Auto merge of #151247 - chenyukang:yukang-fix-const-recover-151149, r=estebank

Try to recover from over-parsing in const item with missing semicolon

Fixes #151149

r? @estebank

Author: bors

compiler/rustc_parse/src/errors.rs

@@ -1514,7 +1514,6 @@ pub(crate) struct HelpIdentifierStartsWithNumber {
 pub(crate) struct ExpectedSemi {
     pub span: Span,
     pub token: Token,
-
     pub unexpected_token_label: Option<Span>,
     pub sugg: ExpectedSemiSugg,
 }

compiler/rustc_parse/src/parser/item.rs

@@ -296,7 +296,7 @@ impl<'a> Parser<'a> {
             // CONST ITEM
             self.recover_const_mut(const_span);
             self.recover_missing_kw_before_item()?;
-            let (ident, generics, ty, rhs_kind) = self.parse_const_item(false)?;
+            let (ident, generics, ty, rhs_kind) = self.parse_const_item(false, const_span)?;
             ItemKind::Const(Box::new(ConstItem {
                 defaultness: def_(),
                 ident,
@@ -317,7 +317,7 @@ impl<'a> Parser<'a> {
                 // TYPE CONST (mgca)
                 self.recover_const_mut(const_span);
                 self.recover_missing_kw_before_item()?;
-                let (ident, generics, ty, rhs_kind) = self.parse_const_item(true)?;
+                let (ident, generics, ty, rhs_kind) = self.parse_const_item(true, const_span)?;
                 // Make sure this is only allowed if the feature gate is enabled.
                 // #![feature(mgca_type_const_syntax)]
                 self.psess.gated_spans.gate(sym::mgca_type_const_syntax, lo.to(const_span));
@@ -1558,6 +1558,7 @@ impl<'a> Parser<'a> {
     fn parse_const_item(
         &mut self,
         const_arg: bool,
+        const_span: Span,
     ) -> PResult<'a, (Ident, Generics, Box<Ty>, ConstItemRhsKind)> {
         let ident = self.parse_ident_or_underscore()?;
 
@@ -1647,6 +1648,9 @@ impl<'a> Parser<'a> {
 
         generics.where_clause = where_clause;
 
+        if let Some(rhs) = self.try_recover_const_missing_semi(&rhs, const_span) {
+            return Ok((ident, generics, ty, ConstItemRhsKind::Body { rhs: Some(rhs) }));
+        }
         self.expect_semi()?;
 
         Ok((ident, generics, ty, rhs))
@@ -2744,8 +2748,21 @@ impl<'a> Parser<'a> {
             *sig_hi = self.prev_token.span;
             (AttrVec::new(), None)
         } else if self.check(exp!(OpenBrace)) || self.token.is_metavar_block() {
-            self.parse_block_common(self.token.span, BlockCheckMode::Default, None)
-                .map(|(attrs, body)| (attrs, Some(body)))?
+            let prev_in_fn_body = self.in_fn_body;
+            self.in_fn_body = true;
+            let res = self.parse_block_common(self.token.span, BlockCheckMode::Default, None).map(
+                |(attrs, mut body)| {
+                    if let Some(guar) = self.fn_body_missing_semi_guar.take() {
+                        body.stmts.push(self.mk_stmt(
+                            body.span,
+                            StmtKind::Expr(self.mk_expr(body.span, ExprKind::Err(guar))),
+                        ));
+                    }
+                    (attrs, Some(body))
+                },
+            );
+            self.in_fn_body = prev_in_fn_body;
+            res?
         } else if self.token == token::Eq {
             // Recover `fn foo() = $expr;`.
             self.bump(); // `=`
@@ -3501,6 +3518,37 @@ impl<'a> Parser<'a> {
             Ok(Some(_))
         )
     }
+
+    /// Try to recover from over-parsing in const item when a semicolon is missing.
+    ///
+    /// This detects cases where we parsed too much because a semicolon was missing
+    /// and the next line started an expression that the parser treated as a continuation
+    /// (e.g., `foo() \n &bar` was parsed as `foo() & bar`).
+    ///
+    /// Returns a corrected expression if recovery is successful.
+    fn try_recover_const_missing_semi(
+        &mut self,
+        rhs: &ConstItemRhsKind,
+        const_span: Span,
+    ) -> Option<Box<Expr>> {
+        if self.token == TokenKind::Semi {
+            return None;
+        }
+        let ConstItemRhsKind::Body { rhs: Some(rhs) } = rhs else {
+            return None;
+        };
+        if !self.in_fn_body || !self.may_recover() || rhs.span.from_expansion() {
+            return None;
+        }
+        if let Some((span, guar)) =
+            self.missing_semi_from_binop("const", rhs, Some(const_span.shrink_to_lo()))
+        {
+            self.fn_body_missing_semi_guar = Some(guar);
+            Some(self.mk_expr(span, ExprKind::Err(guar)))
+        } else {
+            None
+        }
+    }
 }
 
 enum IsMacroRulesItem {

compiler/rustc_parse/src/parser/mod.rs

@@ -32,18 +32,19 @@ use rustc_ast::tokenstream::{
     ParserRange, ParserReplacement, Spacing, TokenCursor, TokenStream, TokenTree, TokenTreeCursor,
 };
 use rustc_ast::util::case::Case;
+use rustc_ast::util::classify;
 use rustc_ast::{
-    self as ast, AnonConst, AttrArgs, AttrId, ByRef, Const, CoroutineKind, DUMMY_NODE_ID,
-    DelimArgs, Expr, ExprKind, Extern, HasAttrs, HasTokens, MgcaDisambiguation, Mutability,
-    Recovered, Safety, StrLit, Visibility, VisibilityKind,
+    self as ast, AnonConst, AttrArgs, AttrId, BinOpKind, ByRef, Const, CoroutineKind,
+    DUMMY_NODE_ID, DelimArgs, Expr, ExprKind, Extern, HasAttrs, HasTokens, MgcaDisambiguation,
+    Mutability, Recovered, Safety, StrLit, Visibility, VisibilityKind,
 };
 use rustc_ast_pretty::pprust;
 use rustc_data_structures::debug_assert_matches;
 use rustc_data_structures::fx::FxHashMap;
 use rustc_errors::{Applicability, Diag, FatalError, MultiSpan, PResult};
 use rustc_index::interval::IntervalSet;
 use rustc_session::parse::ParseSess;
-use rustc_span::{Ident, Span, Symbol, kw, sym};
+use rustc_span::{ErrorGuaranteed, Ident, Span, Symbol, kw, sym};
 use thin_vec::ThinVec;
 use token_type::TokenTypeSet;
 pub use token_type::{ExpKeywordPair, ExpTokenPair, TokenType};
@@ -232,6 +233,10 @@ pub struct Parser<'a> {
     /// Whether the parser is allowed to do recovery.
     /// This is disabled when parsing macro arguments, see #103534
     recovery: Recovery = Recovery::Allowed,
+    /// Whether we're parsing a function body.
+    in_fn_body: bool = false,
+    /// Whether we have detected a missing semicolon in function body.
+    pub fn_body_missing_semi_guar: Option<ErrorGuaranteed> = None,
 }
 
 // This type is used a lot, e.g. it's cloned when matching many declarative macro rules with
@@ -1649,6 +1654,65 @@ impl<'a> Parser<'a> {
             _ => self.prev_token.span,
         }
     }
+
+    fn missing_semi_from_binop(
+        &self,
+        kind_desc: &str,
+        expr: &Expr,
+        decl_lo: Option<Span>,
+    ) -> Option<(Span, ErrorGuaranteed)> {
+        if self.token == TokenKind::Semi {
+            return None;
+        }
+        if !self.may_recover() || expr.span.from_expansion() {
+            return None;
+        }
+        let sm = self.psess.source_map();
+        if let ExprKind::Binary(op, lhs, rhs) = &expr.kind
+            && sm.is_multiline(lhs.span.shrink_to_hi().until(rhs.span.shrink_to_lo()))
+            && matches!(op.node, BinOpKind::Mul | BinOpKind::BitAnd)
+            && classify::expr_requires_semi_to_be_stmt(rhs)
+        {
+            let lhs_end_span = lhs.span.shrink_to_hi();
+            let token_str = token_descr(&self.token);
+            let mut err = self
+                .dcx()
+                .struct_span_err(lhs_end_span, format!("expected `;`, found {token_str}"));
+            err.span_label(self.token.span, "unexpected token");
+
+            // Use the declaration start if provided, otherwise fall back to lhs_end_span.
+            let continuation_start = decl_lo.unwrap_or(lhs_end_span);
+            let continuation_span = continuation_start.until(rhs.span.shrink_to_hi());
+            err.span_label(
+                continuation_span,
+                format!(
+                    "to finish parsing this {kind_desc}, expected this to be followed by a `;`",
+                ),
+            );
+            let op_desc = match op.node {
+                BinOpKind::BitAnd => "a bit-and",
+                BinOpKind::Mul => "a multiplication",
+                _ => "a binary",
+            };
+            let mut note_spans = MultiSpan::new();
+            note_spans.push_span_label(lhs.span, "parsed as the left-hand expression");
+            note_spans.push_span_label(rhs.span, "parsed as the right-hand expression");
+            note_spans.push_span_label(op.span, format!("this was parsed as {op_desc}"));
+            err.span_note(
+                note_spans,
+                format!("the {kind_desc} was parsed as having {op_desc} binary expression"),
+            );
+
+            err.span_suggestion(
+                lhs_end_span,
+                format!("you may have meant to write a `;` to terminate the {kind_desc} earlier"),
+                ";",
+                Applicability::MaybeIncorrect,
+            );
+            return Some((lhs.span, err.emit()));
+        }
+        None
+    }
 }
 
 // Metavar captures of various kinds.

compiler/rustc_parse/src/parser/stmt.rs

@@ -924,6 +924,21 @@ impl<'a> Parser<'a> {
         }
     }
 
+    fn try_recover_let_missing_semi(&mut self, local: &mut Local) -> Option<ErrorGuaranteed> {
+        let expr = match &mut local.kind {
+            LocalKind::Init(expr) | LocalKind::InitElse(expr, _) => expr,
+            LocalKind::Decl => return None,
+        };
+        if let Some((span, guar)) =
+            self.missing_semi_from_binop("`let` binding", expr, Some(local.span.shrink_to_lo()))
+        {
+            self.fn_body_missing_semi_guar = Some(guar);
+            *expr = self.mk_expr(span, ExprKind::Err(guar));
+            return Some(guar);
+        }
+        None
+    }
+
     /// Parses a statement, including the trailing semicolon.
     pub fn parse_full_stmt(
         &mut self,
@@ -1066,71 +1081,74 @@ impl<'a> Parser<'a> {
                 }
             }
             StmtKind::Expr(_) | StmtKind::MacCall(_) => {}
-            StmtKind::Let(local) if let Err(mut e) = self.expect_semi() => {
-                // We might be at the `,` in `let x = foo<bar, baz>;`. Try to recover.
-                match &mut local.kind {
-                    LocalKind::Init(expr) | LocalKind::InitElse(expr, _) => {
-                        self.check_mistyped_turbofish_with_multiple_type_params(e, expr).map_err(
-                            |mut e| {
-                                self.recover_missing_dot(&mut e);
-                                self.recover_missing_let_else(&mut e, &local.pat, stmt.span);
-                                e
-                            },
-                        )?;
-                        // We found `foo<bar, baz>`, have we fully recovered?
-                        self.expect_semi()?;
-                    }
-                    LocalKind::Decl => {
-                        if let Some(colon_sp) = local.colon_sp {
-                            e.span_label(
-                                colon_sp,
-                                format!(
-                                    "while parsing the type for {}",
-                                    local.pat.descr().map_or_else(
-                                        || "the binding".to_string(),
-                                        |n| format!("`{n}`")
-                                    )
-                                ),
-                            );
-                            let suggest_eq = if self.token == token::Dot
-                                && let _ = self.bump()
-                                && let mut snapshot = self.create_snapshot_for_diagnostic()
-                                && let Ok(_) = snapshot
-                                    .parse_dot_suffix_expr(
-                                        colon_sp,
-                                        self.mk_expr_err(
-                                            colon_sp,
-                                            self.dcx()
-                                                .delayed_bug("error during `:` -> `=` recovery"),
-                                        ),
-                                    )
-                                    .map_err(Diag::cancel)
-                            {
-                                true
-                            } else if let Some(op) = self.check_assoc_op()
-                                && op.node.can_continue_expr_unambiguously()
-                            {
-                                true
-                            } else {
-                                false
-                            };
-                            if suggest_eq {
-                                e.span_suggestion_short(
+            StmtKind::Let(local) => {
+                if self.try_recover_let_missing_semi(local).is_some() {
+                    return Ok(Some(stmt));
+                }
+                if let Err(mut e) = self.expect_semi() {
+                    // We might be at the `,` in `let x = foo<bar, baz>;`. Try to recover.
+                    match &mut local.kind {
+                        LocalKind::Init(expr) | LocalKind::InitElse(expr, _) => {
+                            self.check_mistyped_turbofish_with_multiple_type_params(e, expr)
+                                .map_err(|mut e| {
+                                    self.recover_missing_dot(&mut e);
+                                    self.recover_missing_let_else(&mut e, &local.pat, stmt.span);
+                                    e
+                                })?;
+                            // We found `foo<bar, baz>`, have we fully recovered?
+                            self.expect_semi()?;
+                        }
+                        LocalKind::Decl => {
+                            if let Some(colon_sp) = local.colon_sp {
+                                e.span_label(
                                     colon_sp,
-                                    "use `=` if you meant to assign",
-                                    "=",
-                                    Applicability::MaybeIncorrect,
+                                    format!(
+                                        "while parsing the type for {}",
+                                        local.pat.descr().map_or_else(
+                                            || "the binding".to_string(),
+                                            |n| format!("`{n}`")
+                                        )
+                                    ),
                                 );
+                                let suggest_eq = if self.token == token::Dot
+                                    && let _ = self.bump()
+                                    && let mut snapshot = self.create_snapshot_for_diagnostic()
+                                    && let Ok(_) = snapshot
+                                        .parse_dot_suffix_expr(
+                                            colon_sp,
+                                            self.mk_expr_err(
+                                                colon_sp,
+                                                self.dcx().delayed_bug(
+                                                    "error during `:` -> `=` recovery",
+                                                ),
+                                            ),
+                                        )
+                                        .map_err(Diag::cancel)
+                                {
+                                    true
+                                } else if let Some(op) = self.check_assoc_op()
+                                    && op.node.can_continue_expr_unambiguously()
+                                {
+                                    true
+                                } else {
+                                    false
+                                };
+                                if suggest_eq {
+                                    e.span_suggestion_short(
+                                        colon_sp,
+                                        "use `=` if you meant to assign",
+                                        "=",
+                                        Applicability::MaybeIncorrect,
+                                    );
+                                }
                             }
+                            return Err(e);
                         }
-                        return Err(e);
                     }
                 }
                 eat_semi = false;
             }
-            StmtKind::Empty | StmtKind::Item(_) | StmtKind::Let(_) | StmtKind::Semi(_) => {
-                eat_semi = false
-            }
+            StmtKind::Empty | StmtKind::Item(_) | StmtKind::Semi(_) => eat_semi = false,
         }
 
         if add_semi_to_stmt || (eat_semi && self.eat(exp!(Semi))) {

tests/ui/parser/const-recover-semi-issue-151149.fixed

@@ -0,0 +1,36 @@
+//@ run-rustfix
+#![feature(const_trait_impl)]
+#![allow(dead_code)]
+#![allow(unused)]
+
+const trait ConstDefault {
+    fn const_default() -> Self;
+}
+
+impl const ConstDefault for u8 {
+    fn const_default() -> Self { 0 }
+}
+
+const fn val() -> u8 {
+    42
+}
+
+const C: u8 = u8::const_default()
+&1; //~ ERROR expected `;`, found keyword `const`
+
+const fn foo() -> &'static u8 {
+    const C: u8 = u8::const_default(); //~ ERROR expected `;`
+    &C
+}
+
+const fn bar() {
+    const C: u8 = 1
+     + 2; //~ ERROR expected `;`, found `}`
+}
+
+const fn baz() {
+    const C: u8 = 1
+     + val(); //~ ERROR expected `;`, found `}`
+}
+
+fn main() {}