Detect diff markers in the parser

Partly address #32059.

3 files changed, 72 insertions, 3 deletions

diff --git a/compiler/rustc_parse/src/parser/diagnostics.rs b/compiler/rustc_parse/src/parser/diagnostics.rs
index 61fe379c3e9..ffadb77320a 100644
--- a/compiler/rustc_parse/src/parser/diagnostics.rs
+++ b/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 +2557,57 @@ 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 middle = None;
+        let mut end = None;
+        loop {
+            if self.token.kind == TokenKind::Eof {
+                break;
+            }
+            if let Some(span) = self.diff_marker(&TokenKind::EqEq, &TokenKind::Eq) {
+                spans.push(span);
+                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, "start");
+        if let Some(middle) = middle {
+            err.span_label(middle, "middle");
+        }
+        if let Some(end) = end {
+            err.span_label(end, "end");
+        }
+        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, ()> {
diff --git a/compiler/rustc_parse/src/parser/item.rs b/compiler/rustc_parse/src/parser/item.rs
index e1ced2eb965..51de3d40d91 100644
--- a/compiler/rustc_parse/src/parser/item.rs
+++ b/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>> {
@@ -2427,6 +2433,7 @@ 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;
diff --git a/compiler/rustc_parse/src/parser/stmt.rs b/compiler/rustc_parse/src/parser/stmt.rs
index bae7f2670cb..4cbf89169b6 100644
--- a/compiler/rustc_parse/src/parser/stmt.rs
+++ b/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 proactivelly 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))