes) {
state.push_slice(&bytes[0..escape_index]);
bytes = &bytes[escape_index..];
bytes = handle_escape(bytes, &mut state)?;
bytes = handle_double_escape(bytes, &mut state)?;
}
state.push_slice(&bytes[0..]);