Auto merge of #57651 - JohnTitor:give-char-type, r=estebank

Implement new literal type `Err`

Fixes #57384

I removed `return Ok`, otherwise, two errors occur. Any solutions?

r? @estebank

src/librustc/ich/impls_syntax.rs

@@ -164,6 +164,7 @@ impl_stable_hash_for!(enum ::syntax::ast::LitIntType {
 impl_stable_hash_for_spanned!(::syntax::ast::LitKind);
 impl_stable_hash_for!(enum ::syntax::ast::LitKind {
     Str(value, style),
+    Err(value),
     ByteStr(value),
     Byte(value),
     Char(value),
@@ -329,6 +330,7 @@ fn hash_token<'a, 'gcx, W: StableHasherResult>(
             match *lit {
                 token::Lit::Byte(val) |
                 token::Lit::Char(val) |
+                token::Lit::Err(val) |
                 token::Lit::Integer(val) |
                 token::Lit::Float(val) |
                 token::Lit::Str_(val) |

src/librustc_mir/hair/constant.rs

@@ -37,6 +37,14 @@ crate fn lit_to_const<'a, 'gcx, 'tcx>(
             let id = tcx.allocate_bytes(s.as_bytes());
             ConstValue::new_slice(Scalar::Ptr(id.into()), s.len() as u64, &tcx)
         },
+        LitKind::Err(ref s) => {
+            let s = s.as_str();
+            let id = tcx.allocate_bytes(s.as_bytes());
+            return Ok(ty::Const {
+                val: ConstValue::new_slice(Scalar::Ptr(id.into()), s.len() as u64, &tcx),
+                ty: tcx.types.err,
+            });
+        },
         LitKind::ByteStr(ref data) => {
             let id = tcx.allocate_bytes(data);
             ConstValue::Scalar(Scalar::Ptr(id.into()))

src/librustc_typeck/check/mod.rs

@@ -3121,7 +3121,8 @@ impl<'a, 'gcx, 'tcx> FnCtxt<'a, 'gcx, 'tcx> {
                 opt_ty.unwrap_or_else(
                     || tcx.mk_float_var(self.next_float_var_id()))
             }
-            ast::LitKind::Bool(_) => tcx.types.bool
+            ast::LitKind::Bool(_) => tcx.types.bool,
+            ast::LitKind::Err(_) => tcx.types.err,
         }
     }
 

src/librustdoc/html/highlight.rs

@@ -296,7 +296,7 @@ impl<'a> Classifier<'a> {
             token::Literal(lit, _suf) => {
                 match lit {
                     // Text literals.
-                    token::Byte(..) | token::Char(..) |
+                    token::Byte(..) | token::Char(..) | token::Err(..) |
                         token::ByteStr(..) | token::ByteStrRaw(..) |
                         token::Str_(..) | token::StrRaw(..) => Class::String,
 

src/libsyntax/ast.rs

@@ -1285,6 +1285,8 @@ pub enum LitKind {
     FloatUnsuffixed(Symbol),
     /// A boolean literal.
     Bool(bool),
+    /// A recovered character literal that contains mutliple `char`s, most likely a typo.
+    Err(Symbol),
 }
 
 impl LitKind {
@@ -1321,6 +1323,7 @@ impl LitKind {
             | LitKind::ByteStr(..)
             | LitKind::Byte(..)
             | LitKind::Char(..)
+            | LitKind::Err(..)
             | LitKind::Int(_, LitIntType::Unsuffixed)
             | LitKind::FloatUnsuffixed(..)
             | LitKind::Bool(..) => true,

src/libsyntax/attr/mod.rs

@@ -666,6 +666,7 @@ impl LitKind {
             } else {
                 "false"
             })), false),
+            LitKind::Err(val) => Token::Literal(token::Lit::Err(val), None),
         }
     }
 

src/libsyntax/ext/quote.rs

@@ -646,6 +646,7 @@ fn expr_mk_token(cx: &ExtCtxt, sp: Span, tok: &token::Token) -> P<ast::Expr> {
 
         token::Literal(token::Byte(i), suf) => return mk_lit!("Byte", suf, i),
         token::Literal(token::Char(i), suf) => return mk_lit!("Char", suf, i),
+        token::Literal(token::Err(_i), _suf) => return cx.expr(sp, ast::ExprKind::Err),
         token::Literal(token::Integer(i), suf) => return mk_lit!("Integer", suf, i),
         token::Literal(token::Float(i), suf) => return mk_lit!("Float", suf, i),
         token::Literal(token::Str_(i), suf) => return mk_lit!("Str_", suf, i),

src/libsyntax/parse/lexer/mod.rs

@@ -1408,9 +1408,10 @@ impl<'a> StringReader<'a> {
                     // lifetimes shouldn't end with a single quote
                     // if we find one, then this is an invalid character literal
                     if self.ch_is('\'') {
-                        self.fatal_span_verbose(start_with_quote, self.next_pos,
-                                String::from("character literal may only contain one codepoint"))
-                            .raise();
+                        self.err_span_(start_with_quote, self.next_pos,
+                                "character literal may only contain one codepoint");
+                        self.bump();
+                        return Ok(token::Literal(token::Err(Symbol::intern("??")), None))
 
                     }
 
@@ -1445,7 +1446,7 @@ impl<'a> StringReader<'a> {
                                     format!("\"{}\"", &self.src[start..end]),
                                     Applicability::MachineApplicable
                                 ).emit();
-                            return Ok(token::Literal(token::Str_(Symbol::intern("??")), None))
+                            return Ok(token::Literal(token::Err(Symbol::intern("??")), None))
                         }
                         if self.ch_is('\n') || self.is_eof() || self.ch_is('/') {
                             // Only attempt to infer single line string literals. If we encounter

src/libsyntax/parse/mod.rs

@@ -466,6 +466,7 @@ crate fn lit_token(lit: token::Lit, suf: Option<Symbol>, diag: Option<(Span, &Ha
     match lit {
        token::Byte(i) => (true, Some(LitKind::Byte(byte_lit(&i.as_str()).0))),
        token::Char(i) => (true, Some(LitKind::Char(char_lit(&i.as_str(), diag).0))),
+       token::Err(i) => (true, Some(LitKind::Err(i))),
 
         // There are some valid suffixes for integer and float literals,
         // so all the handling is done internally.

src/libsyntax/parse/token.rs

@@ -60,6 +60,7 @@ impl DelimToken {
 pub enum Lit {
     Byte(ast::Name),
     Char(ast::Name),
+    Err(ast::Name),
     Integer(ast::Name),
     Float(ast::Name),
     Str_(ast::Name),
@@ -73,6 +74,7 @@ impl Lit {
         match *self {
             Byte(_) => "byte literal",
             Char(_) => "char literal",
+            Err(_) => "invalid literal",
             Integer(_) => "integer literal",
             Float(_) => "float literal",
             Str_(_) | StrRaw(..) => "string literal",
@@ -471,8 +473,7 @@ impl Token {
 
             Le | EqEq | Ne | Ge | AndAnd | OrOr | Tilde | BinOpEq(..) | At | DotDotDot |
             DotDotEq | Comma | Semi | ModSep | RArrow | LArrow | FatArrow | Pound | Dollar |
-            Question | OpenDelim(..) | CloseDelim(..) => return None,
-
+            Question | OpenDelim(..) | CloseDelim(..) |
             Literal(..) | Ident(..) | Lifetime(..) | Interpolated(..) | DocComment(..) |
             Whitespace | Comment | Shebang(..) | Eof => return None,
         })

src/libsyntax/print/pprust.rs

@@ -224,6 +224,7 @@ pub fn token_to_string(tok: &Token) -> String {
             let mut out = match lit {
                 token::Byte(b)           => format!("b'{}'", b),
                 token::Char(c)           => format!("'{}'", c),
+                token::Err(c)            => format!("'{}'", c),
                 token::Float(c)          |
                 token::Integer(c)        => c.to_string(),
                 token::Str_(s)           => format!("\"{}\"", s),
@@ -603,6 +604,14 @@ pub trait PrintState<'a> {
         }
         match lit.node {
             ast::LitKind::Str(st, style) => self.print_string(&st.as_str(), style),
+            ast::LitKind::Err(st) => {
+                let st = st.as_str().escape_debug();
+                let mut res = String::with_capacity(st.len() + 2);
+                res.push('\'');
+                res.push_str(&st);
+                res.push('\'');
+                self.writer().word(res)
+            }
             ast::LitKind::Byte(byte) => {
                 let mut res = String::from("b'");
                 res.extend(ascii::escape_default(byte).map(|c| c as char));

src/libsyntax_ext/concat.rs

@@ -23,6 +23,7 @@ pub fn expand_syntax_ext(
         match e.node {
             ast::ExprKind::Lit(ref lit) => match lit.node {
                 ast::LitKind::Str(ref s, _)
+                | ast::LitKind::Err(ref s)
                 | ast::LitKind::Float(ref s, _)
                 | ast::LitKind::FloatUnsuffixed(ref s) => {
                     accumulator.push_str(&s.as_str());

src/test/ui/parser/lex-bad-char-literals-2.rs

@@ -1,4 +1,4 @@
 // This test needs to the last one appearing in this file as it kills the parser
 static c: char =
-    'nope' //~ ERROR: character literal may only contain one codepoint: 'nope'
+    'nope' //~ ERROR: character literal may only contain one codepoint
 ;

src/test/ui/parser/lex-bad-char-literals-2.stderr

@@ -1,8 +1,13 @@
-error: character literal may only contain one codepoint: 'nope'
+error: character literal may only contain one codepoint
   --> $DIR/lex-bad-char-literals-2.rs:3:5
    |
-LL |     'nope' //~ ERROR: character literal may only contain one codepoint: 'nope'
+LL |     'nope' //~ ERROR: character literal may only contain one codepoint
    |     ^^^^^^
 
-error: aborting due to previous error
+error[E0601]: `main` function not found in crate `lex_bad_char_literals_2`
+   |
+   = note: consider adding a `main` function to `$DIR/lex-bad-char-literals-2.rs`
 
+error: aborting due to 2 previous errors
+
+For more information about this error, try `rustc --explain E0601`.

src/test/ui/parser/lex-bad-char-literals-3.rs

@@ -1,7 +1,7 @@
-// This test needs to the last one appearing in this file as it kills the parser
-static c: char =
-    '●●' //~ ERROR: character literal may only contain one codepoint
-         //~| ERROR: mismatched types
-;
+static c: char = '●●';
+//~^ ERROR: character literal may only contain one codepoint
 
-fn main() {}
+fn main() {
+    let ch: &str = '●●';
+    //~^ ERROR: character literal may only contain one codepoint
+}

src/test/ui/parser/lex-bad-char-literals-3.stderr

@@ -1,22 +1,22 @@
 error: character literal may only contain one codepoint
-  --> $DIR/lex-bad-char-literals-3.rs:3:5
+  --> $DIR/lex-bad-char-literals-3.rs:1:18
    |
-LL |     '●●' //~ ERROR: character literal may only contain one codepoint
+LL | static c: char = '●●';
    |                  ^^^^
 help: if you meant to write a `str` literal, use double quotes
    |
-LL |     "●●" //~ ERROR: character literal may only contain one codepoint
+LL | static c: char = "●●";
    |                  ^^^^
 
-error[E0308]: mismatched types
-  --> $DIR/lex-bad-char-literals-3.rs:3:5
+error: character literal may only contain one codepoint
+  --> $DIR/lex-bad-char-literals-3.rs:5:20
    |
-LL |     '●●' //~ ERROR: character literal may only contain one codepoint
-   |     ^^^^ expected char, found reference
+LL |     let ch: &str = '●●';
+   |                    ^^^^
+help: if you meant to write a `str` literal, use double quotes
    |
-   = note: expected type `char`
-              found type `&'static str`
+LL |     let ch: &str = "●●";
+   |                    ^^^^
 
 error: aborting due to 2 previous errors
 
-For more information about this error, try `rustc --explain E0308`.

src/test/ui/parser/lex-bad-char-literals-4.rs

@@ -1,5 +1,5 @@
 //
 // This test needs to the last one appearing in this file as it kills the parser
 static c: char =
-    '●  //~ ERROR: character literal may only contain one codepoint: '●
+    '●  //~ ERROR: character literal may only contain one codepoint
 ;

src/test/ui/parser/lex-bad-char-literals-4.stderr

@@ -1,7 +1,7 @@
 error: character literal may only contain one codepoint: '●
   --> $DIR/lex-bad-char-literals-4.rs:4:5
    |
-LL |     '●  //~ ERROR: character literal may only contain one codepoint: '●
+LL |     '●  //~ ERROR: character literal may only contain one codepoint
    |     ^^
 
 error: aborting due to previous error

src/test/ui/parser/lex-bad-char-literals-5.rs

@@ -1,8 +1,7 @@
-//
-// This test needs to the last one appearing in this file as it kills the parser
-static c: char =
-    '\x10\x10'  //~ ERROR: character literal may only contain one codepoint
-                //~| ERROR: mismatched types
-;
+static c: char = '\x10\x10';
+//~^ ERROR: character literal may only contain one codepoint
 
-fn main() {}
+fn main() {
+    let ch: &str = '\x10\x10';
+    //~^ ERROR: character literal may only contain one codepoint
+}

src/test/ui/parser/lex-bad-char-literals-5.stderr

@@ -1,22 +1,22 @@
 error: character literal may only contain one codepoint
-  --> $DIR/lex-bad-char-literals-5.rs:4:5
+  --> $DIR/lex-bad-char-literals-5.rs:1:18
    |
-LL |     '/x10/x10'  //~ ERROR: character literal may only contain one codepoint
+LL | static c: char = '/x10/x10';
    |                  ^^^^^^^^^^
 help: if you meant to write a `str` literal, use double quotes
    |
-LL |     "/x10/x10"  //~ ERROR: character literal may only contain one codepoint
+LL | static c: char = "/x10/x10";
    |                  ^^^^^^^^^^
 
-error[E0308]: mismatched types
-  --> $DIR/lex-bad-char-literals-5.rs:4:5
+error: character literal may only contain one codepoint
+  --> $DIR/lex-bad-char-literals-5.rs:5:20
    |
-LL |     '/x10/x10'  //~ ERROR: character literal may only contain one codepoint
-   |     ^^^^^^^^^^ expected char, found reference
+LL |     let ch: &str = '/x10/x10';
+   |                    ^^^^^^^^^^
+help: if you meant to write a `str` literal, use double quotes
    |
-   = note: expected type `char`
-              found type `&'static str`
+LL |     let ch: &str = "/x10/x10";
+   |                    ^^^^^^^^^^
 
 error: aborting due to 2 previous errors
 
-For more information about this error, try `rustc --explain E0308`.

src/test/ui/parser/lex-bad-char-literals-6.rs

@@ -0,0 +1,17 @@
+fn main() {
+    let x: &str = 'ab';
+    //~^ ERROR: character literal may only contain one codepoint
+    let y: char = 'cd';
+    //~^ ERROR: character literal may only contain one codepoint
+    let z = 'ef';
+    //~^ ERROR: character literal may only contain one codepoint
+
+    if x == y {}
+    //~^ ERROR: can't compare `&str` with `char`
+    if y == z {}  // no error here
+    if x == z {}
+    //~^ ERROR: can't compare `&str` with `char`
+
+    let a: usize = "";
+    //~^ ERROR: mismatched types
+}

src/test/ui/parser/lex-bad-char-literals-6.stderr

@@ -0,0 +1,47 @@
+error: character literal may only contain one codepoint
+  --> $DIR/lex-bad-char-literals-6.rs:2:19
+   |
+LL |     let x: &str = 'ab';
+   |                   ^^^^
+
+error: character literal may only contain one codepoint
+  --> $DIR/lex-bad-char-literals-6.rs:4:19
+   |
+LL |     let y: char = 'cd';
+   |                   ^^^^
+
+error: character literal may only contain one codepoint
+  --> $DIR/lex-bad-char-literals-6.rs:6:13
+   |
+LL |     let z = 'ef';
+   |             ^^^^
+
+error[E0277]: can't compare `&str` with `char`
+  --> $DIR/lex-bad-char-literals-6.rs:9:10
+   |
+LL |     if x == y {}
+   |          ^^ no implementation for `&str == char`
+   |
+   = help: the trait `std::cmp::PartialEq<char>` is not implemented for `&str`
+
+error[E0308]: mismatched types
+  --> $DIR/lex-bad-char-literals-6.rs:15:20
+   |
+LL |     let a: usize = "";
+   |                    ^^ expected usize, found reference
+   |
+   = note: expected type `usize`
+              found type `&'static str`
+
+error[E0277]: can't compare `&str` with `char`
+  --> $DIR/lex-bad-char-literals-6.rs:12:10
+   |
+LL |     if x == z {}
+   |          ^^ no implementation for `&str == char`
+   |
+   = help: the trait `std::cmp::PartialEq<char>` is not implemented for `&str`
+
+error: aborting due to 6 previous errors
+
+Some errors occurred: E0277, E0308.
+For more information about an error, try `rustc --explain E0277`.

src/test/ui/str/str-as-char.fixed

@@ -1,6 +1,6 @@
 // run-rustfix
 
 fn main() {
-    println!("●●");
-    //~^ ERROR character literal may only contain one codepoint
+    println!("{}", "●●"); //~ ERROR character literal may only contain one codepoint
+    //~^ ERROR format argument must be a string literal
 }

src/test/ui/str/str-as-char.rs

@@ -1,6 +1,6 @@
 // run-rustfix
 
 fn main() {
-    println!('●●');
-    //~^ ERROR character literal may only contain one codepoint
+    println!('●●'); //~ ERROR character literal may only contain one codepoint
+    //~^ ERROR format argument must be a string literal
 }

src/test/ui/str/str-as-char.stderr

@@ -1,12 +1,22 @@
 error: character literal may only contain one codepoint
   --> $DIR/str-as-char.rs:4:14
    |
-LL |     println!('●●');
+LL |     println!('●●'); //~ ERROR character literal may only contain one codepoint
    |              ^^^^
 help: if you meant to write a `str` literal, use double quotes
    |
-LL |     println!("●●");
+LL |     println!("●●"); //~ ERROR character literal may only contain one codepoint
    |              ^^^^
 
-error: aborting due to previous error
+error: format argument must be a string literal
+  --> $DIR/str-as-char.rs:4:14
+   |
+LL |     println!('●●'); //~ ERROR character literal may only contain one codepoint
+   |              ^^^^
+help: you might be missing a string literal to format with
+   |
+LL |     println!("{}", '●●'); //~ ERROR character literal may only contain one codepoint
+   |              ^^^^^
+
+error: aborting due to 2 previous errors