Auto merge of #80459 - mark-i-m:or-pat-reg, r=petrochenkov

Implement edition-based macro :pat feature

This PR does two things:
1. Fixes the perf regression from #80100 (comment)
2. Implements `:pat2018` and `:pat2021` matchers, as described by `@joshtriplett`  in #54883 (comment) behind the feature gate `edition_macro_pat`.

r? `@petrochenkov`

cc `@Mark-Simulacrum`

Author: bors

compiler/rustc_ast/src/token.rs

@@ -15,7 +15,7 @@ use rustc_span::hygiene::ExpnKind;
 use rustc_span::source_map::SourceMap;
 use rustc_span::symbol::{kw, sym};
 use rustc_span::symbol::{Ident, Symbol};
-use rustc_span::{self, FileName, RealFileName, Span, DUMMY_SP};
+use rustc_span::{self, edition::Edition, FileName, RealFileName, Span, DUMMY_SP};
 use std::borrow::Cow;
 use std::{fmt, mem};
 
@@ -690,7 +690,16 @@ pub enum NonterminalKind {
     Item,
     Block,
     Stmt,
-    Pat,
+    Pat2018 {
+        /// Keep track of whether the user used `:pat2018` or `:pat` and we inferred it from the
+        /// edition of the span. This is used for diagnostics.
+        inferred: bool,
+    },
+    Pat2021 {
+        /// Keep track of whether the user used `:pat2018` or `:pat` and we inferred it from the
+        /// edition of the span. This is used for diagnostics.
+        inferred: bool,
+    },
     Expr,
     Ty,
     Ident,
@@ -703,12 +712,25 @@ pub enum NonterminalKind {
 }
 
 impl NonterminalKind {
-    pub fn from_symbol(symbol: Symbol) -> Option<NonterminalKind> {
+    /// The `edition` closure is used to get the edition for the given symbol. Doing
+    /// `span.edition()` is expensive, so we do it lazily.
+    pub fn from_symbol(
+        symbol: Symbol,
+        edition: impl FnOnce() -> Edition,
+    ) -> Option<NonterminalKind> {
         Some(match symbol {
             sym::item => NonterminalKind::Item,
             sym::block => NonterminalKind::Block,
             sym::stmt => NonterminalKind::Stmt,
-            sym::pat => NonterminalKind::Pat,
+            sym::pat => match edition() {
+                Edition::Edition2015 | Edition::Edition2018 => {
+                    NonterminalKind::Pat2018 { inferred: true }
+                }
+                // FIXME(mark-i-m): uncomment when 2021 machinery is available.
+                //Edition::Edition2021 => NonterminalKind::Pat2021{inferred:true},
+            },
+            sym::pat2018 => NonterminalKind::Pat2018 { inferred: false },
+            sym::pat2021 => NonterminalKind::Pat2021 { inferred: false },
             sym::expr => NonterminalKind::Expr,
             sym::ty => NonterminalKind::Ty,
             sym::ident => NonterminalKind::Ident,
@@ -726,7 +748,10 @@ impl NonterminalKind {
             NonterminalKind::Item => sym::item,
             NonterminalKind::Block => sym::block,
             NonterminalKind::Stmt => sym::stmt,
-            NonterminalKind::Pat => sym::pat,
+            NonterminalKind::Pat2018 { inferred: false } => sym::pat2018,
+            NonterminalKind::Pat2021 { inferred: false } => sym::pat2021,
+            NonterminalKind::Pat2018 { inferred: true }
+            | NonterminalKind::Pat2021 { inferred: true } => sym::pat,
             NonterminalKind::Expr => sym::expr,
             NonterminalKind::Ty => sym::ty,
             NonterminalKind::Ident => sym::ident,

compiler/rustc_expand/src/mbe/macro_parser.rs

@@ -77,9 +77,9 @@ use TokenTreeOrTokenTreeSlice::*;
 use crate::mbe::{self, TokenTree};
 
 use rustc_ast::token::{self, DocComment, Nonterminal, Token};
-use rustc_parse::parser::{OrPatNonterminalMode, Parser};
+use rustc_parse::parser::Parser;
 use rustc_session::parse::ParseSess;
-use rustc_span::{edition::Edition, symbol::MacroRulesNormalizedIdent};
+use rustc_span::symbol::MacroRulesNormalizedIdent;
 
 use smallvec::{smallvec, SmallVec};
 
@@ -419,18 +419,6 @@ fn token_name_eq(t1: &Token, t2: &Token) -> bool {
     }
 }
 
-/// In edition 2015/18, `:pat` can only match `pat<no_top_alt>` because otherwise, we have
-/// breakage. As of edition 2021, `:pat` matches `top_pat`.
-///
-/// See <https://github.com/rust-lang/rust/issues/54883> for more info.
-fn or_pat_mode(edition: Edition) -> OrPatNonterminalMode {
-    match edition {
-        Edition::Edition2015 | Edition::Edition2018 => OrPatNonterminalMode::NoTopAlt,
-        // FIXME(mark-i-m): uncomment this when edition 2021 machinery is added.
-        // Edition::Edition2021 =>  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`.
 ///
@@ -578,14 +566,13 @@ fn inner_parse_loop<'root, 'tt>(
 
                 // We need to match a metavar with a valid ident... call out to the black-box
                 // parser by adding an item to `bb_items`.
-                TokenTree::MetaVarDecl(span, _, Some(kind)) => {
+                TokenTree::MetaVarDecl(_, _, Some(kind)) => {
                     // Built-in nonterminals never start with these tokens, so we can eliminate
                     // them from consideration.
                     //
                     // We use the span of the metavariable declaration to determine any
                     // edition-specific matching behavior for non-terminals.
-                    if Parser::nonterminal_may_begin_with(kind, token, or_pat_mode(span.edition()))
-                    {
+                    if Parser::nonterminal_may_begin_with(kind, token) {
                         bb_items.push(item);
                     }
                 }
@@ -749,8 +736,7 @@ pub(super) fn parse_tt(parser: &mut Cow<'_, Parser<'_>>, ms: &[TokenTree]) -> Na
                 let match_cur = item.match_cur;
                 // We use the span of the metavariable declaration to determine any
                 // edition-specific matching behavior for non-terminals.
-                let nt = match parser.to_mut().parse_nonterminal(kind, or_pat_mode(span.edition()))
-                {
+                let nt = match parser.to_mut().parse_nonterminal(kind) {
                     Err(mut err) => {
                         err.span_label(
                             span,

compiler/rustc_expand/src/mbe/macro_rules.rs

@@ -476,10 +476,15 @@ pub fn compile_declarative_macro(
             .map(|m| {
                 if let MatchedNonterminal(ref nt) = *m {
                     if let NtTT(ref tt) = **nt {
-                        let tt =
-                            mbe::quoted::parse(tt.clone().into(), true, &sess.parse_sess, def.id)
-                                .pop()
-                                .unwrap();
+                        let tt = mbe::quoted::parse(
+                            tt.clone().into(),
+                            true,
+                            &sess.parse_sess,
+                            def.id,
+                            features,
+                        )
+                        .pop()
+                        .unwrap();
                         valid &= check_lhs_nt_follows(&sess.parse_sess, features, &def.attrs, &tt);
                         return tt;
                     }
@@ -501,6 +506,7 @@ pub fn compile_declarative_macro(
                             false,
                             &sess.parse_sess,
                             def.id,
+                            features,
                         )
                         .pop()
                         .unwrap();
@@ -1090,7 +1096,7 @@ fn is_in_follow(tok: &mbe::TokenTree, kind: NonterminalKind) -> IsInFollow {
                     _ => IsInFollow::No(TOKENS),
                 }
             }
-            NonterminalKind::Pat => {
+            NonterminalKind::Pat2018 { .. } | NonterminalKind::Pat2021 { .. } => {
                 const TOKENS: &[&str] = &["`=>`", "`,`", "`=`", "`|`", "`if`", "`in`"];
                 match tok {
                     TokenTree::Token(token) => match token.kind {

compiler/rustc_expand/src/mbe/quoted.rs

@@ -5,8 +5,9 @@ use rustc_ast::token::{self, Token};
 use rustc_ast::tokenstream;
 use rustc_ast::{NodeId, DUMMY_NODE_ID};
 use rustc_ast_pretty::pprust;
-use rustc_session::parse::ParseSess;
-use rustc_span::symbol::{kw, Ident};
+use rustc_feature::Features;
+use rustc_session::parse::{feature_err, ParseSess};
+use rustc_span::symbol::{kw, sym, Ident};
 
 use rustc_span::Span;
 
@@ -29,10 +30,8 @@ const VALID_FRAGMENT_NAMES_MSG: &str = "valid fragment specifiers are \
 ///   `ident` are "matchers". They are not present in the body of a macro rule -- just in the
 ///   pattern, so we pass a parameter to indicate whether to expect them or not.
 /// - `sess`: the parsing session. Any errors will be emitted to this session.
-/// - `features`, `attrs`: language feature flags and attributes so that we know whether to use
-///   unstable features or not.
-/// - `edition`: which edition are we in.
-/// - `macro_node_id`: the NodeId of the macro we are parsing.
+/// - `node_id`: the NodeId of the macro we are parsing.
+/// - `features`: language features so we can do feature gating.
 ///
 /// # Returns
 ///
@@ -42,6 +41,7 @@ pub(super) fn parse(
     expect_matchers: bool,
     sess: &ParseSess,
     node_id: NodeId,
+    features: &Features,
 ) -> Vec<TokenTree> {
     // Will contain the final collection of `self::TokenTree`
     let mut result = Vec::new();
@@ -52,7 +52,7 @@ pub(super) fn parse(
     while let Some(tree) = trees.next() {
         // Given the parsed tree, if there is a metavar and we are expecting matchers, actually
         // parse out the matcher (i.e., in `$id:ident` this would parse the `:` and `ident`).
-        let tree = parse_tree(tree, &mut trees, expect_matchers, sess, node_id);
+        let tree = parse_tree(tree, &mut trees, expect_matchers, sess, node_id, features);
         match tree {
             TokenTree::MetaVar(start_sp, ident) if expect_matchers => {
                 let span = match trees.next() {
@@ -61,18 +61,39 @@ pub(super) fn parse(
                             Some(tokenstream::TokenTree::Token(token)) => match token.ident() {
                                 Some((frag, _)) => {
                                     let span = token.span.with_lo(start_sp.lo());
-                                    let kind = token::NonterminalKind::from_symbol(frag.name)
-                                        .unwrap_or_else(|| {
-                                            let msg = format!(
-                                                "invalid fragment specifier `{}`",
-                                                frag.name
-                                            );
-                                            sess.span_diagnostic
-                                                .struct_span_err(span, &msg)
-                                                .help(VALID_FRAGMENT_NAMES_MSG)
+
+                                    match frag.name {
+                                        sym::pat2018 | sym::pat2021 => {
+                                            if !features.edition_macro_pats {
+                                                feature_err(
+                                                    sess,
+                                                    sym::edition_macro_pats,
+                                                    frag.span,
+                                                    "`pat2018` and `pat2021` are unstable.",
+                                                )
                                                 .emit();
-                                            token::NonterminalKind::Ident
-                                        });
+                                            }
+                                        }
+                                        _ => {}
+                                    }
+
+                                    let kind =
+                                        token::NonterminalKind::from_symbol(frag.name, || {
+                                            span.edition()
+                                        })
+                                        .unwrap_or_else(
+                                            || {
+                                                let msg = format!(
+                                                    "invalid fragment specifier `{}`",
+                                                    frag.name
+                                                );
+                                                sess.span_diagnostic
+                                                    .struct_span_err(span, &msg)
+                                                    .help(VALID_FRAGMENT_NAMES_MSG)
+                                                    .emit();
+                                                token::NonterminalKind::Ident
+                                            },
+                                        );
                                     result.push(TokenTree::MetaVarDecl(span, ident, Some(kind)));
                                     continue;
                                 }
@@ -110,14 +131,14 @@ pub(super) fn parse(
 ///   converting `tree`
 /// - `expect_matchers`: same as for `parse` (see above).
 /// - `sess`: the parsing session. Any errors will be emitted to this session.
-/// - `features`, `attrs`: language feature flags and attributes so that we know whether to use
-///   unstable features or not.
+/// - `features`: language features so we can do feature gating.
 fn parse_tree(
     tree: tokenstream::TokenTree,
     outer_trees: &mut impl Iterator<Item = tokenstream::TokenTree>,
     expect_matchers: bool,
     sess: &ParseSess,
     node_id: NodeId,
+    features: &Features,
 ) -> TokenTree {
     // Depending on what `tree` is, we could be parsing different parts of a macro
     match tree {
@@ -145,7 +166,7 @@ fn parse_tree(
                         sess.span_diagnostic.span_err(span.entire(), &msg);
                     }
                     // Parse the contents of the sequence itself
-                    let sequence = parse(tts, expect_matchers, sess, node_id);
+                    let sequence = parse(tts, expect_matchers, sess, node_id, features);
                     // Get the Kleene operator and optional separator
                     let (separator, kleene) =
                         parse_sep_and_kleene_op(&mut trees, span.entire(), sess);
@@ -196,7 +217,10 @@ fn parse_tree(
         // descend into the delimited set and further parse it.
         tokenstream::TokenTree::Delimited(span, delim, tts) => TokenTree::Delimited(
             span,
-            Lrc::new(Delimited { delim, tts: parse(tts, expect_matchers, sess, node_id) }),
+            Lrc::new(Delimited {
+                delim,
+                tts: parse(tts, expect_matchers, sess, node_id, features),
+            }),
         ),
     }
 }

compiler/rustc_feature/src/active.rs

@@ -620,6 +620,9 @@ declare_features! (
     /// Allows arbitrary expressions in key-value attributes at parse time.
     (active, extended_key_value_attributes, "1.50.0", Some(78835), None),
 
+    /// `:pat2018` and `:pat2021` macro matchers.
+    (active, edition_macro_pats, "1.51.0", Some(54883), None),
+
     // -------------------------------------------------------------------------
     // feature-group-end: actual feature gates
     // -------------------------------------------------------------------------

compiler/rustc_parse/src/parser/mod.rs

@@ -12,7 +12,6 @@ mod ty;
 use crate::lexer::UnmatchedBrace;
 pub use diagnostics::AttemptLocalParseRecovery;
 use diagnostics::Error;
-pub use pat::OrPatNonterminalMode;
 pub use path::PathStyle;
 
 use rustc_ast::ptr::P;

compiler/rustc_parse/src/parser/nonterminal.rs

@@ -4,19 +4,15 @@ use rustc_ast_pretty::pprust;
 use rustc_errors::PResult;
 use rustc_span::symbol::{kw, Ident};
 
-use crate::parser::pat::{GateOr, OrPatNonterminalMode, RecoverComma};
+use crate::parser::pat::{GateOr, RecoverComma};
 use crate::parser::{FollowedByType, Parser, PathStyle};
 
 impl<'a> Parser<'a> {
     /// Checks whether a non-terminal may begin with a particular token.
     ///
     /// 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,
-        or_pat_mode: OrPatNonterminalMode,
-    ) -> bool {
+    pub fn nonterminal_may_begin_with(kind: NonterminalKind, token: &Token) -> bool {
         /// Checks whether the non-terminal may contain a single (non-keyword) identifier.
         fn may_be_ident(nt: &token::Nonterminal) -> bool {
             match *nt {
@@ -62,7 +58,7 @@ impl<'a> Parser<'a> {
                 },
                 _ => false,
             },
-            NonterminalKind::Pat => match token.kind {
+            NonterminalKind::Pat2018 { .. } | NonterminalKind::Pat2021 { .. } => match token.kind {
                 token::Ident(..) |                  // box, ref, mut, and other identifiers (can stricten)
                 token::OpenDelim(token::Paren) |    // tuple pattern
                 token::OpenDelim(token::Bracket) |  // slice pattern
@@ -76,7 +72,7 @@ impl<'a> Parser<'a> {
                 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::BinOp(token::Or) =>  matches!(kind, NonterminalKind::Pat2021 {..}),
                 token::Interpolated(ref nt) => may_be_ident(nt),
                 _ => false,
             },
@@ -94,11 +90,7 @@ impl<'a> Parser<'a> {
     }
 
     /// 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> {
+    pub fn parse_nonterminal(&mut self, kind: NonterminalKind) -> 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,
@@ -141,12 +133,13 @@ impl<'a> Parser<'a> {
                     }
                 }
             }
-            NonterminalKind::Pat => {
-                let (mut pat, tokens) = self.collect_tokens(|this| match or_pat_mode {
-                    OrPatNonterminalMode::TopPat => {
+            NonterminalKind::Pat2018 { .. } | NonterminalKind::Pat2021 { .. } => {
+                let (mut pat, tokens) = self.collect_tokens(|this| match kind {
+                    NonterminalKind::Pat2018 { .. } => this.parse_pat(None),
+                    NonterminalKind::Pat2021 { .. } => {
                         this.parse_top_pat(GateOr::Yes, RecoverComma::No)
                     }
-                    OrPatNonterminalMode::NoTopAlt => this.parse_pat(None),
+                    _ => unreachable!(),
                 })?;
                 // We have have eaten an NtPat, which could already have tokens
                 if pat.tokens.is_none() {

compiler/rustc_parse/src/parser/pat.rs

@@ -31,13 +31,6 @@ pub(super) enum RecoverComma {
     No,
 }
 
-/// 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<'a> Parser<'a> {
     /// Parses a pattern.
     ///