itsscb/json
1
0
Fork 0
mirror of https://github.com/serde-rs/json.git synced 2025-10-04 16:25:42 +00:00
Code Issues Packages Projects Releases Wiki Activity

Use the same UTF-8/WTF-8 impl for surrogates

Browse Source 
This does not affect performance.
This commit is contained in:
Alisa Sireneva 2024-08-12 19:46:31 +03:00
parent 0e90b61b8c
commit 2f28d106e6
1 changed files with 7 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
src/read.rs
View File

@ -898,20 +898,12 @@ fn parse_unicode_escape<'de, R: Read<'de>>(
validate: bool, validate: bool,
scratch: &mut Vec<u8>, scratch: &mut Vec<u8>,
) -> Result<()> { ) -> Result<()> {
fn encode_surrogate(scratch: &mut Vec<u8>, n: u16) {
scratch.extend_from_slice(&[
(n >> 12 & 0b0000_1111) as u8 | 0b1110_0000,
(n >> 6 & 0b0011_1111) as u8 | 0b1000_0000,
(n & 0b0011_1111) as u8 | 0b1000_0000,
]);
}
let c = match tri!(read.decode_hex_escape()) { let c = match tri!(read.decode_hex_escape()) {
n @ 0xDC00..=0xDFFF => { n @ 0xDC00..=0xDFFF => {
return if validate { return if validate {
error(read, ErrorCode::LoneLeadingSurrogateInHexEscape) error(read, ErrorCode::LoneLeadingSurrogateInHexEscape)
} else { } else {
encode_surrogate(scratch, n); push_wtf8_codepoint(n as u32, scratch);
Ok(()) Ok(())
}; };
} }
@ -928,7 +920,7 @@ fn parse_unicode_escape<'de, R: Read<'de>>(
read.discard(); read.discard();
error(read, ErrorCode::UnexpectedEndOfHexEscape) error(read, ErrorCode::UnexpectedEndOfHexEscape)
} else { } else {
encode_surrogate(scratch, n1); push_wtf8_codepoint(n1 as u32, scratch);
Ok(()) Ok(())
}; };
} }
@ -940,7 +932,7 @@ fn parse_unicode_escape<'de, R: Read<'de>>(
read.discard(); read.discard();
error(read, ErrorCode::UnexpectedEndOfHexEscape) error(read, ErrorCode::UnexpectedEndOfHexEscape)
} else { } else {
encode_surrogate(scratch, n1); push_wtf8_codepoint(n1 as u32, scratch);
// The \ prior to this byte started an escape sequence, // The \ prior to this byte started an escape sequence,
// so we need to parse that now. This recursive call // so we need to parse that now. This recursive call
// does not blow the stack on malicious input because // does not blow the stack on malicious input because
@ -966,17 +958,14 @@ fn parse_unicode_escape<'de, R: Read<'de>>(
n => n as u32, n => n as u32,
}; };
// SAFETY: c is always a codepoint. push_wtf8_codepoint(c, scratch);
unsafe {
push_utf8_codepoint(c, scratch);
}
Ok(()) Ok(())
} }
/// Adds a UTF-8 codepoint to the end of the buffer. This is a more efficient /// Adds a WTF-8 codepoint to the end of the buffer. This is a more efficient
/// implementation of String::push. n must be a valid codepoint. /// implementation of String::push. The codepoint may be a surrogate.
#[inline] #[inline]
unsafe fn push_utf8_codepoint(n: u32, scratch: &mut Vec<u8>) { fn push_wtf8_codepoint(n: u32, scratch: &mut Vec<u8>) {
if n < 0x80 { if n < 0x80 {
scratch.push(n as u8); scratch.push(n as u8);
return; return;