compiler/rustc_parse/src/lexer/unescape_error_reporting.rs

   1 //! Utilities for rendering escape sequence errors as diagnostics.
   2
   3 use std::iter::once;
   4 use std::ops::Range;
   5
   6 use rustc_errors::{pluralize, Applicability, Handler};
   7 use rustc_lexer::unescape::{EscapeError, Mode};
   8 use rustc_span::{BytePos, Span};
   9
  10 pub(crate) fn emit_unescape_error(
  11     handler: &Handler,
  12     // interior part of the literal, without quotes
  13     lit: &str,
  14     // full span of the literal, including quotes
  15     span_with_quotes: Span,
  16     // interior span of the literal, without quotes
  17     span: Span,
  18     mode: Mode,
  19     // range of the error inside `lit`
  20     range: Range<usize>,
  21     error: EscapeError,
  22 ) {
  23     debug!(
  24         "emit_unescape_error: {:?}, {:?}, {:?}, {:?}, {:?}",
  25         lit, span_with_quotes, mode, range, error
  26     );
  27     let last_char = || {
  28         let c = lit[range.clone()].chars().rev().next().unwrap();
  29         let span = span.with_lo(span.hi() - BytePos(c.len_utf8() as u32));
  30         (c, span)
  31     };
  32     match error {
  33         EscapeError::LoneSurrogateUnicodeEscape => {
  34             handler
  35                 .struct_span_err(span, "invalid unicode character escape")
  36                 .span_label(span, "invalid escape")
  37                 .help("unicode escape must not be a surrogate")
  38                 .emit();
  39         }
  40         EscapeError::OutOfRangeUnicodeEscape => {
  41             handler
  42                 .struct_span_err(span, "invalid unicode character escape")
  43                 .span_label(span, "invalid escape")
  44                 .help("unicode escape must be at most 10FFFF")
  45                 .emit();
  46         }
  47         EscapeError::MoreThanOneChar => {
  48             use unicode_normalization::{char::is_combining_mark, UnicodeNormalization};
  49
  50             let mut has_help = false;
  51             let mut handler = handler.struct_span_err(
  52                 span_with_quotes,
  53                 "character literal may only contain one codepoint",
  54             );
  55
  56             if lit.chars().skip(1).all(|c| is_combining_mark(c)) {
  57                 let escaped_marks =
  58                     lit.chars().skip(1).map(|c| c.escape_default().to_string()).collect::<Vec<_>>();
  59                 handler.span_note(
  60                     span,
  61                     &format!(
  62                         "this `{}` is followed by the combining mark{} `{}`",
  63                         lit.chars().next().unwrap(),
  64                         pluralize!(escaped_marks.len()),
  65                         escaped_marks.join(""),
  66                     ),
  67                 );
  68                 let normalized = lit.nfc().to_string();
  69                 if normalized.chars().count() == 1 {
  70                     has_help = true;
  71                     handler.span_suggestion(
  72                         span,
  73                         &format!(
  74                             "consider using the normalized form `{}` of this character",
  75                             normalized.chars().next().unwrap().escape_default()
  76                         ),
  77                         normalized,
  78                         Applicability::MachineApplicable,
  79                     );
  80                 }
  81             } else {
  82                 let printable: Vec<char> = lit
  83                     .chars()
  84                     .filter(|&x| {
  85                         unicode_width::UnicodeWidthChar::width(x).unwrap_or(0) != 0
  86                             && !x.is_whitespace()
  87                     })
  88                     .collect();
  89
  90                 if let [ch] = printable.as_slice() {
  91                     has_help = true;
  92
  93                     handler.span_note(
  94                         span,
  95                         &format!(
  96                             "there are non-printing characters, the full sequence is `{}`",
  97                             lit.escape_default(),
  98                         ),
  99                     );
 100
 101                     handler.span_suggestion(
 102                         span,
 103                         "consider removing the non-printing characters",
 104                         ch,
 105                         Applicability::MaybeIncorrect,
 106                     );
 107                 }
 108             }
 109
 110             if !has_help {
 111                 let (prefix, msg) = if mode.is_bytes() {
 112                     ("b", "if you meant to write a byte string literal, use double quotes")
 113                 } else {
 114                     ("", "if you meant to write a `str` literal, use double quotes")
 115                 };
 116
 117                 handler.span_suggestion(
 118                     span_with_quotes,
 119                     msg,
 120                     format!("{}\"{}\"", prefix, lit),
 121                     Applicability::MachineApplicable,
 122                 );
 123             }
 124
 125             handler.emit();
 126         }
 127         EscapeError::EscapeOnlyChar => {
 128             let (c, char_span) = last_char();
 129
 130             let msg = if mode.is_bytes() {
 131                 "byte constant must be escaped"
 132             } else {
 133                 "character constant must be escaped"
 134             };
 135             handler
 136                 .struct_span_err(span, &format!("{}: `{}`", msg, escaped_char(c)))
 137                 .span_suggestion(
 138                     char_span,
 139                     "escape the character",
 140                     c.escape_default(),
 141                     Applicability::MachineApplicable,
 142                 )
 143                 .emit();
 144         }
 145         EscapeError::BareCarriageReturn => {
 146             let msg = if mode.in_double_quotes() {
 147                 "bare CR not allowed in string, use `\\r` instead"
 148             } else {
 149                 "character constant must be escaped: `\\r`"
 150             };
 151             handler
 152                 .struct_span_err(span, msg)
 153                 .span_suggestion(
 154                     span,
 155                     "escape the character",
 156                     "\\r",
 157                     Applicability::MachineApplicable,
 158                 )
 159                 .emit();
 160         }
 161         EscapeError::BareCarriageReturnInRawString => {
 162             assert!(mode.in_double_quotes());
 163             let msg = "bare CR not allowed in raw string";
 164             handler.span_err(span, msg);
 165         }
 166         EscapeError::InvalidEscape => {
 167             let (c, span) = last_char();
 168
 169             let label =
 170                 if mode.is_bytes() { "unknown byte escape" } else { "unknown character escape" };
 171             let ec = escaped_char(c);
 172             let mut diag = handler.struct_span_err(span, &format!("{}: `{}`", label, ec));
 173             diag.span_label(span, label);
 174             if c == '{' || c == '}' && !mode.is_bytes() {
 175                 diag.help(
 176                     "if used in a formatting string, curly braces are escaped with `{{` and `}}`",
 177                 );
 178             } else if c == '\r' {
 179                 diag.help(
 180                     "this is an isolated carriage return; consider checking your editor and \
 181                      version control settings",
 182                 );
 183             } else {
 184                 if !mode.is_bytes() {
 185                     diag.span_suggestion(
 186                         span_with_quotes,
 187                         "if you meant to write a literal backslash (perhaps escaping in a regular expression), consider a raw string literal",
 188                         format!("r\"{}\"", lit),
 189                         Applicability::MaybeIncorrect,
 190                     );
 191                 }
 192
 193                 diag.help(
 194                     "for more information, visit \
 195                      <https://static.rust-lang.org/doc/master/reference.html#literals>",
 196                 );
 197             }
 198             diag.emit();
 199         }
 200         EscapeError::TooShortHexEscape => {
 201             handler.span_err(span, "numeric character escape is too short");
 202         }
 203         EscapeError::InvalidCharInHexEscape | EscapeError::InvalidCharInUnicodeEscape => {
 204             let (c, span) = last_char();
 205
 206             let msg = if error == EscapeError::InvalidCharInHexEscape {
 207                 "invalid character in numeric character escape"
 208             } else {
 209                 "invalid character in unicode escape"
 210             };
 211             let c = escaped_char(c);
 212
 213             handler
 214                 .struct_span_err(span, &format!("{}: `{}`", msg, c))
 215                 .span_label(span, msg)
 216                 .emit();
 217         }
 218         EscapeError::NonAsciiCharInByte => {
 219             assert!(mode.is_bytes());
 220             let (c, span) = last_char();
 221             let mut err = handler.struct_span_err(span, "non-ASCII character in byte constant");
 222             let postfix = if unicode_width::UnicodeWidthChar::width(c).unwrap_or(1) == 0 {
 223                 format!(" but is {:?}", c)
 224             } else {
 225                 String::new()
 226             };
 227             err.span_label(span, &format!("byte constant must be ASCII{}", postfix));
 228             if (c as u32) <= 0xFF {
 229                 err.span_suggestion(
 230                     span,
 231                     &format!(
 232                         "if you meant to use the unicode code point for {:?}, use a \\xHH escape",
 233                         c
 234                     ),
 235                     format!("\\x{:X}", c as u32),
 236                     Applicability::MaybeIncorrect,
 237                 );
 238             } else if matches!(mode, Mode::Byte) {
 239                 err.span_label(span, "this multibyte character does not fit into a single byte");
 240             } else if matches!(mode, Mode::ByteStr) {
 241                 let mut utf8 = String::new();
 242                 utf8.push(c);
 243                 err.span_suggestion(
 244                     span,
 245                     &format!(
 246                         "if you meant to use the UTF-8 encoding of {:?}, use \\xHH escapes",
 247                         c
 248                     ),
 249                     utf8.as_bytes()
 250                         .iter()
 251                         .map(|b: &u8| format!("\\x{:X}", *b))
 252                         .fold("".to_string(), |a, c| a + &c),
 253                     Applicability::MaybeIncorrect,
 254                 );
 255             }
 256             err.emit();
 257         }
 258         EscapeError::NonAsciiCharInByteString => {
 259             assert!(mode.is_bytes());
 260             let (c, span) = last_char();
 261             let postfix = if unicode_width::UnicodeWidthChar::width(c).unwrap_or(1) == 0 {
 262                 format!(" but is {:?}", c)
 263             } else {
 264                 String::new()
 265             };
 266             handler
 267                 .struct_span_err(span, "raw byte string must be ASCII")
 268                 .span_label(span, &format!("must be ASCII{}", postfix))
 269                 .emit();
 270         }
 271         EscapeError::OutOfRangeHexEscape => {
 272             handler
 273                 .struct_span_err(span, "out of range hex escape")
 274                 .span_label(span, "must be a character in the range [\\x00-\\x7f]")
 275                 .emit();
 276         }
 277         EscapeError::LeadingUnderscoreUnicodeEscape => {
 278             let (c, span) = last_char();
 279             let msg = "invalid start of unicode escape";
 280             handler
 281                 .struct_span_err(span, &format!("{}: `{}`", msg, c))
 282                 .span_label(span, msg)
 283                 .emit();
 284         }
 285         EscapeError::OverlongUnicodeEscape => {
 286             handler
 287                 .struct_span_err(span, "overlong unicode escape")
 288                 .span_label(span, "must have at most 6 hex digits")
 289                 .emit();
 290         }
 291         EscapeError::UnclosedUnicodeEscape => {
 292             handler
 293                 .struct_span_err(span, "unterminated unicode escape")
 294                 .span_label(span, "missing a closing `}`")
 295                 .span_suggestion_verbose(
 296                     span.shrink_to_hi(),
 297                     "terminate the unicode escape",
 298                     "}",
 299                     Applicability::MaybeIncorrect,
 300                 )
 301                 .emit();
 302         }
 303         EscapeError::NoBraceInUnicodeEscape => {
 304             let msg = "incorrect unicode escape sequence";
 305             let mut diag = handler.struct_span_err(span, msg);
 306
 307             let mut suggestion = "\\u{".to_owned();
 308             let mut suggestion_len = 0;
 309             let (c, char_span) = last_char();
 310             let chars = once(c).chain(lit[range.end..].chars());
 311             for c in chars.take(6).take_while(|c| c.is_digit(16)) {
 312                 suggestion.push(c);
 313                 suggestion_len += c.len_utf8();
 314             }
 315
 316             if suggestion_len > 0 {
 317                 suggestion.push('}');
 318                 let hi = char_span.lo() + BytePos(suggestion_len as u32);
 319                 diag.span_suggestion(
 320                     span.with_hi(hi),
 321                     "format of unicode escape sequences uses braces",
 322                     suggestion,
 323                     Applicability::MaybeIncorrect,
 324                 );
 325             } else {
 326                 diag.span_label(span, msg);
 327                 diag.help("format of unicode escape sequences is `\\u{...}`");
 328             }
 329
 330             diag.emit();
 331         }
 332         EscapeError::UnicodeEscapeInByte => {
 333             let msg = "unicode escape in byte string";
 334             handler
 335                 .struct_span_err(span, msg)
 336                 .span_label(span, msg)
 337                 .help("unicode escape sequences cannot be used as a byte or in a byte string")
 338                 .emit();
 339         }
 340         EscapeError::EmptyUnicodeEscape => {
 341             handler
 342                 .struct_span_err(span, "empty unicode escape")
 343                 .span_label(span, "this escape must have at least 1 hex digit")
 344                 .emit();
 345         }
 346         EscapeError::ZeroChars => {
 347             let msg = "empty character literal";
 348             handler.struct_span_err(span, msg).span_label(span, msg).emit();
 349         }
 350         EscapeError::LoneSlash => {
 351             let msg = "invalid trailing slash in literal";
 352             handler.struct_span_err(span, msg).span_label(span, msg).emit();
 353         }
 354         EscapeError::UnskippedWhitespaceWarning => {
 355             let (c, char_span) = last_char();
 356             let msg =
 357                 format!("non-ASCII whitespace symbol '{}' is not skipped", c.escape_unicode());
 358             handler.struct_span_warn(span, &msg).span_label(char_span, &msg).emit();
 359         }
 360         EscapeError::MultipleSkippedLinesWarning => {
 361             let msg = "multiple lines skipped by escaped newline";
 362             let bottom_msg = "skipping everything up to and including this point";
 363             handler.struct_span_warn(span, msg).span_label(span, bottom_msg).emit();
 364         }
 365     }
 366 }
 367
 368 /// Pushes a character to a message string for error reporting
 369 pub(crate) fn escaped_char(c: char) -> String {
 370     match c {
 371         '\u{20}'..='\u{7e}' => {
 372             // Don't escape \, ' or " for user-facing messages
 373             c.to_string()
 374         }
 375         _ => c.escape_default().to_string(),
 376     }
 377 }