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.
57 /// Non-ascii character in byte string literal.
58 NonAsciiCharInByteString,
60 /// After a line ending with '\', the next line contains whitespace
61 /// characters that are not skipped.
62 UnskippedWhitespaceWarning,
64 /// After a line ending with '\', multiple lines are skipped.
65 MultipleSkippedLinesWarning,
69 /// Returns true for actual errors, as opposed to warnings.
70 pub fn is_fatal(&self) -> bool {
73 EscapeError::UnskippedWhitespaceWarning | EscapeError::MultipleSkippedLinesWarning
78 /// Takes a contents of a literal (without quotes) and produces a
79 /// sequence of escaped characters or errors.
80 /// Values are returned through invoking of the provided callback.
81 pub fn unescape_literal<F>(literal_text: &str, mode: Mode, callback: &mut F)
83 F: FnMut(Range<usize>, Result<char, EscapeError>),
86 Mode::Char | Mode::Byte => {
87 let mut chars = literal_text.chars();
88 let result = unescape_char_or_byte(&mut chars, mode);
89 // The Chars iterator moved forward.
90 callback(0..(literal_text.len() - chars.as_str().len()), result);
92 Mode::Str | Mode::ByteStr => unescape_str_or_byte_str(literal_text, mode, callback),
93 // NOTE: Raw strings do not perform any explicit character escaping, here we
94 // only translate CRLF to LF and produce errors on bare CR.
95 Mode::RawStr | Mode::RawByteStr => {
96 unescape_raw_str_or_raw_byte_str(literal_text, mode, callback)
101 /// Takes a contents of a byte, byte string or raw byte string (without quotes)
102 /// and produces a sequence of bytes or errors.
103 /// Values are returned through invoking of the provided callback.
104 pub fn unescape_byte_literal<F>(literal_text: &str, mode: Mode, callback: &mut F)
106 F: FnMut(Range<usize>, Result<u8, EscapeError>),
108 debug_assert!(mode.is_bytes());
109 unescape_literal(literal_text, mode, &mut |range, result| {
110 callback(range, result.map(byte_from_char));
114 /// Takes a contents of a char literal (without quotes), and returns an
115 /// unescaped char or an error
116 pub fn unescape_char(literal_text: &str) -> Result<char, (usize, EscapeError)> {
117 let mut chars = literal_text.chars();
118 unescape_char_or_byte(&mut chars, Mode::Char)
119 .map_err(|err| (literal_text.len() - chars.as_str().len(), err))
122 /// Takes a contents of a byte literal (without quotes), and returns an
123 /// unescaped byte or an error.
124 pub fn unescape_byte(literal_text: &str) -> Result<u8, (usize, EscapeError)> {
125 let mut chars = literal_text.chars();
126 unescape_char_or_byte(&mut chars, Mode::Byte)
128 .map_err(|err| (literal_text.len() - chars.as_str().len(), err))
131 /// What kind of literal do we parse.
132 #[derive(Debug, Clone, Copy, PartialEq)]
143 pub fn in_double_quotes(self) -> bool {
145 Mode::Str | Mode::ByteStr | Mode::RawStr | Mode::RawByteStr => true,
146 Mode::Char | Mode::Byte => false,
150 pub fn is_bytes(self) -> bool {
152 Mode::Byte | Mode::ByteStr | Mode::RawByteStr => true,
153 Mode::Char | Mode::Str | Mode::RawStr => false,
158 fn scan_escape(chars: &mut Chars<'_>, mode: Mode) -> Result<char, EscapeError> {
159 // Previous character was '\\', unescape what follows.
161 let second_char = chars.next().ok_or(EscapeError::LoneSlash)?;
163 let res = match second_char {
173 // Parse hexadecimal character code.
175 let hi = chars.next().ok_or(EscapeError::TooShortHexEscape)?;
176 let hi = hi.to_digit(16).ok_or(EscapeError::InvalidCharInHexEscape)?;
178 let lo = chars.next().ok_or(EscapeError::TooShortHexEscape)?;
179 let lo = lo.to_digit(16).ok_or(EscapeError::InvalidCharInHexEscape)?;
181 let value = hi * 16 + lo;
183 // For a non-byte literal verify that it is within ASCII range.
184 if !mode.is_bytes() && !is_ascii(value) {
185 return Err(EscapeError::OutOfRangeHexEscape);
187 let value = value as u8;
193 // We've parsed '\u', now we have to parse '{..}'.
195 if chars.next() != Some('{') {
196 return Err(EscapeError::NoBraceInUnicodeEscape);
199 // First character must be a hexadecimal digit.
200 let mut n_digits = 1;
201 let mut value: u32 = match chars.next().ok_or(EscapeError::UnclosedUnicodeEscape)? {
202 '_' => return Err(EscapeError::LeadingUnderscoreUnicodeEscape),
203 '}' => return Err(EscapeError::EmptyUnicodeEscape),
204 c => c.to_digit(16).ok_or(EscapeError::InvalidCharInUnicodeEscape)?,
207 // First character is valid, now parse the rest of the number
208 // and closing brace.
211 None => return Err(EscapeError::UnclosedUnicodeEscape),
212 Some('_') => continue,
215 return Err(EscapeError::OverlongUnicodeEscape);
218 // Incorrect syntax has higher priority for error reporting
219 // than unallowed value for a literal.
221 return Err(EscapeError::UnicodeEscapeInByte);
224 break std::char::from_u32(value).ok_or_else(|| {
225 if value > 0x10FFFF {
226 EscapeError::OutOfRangeUnicodeEscape
228 EscapeError::LoneSurrogateUnicodeEscape
234 c.to_digit(16).ok_or(EscapeError::InvalidCharInUnicodeEscape)?;
237 // Stop updating value since we're sure that it's incorrect already.
240 let digit = digit as u32;
241 value = value * 16 + digit;
246 _ => return Err(EscapeError::InvalidEscape),
252 fn ascii_check(first_char: char, mode: Mode) -> Result<char, EscapeError> {
253 if mode.is_bytes() && !first_char.is_ascii() {
254 // Byte literal can't be a non-ascii character.
255 Err(EscapeError::NonAsciiCharInByte)
261 fn unescape_char_or_byte(chars: &mut Chars<'_>, mode: Mode) -> Result<char, EscapeError> {
262 debug_assert!(mode == Mode::Char || mode == Mode::Byte);
263 let first_char = chars.next().ok_or(EscapeError::ZeroChars)?;
264 let res = match first_char {
265 '\\' => scan_escape(chars, mode),
266 '\n' | '\t' | '\'' => Err(EscapeError::EscapeOnlyChar),
267 '\r' => Err(EscapeError::BareCarriageReturn),
268 _ => ascii_check(first_char, mode),
270 if chars.next().is_some() {
271 return Err(EscapeError::MoreThanOneChar);
276 /// Takes a contents of a string literal (without quotes) and produces a
277 /// sequence of escaped characters or errors.
278 fn unescape_str_or_byte_str<F>(src: &str, mode: Mode, callback: &mut F)
280 F: FnMut(Range<usize>, Result<char, EscapeError>),
282 debug_assert!(mode == Mode::Str || mode == Mode::ByteStr);
283 let initial_len = src.len();
284 let mut chars = src.chars();
285 while let Some(first_char) = chars.next() {
286 let start = initial_len - chars.as_str().len() - first_char.len_utf8();
288 let unescaped_char = match first_char {
290 let second_char = chars.clone().next();
293 // Rust language specification requires us to skip whitespaces
294 // if unescaped '\' character is followed by '\n'.
295 // For details see [Rust language reference]
296 // (https://doc.rust-lang.org/reference/tokens.html#string-literals).
297 skip_ascii_whitespace(&mut chars, start, callback);
300 _ => scan_escape(&mut chars, mode),
305 '"' => Err(EscapeError::EscapeOnlyChar),
306 '\r' => Err(EscapeError::BareCarriageReturn),
307 _ => ascii_check(first_char, mode),
309 let end = initial_len - chars.as_str().len();
310 callback(start..end, unescaped_char);
313 fn skip_ascii_whitespace<F>(chars: &mut Chars<'_>, start: usize, callback: &mut F)
315 F: FnMut(Range<usize>, Result<char, EscapeError>),
317 let tail = chars.as_str();
318 let first_non_space = tail
320 .position(|b| b != b' ' && b != b'\t' && b != b'\n' && b != b'\r')
321 .unwrap_or(tail.len());
322 if tail[1..first_non_space].contains('\n') {
323 // The +1 accounts for the escaping slash.
324 let end = start + first_non_space + 1;
325 callback(start..end, Err(EscapeError::MultipleSkippedLinesWarning));
327 let tail = &tail[first_non_space..];
328 if let Some(c) = tail.chars().nth(0) {
329 // For error reporting, we would like the span to contain the character that was not
330 // skipped. The +1 is necessary to account for the leading \ that started the escape.
331 let end = start + first_non_space + c.len_utf8() + 1;
332 if c.is_whitespace() {
333 callback(start..end, Err(EscapeError::UnskippedWhitespaceWarning));
336 *chars = tail.chars();
340 /// Takes a contents of a string literal (without quotes) and produces a
341 /// sequence of characters or errors.
342 /// NOTE: Raw strings do not perform any explicit character escaping, here we
343 /// only translate CRLF to LF and produce errors on bare CR.
344 fn unescape_raw_str_or_raw_byte_str<F>(literal_text: &str, mode: Mode, callback: &mut F)
346 F: FnMut(Range<usize>, Result<char, EscapeError>),
348 debug_assert!(mode == Mode::RawStr || mode == Mode::RawByteStr);
349 let initial_len = literal_text.len();
351 let mut chars = literal_text.chars();
352 while let Some(curr) = chars.next() {
353 let start = initial_len - chars.as_str().len() - curr.len_utf8();
355 let result = match curr {
356 '\r' => Err(EscapeError::BareCarriageReturnInRawString),
357 c if mode.is_bytes() && !c.is_ascii() => Err(EscapeError::NonAsciiCharInByteString),
360 let end = initial_len - chars.as_str().len();
362 callback(start..end, result);
366 fn byte_from_char(c: char) -> u8 {
368 debug_assert!(res <= u8::MAX as u32, "guaranteed because of Mode::ByteStr");
372 fn is_ascii(x: u32) -> bool {