Delay literal unescaping

Cargo.lock

@@ -3428,6 +3428,7 @@ dependencies = [
  "rustc_index",
  "rustc_macros",
  "rustc_middle",
+ "rustc_parse",
  "rustc_session",
  "rustc_span",
  "rustc_target",

compiler/rustc_ast_lowering/Cargo.toml

@@ -17,6 +17,7 @@ rustc_hir = { path = "../rustc_hir" }
 rustc_index = { path = "../rustc_index" }
 rustc_macros = { path = "../rustc_macros" }
 rustc_middle = { path = "../rustc_middle" }
+rustc_parse = { path = "../rustc_parse" }
 rustc_session = { path = "../rustc_session" }
 rustc_span = { path = "../rustc_span" }
 rustc_target = { path = "../rustc_target" }

compiler/rustc_ast_lowering/src/expr.rs

@@ -14,7 +14,7 @@ use rustc_data_structures::stack::ensure_sufficient_stack;
 use rustc_hir as hir;
 use rustc_hir::def::{DefKind, Res};
 use rustc_middle::span_bug;
-use rustc_session::errors::report_lit_error;
+use rustc_parse::parser::token_lit_to_lit_kind_and_report_errs;
 use rustc_span::source_map::{respan, Spanned};
 use rustc_span::symbol::{kw, sym, Ident, Symbol};
 use rustc_span::DUMMY_SP;
@@ -119,13 +119,12 @@ impl<'hir> LoweringContext<'_, 'hir> {
                     hir::ExprKind::Unary(op, ohs)
                 }
                 ExprKind::Lit(token_lit) => {
-                    let lit_kind = match LitKind::from_token_lit(*token_lit) {
-                        Ok(lit_kind) => lit_kind,
-                        Err(err) => {
-                            report_lit_error(&self.tcx.sess.parse_sess, err, *token_lit, e.span);
-                            LitKind::Err
-                        }
-                    };
+                    let lit_kind = token_lit_to_lit_kind_and_report_errs(
+                        &self.tcx.sess.parse_sess,
+                        *token_lit,
+                        e.span,
+                    )
+                    .unwrap_or(LitKind::Err);
                     let lit = self.arena.alloc(respan(self.lower_span(e.span), lit_kind));
                     hir::ExprKind::Lit(lit)
                 }

compiler/rustc_builtin_macros/src/concat.rs

@@ -1,7 +1,7 @@
 use rustc_ast as ast;
 use rustc_ast::tokenstream::TokenStream;
 use rustc_expand::base::{self, DummyResult};
-use rustc_session::errors::report_lit_error;
+use rustc_parse::parser::token_lit_to_lit_kind_and_report_errs;
 use rustc_span::symbol::Symbol;
 
 use crate::errors;
@@ -19,44 +19,43 @@ pub fn expand_concat(
     let mut has_errors = false;
     for e in es {
         match e.kind {
-            ast::ExprKind::Lit(token_lit) => match ast::LitKind::from_token_lit(token_lit) {
-                Ok(ast::LitKind::Str(s, _) | ast::LitKind::Float(s, _)) => {
-                    accumulator.push_str(s.as_str());
-                }
-                Ok(ast::LitKind::Char(c)) => {
-                    accumulator.push(c);
-                }
-                Ok(ast::LitKind::Int(i, _)) => {
-                    accumulator.push_str(&i.to_string());
-                }
-                Ok(ast::LitKind::Bool(b)) => {
-                    accumulator.push_str(&b.to_string());
-                }
-                Ok(ast::LitKind::CStr(..)) => {
-                    cx.emit_err(errors::ConcatCStrLit { span: e.span });
-                    has_errors = true;
-                }
-                Ok(ast::LitKind::Byte(..) | ast::LitKind::ByteStr(..)) => {
-                    cx.emit_err(errors::ConcatBytestr { span: e.span });
-                    has_errors = true;
-                }
-                Ok(ast::LitKind::Err) => {
-                    has_errors = true;
-                }
-                Err(err) => {
-                    report_lit_error(&cx.sess.parse_sess, err, token_lit, e.span);
-                    has_errors = true;
+            ast::ExprKind::Lit(token_lit) => {
+                match token_lit_to_lit_kind_and_report_errs(&cx.sess.parse_sess, token_lit, e.span)
+                {
+                    Ok(ast::LitKind::Str(s, _) | ast::LitKind::Float(s, _)) => {
+                        accumulator.push_str(s.as_str());
+                    }
+                    Ok(ast::LitKind::Char(c)) => {
+                        accumulator.push(c);
+                    }
+                    Ok(ast::LitKind::Int(i, _)) => {
+                        accumulator.push_str(&i.to_string());
+                    }
+                    Ok(ast::LitKind::Bool(b)) => {
+                        accumulator.push_str(&b.to_string());
+                    }
+                    Ok(ast::LitKind::CStr(..)) => {
+                        cx.emit_err(errors::ConcatCStrLit { span: e.span });
+                        has_errors = true;
+                    }
+                    Ok(ast::LitKind::Byte(..) | ast::LitKind::ByteStr(..)) => {
+                        cx.emit_err(errors::ConcatBytestr { span: e.span });
+                        has_errors = true;
+                    }
+                    Ok(ast::LitKind::Err) | Err(()) => {
+                        has_errors = true;
+                    }
                 }
-            },
+            }
             // We also want to allow negative numeric literals.
             ast::ExprKind::Unary(ast::UnOp::Neg, ref expr)
                 if let ast::ExprKind::Lit(token_lit) = expr.kind =>
             {
-                match ast::LitKind::from_token_lit(token_lit) {
+                match token_lit_to_lit_kind_and_report_errs(&cx.sess.parse_sess, token_lit, e.span)
+                {
                     Ok(ast::LitKind::Int(i, _)) => accumulator.push_str(&format!("-{i}")),
                     Ok(ast::LitKind::Float(f, _)) => accumulator.push_str(&format!("-{f}")),
-                    Err(err) => {
-                        report_lit_error(&cx.sess.parse_sess, err, token_lit, e.span);
+                    Err(()) => {
                         has_errors = true;
                     }
                     _ => missing_literal.push(e.span),

compiler/rustc_builtin_macros/src/concat_bytes.rs

@@ -1,7 +1,7 @@
 use rustc_ast as ast;
 use rustc_ast::{ptr::P, tokenstream::TokenStream};
 use rustc_expand::base::{self, DummyResult};
-use rustc_session::errors::report_lit_error;
+use rustc_parse::parser::token_lit_to_lit_kind_and_report_errs;
 use rustc_span::Span;
 
 use crate::errors;
@@ -17,7 +17,7 @@ fn invalid_type_err(
         ConcatBytesInvalid, ConcatBytesInvalidSuggestion, ConcatBytesNonU8, ConcatBytesOob,
     };
     let snippet = cx.sess.source_map().span_to_snippet(span).ok();
-    match ast::LitKind::from_token_lit(token_lit) {
+    match token_lit_to_lit_kind_and_report_errs(&cx.sess.parse_sess, token_lit, span) {
         Ok(ast::LitKind::CStr(_, _)) => {
             // Avoid ambiguity in handling of terminal `NUL` by refusing to
             // concatenate C string literals as bytes.
@@ -60,9 +60,7 @@ fn invalid_type_err(
             cx.emit_err(ConcatBytesNonU8 { span });
         }
         Ok(ast::LitKind::ByteStr(..) | ast::LitKind::Byte(_)) => unreachable!(),
-        Err(err) => {
-            report_lit_error(&cx.sess.parse_sess, err, token_lit, span);
-        }
+        Err(()) => {}
     }
 }
 

compiler/rustc_expand/src/base.rs

@@ -21,8 +21,7 @@ use rustc_errors::{
 use rustc_feature::Features;
 use rustc_lint_defs::builtin::PROC_MACRO_BACK_COMPAT;
 use rustc_lint_defs::{BufferedEarlyLint, BuiltinLintDiagnostics, RegisteredTools};
-use rustc_parse::{parser, MACRO_ARGUMENTS};
-use rustc_session::errors::report_lit_error;
+use rustc_parse::{self, parser, MACRO_ARGUMENTS};
 use rustc_session::{parse::ParseSess, Limit, Session};
 use rustc_span::def_id::{CrateNum, DefId, LocalDefId};
 use rustc_span::edition::Edition;
@@ -1236,26 +1235,30 @@ pub fn expr_to_spanned_string<'a>(
     let expr = cx.expander().fully_expand_fragment(AstFragment::Expr(expr)).make_expr();
 
     Err(match expr.kind {
-        ast::ExprKind::Lit(token_lit) => match ast::LitKind::from_token_lit(token_lit) {
-            Ok(ast::LitKind::Str(s, style)) => return Ok((s, style, expr.span)),
-            Ok(ast::LitKind::ByteStr(..)) => {
-                let mut err = cx.struct_span_err(expr.span, err_msg);
-                let span = expr.span.shrink_to_lo();
-                err.span_suggestion(
-                    span.with_hi(span.lo() + BytePos(1)),
-                    "consider removing the leading `b`",
-                    "",
-                    Applicability::MaybeIncorrect,
-                );
-                Some((err, true))
-            }
-            Ok(ast::LitKind::Err) => None,
-            Err(err) => {
-                report_lit_error(&cx.sess.parse_sess, err, token_lit, expr.span);
-                None
-            }
-            _ => Some((cx.struct_span_err(expr.span, err_msg), false)),
-        },
+        ast::ExprKind::Lit(token_lit) => {
+            let res = match parser::token_lit_to_lit_kind_and_report_errs(
+                &cx.sess.parse_sess,
+                token_lit,
+                expr.span,
+            ) {
+                Ok(ast::LitKind::Str(s, style)) => return Ok((s, style, expr.span)),
+                Ok(ast::LitKind::ByteStr(..)) => {
+                    let mut err = cx.struct_span_err(expr.span, err_msg);
+                    let span = expr.span.shrink_to_lo();
+                    err.span_suggestion(
+                        span.with_hi(span.lo() + BytePos(1)),
+                        "consider removing the leading `b`",
+                        "",
+                        Applicability::MaybeIncorrect,
+                    );
+                    Some((err, true))
+                }
+                Ok(ast::LitKind::Err) => None,
+                Err(()) => None,
+                _ => Some((cx.struct_span_err(expr.span, err_msg), false)),
+            };
+            res
+        }
         ast::ExprKind::Err => None,
         _ => Some((cx.struct_span_err(expr.span, err_msg), false)),
     })

compiler/rustc_lexer/src/unescape.rs

@@ -59,6 +59,9 @@ pub enum EscapeError {
     /// Non-ascii character in byte literal, byte string literal, or raw byte string literal.
     NonAsciiCharInByte,
 
+    // `\0` in a C string literal.
+    NulInCStr,
+
     /// After a line ending with '\', the next line contains whitespace
     /// characters that are not skipped.
     UnskippedWhitespaceWarning,
@@ -122,10 +125,20 @@ where
 {
     match mode {
         CStr => {
-            unescape_non_raw_common(src, mode, callback);
+            unescape_non_raw_common(src, mode, &mut |r, mut result| {
+                if let Ok(CStrUnit::Byte(0) | CStrUnit::Char('\0')) = result {
+                    result = Err(EscapeError::NulInCStr);
+                }
+                callback(r, result)
+            });
         }
         RawCStr => {
-            check_raw_common(src, mode, &mut |r, result| callback(r, result.map(CStrUnit::Char)));
+            check_raw_common(src, mode, &mut |r, mut result| {
+                if let Ok('\0') = result {
+                    result = Err(EscapeError::NulInCStr);
+                }
+                callback(r, result.map(CStrUnit::Char))
+            });
         }
         Char | Byte | Str | RawStr | ByteStr | RawByteStr => unreachable!(),
     }
@@ -335,7 +348,7 @@ where
     // them in the range computation.
     while let Some(c) = chars.next() {
         let start = src.len() - chars.as_str().len() - c.len_utf8();
-        let res = match c {
+        let res: Result<T, EscapeError> = match c {
             '\\' => {
                 match chars.clone().next() {
                     Some('\n') => {

compiler/rustc_parse/messages.ftl

@@ -59,6 +59,8 @@ parse_bare_cr = {$double_quotes ->
 
 parse_bare_cr_in_raw_string = bare CR not allowed in raw string
 
+parse_binary_float_literal_not_supported = binary float literal is not supported
+
 parse_bounds_not_allowed_on_trait_aliases = bounds are not allowed on trait aliases
 
 parse_box_not_pat = expected pattern, found {$descr}
@@ -292,7 +294,11 @@ parse_generic_parameters_without_angle_brackets = generic parameters without sur
 parse_generics_in_path = unexpected generic arguments in path
 
 parse_help_set_edition_cargo = set `edition = "{$edition}"` in `Cargo.toml`
+
 parse_help_set_edition_standalone = pass `--edition {$edition}` to `rustc`
+
+parse_hexadecimal_float_literal_not_supported = hexadecimal float literal is not supported
+
 parse_if_expression_missing_condition = missing condition for `if` expression
     .condition_label = expected condition here
     .block_label = if this block is the condition of the `if` expression, then it must be followed by another block
@@ -364,6 +370,9 @@ parse_inner_doc_comment_not_permitted = expected outer doc comment
     .label_does_not_annotate_this = the inner doc comment doesn't annotate this {$item}
     .sugg_change_inner_to_outer = to annotate the {$item}, change the doc comment from inner to outer style
 
+parse_int_literal_too_large = integer literal is too large
+    .note = value exceeds limit of `{$limit}`
+
 parse_invalid_block_macro_segment = cannot use a `block` macro fragment here
     .label = the `block` fragment is within this context
     .suggestion = wrap this in another block
@@ -388,8 +397,18 @@ parse_invalid_dyn_keyword = invalid `dyn` keyword
     .suggestion = remove this keyword
 
 parse_invalid_expression_in_let_else = a `{$operator}` expression cannot be directly assigned in `let...else`
+parse_invalid_float_literal_suffix = invalid suffix `{$suffix}` for float literal
+    .label = invalid suffix `{$suffix}`
+    .help = valid suffixes are `f32` and `f64`
+
+parse_invalid_float_literal_width = invalid width `{$width}` for float literal
+    .help = valid widths are 32 and 64
+
 parse_invalid_identifier_with_leading_number = identifiers cannot start with a number
 
+parse_invalid_int_literal_width = invalid width `{$width}` for integer literal
+    .help = valid widths are 8, 16, 32, 64 and 128
+
 parse_invalid_interpolated_expression = invalid interpolated expression
 
 parse_invalid_literal_suffix = suffixes on {$kind} literals are invalid
@@ -408,6 +427,14 @@ parse_invalid_logical_operator = `{$incorrect}` is not a logical operator
 
 parse_invalid_meta_item = expected unsuffixed literal or identifier, found `{$token}`
 
+parse_invalid_num_literal_base_prefix = invalid base prefix for number literal
+    .note = base prefixes (`0xff`, `0b1010`, `0o755`) are lowercase
+    .suggestion = try making the prefix lowercase
+
+parse_invalid_num_literal_suffix = invalid suffix `{$suffix}` for number literal
+    .label = invalid suffix `{$suffix}`
+    .help = the suffix must be one of the numeric types (`u32`, `isize`, `f32`, etc.)
+
 parse_invalid_unicode_escape = invalid unicode character escape
     .label = invalid escape
     .help = unicode escape must {$surrogate ->
@@ -612,6 +639,10 @@ parse_note_mut_pattern_usage = `mut` may be followed by `variable` and `variable
 
 parse_note_pattern_alternatives_use_single_vert = alternatives in or-patterns are separated with `|`, not `||`
 
+parse_nul_in_c_str = null characters in C string literals are not supported
+
+parse_octal_float_literal_not_supported = octal float literal is not supported
+
 parse_or_pattern_not_allowed_in_fn_parameters = top-level or-patterns are not allowed in function parameters
 parse_or_pattern_not_allowed_in_let_binding = top-level or-patterns are not allowed in `let` bindings
 parse_out_of_range_hex_escape = out of range hex escape