Rollup merge of rust-lang#106242 - estebank:diff-markers, r=jyn514

Detect diff markers in the parser

Partly address rust-lang#32059.

compiler/rustc_parse/src/parser/diagnostics.rs

@@ -31,7 +31,8 @@ use rustc_ast::{
 use rustc_ast_pretty::pprust;
 use rustc_data_structures::fx::FxHashSet;
 use rustc_errors::{
-    fluent, Applicability, DiagnosticBuilder, DiagnosticMessage, Handler, MultiSpan, PResult,
+    fluent, Applicability, DiagnosticBuilder, DiagnosticMessage, FatalError, Handler, MultiSpan,
+    PResult,
 };
 use rustc_errors::{pluralize, Diagnostic, ErrorGuaranteed, IntoDiagnostic};
 use rustc_session::errors::ExprParenthesesNeeded;
@@ -2556,6 +2557,75 @@ impl<'a> Parser<'a> {
         Ok(())
     }
 
+    pub fn is_diff_marker(&mut self, long_kind: &TokenKind, short_kind: &TokenKind) -> bool {
+        (0..3).all(|i| self.look_ahead(i, |tok| tok == long_kind))
+            && self.look_ahead(3, |tok| tok == short_kind)
+    }
+
+    fn diff_marker(&mut self, long_kind: &TokenKind, short_kind: &TokenKind) -> Option<Span> {
+        if self.is_diff_marker(long_kind, short_kind) {
+            let lo = self.token.span;
+            for _ in 0..4 {
+                self.bump();
+            }
+            return Some(lo.to(self.prev_token.span));
+        }
+        None
+    }
+
+    pub fn recover_diff_marker(&mut self) {
+        let Some(start) = self.diff_marker(&TokenKind::BinOp(token::Shl), &TokenKind::Lt) else {
+            return;
+        };
+        let mut spans = Vec::with_capacity(3);
+        spans.push(start);
+        let mut middlediff3 = None;
+        let mut middle = None;
+        let mut end = None;
+        loop {
+            if self.token.kind == TokenKind::Eof {
+                break;
+            }
+            if let Some(span) = self.diff_marker(&TokenKind::OrOr, &TokenKind::BinOp(token::Or)) {
+                middlediff3 = Some(span);
+            }
+            if let Some(span) = self.diff_marker(&TokenKind::EqEq, &TokenKind::Eq) {
+                middle = Some(span);
+            }
+            if let Some(span) = self.diff_marker(&TokenKind::BinOp(token::Shr), &TokenKind::Gt) {
+                spans.push(span);
+                end = Some(span);
+                break;
+            }
+            self.bump();
+        }
+        let mut err = self.struct_span_err(spans, "encountered diff marker");
+        err.span_label(start, "after this is the code before the merge");
+        if let Some(middle) = middlediff3 {
+            err.span_label(middle, "");
+        }
+        if let Some(middle) = middle {
+            err.span_label(middle, "");
+        }
+        if let Some(end) = end {
+            err.span_label(end, "above this are the incoming code changes");
+        }
+        err.help(
+            "if you're having merge conflicts after pulling new code, the top section is the code \
+             you already had and the bottom section is the remote code",
+        );
+        err.help(
+            "if you're in the middle of a rebase, the top section is the code being rebased onto \
+             and the bottom section is the code coming from the current commit being rebased",
+        );
+        err.note(
+            "for an explanation on these markers from the `git` documentation, visit \
+             <https://git-scm.com/book/en/v2/Git-Tools-Advanced-Merging#_checking_out_conflicts>",
+        );
+        err.emit();
+        FatalError.raise()
+    }
+
     /// Parse and throw away a parenthesized comma separated
     /// sequence of patterns until `)` is reached.
     fn skip_pat_list(&mut self) -> PResult<'a, ()> {

compiler/rustc_parse/src/parser/expr.rs

@@ -3039,6 +3039,7 @@ impl<'a> Parser<'a> {
     /// Parses `ident (COLON expr)?`.
     fn parse_expr_field(&mut self) -> PResult<'a, ExprField> {
         let attrs = self.parse_outer_attributes()?;
+        self.recover_diff_marker();
         self.collect_tokens_trailing_token(attrs, ForceCollect::No, |this, attrs| {
             let lo = this.token.span;
 

compiler/rustc_parse/src/parser/item.rs

@@ -98,7 +98,9 @@ impl<'a> Parser<'a> {
         fn_parse_mode: FnParseMode,
         force_collect: ForceCollect,
     ) -> PResult<'a, Option<Item>> {
+        self.recover_diff_marker();
         let attrs = self.parse_outer_attributes()?;
+        self.recover_diff_marker();
         self.parse_item_common(attrs, true, false, fn_parse_mode, force_collect)
     }
 
@@ -704,6 +706,7 @@ impl<'a> Parser<'a> {
             if self.recover_doc_comment_before_brace() {
                 continue;
             }
+            self.recover_diff_marker();
             match parse_item(self) {
                 Ok(None) => {
                     let mut is_unnecessary_semicolon = !items.is_empty()
@@ -1039,8 +1042,11 @@ impl<'a> Parser<'a> {
     /// USE_TREE_LIST = Ø | (USE_TREE `,`)* USE_TREE [`,`]
     /// ```
     fn parse_use_tree_list(&mut self) -> PResult<'a, Vec<(UseTree, ast::NodeId)>> {
-        self.parse_delim_comma_seq(Delimiter::Brace, |p| Ok((p.parse_use_tree()?, DUMMY_NODE_ID)))
-            .map(|(r, _)| r)
+        self.parse_delim_comma_seq(Delimiter::Brace, |p| {
+            p.recover_diff_marker();
+            Ok((p.parse_use_tree()?, DUMMY_NODE_ID))
+        })
+        .map(|(r, _)| r)
     }
 
     fn parse_rename(&mut self) -> PResult<'a, Option<Ident>> {
@@ -1379,7 +1385,9 @@ impl<'a> Parser<'a> {
     }
 
     fn parse_enum_variant(&mut self) -> PResult<'a, Option<Variant>> {
+        self.recover_diff_marker();
         let variant_attrs = self.parse_outer_attributes()?;
+        self.recover_diff_marker();
         self.collect_tokens_trailing_token(
             variant_attrs,
             ForceCollect::No,
@@ -1573,9 +1581,32 @@ impl<'a> Parser<'a> {
         self.parse_paren_comma_seq(|p| {
             let attrs = p.parse_outer_attributes()?;
             p.collect_tokens_trailing_token(attrs, ForceCollect::No, |p, attrs| {
+                let mut snapshot = None;
+                if p.is_diff_marker(&TokenKind::BinOp(token::Shl), &TokenKind::Lt) {
+                    // Account for `<<<<<<<` diff markers. We can't proactively error here because
+                    // that can be a valid type start, so we snapshot and reparse only we've
+                    // encountered another parse error.
+                    snapshot = Some(p.create_snapshot_for_diagnostic());
+                }
                 let lo = p.token.span;
-                let vis = p.parse_visibility(FollowedByType::Yes)?;
-                let ty = p.parse_ty()?;
+                let vis = match p.parse_visibility(FollowedByType::Yes) {
+                    Ok(vis) => vis,
+                    Err(err) => {
+                        if let Some(ref mut snapshot) = snapshot {
+                            snapshot.recover_diff_marker();
+                        }
+                        return Err(err);
+                    }
+                };
+                let ty = match p.parse_ty() {
+                    Ok(ty) => ty,
+                    Err(err) => {
+                        if let Some(ref mut snapshot) = snapshot {
+                            snapshot.recover_diff_marker();
+                        }
+                        return Err(err);
+                    }
+                };
 
                 Ok((
                     FieldDef {
@@ -1596,7 +1627,9 @@ impl<'a> Parser<'a> {
 
     /// Parses an element of a struct declaration.
     fn parse_field_def(&mut self, adt_ty: &str) -> PResult<'a, FieldDef> {
+        self.recover_diff_marker();
         let attrs = self.parse_outer_attributes()?;
+        self.recover_diff_marker();
         self.collect_tokens_trailing_token(attrs, ForceCollect::No, |this, attrs| {
             let lo = this.token.span;
             let vis = this.parse_visibility(FollowedByType::No)?;
@@ -2427,6 +2460,7 @@ impl<'a> Parser<'a> {
         let mut first_param = true;
         // Parse the arguments, starting out with `self` being allowed...
         let (mut params, _) = self.parse_paren_comma_seq(|p| {
+            p.recover_diff_marker();
             let param = p.parse_param_general(req_name, first_param).or_else(|mut e| {
                 e.emit();
                 let lo = p.prev_token.span;

compiler/rustc_parse/src/parser/stmt.rs

@@ -531,13 +531,23 @@ impl<'a> Parser<'a> {
         recover: AttemptLocalParseRecovery,
     ) -> PResult<'a, P<Block>> {
         let mut stmts = vec![];
+        let mut snapshot = None;
         while !self.eat(&token::CloseDelim(Delimiter::Brace)) {
             if self.token == token::Eof {
                 break;
             }
+            if self.is_diff_marker(&TokenKind::BinOp(token::Shl), &TokenKind::Lt) {
+                // Account for `<<<<<<<` diff markers. We can't proactively error here because
+                // that can be a valid path start, so we snapshot and reparse only we've
+                // encountered another parse error.
+                snapshot = Some(self.create_snapshot_for_diagnostic());
+            }
             let stmt = match self.parse_full_stmt(recover) {
                 Err(mut err) if recover.yes() => {
                     self.maybe_annotate_with_ascription(&mut err, false);
+                    if let Some(ref mut snapshot) = snapshot {
+                        snapshot.recover_diff_marker();
+                    }
                     err.emit();
                     self.recover_stmt_(SemiColonMode::Ignore, BlockMode::Ignore);
                     Some(self.mk_stmt_err(self.token.span))

src/test/ui/parser/diff-markers/enum-2.rs

@@ -0,0 +1,11 @@
+enum E {
+    Foo {
+<<<<<<< HEAD //~ ERROR encountered diff marker
+        x: u8,
+|||||||
+        z: (),
+=======
+        y: i8,
+>>>>>>> branch
+    }
+}

src/test/ui/parser/diff-markers/enum-2.stderr

@@ -0,0 +1,21 @@
+error: encountered diff marker
+  --> $DIR/enum-2.rs:3:1
+   |
+LL | <<<<<<< HEAD
+   | ^^^^^^^ after this is the code before the merge
+LL |         x: u8,
+LL | |||||||
+   | -------
+LL |         z: (),
+LL | =======
+   | -------
+LL |         y: i8,
+LL | >>>>>>> branch
+   | ^^^^^^^ above this are the incoming code changes
+   |
+   = help: if you're having merge conflicts after pulling new code, the top section is the code you already had and the bottom section is the remote code
+   = help: if you're in the middle of a rebase, the top section is the code being rebased onto and the bottom section is the code coming from the current commit being rebased
+   = note: for an explanation on these markers from the `git` documentation, visit <https://git-scm.com/book/en/v2/Git-Tools-Advanced-Merging#_checking_out_conflicts>
+
+error: aborting due to previous error
+

src/test/ui/parser/diff-markers/enum.rs

@@ -0,0 +1,7 @@
+enum E {
+<<<<<<< HEAD //~ ERROR encountered diff marker
+    Foo(u8),
+=======
+    Bar(i8),
+>>>>>>> branch
+}

src/test/ui/parser/diff-markers/enum.stderr

@@ -0,0 +1,18 @@
+error: encountered diff marker
+  --> $DIR/enum.rs:2:1
+   |
+LL | <<<<<<< HEAD
+   | ^^^^^^^ after this is the code before the merge
+LL |     Foo(u8),
+LL | =======
+   | -------
+LL |     Bar(i8),
+LL | >>>>>>> branch
+   | ^^^^^^^ above this are the incoming code changes
+   |
+   = help: if you're having merge conflicts after pulling new code, the top section is the code you already had and the bottom section is the remote code
+   = help: if you're in the middle of a rebase, the top section is the code being rebased onto and the bottom section is the code coming from the current commit being rebased
+   = note: for an explanation on these markers from the `git` documentation, visit <https://git-scm.com/book/en/v2/Git-Tools-Advanced-Merging#_checking_out_conflicts>
+
+error: aborting due to previous error
+

src/test/ui/parser/diff-markers/fn-arg.rs

@@ -0,0 +1,16 @@
+trait T {
+    fn foo(
+<<<<<<< HEAD //~ ERROR encountered diff marker
+        x: u8,
+=======
+        x: i8,
+>>>>>>> branch
+    ) {}
+}
+
+struct S;
+impl T for S {}
+
+fn main() {
+    S::foo(42);
+}

src/test/ui/parser/diff-markers/fn-arg.stderr

@@ -0,0 +1,18 @@
+error: encountered diff marker
+  --> $DIR/fn-arg.rs:3:1
+   |
+LL | <<<<<<< HEAD
+   | ^^^^^^^ after this is the code before the merge
+LL |         x: u8,
+LL | =======
+   | -------
+LL |         x: i8,
+LL | >>>>>>> branch
+   | ^^^^^^^ above this are the incoming code changes
+   |
+   = help: if you're having merge conflicts after pulling new code, the top section is the code you already had and the bottom section is the remote code
+   = help: if you're in the middle of a rebase, the top section is the code being rebased onto and the bottom section is the code coming from the current commit being rebased
+   = note: for an explanation on these markers from the `git` documentation, visit <https://git-scm.com/book/en/v2/Git-Tools-Advanced-Merging#_checking_out_conflicts>
+
+error: aborting due to previous error
+

src/test/ui/parser/diff-markers/item-with-attr.rs

@@ -0,0 +1,10 @@
+#[attribute]
+<<<<<<< HEAD //~ ERROR encountered diff marker
+fn foo() {}
+=======
+fn bar() {}
+>>>>>>> branch
+
+fn main() {
+    foo();
+}

src/test/ui/parser/diff-markers/item-with-attr.stderr

@@ -0,0 +1,18 @@
+error: encountered diff marker
+  --> $DIR/item-with-attr.rs:2:1
+   |
+LL | <<<<<<< HEAD
+   | ^^^^^^^ after this is the code before the merge
+LL | fn foo() {}
+LL | =======
+   | -------
+LL | fn bar() {}
+LL | >>>>>>> branch
+   | ^^^^^^^ above this are the incoming code changes
+   |
+   = help: if you're having merge conflicts after pulling new code, the top section is the code you already had and the bottom section is the remote code
+   = help: if you're in the middle of a rebase, the top section is the code being rebased onto and the bottom section is the code coming from the current commit being rebased
+   = note: for an explanation on these markers from the `git` documentation, visit <https://git-scm.com/book/en/v2/Git-Tools-Advanced-Merging#_checking_out_conflicts>
+
+error: aborting due to previous error
+

src/test/ui/parser/diff-markers/item.rs

@@ -0,0 +1,9 @@
+<<<<<<< HEAD //~ ERROR encountered diff marker
+fn foo() {}
+=======
+fn bar() {}
+>>>>>>> branch
+
+fn main() {
+    foo();
+}

src/test/ui/parser/diff-markers/item.stderr

@@ -0,0 +1,18 @@
+error: encountered diff marker
+  --> $DIR/item.rs:1:1
+   |
+LL | <<<<<<< HEAD
+   | ^^^^^^^ after this is the code before the merge
+LL | fn foo() {}
+LL | =======
+   | -------
+LL | fn bar() {}
+LL | >>>>>>> branch
+   | ^^^^^^^ above this are the incoming code changes
+   |
+   = help: if you're having merge conflicts after pulling new code, the top section is the code you already had and the bottom section is the remote code
+   = help: if you're in the middle of a rebase, the top section is the code being rebased onto and the bottom section is the code coming from the current commit being rebased
+   = note: for an explanation on these markers from the `git` documentation, visit <https://git-scm.com/book/en/v2/Git-Tools-Advanced-Merging#_checking_out_conflicts>
+
+error: aborting due to previous error
+

src/test/ui/parser/diff-markers/statement.rs

@@ -0,0 +1,15 @@
+trait T {
+    fn foo() {}
+    fn bar() {}
+}
+
+struct S;
+impl T for S {}
+
+fn main() {
+<<<<<<< HEAD //~ ERROR encountered diff marker
+    S::foo();
+=======
+    S::bar();
+>>>>>>> branch
+}