implement or-pattern macro fallback

Author: mark-i-m

compiler/rustc_ast/src/token.rs

@@ -717,6 +717,13 @@ pub enum NonterminalKind {
     TT,
 }
 
+/// Used when parsing a non-terminal (see `parse_nonterminal`) to determine if `:pat` should match
+/// `top_pat` or `pat<no_top_alt>`. See issue https://github.com/rust-lang/rust/pull/78935.
+pub enum OrPatNonterminalMode {
+    TopPat,
+    NoTopAlt,
+}
+
 impl NonterminalKind {
     pub fn from_symbol(symbol: Symbol) -> Option<NonterminalKind> {
         Some(match symbol {

compiler/rustc_expand/src/mbe/macro_parser.rs

@@ -76,7 +76,9 @@ use TokenTreeOrTokenTreeSlice::*;
 
 use crate::mbe::{self, TokenTree};
 
-use rustc_ast::token::{self, DocComment, Nonterminal, Token};
+use rustc_ast::token::{
+    self, BinOpToken, DocComment, Nonterminal, OrPatNonterminalMode, Token, TokenKind,
+};
 use rustc_parse::parser::Parser;
 use rustc_session::parse::ParseSess;
 use rustc_span::symbol::MacroRulesNormalizedIdent;
@@ -414,6 +416,27 @@ fn token_name_eq(t1: &Token, t2: &Token) -> bool {
     }
 }
 
+/// Check whether the next token in the matcher is `|` (i.e. if we are matching against `$foo:type
+/// |`). The reason is that when we stabilized the `or_patterns` feature, we wanted `:pat` to match
+/// top-level or-patterns too.  But doing that would cause a lot of breakage, since `|` could
+/// previously be a separator that follows `:pat`. Thus, we make a special case to parse `:pat` the
+/// old way if it happens to be followed by `|` in the matcher.
+///
+/// See https://github.com/rust-lang/rust/issues/54883 for more info.
+fn or_pat_mode(item: &MatcherPosHandle<'_, '_>) -> OrPatNonterminalMode {
+    if item.idx < item.top_elts.len() - 1 {
+        // Look at the token after the current one to see if it is `|`.
+        let tt = item.top_elts.get_tt(item.idx + 1);
+        if let TokenTree::Token(Token { kind: TokenKind::BinOp(BinOpToken::Or), .. }) = tt {
+            OrPatNonterminalMode::NoTopAlt
+        } else {
+            OrPatNonterminalMode::TopPat
+        }
+    } else {
+        OrPatNonterminalMode::TopPat
+    }
+}
+
 /// Process the matcher positions of `cur_items` until it is empty. In the process, this will
 /// produce more items in `next_items`, `eof_items`, and `bb_items`.
 ///
@@ -518,6 +541,8 @@ fn inner_parse_loop<'root, 'tt>(
         }
         // We are in the middle of a matcher.
         else {
+            let or_pat_mode = or_pat_mode(&item);
+
             // Look at what token in the matcher we are trying to match the current token (`token`)
             // against. Depending on that, we may generate new items.
             match item.top_elts.get_tt(idx) {
@@ -559,7 +584,7 @@ fn inner_parse_loop<'root, 'tt>(
                 TokenTree::MetaVarDecl(_, _, kind) => {
                     // Built-in nonterminals never start with these tokens,
                     // so we can eliminate them from consideration.
-                    if Parser::nonterminal_may_begin_with(kind, token) {
+                    if Parser::nonterminal_may_begin_with(kind, token, or_pat_mode) {
                         bb_items.push(item);
                     }
                 }
@@ -718,9 +743,10 @@ pub(super) fn parse_tt(parser: &mut Cow<'_, Parser<'_>>, ms: &[TokenTree]) -> Na
             assert_eq!(bb_items.len(), 1);
 
             let mut item = bb_items.pop().unwrap();
+            let or_pat_mode = or_pat_mode(&item);
             if let TokenTree::MetaVarDecl(span, _, kind) = item.top_elts.get_tt(item.idx) {
                 let match_cur = item.match_cur;
-                let nt = match parser.to_mut().parse_nonterminal(kind) {
+                let nt = match parser.to_mut().parse_nonterminal(kind, or_pat_mode) {
                     Err(mut err) => {
                         err.span_label(
                             span,

compiler/rustc_parse/src/parser/nonterminal.rs

@@ -1,5 +1,5 @@
 use rustc_ast::ptr::P;
-use rustc_ast::token::{self, Nonterminal, NonterminalKind, Token};
+use rustc_ast::token::{self, Nonterminal, NonterminalKind, OrPatNonterminalMode, Token};
 use rustc_ast_pretty::pprust;
 use rustc_errors::PResult;
 use rustc_span::symbol::{kw, Ident};
@@ -11,7 +11,11 @@ impl<'a> Parser<'a> {
     ///
     /// Returning `false` is a *stability guarantee* that such a matcher will *never* begin with that
     /// token. Be conservative (return true) if not sure.
-    pub fn nonterminal_may_begin_with(kind: NonterminalKind, token: &Token) -> bool {
+    pub fn nonterminal_may_begin_with(
+        kind: NonterminalKind,
+        token: &Token,
+        or_pat_mode: OrPatNonterminalMode,
+    ) -> bool {
         /// Checks whether the non-terminal may contain a single (non-keyword) identifier.
         fn may_be_ident(nt: &token::Nonterminal) -> bool {
             match *nt {
@@ -61,14 +65,15 @@ impl<'a> Parser<'a> {
                 token::OpenDelim(token::Bracket) |  // slice pattern
                 token::BinOp(token::And) |          // reference
                 token::BinOp(token::Minus) |        // negative literal
-                token::BinOp(token::Or) |           // leading vert `|` or-pattern
                 token::AndAnd |                     // double reference
                 token::Literal(..) |                // literal
                 token::DotDot |                     // range pattern (future compat)
                 token::DotDotDot |                  // range pattern (future compat)
                 token::ModSep |                     // path
                 token::Lt |                         // path (UFCS constant)
                 token::BinOp(token::Shl) => true,   // path (double UFCS)
+                // leading vert `|` or-pattern
+                token::BinOp(token::Or) =>  matches!(or_pat_mode, OrPatNonterminalMode::TopPat),
                 token::Interpolated(ref nt) => may_be_ident(nt),
                 _ => false,
             },
@@ -85,7 +90,12 @@ impl<'a> Parser<'a> {
         }
     }
 
-    pub fn parse_nonterminal(&mut self, kind: NonterminalKind) -> PResult<'a, Nonterminal> {
+    /// Parse a non-terminal (e.g. MBE `:pat` or `:ident`).
+    pub fn parse_nonterminal(
+        &mut self,
+        kind: NonterminalKind,
+        or_pat_mode: OrPatNonterminalMode,
+    ) -> PResult<'a, Nonterminal> {
         // Any `Nonterminal` which stores its tokens (currently `NtItem` and `NtExpr`)
         // needs to have them force-captured here.
         // A `macro_rules!` invocation may pass a captured item/expr to a proc-macro,
@@ -129,8 +139,14 @@ impl<'a> Parser<'a> {
                 }
             }
             NonterminalKind::Pat => {
-                let (mut pat, tokens) =
-                    self.collect_tokens(|this| this.parse_top_pat_no_commas())?;
+                let (mut pat, tokens) = match or_pat_mode {
+                    OrPatNonterminalMode::TopPat => {
+                        self.collect_tokens(|this| this.parse_top_pat_no_commas())?
+                    }
+                    OrPatNonterminalMode::NoTopAlt => {
+                        self.collect_tokens(|this| this.parse_pat(None))?
+                    }
+                };
                 // We have have eaten an NtPat, which could already have tokens
                 if pat.tokens.is_none() {
                     pat.tokens = tokens;

src/test/ui/pattern/or-pattern-macro-pat-fallback.rs

@@ -0,0 +1,40 @@
+// run-pass
+
+//#![feature(or_patterns)]
+
+macro_rules! foo {
+    ($orpat:pat, $val:expr) => {
+        match $val {
+            x @ ($orpat) => x,
+            _ => 0xDEADBEEF,
+        }
+    };
+    ($nonor:pat | $val:expr, 3) => {
+        match $val {
+            x @ ($orpat) => x,
+            _ => 0xDEADBEEF,
+        }
+    };
+}
+
+macro_rules! bar {
+    ($nonor:pat |) => {};
+}
+
+macro_rules! baz {
+    ($nonor:pat) => {};
+}
+
+fn main() {
+    // Test ambiguity.
+    foo!(1 | 2, 3); //~ERROR: multiple matchers
+
+    // Leading vert not allowed in pat<no_top_alt>
+    bar!(1 | 2 | 3 |); // ok
+    bar!(|1| 2 | 3 |); //~ERROR: no rules expected
+    bar!(1 | 2 | 3); //~ERROR: unexpected end
+
+    baz!(1 | 2 | 3); // ok
+    baz!(|1| 2 | 3); // ok
+    baz!(|1| 2 | 3 |); //~ERROR: no rules expected
+}

src/test/ui/pattern/or-pattern-macro-pat.rs

@@ -8,25 +8,50 @@ use Foo::*;
 enum Foo {
     A(u64),
     B(u64),
-    C, D
+    C,
+    D,
 }
 
 macro_rules! foo {
     ($orpat:pat, $val:expr) => {
         match $val {
-            x @ ($orpat) => x,
+            x @ ($orpat) => x, // leading vert would not be allowed in $orpat
             _ => B(0xDEADBEEF),
         }
     };
 }
 
 macro_rules! bar {
+    ($orpat:pat, $val:expr) => {
+        match $val {
+            $orpat => 42, // leading vert allowed here
+            _ => 0xDEADBEEF,
+        }
+    };
+}
+
+macro_rules! quux {
+    ($orpat1:pat | $orpat2:pat, $val:expr) => {
+        match $val {
+            x @ ($orpat1) => x,
+            _ => B(0xDEADBEEF),
+        }
+    };
+}
+
+macro_rules! baz {
     ($orpat:pat, $val:expr) => {
         match $val {
             $orpat => 42,
             _ => 0xDEADBEEF,
         }
     };
+    ($nonor:pat | $val:expr, C) => {
+        match $val {
+            x @ ($orpat) => x,
+            _ => 0xDEADBEEF,
+        }
+    };
 }
 
 fn main() {
@@ -35,6 +60,14 @@ fn main() {
     assert_eq!(y, A(32));
 
     // Leading vert in or-pattern.
-    let y = bar!(| C | D, C);
+    let y = bar!(|C| D, C);
     assert_eq!(y, 42u64);
+
+    // Leading vert in or-pattern makes baz! unambiguous.
+    let y = baz!(|C| D, C);
+    assert_eq!(y, 42u64);
+
+    // Or-separator fallback.
+    let y = quux!(C | D, D);
+    assert_eq!(y, B(0xDEADBEEF));
 }