Auto merge of #117565 - estebank:issue-100825, r=Nilstrieb

Tweak parsing recovery of enums, for exprs and match arm patterns

Tweak recovery of `for (pat in expr) {}` for more accurate spans.

When encountering `match` arm `(pat if expr) => {}`, recover and suggest removing parentheses. Fix #100825.

When encountering malformed enums, try more localized per-variant parse recovery.

Move parser recovery tests to subdirectory.

6 files changed, 233 insertions, 131 deletions

diff --git a/compiler/rustc_parse/src/errors.rs b/compiler/rustc_parse/src/errors.rs
index e5d4cb6f4da..03e047b297d 100644
--- a/compiler/rustc_parse/src/errors.rs
+++ b/compiler/rustc_parse/src/errors.rs
@@ -1275,12 +1275,28 @@ pub(crate) struct ParenthesesInForHead {
 #[derive(Subdiagnostic)]
 #[multipart_suggestion(parse_suggestion, applicability = "machine-applicable")]
 pub(crate) struct ParenthesesInForHeadSugg {
-    #[suggestion_part(code = "{left_snippet}")]
+    #[suggestion_part(code = " ")]
+    pub left: Span,
+    #[suggestion_part(code = " ")]
+    pub right: Span,
+}
+
+#[derive(Diagnostic)]
+#[diag(parse_unexpected_parentheses_in_match_arm_pattern)]
+pub(crate) struct ParenthesesInMatchPat {
+    #[primary_span]
+    pub span: Vec<Span>,
+    #[subdiagnostic]
+    pub sugg: ParenthesesInMatchPatSugg,
+}
+
+#[derive(Subdiagnostic)]
+#[multipart_suggestion(parse_suggestion, applicability = "machine-applicable")]
+pub(crate) struct ParenthesesInMatchPatSugg {
+    #[suggestion_part(code = "")]
     pub left: Span,
-    pub left_snippet: String,
-    #[suggestion_part(code = "{right_snippet}")]
+    #[suggestion_part(code = "")]
     pub right: Span,
-    pub right_snippet: String,
 }
 
 #[derive(Diagnostic)]
diff --git a/compiler/rustc_parse/src/parser/diagnostics.rs b/compiler/rustc_parse/src/parser/diagnostics.rs
index ecb840f067e..8921c1c6a03 100644
--- a/compiler/rustc_parse/src/parser/diagnostics.rs
+++ b/compiler/rustc_parse/src/parser/diagnostics.rs
@@ -11,12 +11,12 @@ use crate::errors::{
     DoubleColonInBound, ExpectedIdentifier, ExpectedSemi, ExpectedSemiSugg,
     GenericParamsWithoutAngleBrackets, GenericParamsWithoutAngleBracketsSugg,
     HelpIdentifierStartsWithNumber, InInTypo, IncorrectAwait, IncorrectSemicolon,
-    IncorrectUseOfAwait, ParenthesesInForHead, ParenthesesInForHeadSugg,
-    PatternMethodParamWithoutBody, QuestionMarkInType, QuestionMarkInTypeSugg, SelfParamNotFirst,
-    StructLiteralBodyWithoutPath, StructLiteralBodyWithoutPathSugg, StructLiteralNeedingParens,
-    StructLiteralNeedingParensSugg, SuggAddMissingLetStmt, SuggEscapeIdentifier, SuggRemoveComma,
-    TernaryOperator, UnexpectedConstInGenericParam, UnexpectedConstParamDeclaration,
-    UnexpectedConstParamDeclarationSugg, UnmatchedAngleBrackets, UseEqInstead, WrapType,
+    IncorrectUseOfAwait, PatternMethodParamWithoutBody, QuestionMarkInType, QuestionMarkInTypeSugg,
+    SelfParamNotFirst, StructLiteralBodyWithoutPath, StructLiteralBodyWithoutPathSugg,
+    StructLiteralNeedingParens, StructLiteralNeedingParensSugg, SuggAddMissingLetStmt,
+    SuggEscapeIdentifier, SuggRemoveComma, TernaryOperator, UnexpectedConstInGenericParam,
+    UnexpectedConstParamDeclaration, UnexpectedConstParamDeclarationSugg, UnmatchedAngleBrackets,
+    UseEqInstead, WrapType,
 };
 
 use crate::fluent_generated as fluent;
@@ -1994,56 +1994,6 @@ impl<'a> Parser<'a> {
         }
     }
 
-    /// Recovers a situation like `for ( $pat in $expr )`
-    /// and suggest writing `for $pat in $expr` instead.
-    ///
-    /// This should be called before parsing the `$block`.
-    pub(super) fn recover_parens_around_for_head(
-        &mut self,
-        pat: P<Pat>,
-        begin_paren: Option<Span>,
-    ) -> P<Pat> {
-        match (&self.token.kind, begin_paren) {
-            (token::CloseDelim(Delimiter::Parenthesis), Some(begin_par_sp)) => {
-                self.bump();
-
-                let sm = self.sess.source_map();
-                let left = begin_par_sp;
-                let right = self.prev_token.span;
-                let left_snippet = if let Ok(snip) = sm.span_to_prev_source(left)
-                    && !snip.ends_with(' ')
-                {
-                    " ".to_string()
-                } else {
-                    "".to_string()
-                };
-
-                let right_snippet = if let Ok(snip) = sm.span_to_next_source(right)
-                    && !snip.starts_with(' ')
-                {
-                    " ".to_string()
-                } else {
-                    "".to_string()
-                };
-
-                self.sess.emit_err(ParenthesesInForHead {
-                    span: vec![left, right],
-                    // With e.g. `for (x) in y)` this would replace `(x) in y)`
-                    // with `x) in y)` which is syntactically invalid.
-                    // However, this is prevented before we get here.
-                    sugg: ParenthesesInForHeadSugg { left, right, left_snippet, right_snippet },
-                });
-
-                // Unwrap `(pat)` into `pat` to avoid the `unused_parens` lint.
-                pat.and_then(|pat| match pat.kind {
-                    PatKind::Paren(pat) => pat,
-                    _ => P(pat),
-                })
-            }
-            _ => pat,
-        }
-    }
-
     pub(super) fn recover_seq_parse_error(
         &mut self,
         delim: Delimiter,
diff --git a/compiler/rustc_parse/src/parser/expr.rs b/compiler/rustc_parse/src/parser/expr.rs
index e334837c1d4..b1b77305e4f 100644
--- a/compiler/rustc_parse/src/parser/expr.rs
+++ b/compiler/rustc_parse/src/parser/expr.rs
@@ -10,7 +10,7 @@ use super::{
 use crate::errors;
 use crate::maybe_recover_from_interpolated_ty_qpath;
 use ast::mut_visit::{noop_visit_expr, MutVisitor};
-use ast::{GenBlockKind, Path, PathSegment};
+use ast::{GenBlockKind, Pat, Path, PathSegment};
 use core::mem;
 use rustc_ast::ptr::P;
 use rustc_ast::token::{self, Delimiter, Token, TokenKind};
@@ -2609,30 +2609,72 @@ impl<'a> Parser<'a> {
         }
     }
 
-    /// Parses `for <src_pat> in <src_expr> <src_loop_block>` (`for` token already eaten).
-    fn parse_expr_for(&mut self, opt_label: Option<Label>, lo: Span) -> PResult<'a, P<Expr>> {
-        // Record whether we are about to parse `for (`.
-        // This is used below for recovery in case of `for ( $stuff ) $block`
-        // in which case we will suggest `for $stuff $block`.
-        let begin_paren = match self.token.kind {
-            token::OpenDelim(Delimiter::Parenthesis) => Some(self.token.span),
-            _ => None,
+    fn parse_for_head(&mut self) -> PResult<'a, (P<Pat>, P<Expr>)> {
+        let begin_paren = if self.token.kind == token::OpenDelim(Delimiter::Parenthesis) {
+            // Record whether we are about to parse `for (`.
+            // This is used below for recovery in case of `for ( $stuff ) $block`
+            // in which case we will suggest `for $stuff $block`.
+            let start_span = self.token.span;
+            let left = self.prev_token.span.between(self.look_ahead(1, |t| t.span));
+            Some((start_span, left))
+        } else {
+            None
+        };
+        // Try to parse the pattern `for ($PAT) in $EXPR`.
+        let pat = match (
+            self.parse_pat_allow_top_alt(
+                None,
+                RecoverComma::Yes,
+                RecoverColon::Yes,
+                CommaRecoveryMode::LikelyTuple,
+            ),
+            begin_paren,
+        ) {
+            (Ok(pat), _) => pat, // Happy path.
+            (Err(err), Some((start_span, left))) if self.eat_keyword(kw::In) => {
+                // We know for sure we have seen `for ($SOMETHING in`. In the happy path this would
+                // happen right before the return of this method.
+                let expr = match self.parse_expr_res(Restrictions::NO_STRUCT_LITERAL, None) {
+                    Ok(expr) => expr,
+                    Err(expr_err) => {
+                        // We don't know what followed the `in`, so cancel and bubble up the
+                        // original error.
+                        expr_err.cancel();
+                        return Err(err);
+                    }
+                };
+                return if self.token.kind == token::CloseDelim(Delimiter::Parenthesis) {
+                    // We know for sure we have seen `for ($SOMETHING in $EXPR)`, so we recover the
+                    // parser state and emit a targetted suggestion.
+                    let span = vec![start_span, self.token.span];
+                    let right = self.prev_token.span.between(self.look_ahead(1, |t| t.span));
+                    self.bump(); // )
+                    err.cancel();
+                    self.sess.emit_err(errors::ParenthesesInForHead {
+                        span,
+                        // With e.g. `for (x) in y)` this would replace `(x) in y)`
+                        // with `x) in y)` which is syntactically invalid.
+                        // However, this is prevented before we get here.
+                        sugg: errors::ParenthesesInForHeadSugg { left, right },
+                    });
+                    Ok((self.mk_pat(start_span.to(right), ast::PatKind::Wild), expr))
+                } else {
+                    Err(err) // Some other error, bubble up.
+                };
+            }
+            (Err(err), _) => return Err(err), // Some other error, bubble up.
         };
-
-        let pat = self.parse_pat_allow_top_alt(
-            None,
-            RecoverComma::Yes,
-            RecoverColon::Yes,
-            CommaRecoveryMode::LikelyTuple,
-        )?;
         if !self.eat_keyword(kw::In) {
             self.error_missing_in_for_loop();
         }
         self.check_for_for_in_in_typo(self.prev_token.span);
         let expr = self.parse_expr_res(Restrictions::NO_STRUCT_LITERAL, None)?;
+        Ok((pat, expr))
+    }
 
-        let pat = self.recover_parens_around_for_head(pat, begin_paren);
-
+    /// Parses `for <src_pat> in <src_expr> <src_loop_block>` (`for` token already eaten).
+    fn parse_expr_for(&mut self, opt_label: Option<Label>, lo: Span) -> PResult<'a, P<Expr>> {
+        let (pat, expr) = self.parse_for_head()?;
         // Recover from missing expression in `for` loop
         if matches!(expr.kind, ExprKind::Block(..))
             && !matches!(self.token.kind, token::OpenDelim(Delimiter::Brace))
@@ -2853,47 +2895,10 @@ impl<'a> Parser<'a> {
     }
 
     pub(super) fn parse_arm(&mut self) -> PResult<'a, Arm> {
-        // Used to check the `let_chains` and `if_let_guard` features mostly by scanning
-        // `&&` tokens.
-        fn check_let_expr(expr: &Expr) -> (bool, bool) {
-            match &expr.kind {
-                ExprKind::Binary(BinOp { node: BinOpKind::And, .. }, lhs, rhs) => {
-                    let lhs_rslt = check_let_expr(lhs);
-                    let rhs_rslt = check_let_expr(rhs);
-                    (lhs_rslt.0 || rhs_rslt.0, false)
-                }
-                ExprKind::Let(..) => (true, true),
-                _ => (false, true),
-            }
-        }
         let attrs = self.parse_outer_attributes()?;
         self.collect_tokens_trailing_token(attrs, ForceCollect::No, |this, attrs| {
             let lo = this.token.span;
-            let pat = this.parse_pat_allow_top_alt(
-                None,
-                RecoverComma::Yes,
-                RecoverColon::Yes,
-                CommaRecoveryMode::EitherTupleOrPipe,
-            )?;
-            let guard = if this.eat_keyword(kw::If) {
-                let if_span = this.prev_token.span;
-                let mut cond = this.parse_match_guard_condition()?;
-
-                CondChecker::new(this).visit_expr(&mut cond);
-
-                let (has_let_expr, does_not_have_bin_op) = check_let_expr(&cond);
-                if has_let_expr {
-                    if does_not_have_bin_op {
-                        // Remove the last feature gating of a `let` expression since it's stable.
-                        this.sess.gated_spans.ungate_last(sym::let_chains, cond.span);
-                    }
-                    let span = if_span.to(cond.span);
-                    this.sess.gated_spans.gate(sym::if_let_guard, span);
-                }
-                Some(cond)
-            } else {
-                None
-            };
+            let (pat, guard) = this.parse_match_arm_pat_and_guard()?;
             let arrow_span = this.token.span;
             if let Err(mut err) = this.expect(&token::FatArrow) {
                 // We might have a `=>` -> `=` or `->` typo (issue #89396).
@@ -3023,6 +3028,90 @@ impl<'a> Parser<'a> {
         })
     }
 
+    fn parse_match_arm_guard(&mut self) -> PResult<'a, Option<P<Expr>>> {
+        // Used to check the `let_chains` and `if_let_guard` features mostly by scanning
+        // `&&` tokens.
+        fn check_let_expr(expr: &Expr) -> (bool, bool) {
+            match &expr.kind {
+                ExprKind::Binary(BinOp { node: BinOpKind::And, .. }, lhs, rhs) => {
+                    let lhs_rslt = check_let_expr(lhs);
+                    let rhs_rslt = check_let_expr(rhs);
+                    (lhs_rslt.0 || rhs_rslt.0, false)
+                }
+                ExprKind::Let(..) => (true, true),
+                _ => (false, true),
+            }
+        }
+        if !self.eat_keyword(kw::If) {
+            // No match arm guard present.
+            return Ok(None);
+        }
+
+        let if_span = self.prev_token.span;
+        let mut cond = self.parse_match_guard_condition()?;
+
+        CondChecker::new(self).visit_expr(&mut cond);
+
+        let (has_let_expr, does_not_have_bin_op) = check_let_expr(&cond);
+        if has_let_expr {
+            if does_not_have_bin_op {
+                // Remove the last feature gating of a `let` expression since it's stable.
+                self.sess.gated_spans.ungate_last(sym::let_chains, cond.span);
+            }
+            let span = if_span.to(cond.span);
+            self.sess.gated_spans.gate(sym::if_let_guard, span);
+        }
+        Ok(Some(cond))
+    }
+
+    fn parse_match_arm_pat_and_guard(&mut self) -> PResult<'a, (P<Pat>, Option<P<Expr>>)> {
+        if self.token.kind == token::OpenDelim(Delimiter::Parenthesis) {
+            // Detect and recover from `($pat if $cond) => $arm`.
+            let left = self.token.span;
+            match self.parse_pat_allow_top_alt(
+                None,
+                RecoverComma::Yes,
+                RecoverColon::Yes,
+                CommaRecoveryMode::EitherTupleOrPipe,
+            ) {
+                Ok(pat) => Ok((pat, self.parse_match_arm_guard()?)),
+                Err(err)
+                    if let prev_sp = self.prev_token.span
+                        && let true = self.eat_keyword(kw::If) =>
+                {
+                    // We know for certain we've found `($pat if` so far.
+                    let mut cond = match self.parse_match_guard_condition() {
+                        Ok(cond) => cond,
+                        Err(cond_err) => {
+                            cond_err.cancel();
+                            return Err(err);
+                        }
+                    };
+                    err.cancel();
+                    CondChecker::new(self).visit_expr(&mut cond);
+                    self.eat_to_tokens(&[&token::CloseDelim(Delimiter::Parenthesis)]);
+                    self.expect(&token::CloseDelim(Delimiter::Parenthesis))?;
+                    let right = self.prev_token.span;
+                    self.sess.emit_err(errors::ParenthesesInMatchPat {
+                        span: vec![left, right],
+                        sugg: errors::ParenthesesInMatchPatSugg { left, right },
+                    });
+                    Ok((self.mk_pat(left.to(prev_sp), ast::PatKind::Wild), Some(cond)))
+                }
+                Err(err) => Err(err),
+            }
+        } else {
+            // Regular parser flow:
+            let pat = self.parse_pat_allow_top_alt(
+                None,
+                RecoverComma::Yes,
+                RecoverColon::Yes,
+                CommaRecoveryMode::EitherTupleOrPipe,
+            )?;
+            Ok((pat, self.parse_match_arm_guard()?))
+        }
+    }
+
     fn parse_match_guard_condition(&mut self) -> PResult<'a, P<Expr>> {
         self.parse_expr_res(Restrictions::ALLOW_LET | Restrictions::IN_IF_GUARD, None).map_err(
             |mut err| {
diff --git a/compiler/rustc_parse/src/parser/item.rs b/compiler/rustc_parse/src/parser/item.rs
index 23886b1208b..ab18e40482e 100644
--- a/compiler/rustc_parse/src/parser/item.rs
+++ b/compiler/rustc_parse/src/parser/item.rs
@@ -1415,8 +1415,8 @@ impl<'a> Parser<'a> {
             self.bump();
             (thin_vec![], false)
         } else {
-            self.parse_delim_comma_seq(Delimiter::Brace, |p| p.parse_enum_variant()).map_err(
-                |mut err| {
+            self.parse_delim_comma_seq(Delimiter::Brace, |p| p.parse_enum_variant(id.span))
+                .map_err(|mut err| {
                     err.span_label(id.span, "while parsing this enum");
                     if self.token == token::Colon {
                         let snapshot = self.create_snapshot_for_diagnostic();
@@ -1436,20 +1436,22 @@ impl<'a> Parser<'a> {
                         }
                         self.restore_snapshot(snapshot);
                     }
-                    self.recover_stmt();
+                    self.eat_to_tokens(&[&token::CloseDelim(Delimiter::Brace)]);
+                    self.bump(); // }
                     err
-                },
-            )?
+                })?
         };
 
         let enum_definition = EnumDef { variants: variants.into_iter().flatten().collect() };
         Ok((id, ItemKind::Enum(enum_definition, generics)))
     }
 
-    fn parse_enum_variant(&mut self) -> PResult<'a, Option<Variant>> {
+    fn parse_enum_variant(&mut self, span: Span) -> PResult<'a, Option<Variant>> {
         self.recover_diff_marker();
         let variant_attrs = self.parse_outer_attributes()?;
         self.recover_diff_marker();
+        let help = "enum variants can be `Variant`, `Variant = <integer>`, \
+                    `Variant(Type, ..., TypeN)` or `Variant { fields: Types }`";
         self.collect_tokens_trailing_token(
             variant_attrs,
             ForceCollect::No,
@@ -1476,10 +1478,39 @@ impl<'a> Parser<'a> {
                 let struct_def = if this.check(&token::OpenDelim(Delimiter::Brace)) {
                     // Parse a struct variant.
                     let (fields, recovered) =
-                        this.parse_record_struct_body("struct", ident.span, false)?;
+                        match this.parse_record_struct_body("struct", ident.span, false) {
+                            Ok((fields, recovered)) => (fields, recovered),
+                            Err(mut err) => {
+                                if this.token == token::Colon {
+                                    // We handle `enum` to `struct` suggestion in the caller.
+                                    return Err(err);
+                                }
+                                this.eat_to_tokens(&[&token::CloseDelim(Delimiter::Brace)]);
+                                this.bump(); // }
+                                err.span_label(span, "while parsing this enum");
+                                err.help(help);
+                                err.emit();
+                                (thin_vec![], true)
+                            }
+                        };
                     VariantData::Struct(fields, recovered)
                 } else if this.check(&token::OpenDelim(Delimiter::Parenthesis)) {
-                    VariantData::Tuple(this.parse_tuple_struct_body()?, DUMMY_NODE_ID)
+                    let body = match this.parse_tuple_struct_body() {
+                        Ok(body) => body,
+                        Err(mut err) => {
+                            if this.token == token::Colon {
+                                // We handle `enum` to `struct` suggestion in the caller.
+                                return Err(err);
+                            }
+                            this.eat_to_tokens(&[&token::CloseDelim(Delimiter::Parenthesis)]);
+                            this.bump(); // )
+                            err.span_label(span, "while parsing this enum");
+                            err.help(help);
+                            err.emit();
+                            thin_vec![]
+                        }
+                    };
+                    VariantData::Tuple(body, DUMMY_NODE_ID)
                 } else {
                     VariantData::Unit(DUMMY_NODE_ID)
                 };
@@ -1500,8 +1531,9 @@ impl<'a> Parser<'a> {
 
                 Ok((Some(vr), TrailingToken::MaybeComma))
             },
-        ).map_err(|mut err| {
-            err.help("enum variants can be `Variant`, `Variant = <integer>`, `Variant(Type, ..., TypeN)` or `Variant { fields: Types }`");
+        )
+        .map_err(|mut err| {
+            err.help(help);
             err
         })
     }
diff --git a/compiler/rustc_parse/src/parser/mod.rs b/compiler/rustc_parse/src/parser/mod.rs
index 9bd436f01ac..c680a950584 100644
--- a/compiler/rustc_parse/src/parser/mod.rs
+++ b/compiler/rustc_parse/src/parser/mod.rs
@@ -875,6 +875,9 @@ impl<'a> Parser<'a> {
                                     if self.token == token::Colon {
                                         // we will try to recover in `maybe_recover_struct_lit_bad_delims`
                                         return Err(expect_err);
+                                    } else if let [token::CloseDelim(Delimiter::Parenthesis)] = kets
+                                    {
+                                        return Err(expect_err);
                                     } else {
                                         expect_err.emit();
                                         break;
diff --git a/compiler/rustc_parse/src/parser/pat.rs b/compiler/rustc_parse/src/parser/pat.rs
index b05868e235a..3d1d1ec8108 100644
--- a/compiler/rustc_parse/src/parser/pat.rs
+++ b/compiler/rustc_parse/src/parser/pat.rs
@@ -141,7 +141,19 @@ impl<'a> Parser<'a> {
         };
 
         // Parse the first pattern (`p_0`).
-        let mut first_pat = self.parse_pat_no_top_alt(expected, syntax_loc)?;
+        let mut first_pat = match self.parse_pat_no_top_alt(expected, syntax_loc) {
+            Ok(pat) => pat,
+            Err(mut err)
+                if self.token.is_reserved_ident()
+                    && !self.token.is_keyword(kw::In)
+                    && !self.token.is_keyword(kw::If) =>
+            {
+                err.emit();
+                self.bump();
+                self.mk_pat(self.token.span, PatKind::Wild)
+            }
+            Err(err) => return Err(err),
+        };
         if rc == RecoverComma::Yes {
             self.maybe_recover_unexpected_comma(
                 first_pat.span,
@@ -379,10 +391,10 @@ impl<'a> Parser<'a> {
             self.parse_pat_ident_mut(syntax_loc)?
         } else if self.eat_keyword(kw::Ref) {
             if self.check_keyword(kw::Box) {
-                // Suggest `box ref` and quit parsing pattern to prevent series of
-                // misguided diagnostics from later stages of the compiler.
+                // Suggest `box ref`.
                 let span = self.prev_token.span.to(self.token.span);
-                return Err(self.sess.create_err(SwitchRefBoxOrder { span }));
+                self.bump();
+                self.sess.emit_err(SwitchRefBoxOrder { span });
             }
             // Parse ref ident @ pat / ref mut ident @ pat
             let mutbl = self.parse_mutability();
@@ -839,7 +851,7 @@ impl<'a> Parser<'a> {
         binding_annotation: BindingAnnotation,
         syntax_loc: Option<PatternLocation>,
     ) -> PResult<'a, PatKind> {
-        let ident = self.parse_ident()?;
+        let ident = self.parse_ident_common(false)?;
 
         if self.may_recover()
             && !matches!(syntax_loc, Some(PatternLocation::FunctionParameter))