src/libsyntax/parse/literal.rs

   1 //! Code related to parsing literals.
   2
   3 use crate::ast::{self, Lit, LitKind};
   4 use crate::parse::token::{self, Token};
   5 use crate::symbol::{kw, sym, Symbol};
   6 use crate::tokenstream::TokenTree;
   7
   8 use log::debug;
   9 use rustc_data_structures::sync::Lrc;
  10 use syntax_pos::Span;
  11 use rustc_lexer::unescape::{unescape_char, unescape_byte};
  12 use rustc_lexer::unescape::{unescape_str, unescape_byte_str};
  13 use rustc_lexer::unescape::{unescape_raw_str, unescape_raw_byte_str};
  14
  15 use std::ascii;
  16
  17 crate enum LitError {
  18     NotLiteral,
  19     LexerError,
  20     InvalidSuffix,
  21     InvalidIntSuffix,
  22     InvalidFloatSuffix,
  23     NonDecimalFloat(u32),
  24     IntTooLarge,
  25 }
  26
  27 impl LitKind {
  28     /// Converts literal token into a semantic literal.
  29     fn from_lit_token(lit: token::Lit) -> Result<LitKind, LitError> {
  30         let token::Lit { kind, symbol, suffix } = lit;
  31         if suffix.is_some() && !kind.may_have_suffix() {
  32             return Err(LitError::InvalidSuffix);
  33         }
  34
  35         Ok(match kind {
  36             token::Bool => {
  37                 assert!(symbol.is_bool_lit());
  38                 LitKind::Bool(symbol == kw::True)
  39             }
  40             token::Byte => return unescape_byte(&symbol.as_str())
  41                 .map(LitKind::Byte).map_err(|_| LitError::LexerError),
  42             token::Char => return unescape_char(&symbol.as_str())
  43                 .map(LitKind::Char).map_err(|_| LitError::LexerError),
  44
  45             // There are some valid suffixes for integer and float literals,
  46             // so all the handling is done internally.
  47             token::Integer => return integer_lit(symbol, suffix),
  48             token::Float => return float_lit(symbol, suffix),
  49
  50             token::Str => {
  51                 // If there are no characters requiring special treatment we can
  52                 // reuse the symbol from the token. Otherwise, we must generate a
  53                 // new symbol because the string in the LitKind is different to the
  54                 // string in the token.
  55                 let s = symbol.as_str();
  56                 let symbol = if s.contains(&['\\', '\r'][..]) {
  57                     let mut buf = String::with_capacity(s.len());
  58                     let mut error = Ok(());
  59                     unescape_str(&s, &mut |_, unescaped_char| {
  60                         match unescaped_char {
  61                             Ok(c) => buf.push(c),
  62                             Err(_) => error = Err(LitError::LexerError),
  63                         }
  64                     });
  65                     error?;
  66                     Symbol::intern(&buf)
  67                 } else {
  68                     symbol
  69                 };
  70                 LitKind::Str(symbol, ast::StrStyle::Cooked)
  71             }
  72             token::StrRaw(n) => {
  73                 // Ditto.
  74                 let s = symbol.as_str();
  75                 let symbol = if s.contains('\r') {
  76                     let mut buf = String::with_capacity(s.len());
  77                     let mut error = Ok(());
  78                     unescape_raw_str(&s, &mut |_, unescaped_char| {
  79                         match unescaped_char {
  80                             Ok(c) => buf.push(c),
  81                             Err(_) => error = Err(LitError::LexerError),
  82                         }
  83                     });
  84                     error?;
  85                     buf.shrink_to_fit();
  86                     Symbol::intern(&buf)
  87                 } else {
  88                     symbol
  89                 };
  90                 LitKind::Str(symbol, ast::StrStyle::Raw(n))
  91             }
  92             token::ByteStr => {
  93                 let s = symbol.as_str();
  94                 let mut buf = Vec::with_capacity(s.len());
  95                 let mut error = Ok(());
  96                 unescape_byte_str(&s, &mut |_, unescaped_byte| {
  97                     match unescaped_byte {
  98                         Ok(c) => buf.push(c),
  99                         Err(_) => error = Err(LitError::LexerError),
 100                     }
 101                 });
 102                 error?;
 103                 buf.shrink_to_fit();
 104                 LitKind::ByteStr(Lrc::new(buf))
 105             }
 106             token::ByteStrRaw(_) => {
 107                 let s = symbol.as_str();
 108                 let bytes = if s.contains('\r') {
 109                     let mut buf = Vec::with_capacity(s.len());
 110                     let mut error = Ok(());
 111                     unescape_raw_byte_str(&s, &mut |_, unescaped_byte| {
 112                         match unescaped_byte {
 113                             Ok(c) => buf.push(c),
 114                             Err(_) => error = Err(LitError::LexerError),
 115                         }
 116                     });
 117                     error?;
 118                     buf.shrink_to_fit();
 119                     buf
 120                 } else {
 121                     symbol.to_string().into_bytes()
 122                 };
 123
 124                 LitKind::ByteStr(Lrc::new(bytes))
 125             },
 126             token::Err => LitKind::Err(symbol),
 127         })
 128     }
 129
 130     /// Attempts to recover a token from semantic literal.
 131     /// This function is used when the original token doesn't exist (e.g. the literal is created
 132     /// by an AST-based macro) or unavailable (e.g. from HIR pretty-printing).
 133     pub fn to_lit_token(&self) -> token::Lit {
 134         let (kind, symbol, suffix) = match *self {
 135             LitKind::Str(symbol, ast::StrStyle::Cooked) => {
 136                 // Don't re-intern unless the escaped string is different.
 137                 let s: &str = &symbol.as_str();
 138                 let escaped = s.escape_default().to_string();
 139                 let symbol = if escaped == *s { symbol } else { Symbol::intern(&escaped) };
 140                 (token::Str, symbol, None)
 141             }
 142             LitKind::Str(symbol, ast::StrStyle::Raw(n)) => {
 143                 (token::StrRaw(n), symbol, None)
 144             }
 145             LitKind::ByteStr(ref bytes) => {
 146                 let string = bytes.iter().cloned().flat_map(ascii::escape_default)
 147                     .map(Into::<char>::into).collect::<String>();
 148                 (token::ByteStr, Symbol::intern(&string), None)
 149             }
 150             LitKind::Byte(byte) => {
 151                 let string: String = ascii::escape_default(byte).map(Into::<char>::into).collect();
 152                 (token::Byte, Symbol::intern(&string), None)
 153             }
 154             LitKind::Char(ch) => {
 155                 let string: String = ch.escape_default().map(Into::<char>::into).collect();
 156                 (token::Char, Symbol::intern(&string), None)
 157             }
 158             LitKind::Int(n, ty) => {
 159                 let suffix = match ty {
 160                     ast::LitIntType::Unsigned(ty) => Some(ty.to_symbol()),
 161                     ast::LitIntType::Signed(ty) => Some(ty.to_symbol()),
 162                     ast::LitIntType::Unsuffixed => None,
 163                 };
 164                 (token::Integer, sym::integer(n), suffix)
 165             }
 166             LitKind::Float(symbol, ty) => {
 167                 (token::Float, symbol, Some(ty.to_symbol()))
 168             }
 169             LitKind::FloatUnsuffixed(symbol) => {
 170                 (token::Float, symbol, None)
 171             }
 172             LitKind::Bool(value) => {
 173                 let symbol = if value { kw::True } else { kw::False };
 174                 (token::Bool, symbol, None)
 175             }
 176             LitKind::Err(symbol) => {
 177                 (token::Err, symbol, None)
 178             }
 179         };
 180
 181         token::Lit::new(kind, symbol, suffix)
 182     }
 183 }
 184
 185 impl Lit {
 186     /// Converts literal token into an AST literal.
 187     crate fn from_lit_token(token: token::Lit, span: Span) -> Result<Lit, LitError> {
 188         Ok(Lit { token, kind: LitKind::from_lit_token(token)?, span })
 189     }
 190
 191     /// Converts arbitrary token into an AST literal.
 192     crate fn from_token(token: &Token) -> Result<Lit, LitError> {
 193         let lit = match token.kind {
 194             token::Ident(name, false) if name.is_bool_lit() =>
 195                 token::Lit::new(token::Bool, name, None),
 196             token::Literal(lit) =>
 197                 lit,
 198             token::Interpolated(ref nt) => {
 199                 if let token::NtExpr(expr) | token::NtLiteral(expr) = &**nt {
 200                     if let ast::ExprKind::Lit(lit) = &expr.kind {
 201                         return Ok(lit.clone());
 202                     }
 203                 }
 204                 return Err(LitError::NotLiteral);
 205             }
 206             _ => return Err(LitError::NotLiteral)
 207         };
 208
 209         Lit::from_lit_token(lit, token.span)
 210     }
 211
 212     /// Attempts to recover an AST literal from semantic literal.
 213     /// This function is used when the original token doesn't exist (e.g. the literal is created
 214     /// by an AST-based macro) or unavailable (e.g. from HIR pretty-printing).
 215     pub fn from_lit_kind(kind: LitKind, span: Span) -> Lit {
 216         Lit { token: kind.to_lit_token(), kind, span }
 217     }
 218
 219     /// Losslessly convert an AST literal into a token tree.
 220     crate fn token_tree(&self) -> TokenTree {
 221         let token = match self.token.kind {
 222             token::Bool => token::Ident(self.token.symbol, false),
 223             _ => token::Literal(self.token),
 224         };
 225         TokenTree::token(token, self.span)
 226     }
 227 }
 228
 229 fn strip_underscores(symbol: Symbol) -> Symbol {
 230     // Do not allocate a new string unless necessary.
 231     let s = symbol.as_str();
 232     if s.contains('_') {
 233         let mut s = s.to_string();
 234         s.retain(|c| c != '_');
 235         return Symbol::intern(&s);
 236     }
 237     symbol
 238 }
 239
 240 fn filtered_float_lit(symbol: Symbol, suffix: Option<Symbol>, base: u32)
 241                       -> Result<LitKind, LitError> {
 242     debug!("filtered_float_lit: {:?}, {:?}, {:?}", symbol, suffix, base);
 243     if base != 10 {
 244         return Err(LitError::NonDecimalFloat(base));
 245     }
 246     Ok(match suffix {
 247         Some(suf) => match suf {
 248             sym::f32 => LitKind::Float(symbol, ast::FloatTy::F32),
 249             sym::f64 => LitKind::Float(symbol, ast::FloatTy::F64),
 250             _ => return Err(LitError::InvalidFloatSuffix),
 251         }
 252         None => LitKind::FloatUnsuffixed(symbol)
 253     })
 254 }
 255
 256 fn float_lit(symbol: Symbol, suffix: Option<Symbol>) -> Result<LitKind, LitError> {
 257     debug!("float_lit: {:?}, {:?}", symbol, suffix);
 258     filtered_float_lit(strip_underscores(symbol), suffix, 10)
 259 }
 260
 261 fn integer_lit(symbol: Symbol, suffix: Option<Symbol>) -> Result<LitKind, LitError> {
 262     debug!("integer_lit: {:?}, {:?}", symbol, suffix);
 263     let symbol = strip_underscores(symbol);
 264     let s = symbol.as_str();
 265
 266     let base = match s.as_bytes() {
 267         [b'0', b'x', ..] => 16,
 268         [b'0', b'o', ..] => 8,
 269         [b'0', b'b', ..] => 2,
 270         _ => 10,
 271     };
 272
 273     let ty = match suffix {
 274         Some(suf) => match suf {
 275             sym::isize => ast::LitIntType::Signed(ast::IntTy::Isize),
 276             sym::i8  => ast::LitIntType::Signed(ast::IntTy::I8),
 277             sym::i16 => ast::LitIntType::Signed(ast::IntTy::I16),
 278             sym::i32 => ast::LitIntType::Signed(ast::IntTy::I32),
 279             sym::i64 => ast::LitIntType::Signed(ast::IntTy::I64),
 280             sym::i128 => ast::LitIntType::Signed(ast::IntTy::I128),
 281             sym::usize => ast::LitIntType::Unsigned(ast::UintTy::Usize),
 282             sym::u8  => ast::LitIntType::Unsigned(ast::UintTy::U8),
 283             sym::u16 => ast::LitIntType::Unsigned(ast::UintTy::U16),
 284             sym::u32 => ast::LitIntType::Unsigned(ast::UintTy::U32),
 285             sym::u64 => ast::LitIntType::Unsigned(ast::UintTy::U64),
 286             sym::u128 => ast::LitIntType::Unsigned(ast::UintTy::U128),
 287             // `1f64` and `2f32` etc. are valid float literals, and
 288             // `fxxx` looks more like an invalid float literal than invalid integer literal.
 289             _ if suf.as_str().starts_with('f') => return filtered_float_lit(symbol, suffix, base),
 290             _ => return Err(LitError::InvalidIntSuffix),
 291         }
 292         _ => ast::LitIntType::Unsuffixed
 293     };
 294
 295     let s = &s[if base != 10 { 2 } else { 0 } ..];
 296     u128::from_str_radix(s, base).map(|i| LitKind::Int(i, ty)).map_err(|_| {
 297         // Small bases are lexed as if they were base 10, e.g, the string
 298         // might be `0b10201`. This will cause the conversion above to fail,
 299         // but these kinds of errors are already reported by the lexer.
 300         let from_lexer =
 301             base < 10 && s.chars().any(|c| c.to_digit(10).map_or(false, |d| d >= base));
 302         if from_lexer { LitError::LexerError } else { LitError::IntTooLarge }
 303     })
 304 }