mbe: Refactor diagnostics for invalid metavar expression syntax

Give a more user-friendly diagnostic about the following:

* Invalid syntax within the `${...}` braces, including missing
  parentheses or trailing tokens.
* Incorrect number of arguments passed to specific metavariable
  expressions.

Author: tgross35

compiler/rustc_expand/messages.ftl

@@ -133,6 +133,22 @@ expand_module_multiple_candidates =
 expand_must_repeat_once =
     this must repeat at least once
 
+expand_mve_extra_tokens =
+    unexpected trailing tokens
+    .label = for this metavariable expression
+    .note = the `{$name}` metavariable expression takes up to {$max_args} arguments
+    .suggestion = try removing {$extra_count ->
+        [one] this token
+        *[other] these tokens
+    }
+
+expand_mve_missing_paren =
+    expected `(`
+    .label = for this this metavariable expression
+    .unexpected = unexpected token
+    .note = metavariable expressions use function-like parentheses syntax
+    .suggestion = try adding parentheses
+
 expand_mve_unrecognized_var =
     variable `{$key}` is not recognized in meta-variable expression
 

compiler/rustc_expand/src/errors.rs

@@ -496,6 +496,37 @@ pub(crate) use metavar_exprs::*;
 mod metavar_exprs {
     use super::*;
 
+    #[derive(Diagnostic)]
+    #[diag(expand_mve_extra_tokens)]
+    pub(crate) struct MveExtraTokens {
+        #[primary_span]
+        #[suggestion(code = "", applicability = "machine-applicable")]
+        pub span: Span,
+        #[label]
+        pub ident_span: Span,
+        pub extra_count: usize,
+
+        // Only used for extra tokens in the arg position. These should be `Option`al but
+        // `derive(Diagnostic)` doesn't allow that.
+        #[note]
+        pub args_note: Option<()>,
+        pub max_args: usize,
+        pub name: &'static str,
+    }
+
+    #[derive(Diagnostic)]
+    #[note]
+    #[diag(expand_mve_missing_paren)]
+    pub(crate) struct MveMissingParen {
+        #[primary_span]
+        #[label]
+        pub ident_span: Span,
+        #[label(expand_unexpected)]
+        pub unexpected_span: Option<Span>,
+        #[suggestion(code = "( /* ... */ )", applicability = "has-placeholders")]
+        pub insert_span: Option<Span>,
+    }
+
     #[derive(Diagnostic)]
     #[diag(expand_mve_unrecognized_var)]
     pub(crate) struct MveUnrecognizedVar {

compiler/rustc_expand/src/mbe/metavar_expr.rs

@@ -7,9 +7,20 @@ use rustc_macros::{Decodable, Encodable};
 use rustc_session::parse::ParseSess;
 use rustc_span::{Ident, Span, Symbol};
 
+use crate::errors;
+
 pub(crate) const RAW_IDENT_ERR: &str = "`${concat(..)}` currently does not support raw identifiers";
 pub(crate) const UNSUPPORTED_CONCAT_ELEM_ERR: &str = "expected identifier or string literal";
 
+/// Map from expression names to the maximum arg count.
+const EXPR_NAME_ARG_MAP: &[(&str, Option<usize>)] = &[
+    ("concat", None),
+    ("count", Some(2)),
+    ("ignore", Some(1)),
+    ("index", Some(2)),
+    ("len", Some(2)),
+];
+
 /// A meta-variable expression, for expansions based on properties of meta-variables.
 #[derive(Debug, PartialEq, Encodable, Decodable)]
 pub(crate) enum MetaVarExpr {
@@ -40,11 +51,33 @@ impl MetaVarExpr {
     ) -> PResult<'psess, MetaVarExpr> {
         let mut iter = input.iter();
         let ident = parse_ident(&mut iter, psess, outer_span)?;
-        let Some(TokenTree::Delimited(.., Delimiter::Parenthesis, args)) = iter.next() else {
-            let msg = "meta-variable expression parameter must be wrapped in parentheses";
-            return Err(psess.dcx().struct_span_err(ident.span, msg));
+        let next = iter.next();
+        let Some(TokenTree::Delimited(.., Delimiter::Parenthesis, args)) = next else {
+            // No `()`; wrong or no delimiters
+            let (unexpected_span, insert_span) = match next {
+                Some(TokenTree::Delimited(..)) => (None, None),
+                Some(tt) => (Some(tt.span()), None),
+                None => (None, Some(ident.span.shrink_to_hi())),
+            };
+            let err =
+                errors::MveMissingParen { ident_span: ident.span, unexpected_span, insert_span };
+            return Err(psess.dcx().create_err(err));
         };
-        check_trailing_token(&mut iter, psess)?;
+
+        // Ensure there are no other tokens in the
+        if iter.peek().is_some() {
+            let span = iter_span(&iter).expect("checked is_some above");
+            let err = errors::MveExtraTokens {
+                span,
+                ident_span: ident.span,
+                extra_count: iter.count(),
+                args_note: None,
+                max_args: 0,
+                name: "",
+            };
+            return Err(psess.dcx().create_err(err));
+        }
+
         let mut iter = args.iter();
         let rslt = match ident.as_str() {
             "concat" => parse_concat(&mut iter, psess, outer_span, ident.span)?,
@@ -67,7 +100,7 @@ impl MetaVarExpr {
                 return Err(err);
             }
         };
-        check_trailing_token(&mut iter, psess)?;
+        check_trailing_tokens(&mut iter, psess, ident)?;
         Ok(rslt)
     }
 
@@ -87,20 +120,44 @@ impl MetaVarExpr {
     }
 }
 
-// Checks if there are any remaining tokens. For example, `${ignore(ident ... a b c ...)}`
-fn check_trailing_token<'psess>(
+/// Checks if there are any remaining tokens. For example, `${ignore(ident ... a b c ...)}`
+fn check_trailing_tokens<'psess>(
     iter: &mut TokenStreamIter<'_>,
     psess: &'psess ParseSess,
+    ident: Ident,
 ) -> PResult<'psess, ()> {
-    if let Some(tt) = iter.next() {
-        let mut diag = psess
-            .dcx()
-            .struct_span_err(tt.span(), format!("unexpected token: {}", pprust::tt_to_string(tt)));
-        diag.span_note(tt.span(), "meta-variable expression must not have trailing tokens");
-        Err(diag)
-    } else {
-        Ok(())
+    if iter.peek().is_none() {
+        // All tokens used, no problem
+        return Ok(());
     }
+
+    let (name, max) = EXPR_NAME_ARG_MAP
+        .iter()
+        .find(|(name, _)| *name == ident.as_str())
+        .expect("called with an invalid name");
+
+    // For expressions like `concat`, all tokens should be consumed already
+    let max =
+        max.unwrap_or_else(|| panic!("{name} takes unlimited tokens but didn't eat them all"));
+
+    let err = errors::MveExtraTokens {
+        span: iter_span(iter).expect("checked is_none above"),
+        ident_span: ident.span,
+        extra_count: iter.count(),
+        args_note: Some(()),
+        max_args: max,
+        name,
+    };
+    Err(psess.dcx().create_err(err))
+}
+
+/// Returns a span encompassing all tokens in the iterator if there is at least one item.
+fn iter_span(iter: &TokenStreamIter<'_>) -> Option<Span> {
+    let mut iter = iter.clone(); // cloning is cheap
+    let first_sp = iter.next()?.span();
+    let last_sp = iter.last().map(TokenTree::span).unwrap_or(first_sp);
+    let span = first_sp.with_hi(last_sp.hi());
+    Some(span)
 }
 
 /// Indicates what is placed in a `concat` parameter. For example, literals

tests/ui/macros/metavar-expressions/syntax-errors.rs

@@ -30,7 +30,7 @@ macro_rules! metavar_with_literal_suffix {
 
 macro_rules! mve_without_parens {
     ( $( $i:ident ),* ) => { ${ count } };
-    //~^ ERROR meta-variable expression parameter must be wrapped in parentheses
+    //~^ ERROR expected `(`
 }
 
 #[rustfmt::skip]
@@ -45,9 +45,14 @@ macro_rules! open_brackets_with_lit {
      //~^ ERROR expected identifier
  }
 
+macro_rules! mvs_missing_paren {
+    ( $( $i:ident ),* ) => { ${ count $i ($i) } };
+    //~^ ERROR expected `(`
+}
+
 macro_rules! mve_wrong_delim {
     ( $( $i:ident ),* ) => { ${ count{i} } };
-    //~^ ERROR meta-variable expression parameter must be wrapped in parentheses
+    //~^ ERROR expected `(`
 }
 
 macro_rules! invalid_metavar {
@@ -64,28 +69,28 @@ macro_rules! open_brackets_with_group {
 macro_rules! extra_garbage_after_metavar {
     ( $( $i:ident ),* ) => {
         ${count() a b c}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
         ${count($i a b c)}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
         ${count($i, 1 a b c)}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
         ${count($i) a b c}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
 
         ${ignore($i) a b c}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
         ${ignore($i a b c)}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
 
         ${index() a b c}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
         ${index(1 a b c)}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
 
         ${index() a b c}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
         ${index(1 a b c)}
-        //~^ ERROR unexpected token: a
+        //~^ ERROR unexpected trailing tokens
     };
 }