Add support for literals

c410-f3r · c410-f3r · commit 49e479a80c3b · 2024-07-01T10:33:01.000-03:00
diff --git a/compiler/rustc_expand/src/mbe/metavar_expr.rs b/compiler/rustc_expand/src/mbe/metavar_expr.rs
@@ -1,12 +1,12 @@
-use rustc_ast::token::{self, Delimiter, IdentIsRaw};
+use rustc_ast::token::{self, Delimiter, IdentIsRaw, Lit, Token, TokenKind};
 use rustc_ast::tokenstream::{RefTokenTreeCursor, TokenStream, TokenTree};
 use rustc_ast::{LitIntType, LitKind};
 use rustc_ast_pretty::pprust;
 use rustc_errors::{Applicability, PResult};
 use rustc_macros::{Decodable, Encodable};
 use rustc_session::parse::ParseSess;
 use rustc_span::symbol::Ident;
-use rustc_span::Span;
+use rustc_span::{Span, Symbol};
 
 pub(crate) const RAW_IDENT_ERR: &str = "`${concat(..)}` currently does not support raw identifiers";
 
@@ -51,11 +51,18 @@ impl MetaVarExpr {
                 let mut result = Vec::new();
                 loop {
                     let is_var = try_eat_dollar(&mut iter);
-                    let element_ident = parse_ident(&mut iter, psess, outer_span)?;
+                    let token = parse_token(&mut iter, psess, outer_span)?;
                     let element = if is_var {
-                        MetaVarExprConcatElem::Var(element_ident)
+                        MetaVarExprConcatElem::Var(parse_ident_from_token(psess, token)?)
+                    } else if let TokenKind::Literal(Lit {
+                        kind: token::LitKind::Char | token::LitKind::Integer | token::LitKind::Str,
+                        symbol,
+                        suffix: None,
+                    }) = token.kind
+                    {
+                        MetaVarExprConcatElem::Literal(symbol)
                     } else {
-                        MetaVarExprConcatElem::Ident(element_ident)
+                        MetaVarExprConcatElem::Ident(parse_ident_from_token(psess, token)?)
                     };
                     result.push(element);
                     if iter.look_ahead(0).is_none() {
@@ -105,11 +112,13 @@ impl MetaVarExpr {
 
 #[derive(Debug, Decodable, Encodable, PartialEq)]
 pub(crate) enum MetaVarExprConcatElem {
-    /// There is NO preceding dollar sign, which means that this identifier should be interpreted
-    /// as a literal.
+    /// Identifier WITHOUT a preceding dollar sign, which means that this identifier should be
+    /// interpreted as a literal.
     Ident(Ident),
-    /// There is a preceding dollar sign, which means that this identifier should be expanded
-    /// and interpreted as a variable.
+    /// For example, a number or a string.
+    Literal(Symbol),
+    /// Identifier WITH a preceding dollar sign, which means that this identifier should be
+    /// expanded and interpreted as a variable.
     Var(Ident),
 }
 
@@ -158,7 +167,7 @@ fn parse_depth<'psess>(
     span: Span,
 ) -> PResult<'psess, usize> {
     let Some(tt) = iter.next() else { return Ok(0) };
-    let TokenTree::Token(token::Token { kind: token::TokenKind::Literal(lit), .. }, _) = tt else {
+    let TokenTree::Token(Token { kind: TokenKind::Literal(lit), .. }, _) = tt else {
         return Err(psess
             .dcx()
             .struct_span_err(span, "meta-variable expression depth must be a literal"));
@@ -180,12 +189,14 @@ fn parse_ident<'psess>(
     psess: &'psess ParseSess,
     fallback_span: Span,
 ) -> PResult<'psess, Ident> {
-    let Some(tt) = iter.next() else {
-        return Err(psess.dcx().struct_span_err(fallback_span, "expected identifier"));
-    };
-    let TokenTree::Token(token, _) = tt else {
-        return Err(psess.dcx().struct_span_err(tt.span(), "expected identifier"));
-    };
+    let token = parse_token(iter, psess, fallback_span)?;
+    parse_ident_from_token(psess, token)
+}
+
+fn parse_ident_from_token<'psess>(
+    psess: &'psess ParseSess,
+    token: &Token,
+) -> PResult<'psess, Ident> {
     if let Some((elem, is_raw)) = token.ident() {
         if let IdentIsRaw::Yes = is_raw {
             return Err(psess.dcx().struct_span_err(elem.span, RAW_IDENT_ERR));
@@ -205,10 +216,24 @@ fn parse_ident<'psess>(
     Err(err)
 }
 
+fn parse_token<'psess, 't>(
+    iter: &mut RefTokenTreeCursor<'t>,
+    psess: &'psess ParseSess,
+    fallback_span: Span,
+) -> PResult<'psess, &'t Token> {
+    let Some(tt) = iter.next() else {
+        return Err(psess.dcx().struct_span_err(fallback_span, "expected identifier or literal"));
+    };
+    let TokenTree::Token(token, _) = tt else {
+        return Err(psess.dcx().struct_span_err(tt.span(), "expected identifier or literal"));
+    };
+    Ok(token)
+}
+
 /// Tries to move the iterator forward returning `true` if there is a comma. If not, then the
 /// iterator is not modified and the result is `false`.
 fn try_eat_comma(iter: &mut RefTokenTreeCursor<'_>) -> bool {
-    if let Some(TokenTree::Token(token::Token { kind: token::Comma, .. }, _)) = iter.look_ahead(0) {
+    if let Some(TokenTree::Token(Token { kind: token::Comma, .. }, _)) = iter.look_ahead(0) {
         let _ = iter.next();
         return true;
     }
@@ -218,8 +243,7 @@ fn try_eat_comma(iter: &mut RefTokenTreeCursor<'_>) -> bool {
 /// Tries to move the iterator forward returning `true` if there is a dollar sign. If not, then the
 /// iterator is not modified and the result is `false`.
 fn try_eat_dollar(iter: &mut RefTokenTreeCursor<'_>) -> bool {
-    if let Some(TokenTree::Token(token::Token { kind: token::Dollar, .. }, _)) = iter.look_ahead(0)
-    {
+    if let Some(TokenTree::Token(Token { kind: token::Dollar, .. }, _)) = iter.look_ahead(0) {
         let _ = iter.next();
         return true;
     }
@@ -232,8 +256,7 @@ fn eat_dollar<'psess>(
     psess: &'psess ParseSess,
     span: Span,
 ) -> PResult<'psess, ()> {
-    if let Some(TokenTree::Token(token::Token { kind: token::Dollar, .. }, _)) = iter.look_ahead(0)
-    {
+    if let Some(TokenTree::Token(Token { kind: token::Dollar, .. }, _)) = iter.look_ahead(0) {
         let _ = iter.next();
         return Ok(());
     }
diff --git a/compiler/rustc_expand/src/mbe/transcribe.rs b/compiler/rustc_expand/src/mbe/transcribe.rs
@@ -11,11 +11,13 @@ use rustc_ast::token::{self, Delimiter, Token, TokenKind};
 use rustc_ast::tokenstream::{DelimSpacing, DelimSpan, Spacing, TokenStream, TokenTree};
 use rustc_data_structures::fx::FxHashMap;
 use rustc_errors::{pluralize, Diag, DiagCtxtHandle, PResult};
+use rustc_parse::lexer::nfc_normalize;
 use rustc_parse::parser::ParseNtResult;
 use rustc_session::parse::ParseSess;
+use rustc_session::parse::SymbolGallery;
 use rustc_span::hygiene::{LocalExpnId, Transparency};
 use rustc_span::symbol::{sym, Ident, MacroRulesNormalizedIdent};
-use rustc_span::{with_metavar_spans, Span, Symbol, SyntaxContext};
+use rustc_span::{with_metavar_spans, Span, SyntaxContext};
 use smallvec::{smallvec, SmallVec};
 use std::mem;
 
@@ -312,7 +314,16 @@ pub(super) fn transcribe<'a>(
 
             // Replace meta-variable expressions with the result of their expansion.
             mbe::TokenTree::MetaVarExpr(sp, expr) => {
-                transcribe_metavar_expr(dcx, expr, interp, &mut marker, &repeats, &mut result, sp)?;
+                transcribe_metavar_expr(
+                    dcx,
+                    expr,
+                    interp,
+                    &mut marker,
+                    &repeats,
+                    &mut result,
+                    sp,
+                    &psess.symbol_gallery,
+                )?;
             }
 
             // If we are entering a new delimiter, we push its contents to the `stack` to be
@@ -669,6 +680,7 @@ fn transcribe_metavar_expr<'a>(
     repeats: &[(usize, usize)],
     result: &mut Vec<TokenTree>,
     sp: &DelimSpan,
+    symbol_gallery: &SymbolGallery,
 ) -> PResult<'a, ()> {
     let mut visited_span = || {
         let mut span = sp.entire();
@@ -680,16 +692,26 @@ fn transcribe_metavar_expr<'a>(
             let mut concatenated = String::new();
             for element in elements.into_iter() {
                 let string = match element {
-                    MetaVarExprConcatElem::Ident(ident) => ident.to_string(),
-                    MetaVarExprConcatElem::Var(ident) => extract_ident(dcx, *ident, interp)?,
+                    MetaVarExprConcatElem::Ident(elem) => elem.to_string(),
+                    MetaVarExprConcatElem::Literal(elem) => elem.as_str().into(),
+                    MetaVarExprConcatElem::Var(elem) => extract_ident(dcx, *elem, interp)?,
                 };
                 concatenated.push_str(&string);
             }
+            let symbol = nfc_normalize(&concatenated);
+            let concatenated_span = visited_span();
+            if !rustc_lexer::is_ident(symbol.as_str()) {
+                return Err(dcx.struct_span_err(
+                    concatenated_span,
+                    "`${concat(..)}` is not generating a valid identifier",
+                ));
+            }
+            symbol_gallery.insert(symbol, concatenated_span);
             // The current implementation marks the span as coming from the macro regardless of
             // contexts of the concatenated identifiers but this behavior may change in the
             // future.
             result.push(TokenTree::Token(
-                Token::from_ast_ident(Ident::new(Symbol::intern(&concatenated), visited_span())),
+                Token::from_ast_ident(Ident::new(symbol, concatenated_span)),
                 Spacing::Alone,
             ));
         }
diff --git a/tests/ui/macros/macro-metavar-expr-concat/allowed-operations.rs b/tests/ui/macros/macro-metavar-expr-concat/allowed-operations.rs
@@ -37,6 +37,18 @@ macro_rules! without_dollar_sign_is_an_ident {
     };
 }
 
+macro_rules! literals {
+    ($ident:ident) => {{
+        let ${concat(_a, 'b')}: () = ();
+        let ${concat(_a, 1)}: () = ();
+        let ${concat(_a, "b")}: () = ();
+
+        let ${concat($ident, 'b')}: () = ();
+        let ${concat($ident, 1)}: () = ();
+        let ${concat($ident, "b")}: () = ();
+    }};
+}
+
 fn main() {
     create_things!(behold);
     behold_separated_idents_in_a_fn();
@@ -55,4 +67,6 @@ fn main() {
     without_dollar_sign_is_an_ident!(_123);
     assert_eq!(VARident, 1);
     assert_eq!(VAR_123, 2);
+
+    literals!(_hello);
 }
diff --git a/tests/ui/macros/macro-metavar-expr-concat/syntax-errors.rs b/tests/ui/macros/macro-metavar-expr-concat/syntax-errors.rs
@@ -11,9 +11,6 @@ macro_rules! wrong_concat_declarations {
         ${concat(aaaa,)}
         //~^ ERROR expected identifier
 
-        ${concat(aaaa, 1)}
-        //~^ ERROR expected identifier
-
         ${concat(_, aaaa)}
 
         ${concat(aaaa aaaa)}
@@ -30,9 +27,6 @@ macro_rules! wrong_concat_declarations {
 
         ${concat($ex, aaaa,)}
         //~^ ERROR expected identifier
-
-        ${concat($ex, aaaa, 123)}
-        //~^ ERROR expected identifier
     };
 }
 
@@ -43,8 +37,69 @@ macro_rules! dollar_sign_without_referenced_ident {
     };
 }
 
+macro_rules! starting_number {
+    ($ident:ident) => {{
+        let ${concat(1, $ident)}: () = ();
+        //~^ ERROR `${concat(..)}` is not generating a valid identifier
+    }};
+}
+
+macro_rules! starting_valid_unicode {
+    ($ident:ident) => {{
+        let ${concat('Ý', $ident)}: () = ();
+    }};
+}
+
+macro_rules! starting_invalid_unicode {
+    ($ident:ident) => {{
+        let ${concat("\u{999999}", $ident)}: () = ();
+        //~^ ERROR invalid unicode character escape
+        //~| ERROR expected identifier, found
+        //~| ERROR expected pattern, found
+    }};
+}
+
+macro_rules! ending_number {
+    ($ident:ident) => {{
+        let ${concat($ident, 1)}: () = ();
+    }};
+}
+
+macro_rules! ending_valid_unicode {
+    ($ident:ident) => {{
+        let ${concat($ident, 'Ý')}: () = ();
+    }};
+}
+
+macro_rules! ending_invalid_unicode {
+    ($ident:ident) => {{
+        let ${concat($ident, "\u{999999}")}: () = ();
+        //~^ ERROR invalid unicode character escape
+        //~| ERROR expected identifier, found
+        //~| ERROR expected pattern, found
+    }};
+}
+
+macro_rules! empty {
+    () => {{
+        let ${concat("", "")}: () = ();
+        //~^ ERROR `${concat(..)}` is not generating a valid identifier
+    }};
+}
+
+
 fn main() {
     wrong_concat_declarations!(1);
 
     dollar_sign_without_referenced_ident!(VAR);
+
+    starting_number!(_abc);
+    starting_valid_unicode!(_abc);
+    starting_invalid_unicode!(_abc);
+
+    ending_number!(_abc);
+    ending_valid_unicode!(_abc);
+    ending_invalid_unicode!(_abc);
+
+    empty!();
 }
diff --git a/tests/ui/macros/macro-metavar-expr-concat/syntax-errors.stderr b/tests/ui/macros/macro-metavar-expr-concat/syntax-errors.stderr
diff --git a/tests/ui/macros/rfc-3086-metavar-expr/syntax-errors.stderr b/tests/ui/macros/rfc-3086-metavar-expr/syntax-errors.stderr
