Auto merge of #96546 - nnethercote:overhaul-MacArgs, r=petrochenkov

Overhaul `MacArgs`

Motivation:
- Clarify some code that I found hard to understand.
- Eliminate one use of three places where `TokenKind::Interpolated` values are created.

r? `@petrochenkov`

compiler/rustc_ast/src/ast.rs

@@ -23,7 +23,7 @@ pub use GenericArgs::*;
 pub use UnsafeSource::*;
 
 use crate::ptr::P;
-use crate::token::{self, CommentKind, Delimiter, Token};
+use crate::token::{self, CommentKind, Delimiter, Token, TokenKind};
 use crate::tokenstream::{DelimSpan, LazyTokenStream, TokenStream, TokenTree};
 
 use rustc_data_structures::stable_hasher::{HashStable, StableHasher};
@@ -1532,7 +1532,7 @@ impl MacCall {
 }
 
 /// Arguments passed to an attribute or a function-like macro.
-#[derive(Clone, Encodable, Decodable, Debug, HashStable_Generic)]
+#[derive(Clone, Encodable, Decodable, Debug)]
 pub enum MacArgs {
     /// No arguments - `#[attr]`.
     Empty,
@@ -1542,11 +1542,20 @@ pub enum MacArgs {
     Eq(
         /// Span of the `=` token.
         Span,
-        /// "value" as a nonterminal token.
-        Token,
+        /// The "value".
+        MacArgsEq,
     ),
 }
 
+// The RHS of a `MacArgs::Eq` starts out as an expression. Once macro expansion
+// is completed, all cases end up either as a literal, which is the form used
+// after lowering to HIR, or as an error.
+#[derive(Clone, Encodable, Decodable, Debug)]
+pub enum MacArgsEq {
+    Ast(P<Expr>),
+    Hir(Lit),
+}
+
 impl MacArgs {
     pub fn delim(&self) -> Option<Delimiter> {
         match self {
@@ -1559,7 +1568,10 @@ impl MacArgs {
         match self {
             MacArgs::Empty => None,
             MacArgs::Delimited(dspan, ..) => Some(dspan.entire()),
-            MacArgs::Eq(eq_span, token) => Some(eq_span.to(token.span)),
+            MacArgs::Eq(eq_span, MacArgsEq::Ast(expr)) => Some(eq_span.to(expr.span)),
+            MacArgs::Eq(_, MacArgsEq::Hir(lit)) => {
+                unreachable!("in literal form when getting span: {:?}", lit);
+            }
         }
     }
 
@@ -1569,7 +1581,23 @@ impl MacArgs {
         match self {
             MacArgs::Empty => TokenStream::default(),
             MacArgs::Delimited(.., tokens) => tokens.clone(),
-            MacArgs::Eq(.., token) => TokenTree::Token(token.clone()).into(),
+            MacArgs::Eq(_, MacArgsEq::Ast(expr)) => {
+                // Currently only literals are allowed here. If more complex expression kinds are
+                // allowed in the future, then `nt_to_tokenstream` should be used to extract the
+                // token stream. This will require some cleverness, perhaps with a function
+                // pointer, because `nt_to_tokenstream` is not directly usable from this crate.
+                // It will also require changing the `parse_expr` call in `parse_mac_args_common`
+                // to `parse_expr_force_collect`.
+                if let ExprKind::Lit(lit) = &expr.kind {
+                    let token = Token::new(TokenKind::Literal(lit.token), lit.span);
+                    TokenTree::Token(token).into()
+                } else {
+                    unreachable!("couldn't extract literal when getting inner tokens: {:?}", expr)
+                }
+            }
+            MacArgs::Eq(_, MacArgsEq::Hir(lit)) => {
+                unreachable!("in literal form when getting inner tokens: {:?}", lit)
+            }
         }
     }
 
@@ -1580,6 +1608,30 @@ impl MacArgs {
     }
 }
 
+impl<CTX> HashStable<CTX> for MacArgs
+where
+    CTX: crate::HashStableContext,
+{
+    fn hash_stable(&self, ctx: &mut CTX, hasher: &mut StableHasher) {
+        mem::discriminant(self).hash_stable(ctx, hasher);
+        match self {
+            MacArgs::Empty => {}
+            MacArgs::Delimited(dspan, delim, tokens) => {
+                dspan.hash_stable(ctx, hasher);
+                delim.hash_stable(ctx, hasher);
+                tokens.hash_stable(ctx, hasher);
+            }
+            MacArgs::Eq(_eq_span, MacArgsEq::Ast(expr)) => {
+                unreachable!("hash_stable {:?}", expr);
+            }
+            MacArgs::Eq(eq_span, MacArgsEq::Hir(lit)) => {
+                eq_span.hash_stable(ctx, hasher);
+                lit.hash_stable(ctx, hasher);
+            }
+        }
+    }
+}
+
 #[derive(Copy, Clone, PartialEq, Eq, Encodable, Decodable, Debug, HashStable_Generic)]
 pub enum MacDelimiter {
     Parenthesis,

compiler/rustc_ast/src/attr/mod.rs

@@ -3,14 +3,16 @@
 use crate::ast;
 use crate::ast::{AttrId, AttrItem, AttrKind, AttrStyle, Attribute};
 use crate::ast::{Lit, LitKind};
-use crate::ast::{MacArgs, MacDelimiter, MetaItem, MetaItemKind, NestedMetaItem};
+use crate::ast::{MacArgs, MacArgsEq, MacDelimiter, MetaItem, MetaItemKind, NestedMetaItem};
 use crate::ast::{Path, PathSegment};
+use crate::ptr::P;
 use crate::token::{self, CommentKind, Delimiter, Token};
 use crate::tokenstream::{AttrAnnotatedTokenStream, AttrAnnotatedTokenTree};
 use crate::tokenstream::{DelimSpan, Spacing, TokenTree, TreeAndSpacing};
 use crate::tokenstream::{LazyTokenStream, TokenStream};
 use crate::util::comments;
 
+use rustc_data_structures::thin_vec::ThinVec;
 use rustc_index::bit_set::GrowableBitSet;
 use rustc_span::source_map::BytePos;
 use rustc_span::symbol::{sym, Ident, Symbol};
@@ -475,7 +477,16 @@ impl MetaItemKind {
     pub fn mac_args(&self, span: Span) -> MacArgs {
         match self {
             MetaItemKind::Word => MacArgs::Empty,
-            MetaItemKind::NameValue(lit) => MacArgs::Eq(span, lit.to_token()),
+            MetaItemKind::NameValue(lit) => {
+                let expr = P(ast::Expr {
+                    id: ast::DUMMY_NODE_ID,
+                    kind: ast::ExprKind::Lit(lit.clone()),
+                    span: lit.span,
+                    attrs: ThinVec::new(),
+                    tokens: None,
+                });
+                MacArgs::Eq(span, MacArgsEq::Ast(expr))
+            }
             MetaItemKind::List(list) => {
                 let mut tts = Vec::new();
                 for (i, item) in list.iter().enumerate() {
@@ -552,12 +563,16 @@ impl MetaItemKind {
 
     fn from_mac_args(args: &MacArgs) -> Option<MetaItemKind> {
         match args {
+            MacArgs::Empty => Some(MetaItemKind::Word),
             MacArgs::Delimited(_, MacDelimiter::Parenthesis, tokens) => {
                 MetaItemKind::list_from_tokens(tokens.clone())
             }
             MacArgs::Delimited(..) => None,
-            MacArgs::Eq(_, token) => Lit::from_token(token).ok().map(MetaItemKind::NameValue),
-            MacArgs::Empty => Some(MetaItemKind::Word),
+            MacArgs::Eq(_, MacArgsEq::Ast(expr)) => match &expr.kind {
+                ast::ExprKind::Lit(lit) => Some(MetaItemKind::NameValue(lit.clone())),
+                _ => None,
+            },
+            MacArgs::Eq(_, MacArgsEq::Hir(lit)) => Some(MetaItemKind::NameValue(lit.clone())),
         }
     }
 

compiler/rustc_ast/src/mut_visit.rs

@@ -370,21 +370,12 @@ pub fn visit_mac_args<T: MutVisitor>(args: &mut MacArgs, vis: &mut T) {
             visit_delim_span(dspan, vis);
             visit_tts(tokens, vis);
         }
-        MacArgs::Eq(eq_span, token) => {
+        MacArgs::Eq(eq_span, MacArgsEq::Ast(expr)) => {
             vis.visit_span(eq_span);
-            if T::VISIT_TOKENS {
-                visit_token(token, vis);
-            } else {
-                // The value in `#[key = VALUE]` must be visited as an expression for backward
-                // compatibility, so that macros can be expanded in that position.
-                match &mut token.kind {
-                    token::Interpolated(nt) => match Lrc::make_mut(nt) {
-                        token::NtExpr(expr) => vis.visit_expr(expr),
-                        t => panic!("unexpected token in key-value attribute: {:?}", t),
-                    },
-                    t => panic!("unexpected token in key-value attribute: {:?}", t),
-                }
-            }
+            vis.visit_expr(expr);
+        }
+        MacArgs::Eq(_, MacArgsEq::Hir(lit)) => {
+            unreachable!("in literal form when visiting mac args eq: {:?}", lit)
         }
     }
 }
@@ -738,7 +729,7 @@ pub fn visit_token<T: MutVisitor>(t: &mut Token, vis: &mut T) {
         }
         token::Interpolated(nt) => {
             let mut nt = Lrc::make_mut(nt);
-            visit_interpolated(&mut nt, vis);
+            visit_nonterminal(&mut nt, vis);
         }
         _ => {}
     }
@@ -769,7 +760,7 @@ pub fn visit_token<T: MutVisitor>(t: &mut Token, vis: &mut T) {
 // contain multiple items, but decided against it when I looked at
 // `parse_item_or_view_item` and tried to figure out what I would do with
 // multiple items there....
-pub fn visit_interpolated<T: MutVisitor>(nt: &mut token::Nonterminal, vis: &mut T) {
+pub fn visit_nonterminal<T: MutVisitor>(nt: &mut token::Nonterminal, vis: &mut T) {
     match nt {
         token::NtItem(item) => visit_clobber(item, |item| {
             // This is probably okay, because the only visitors likely to

compiler/rustc_ast/src/token.rs

@@ -237,6 +237,15 @@ pub enum TokenKind {
     /// treat regular and interpolated lifetime identifiers in the same way.
     Lifetime(Symbol),
 
+    /// An embedded AST node, as produced by a macro. This only exists for
+    /// historical reasons. We'd like to get rid of it, for multiple reasons.
+    /// - It's conceptually very strange. Saying a token can contain an AST
+    ///   node is like saying, in natural language, that a word can contain a
+    ///   sentence.
+    /// - It requires special handling in a bunch of places in the parser.
+    /// - It prevents `Token` from implementing `Copy`.
+    /// It adds complexity and likely slows things down. Please don't add new
+    /// occurrences of this token kind!
     Interpolated(Lrc<Nonterminal>),
 
     /// A doc comment token.

compiler/rustc_ast/src/visit.rs

@@ -14,7 +14,6 @@
 //! those that are created by the expansion of a macro.
 
 use crate::ast::*;
-use crate::token;
 
 use rustc_span::symbol::{Ident, Symbol};
 use rustc_span::Span;
@@ -940,14 +939,9 @@ pub fn walk_mac_args<'a, V: Visitor<'a>>(visitor: &mut V, args: &'a MacArgs) {
     match args {
         MacArgs::Empty => {}
         MacArgs::Delimited(_dspan, _delim, _tokens) => {}
-        // The value in `#[key = VALUE]` must be visited as an expression for backward
-        // compatibility, so that macros can be expanded in that position.
-        MacArgs::Eq(_eq_span, token) => match &token.kind {
-            token::Interpolated(nt) => match &**nt {
-                token::NtExpr(expr) => visitor.visit_expr(expr),
-                t => panic!("unexpected token in key-value attribute: {:?}", t),
-            },
-            t => panic!("unexpected token in key-value attribute: {:?}", t),
-        },
+        MacArgs::Eq(_eq_span, MacArgsEq::Ast(expr)) => visitor.visit_expr(expr),
+        MacArgs::Eq(_, MacArgsEq::Hir(lit)) => {
+            unreachable!("in literal form when walking mac args eq: {:?}", lit)
+        }
     }
 }