1 //! Utilities for validating string and char literals and turning them into
2 //! values they represent.
10 /// Errors and warnings that can occur during string unescaping.
11 #[derive(Debug, PartialEq, Eq)]
12 pub enum EscapeError {
13 /// Expected 1 char, but 0 were found.
15 /// Expected 1 char, but more than 1 were found.
18 /// Escaped '\' character without continuation.
20 /// Invalid escape character (e.g. '\z').
22 /// Raw '\r' encountered.
24 /// Raw '\r' encountered in raw string.
25 BareCarriageReturnInRawString,
26 /// Unescaped character that was expected to be escaped (e.g. raw '\t').
29 /// Numeric character escape is too short (e.g. '\x1').
31 /// Invalid character in numeric escape (e.g. '\xz')
32 InvalidCharInHexEscape,
33 /// Character code in numeric escape is non-ascii (e.g. '\xFF').
36 /// '\u' not followed by '{'.
37 NoBraceInUnicodeEscape,
38 /// Non-hexadecimal value in '\u{..}'.
39 InvalidCharInUnicodeEscape,
42 /// No closing brace in '\u{..}', e.g. '\u{12'.
43 UnclosedUnicodeEscape,
45 LeadingUnderscoreUnicodeEscape,
46 /// More than 6 characters in '\u{..}', e.g. '\u{10FFFF_FF}'
47 OverlongUnicodeEscape,
48 /// Invalid in-bound unicode character code, e.g. '\u{DFFF}'.
49 LoneSurrogateUnicodeEscape,
50 /// Out of bounds unicode character code, e.g. '\u{FFFFFF}'.
51 OutOfRangeUnicodeEscape,
53 /// Unicode escape code in byte literal.
55 /// Non-ascii character in byte literal, byte string literal, or raw byte string literal.
58 /// After a line ending with '\', the next line contains whitespace
59 /// characters that are not skipped.
60 UnskippedWhitespaceWarning,
62 /// After a line ending with '\', multiple lines are skipped.
63 MultipleSkippedLinesWarning,
67 /// Returns true for actual errors, as opposed to warnings.
68 pub fn is_fatal(&self) -> bool {
71 EscapeError::UnskippedWhitespaceWarning | EscapeError::MultipleSkippedLinesWarning
76 /// Takes a contents of a literal (without quotes) and produces a
77 /// sequence of escaped characters or errors.
78 /// Values are returned through invoking of the provided callback.
79 pub fn unescape_literal<F>(src: &str, mode: Mode, callback: &mut F)
81 F: FnMut(Range<usize>, Result<char, EscapeError>),
84 Mode::Char | Mode::Byte => {
85 let mut chars = src.chars();
86 let result = unescape_char_or_byte(&mut chars, mode == Mode::Byte);
87 // The Chars iterator moved forward.
88 callback(0..(src.len() - chars.as_str().len()), result);
90 Mode::Str | Mode::ByteStr => unescape_str_or_byte_str(src, mode == Mode::ByteStr, callback),
91 // NOTE: Raw strings do not perform any explicit character escaping, here we
92 // only translate CRLF to LF and produce errors on bare CR.
93 Mode::RawStr | Mode::RawByteStr => {
94 unescape_raw_str_or_raw_byte_str(src, mode == Mode::RawByteStr, callback)
99 /// Takes a contents of a byte, byte string or raw byte string (without quotes)
100 /// and produces a sequence of bytes or errors.
101 /// Values are returned through invoking of the provided callback.
102 pub fn unescape_byte_literal<F>(src: &str, mode: Mode, callback: &mut F)
104 F: FnMut(Range<usize>, Result<u8, EscapeError>),
106 debug_assert!(mode.is_byte());
107 unescape_literal(src, mode, &mut |range, result| {
108 callback(range, result.map(byte_from_char));
112 /// Takes a contents of a char literal (without quotes), and returns an
113 /// unescaped char or an error
114 pub fn unescape_char(src: &str) -> Result<char, (usize, EscapeError)> {
115 let mut chars = src.chars();
116 unescape_char_or_byte(&mut chars, false).map_err(|err| (src.len() - chars.as_str().len(), err))
119 /// Takes a contents of a byte literal (without quotes), and returns an
120 /// unescaped byte or an error.
121 pub fn unescape_byte(src: &str) -> Result<u8, (usize, EscapeError)> {
122 let mut chars = src.chars();
123 unescape_char_or_byte(&mut chars, true)
125 .map_err(|err| (src.len() - chars.as_str().len(), err))
128 /// What kind of literal do we parse.
129 #[derive(Debug, Clone, Copy, PartialEq)]
140 pub fn in_double_quotes(self) -> bool {
142 Mode::Str | Mode::ByteStr | Mode::RawStr | Mode::RawByteStr => true,
143 Mode::Char | Mode::Byte => false,
147 pub fn is_byte(self) -> bool {
149 Mode::Byte | Mode::ByteStr | Mode::RawByteStr => true,
150 Mode::Char | Mode::Str | Mode::RawStr => false,
155 fn scan_escape(chars: &mut Chars<'_>, is_byte: bool) -> Result<char, EscapeError> {
156 // Previous character was '\\', unescape what follows.
157 let res = match chars.next().ok_or(EscapeError::LoneSlash)? {
167 // Parse hexadecimal character code.
169 let hi = chars.next().ok_or(EscapeError::TooShortHexEscape)?;
170 let hi = hi.to_digit(16).ok_or(EscapeError::InvalidCharInHexEscape)?;
172 let lo = chars.next().ok_or(EscapeError::TooShortHexEscape)?;
173 let lo = lo.to_digit(16).ok_or(EscapeError::InvalidCharInHexEscape)?;
175 let value = hi * 16 + lo;
177 // For a non-byte literal verify that it is within ASCII range.
178 if !is_byte && !is_ascii(value) {
179 return Err(EscapeError::OutOfRangeHexEscape);
181 let value = value as u8;
187 // We've parsed '\u', now we have to parse '{..}'.
189 if chars.next() != Some('{') {
190 return Err(EscapeError::NoBraceInUnicodeEscape);
193 // First character must be a hexadecimal digit.
194 let mut n_digits = 1;
195 let mut value: u32 = match chars.next().ok_or(EscapeError::UnclosedUnicodeEscape)? {
196 '_' => return Err(EscapeError::LeadingUnderscoreUnicodeEscape),
197 '}' => return Err(EscapeError::EmptyUnicodeEscape),
198 c => c.to_digit(16).ok_or(EscapeError::InvalidCharInUnicodeEscape)?,
201 // First character is valid, now parse the rest of the number
202 // and closing brace.
205 None => return Err(EscapeError::UnclosedUnicodeEscape),
206 Some('_') => continue,
209 return Err(EscapeError::OverlongUnicodeEscape);
212 // Incorrect syntax has higher priority for error reporting
213 // than unallowed value for a literal.
215 return Err(EscapeError::UnicodeEscapeInByte);
218 break std::char::from_u32(value).ok_or_else(|| {
219 if value > 0x10FFFF {
220 EscapeError::OutOfRangeUnicodeEscape
222 EscapeError::LoneSurrogateUnicodeEscape
228 c.to_digit(16).ok_or(EscapeError::InvalidCharInUnicodeEscape)?;
231 // Stop updating value since we're sure that it's incorrect already.
234 let digit = digit as u32;
235 value = value * 16 + digit;
240 _ => return Err(EscapeError::InvalidEscape),
246 fn ascii_check(c: char, is_byte: bool) -> Result<char, EscapeError> {
247 if is_byte && !c.is_ascii() {
248 // Byte literal can't be a non-ascii character.
249 Err(EscapeError::NonAsciiCharInByte)
255 fn unescape_char_or_byte(chars: &mut Chars<'_>, is_byte: bool) -> Result<char, EscapeError> {
256 let c = chars.next().ok_or(EscapeError::ZeroChars)?;
258 '\\' => scan_escape(chars, is_byte),
259 '\n' | '\t' | '\'' => Err(EscapeError::EscapeOnlyChar),
260 '\r' => Err(EscapeError::BareCarriageReturn),
261 _ => ascii_check(c, is_byte),
263 if chars.next().is_some() {
264 return Err(EscapeError::MoreThanOneChar);
269 /// Takes a contents of a string literal (without quotes) and produces a
270 /// sequence of escaped characters or errors.
271 fn unescape_str_or_byte_str<F>(src: &str, is_byte: bool, callback: &mut F)
273 F: FnMut(Range<usize>, Result<char, EscapeError>),
275 let mut chars = src.chars();
277 // The `start` and `end` computation here is complicated because
278 // `skip_ascii_whitespace` makes us to skip over chars without counting
279 // them in the range computation.
280 while let Some(c) = chars.next() {
281 let start = src.len() - chars.as_str().len() - c.len_utf8();
282 let result = match c {
284 match chars.clone().next() {
286 // Rust language specification requires us to skip whitespaces
287 // if unescaped '\' character is followed by '\n'.
288 // For details see [Rust language reference]
289 // (https://doc.rust-lang.org/reference/tokens.html#string-literals).
290 skip_ascii_whitespace(&mut chars, start, callback);
293 _ => scan_escape(&mut chars, is_byte),
298 '"' => Err(EscapeError::EscapeOnlyChar),
299 '\r' => Err(EscapeError::BareCarriageReturn),
300 _ => ascii_check(c, is_byte),
302 let end = src.len() - chars.as_str().len();
303 callback(start..end, result);
306 fn skip_ascii_whitespace<F>(chars: &mut Chars<'_>, start: usize, callback: &mut F)
308 F: FnMut(Range<usize>, Result<char, EscapeError>),
310 let tail = chars.as_str();
311 let first_non_space = tail
313 .position(|b| b != b' ' && b != b'\t' && b != b'\n' && b != b'\r')
314 .unwrap_or(tail.len());
315 if tail[1..first_non_space].contains('\n') {
316 // The +1 accounts for the escaping slash.
317 let end = start + first_non_space + 1;
318 callback(start..end, Err(EscapeError::MultipleSkippedLinesWarning));
320 let tail = &tail[first_non_space..];
321 if let Some(c) = tail.chars().nth(0) {
322 // For error reporting, we would like the span to contain the character that was not
323 // skipped. The +1 is necessary to account for the leading \ that started the escape.
324 let end = start + first_non_space + c.len_utf8() + 1;
325 if c.is_whitespace() {
326 callback(start..end, Err(EscapeError::UnskippedWhitespaceWarning));
329 *chars = tail.chars();
333 /// Takes a contents of a string literal (without quotes) and produces a
334 /// sequence of characters or errors.
335 /// NOTE: Raw strings do not perform any explicit character escaping, here we
336 /// only translate CRLF to LF and produce errors on bare CR.
337 fn unescape_raw_str_or_raw_byte_str<F>(src: &str, is_byte: bool, callback: &mut F)
339 F: FnMut(Range<usize>, Result<char, EscapeError>),
341 let mut chars = src.chars();
343 // The `start` and `end` computation here matches the one in
344 // `unescape_str_or_byte_str` for consistency, even though this function
345 // doesn't have to worry about skipping any chars.
346 while let Some(c) = chars.next() {
347 let start = src.len() - chars.as_str().len() - c.len_utf8();
348 let result = match c {
349 '\r' => Err(EscapeError::BareCarriageReturnInRawString),
350 _ => ascii_check(c, is_byte),
352 let end = src.len() - chars.as_str().len();
353 callback(start..end, result);
357 fn byte_from_char(c: char) -> u8 {
359 debug_assert!(res <= u8::MAX as u32, "guaranteed because of Mode::ByteStr");
363 fn is_ascii(x: u32) -> bool {