From 4547b622d8d29df964fa2914213088b148c498fc Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Wed, 17 Apr 2024 14:18:32 +0200 Subject: Merging upstream version 1.67.1+dfsg1. Signed-off-by: Daniel Baumann --- compiler/rustc_lexer/src/unescape.rs | 130 ++++++++++++++--------------------- 1 file changed, 51 insertions(+), 79 deletions(-) (limited to 'compiler/rustc_lexer/src/unescape.rs') diff --git a/compiler/rustc_lexer/src/unescape.rs b/compiler/rustc_lexer/src/unescape.rs index 8f64b5f51..e405013dc 100644 --- a/compiler/rustc_lexer/src/unescape.rs +++ b/compiler/rustc_lexer/src/unescape.rs @@ -52,10 +52,8 @@ pub enum EscapeError { /// Unicode escape code in byte literal. UnicodeEscapeInByte, - /// Non-ascii character in byte literal. + /// Non-ascii character in byte literal, byte string literal, or raw byte string literal. NonAsciiCharInByte, - /// Non-ascii character in byte string literal. - NonAsciiCharInByteString, /// After a line ending with '\', the next line contains whitespace /// characters that are not skipped. @@ -78,54 +76,33 @@ impl EscapeError { /// Takes a contents of a literal (without quotes) and produces a /// sequence of escaped characters or errors. /// Values are returned through invoking of the provided callback. -pub fn unescape_literal(literal_text: &str, mode: Mode, callback: &mut F) +pub fn unescape_literal(src: &str, mode: Mode, callback: &mut F) where F: FnMut(Range, Result), { match mode { Mode::Char | Mode::Byte => { - let mut chars = literal_text.chars(); - let result = unescape_char_or_byte(&mut chars, mode); - // The Chars iterator moved forward. - callback(0..(literal_text.len() - chars.as_str().len()), result); + let mut chars = src.chars(); + let res = unescape_char_or_byte(&mut chars, mode == Mode::Byte); + callback(0..(src.len() - chars.as_str().len()), res); } - Mode::Str | Mode::ByteStr => unescape_str_or_byte_str(literal_text, mode, callback), - // NOTE: Raw strings do not perform any explicit character escaping, here we - // only translate CRLF to LF and produce errors on bare CR. + Mode::Str | Mode::ByteStr => unescape_str_or_byte_str(src, mode == Mode::ByteStr, callback), Mode::RawStr | Mode::RawByteStr => { - unescape_raw_str_or_raw_byte_str(literal_text, mode, callback) + unescape_raw_str_or_raw_byte_str(src, mode == Mode::RawByteStr, callback) } } } -/// Takes a contents of a byte, byte string or raw byte string (without quotes) -/// and produces a sequence of bytes or errors. -/// Values are returned through invoking of the provided callback. -pub fn unescape_byte_literal(literal_text: &str, mode: Mode, callback: &mut F) -where - F: FnMut(Range, Result), -{ - debug_assert!(mode.is_bytes()); - unescape_literal(literal_text, mode, &mut |range, result| { - callback(range, result.map(byte_from_char)); - }) -} - /// Takes a contents of a char literal (without quotes), and returns an -/// unescaped char or an error -pub fn unescape_char(literal_text: &str) -> Result { - let mut chars = literal_text.chars(); - unescape_char_or_byte(&mut chars, Mode::Char) - .map_err(|err| (literal_text.len() - chars.as_str().len(), err)) +/// unescaped char or an error. +pub fn unescape_char(src: &str) -> Result { + unescape_char_or_byte(&mut src.chars(), false) } /// Takes a contents of a byte literal (without quotes), and returns an /// unescaped byte or an error. -pub fn unescape_byte(literal_text: &str) -> Result { - let mut chars = literal_text.chars(); - unescape_char_or_byte(&mut chars, Mode::Byte) - .map(byte_from_char) - .map_err(|err| (literal_text.len() - chars.as_str().len(), err)) +pub fn unescape_byte(src: &str) -> Result { + unescape_char_or_byte(&mut src.chars(), true).map(byte_from_char) } /// What kind of literal do we parse. @@ -147,7 +124,7 @@ impl Mode { } } - pub fn is_bytes(self) -> bool { + pub fn is_byte(self) -> bool { match self { Mode::Byte | Mode::ByteStr | Mode::RawByteStr => true, Mode::Char | Mode::Str | Mode::RawStr => false, @@ -155,12 +132,9 @@ impl Mode { } } -fn scan_escape(chars: &mut Chars<'_>, mode: Mode) -> Result { +fn scan_escape(chars: &mut Chars<'_>, is_byte: bool) -> Result { // Previous character was '\\', unescape what follows. - - let second_char = chars.next().ok_or(EscapeError::LoneSlash)?; - - let res = match second_char { + let res = match chars.next().ok_or(EscapeError::LoneSlash)? { '"' => '"', 'n' => '\n', 'r' => '\r', @@ -181,7 +155,7 @@ fn scan_escape(chars: &mut Chars<'_>, mode: Mode) -> Result { let value = hi * 16 + lo; // For a non-byte literal verify that it is within ASCII range. - if !mode.is_bytes() && !is_ascii(value) { + if !is_byte && !is_ascii(value) { return Err(EscapeError::OutOfRangeHexEscape); } let value = value as u8; @@ -217,7 +191,7 @@ fn scan_escape(chars: &mut Chars<'_>, mode: Mode) -> Result { // Incorrect syntax has higher priority for error reporting // than unallowed value for a literal. - if mode.is_bytes() { + if is_byte { return Err(EscapeError::UnicodeEscapeInByte); } @@ -249,23 +223,22 @@ fn scan_escape(chars: &mut Chars<'_>, mode: Mode) -> Result { } #[inline] -fn ascii_check(first_char: char, mode: Mode) -> Result { - if mode.is_bytes() && !first_char.is_ascii() { +fn ascii_check(c: char, is_byte: bool) -> Result { + if is_byte && !c.is_ascii() { // Byte literal can't be a non-ascii character. Err(EscapeError::NonAsciiCharInByte) } else { - Ok(first_char) + Ok(c) } } -fn unescape_char_or_byte(chars: &mut Chars<'_>, mode: Mode) -> Result { - debug_assert!(mode == Mode::Char || mode == Mode::Byte); - let first_char = chars.next().ok_or(EscapeError::ZeroChars)?; - let res = match first_char { - '\\' => scan_escape(chars, mode), +fn unescape_char_or_byte(chars: &mut Chars<'_>, is_byte: bool) -> Result { + let c = chars.next().ok_or(EscapeError::ZeroChars)?; + let res = match c { + '\\' => scan_escape(chars, is_byte), '\n' | '\t' | '\'' => Err(EscapeError::EscapeOnlyChar), '\r' => Err(EscapeError::BareCarriageReturn), - _ => ascii_check(first_char, mode), + _ => ascii_check(c, is_byte), }?; if chars.next().is_some() { return Err(EscapeError::MoreThanOneChar); @@ -275,20 +248,20 @@ fn unescape_char_or_byte(chars: &mut Chars<'_>, mode: Mode) -> Result(src: &str, mode: Mode, callback: &mut F) +fn unescape_str_or_byte_str(src: &str, is_byte: bool, callback: &mut F) where F: FnMut(Range, Result), { - debug_assert!(mode == Mode::Str || mode == Mode::ByteStr); - let initial_len = src.len(); let mut chars = src.chars(); - while let Some(first_char) = chars.next() { - let start = initial_len - chars.as_str().len() - first_char.len_utf8(); - let unescaped_char = match first_char { + // The `start` and `end` computation here is complicated because + // `skip_ascii_whitespace` makes us to skip over chars without counting + // them in the range computation. + while let Some(c) = chars.next() { + let start = src.len() - chars.as_str().len() - c.len_utf8(); + let res = match c { '\\' => { - let second_char = chars.clone().next(); - match second_char { + match chars.clone().next() { Some('\n') => { // Rust language specification requires us to skip whitespaces // if unescaped '\' character is followed by '\n'. @@ -297,17 +270,17 @@ where skip_ascii_whitespace(&mut chars, start, callback); continue; } - _ => scan_escape(&mut chars, mode), + _ => scan_escape(&mut chars, is_byte), } } '\n' => Ok('\n'), '\t' => Ok('\t'), '"' => Err(EscapeError::EscapeOnlyChar), '\r' => Err(EscapeError::BareCarriageReturn), - _ => ascii_check(first_char, mode), + _ => ascii_check(c, is_byte), }; - let end = initial_len - chars.as_str().len(); - callback(start..end, unescaped_char); + let end = src.len() - chars.as_str().len(); + callback(start..end, res); } fn skip_ascii_whitespace(chars: &mut Chars<'_>, start: usize, callback: &mut F) @@ -340,30 +313,29 @@ where /// Takes a contents of a string literal (without quotes) and produces a /// sequence of characters or errors. /// NOTE: Raw strings do not perform any explicit character escaping, here we -/// only translate CRLF to LF and produce errors on bare CR. -fn unescape_raw_str_or_raw_byte_str(literal_text: &str, mode: Mode, callback: &mut F) +/// only produce errors on bare CR. +fn unescape_raw_str_or_raw_byte_str(src: &str, is_byte: bool, callback: &mut F) where F: FnMut(Range, Result), { - debug_assert!(mode == Mode::RawStr || mode == Mode::RawByteStr); - let initial_len = literal_text.len(); - - let mut chars = literal_text.chars(); - while let Some(curr) = chars.next() { - let start = initial_len - chars.as_str().len() - curr.len_utf8(); + let mut chars = src.chars(); - let result = match curr { + // The `start` and `end` computation here matches the one in + // `unescape_str_or_byte_str` for consistency, even though this function + // doesn't have to worry about skipping any chars. + while let Some(c) = chars.next() { + let start = src.len() - chars.as_str().len() - c.len_utf8(); + let res = match c { '\r' => Err(EscapeError::BareCarriageReturnInRawString), - c if mode.is_bytes() && !c.is_ascii() => Err(EscapeError::NonAsciiCharInByteString), - c => Ok(c), + _ => ascii_check(c, is_byte), }; - let end = initial_len - chars.as_str().len(); - - callback(start..end, result); + let end = src.len() - chars.as_str().len(); + callback(start..end, res); } } -fn byte_from_char(c: char) -> u8 { +#[inline] +pub fn byte_from_char(c: char) -> u8 { let res = c as u32; debug_assert!(res <= u8::MAX as u32, "guaranteed because of Mode::ByteStr"); res as u8 -- cgit v1.2.3