Support the `${concat(...)}` metavariable expression

I didn't follow rustc precisely, because I think it does some things wrongly (or they are FIXME), but I only allowed more code, not less. So we're all fine.

9 files changed, 304 insertions, 5 deletions

diff --git a/src/tools/rust-analyzer/Cargo.lock b/src/tools/rust-analyzer/Cargo.lock
index 6df73c2fd1b..5df48778e3d 100644
--- a/src/tools/rust-analyzer/Cargo.lock
+++ b/src/tools/rust-analyzer/Cargo.lock
@@ -1047,6 +1047,7 @@ dependencies = [
  "expect-test",
  "intern",
  "parser",
+ "ra-ap-rustc_lexer",
  "rustc-hash",
  "smallvec",
  "span",
diff --git a/src/tools/rust-analyzer/crates/hir-def/src/macro_expansion_tests/mbe/metavar_expr.rs b/src/tools/rust-analyzer/crates/hir-def/src/macro_expansion_tests/mbe/metavar_expr.rs
index bf701198387..1430e2a9a99 100644
--- a/src/tools/rust-analyzer/crates/hir-def/src/macro_expansion_tests/mbe/metavar_expr.rs
+++ b/src/tools/rust-analyzer/crates/hir-def/src/macro_expansion_tests/mbe/metavar_expr.rs
@@ -311,3 +311,150 @@ fn test() {
 "#]],
     );
 }
+
+#[test]
+fn concat() {
+    // FIXME: Should this error? rustc currently accepts it.
+    check(
+        r#"
+macro_rules! m {
+    ( $a:ident, $b:literal ) => {
+        let ${concat($a, _, "123", _foo, $b, _, 123)};
+    };
+}
+
+fn test() {
+    m!( abc, 456 );
+    m!( def, "hello" );
+}
+"#,
+        expect![[r#"
+macro_rules! m {
+    ( $a:ident, $b:literal ) => {
+        let ${concat($a, _, "123", _foo, $b, _, 123)};
+    };
+}
+
+fn test() {
+    let abc_123_foo456_123;;
+    let def_123_foohello_123;;
+}
+"#]],
+    );
+}
+
+#[test]
+fn concat_less_than_two_elements() {
+    // FIXME: Should this error? rustc currently accepts it.
+    check(
+        r#"
+macro_rules! m {
+    () => {
+        let ${concat(abc)};
+    };
+}
+
+fn test() {
+    m!()
+}
+"#,
+        expect![[r#"
+macro_rules! m {
+    () => {
+        let ${concat(abc)};
+    };
+}
+
+fn test() {
+    /* error: macro definition has parse errors */
+}
+"#]],
+    );
+}
+
+#[test]
+fn concat_invalid_ident() {
+    // FIXME: Should this error? rustc currently accepts it.
+    check(
+        r#"
+macro_rules! m {
+    () => {
+        let ${concat(abc, '"')};
+    };
+}
+
+fn test() {
+    m!()
+}
+"#,
+        expect![[r#"
+macro_rules! m {
+    () => {
+        let ${concat(abc, '"')};
+    };
+}
+
+fn test() {
+    /* error: `${concat(..)}` is not generating a valid identifier */let __ra_concat_dummy;
+}
+"#]],
+    );
+}
+
+#[test]
+fn concat_invalid_fragment() {
+    // FIXME: Should this error? rustc currently accepts it.
+    check(
+        r#"
+macro_rules! m {
+    ( $e:expr ) => {
+        let ${concat(abc, $e)};
+    };
+}
+
+fn test() {
+    m!(())
+}
+"#,
+        expect![[r#"
+macro_rules! m {
+    ( $e:expr ) => {
+        let ${concat(abc, $e)};
+    };
+}
+
+fn test() {
+    /* error: metavariables of `${concat(..)}` must be of type `ident`, `literal` or `tt` */let abc;
+}
+"#]],
+    );
+}
+
+#[test]
+fn concat_repetition() {
+    // FIXME: Should this error? rustc currently accepts it.
+    check(
+        r#"
+macro_rules! m {
+    ( $($i:ident)* ) => {
+        let ${concat(abc, $i)};
+    };
+}
+
+fn test() {
+    m!(a b c)
+}
+"#,
+        expect![[r#"
+macro_rules! m {
+    ( $($i:ident)* ) => {
+        let ${concat(abc, $i)};
+    };
+}
+
+fn test() {
+    /* error: expected simple binding, found nested binding `i` */let abc;
+}
+"#]],
+    );
+}
diff --git a/src/tools/rust-analyzer/crates/mbe/Cargo.toml b/src/tools/rust-analyzer/crates/mbe/Cargo.toml
index 413a0254a61..b37d57f2801 100644
--- a/src/tools/rust-analyzer/crates/mbe/Cargo.toml
+++ b/src/tools/rust-analyzer/crates/mbe/Cargo.toml
@@ -18,6 +18,7 @@ rustc-hash.workspace = true
 smallvec.workspace = true
 tracing.workspace = true
 arrayvec.workspace = true
+ra-ap-rustc_lexer.workspace = true
 
 # local deps
 syntax.workspace = true
diff --git a/src/tools/rust-analyzer/crates/mbe/src/benchmark.rs b/src/tools/rust-analyzer/crates/mbe/src/benchmark.rs
index f8c1d027c60..9e4ef140740 100644
--- a/src/tools/rust-analyzer/crates/mbe/src/benchmark.rs
+++ b/src/tools/rust-analyzer/crates/mbe/src/benchmark.rs
@@ -216,7 +216,11 @@ fn invocation_fixtures(
 
                 token_trees.push(subtree.into());
             }
-            Op::Ignore { .. } | Op::Index { .. } | Op::Count { .. } | Op::Len { .. } => {}
+            Op::Ignore { .. }
+            | Op::Index { .. }
+            | Op::Count { .. }
+            | Op::Len { .. }
+            | Op::Concat { .. } => {}
         };
 
         // Simple linear congruential generator for deterministic result
diff --git a/src/tools/rust-analyzer/crates/mbe/src/expander/matcher.rs b/src/tools/rust-analyzer/crates/mbe/src/expander/matcher.rs
index 90f56cc31da..7a009a4f6f3 100644
--- a/src/tools/rust-analyzer/crates/mbe/src/expander/matcher.rs
+++ b/src/tools/rust-analyzer/crates/mbe/src/expander/matcher.rs
@@ -584,7 +584,11 @@ fn match_loop_inner<'t>(
                 error_items.push(item);
             }
             OpDelimited::Op(
-                Op::Ignore { .. } | Op::Index { .. } | Op::Count { .. } | Op::Len { .. },
+                Op::Ignore { .. }
+                | Op::Index { .. }
+                | Op::Count { .. }
+                | Op::Len { .. }
+                | Op::Concat { .. },
             ) => {
                 stdx::never!("metavariable expression in lhs found");
             }
@@ -879,7 +883,11 @@ fn collect_vars(collector_fun: &mut impl FnMut(Symbol), pattern: &MetaTemplate)
             Op::Subtree { tokens, .. } => collect_vars(collector_fun, tokens),
             Op::Repeat { tokens, .. } => collect_vars(collector_fun, tokens),
             Op::Literal(_) | Op::Ident(_) | Op::Punct(_) => {}
-            Op::Ignore { .. } | Op::Index { .. } | Op::Count { .. } | Op::Len { .. } => {
+            Op::Ignore { .. }
+            | Op::Index { .. }
+            | Op::Count { .. }
+            | Op::Len { .. }
+            | Op::Concat { .. } => {
                 stdx::never!("metavariable expression in lhs found");
             }
         }
diff --git a/src/tools/rust-analyzer/crates/mbe/src/expander/transcriber.rs b/src/tools/rust-analyzer/crates/mbe/src/expander/transcriber.rs
index 00844115355..1db2f35d262 100644
--- a/src/tools/rust-analyzer/crates/mbe/src/expander/transcriber.rs
+++ b/src/tools/rust-analyzer/crates/mbe/src/expander/transcriber.rs
@@ -2,12 +2,12 @@
 //! `$ident => foo`, interpolates variables in the template, to get `fn foo() {}`
 
 use intern::{sym, Symbol};
-use span::Span;
+use span::{Edition, Span};
 use tt::Delimiter;
 
 use crate::{
     expander::{Binding, Bindings, Fragment},
-    parser::{MetaVarKind, Op, RepeatKind, Separator},
+    parser::{ConcatMetaVarExprElem, MetaVarKind, Op, RepeatKind, Separator},
     ExpandError, ExpandErrorKind, ExpandResult, MetaTemplate,
 };
 
@@ -312,6 +312,82 @@ fn expand_subtree(
                     .into(),
                 );
             }
+            Op::Concat { elements, span: concat_span } => {
+                let mut concatenated = String::new();
+                for element in elements {
+                    match element {
+                        ConcatMetaVarExprElem::Ident(ident) => {
+                            concatenated.push_str(ident.sym.as_str())
+                        }
+                        ConcatMetaVarExprElem::Literal(lit) => {
+                            // FIXME: This isn't really correct wrt. escaping, but that's what rustc does and anyway
+                            // escaping is used most of the times for characters that are invalid in identifiers.
+                            concatenated.push_str(lit.symbol.as_str())
+                        }
+                        ConcatMetaVarExprElem::Var(var) => {
+                            // Handling of repetitions in `${concat}` isn't fleshed out in rustc, so we currently
+                            // err at it.
+                            // FIXME: Do what rustc does for repetitions.
+                            let var_value = match ctx.bindings.get_fragment(
+                                &var.sym,
+                                var.span,
+                                &mut ctx.nesting,
+                                marker,
+                            ) {
+                                Ok(var) => var,
+                                Err(e) => {
+                                    if err.is_none() {
+                                        err = Some(e);
+                                    };
+                                    continue;
+                                }
+                            };
+                            let value = match &var_value {
+                                Fragment::Tokens(tt::TokenTree::Leaf(tt::Leaf::Ident(ident))) => {
+                                    ident.sym.as_str()
+                                }
+                                Fragment::Tokens(tt::TokenTree::Leaf(tt::Leaf::Literal(lit))) => {
+                                    lit.symbol.as_str()
+                                }
+                                _ => {
+                                    if err.is_none() {
+                                        err = Some(ExpandError::binding_error(var.span, "metavariables of `${concat(..)}` must be of type `ident`, `literal` or `tt`"))
+                                    }
+                                    continue;
+                                }
+                            };
+                            concatenated.push_str(value);
+                        }
+                    }
+                }
+
+                // `${concat}` span comes from the macro (at least for now).
+                // See https://github.com/rust-lang/rust/blob/b0af276da341/compiler/rustc_expand/src/mbe/transcribe.rs#L724-L726.
+                let mut result_span = *concat_span;
+                marker(&mut result_span);
+
+                // FIXME: NFC normalize the result.
+                if !rustc_lexer::is_ident(&concatenated) {
+                    if err.is_none() {
+                        err = Some(ExpandError::binding_error(
+                            *concat_span,
+                            "`${concat(..)}` is not generating a valid identifier",
+                        ));
+                    }
+                    // Insert a dummy identifier for better parsing.
+                    concatenated.clear();
+                    concatenated.push_str("__ra_concat_dummy");
+                }
+
+                let needs_raw =
+                    parser::SyntaxKind::from_keyword(&concatenated, Edition::LATEST).is_some();
+                let is_raw = if needs_raw { tt::IdentIsRaw::Yes } else { tt::IdentIsRaw::No };
+                arena.push(tt::TokenTree::Leaf(tt::Leaf::Ident(tt::Ident {
+                    is_raw,
+                    span: result_span,
+                    sym: Symbol::intern(&concatenated),
+                })));
+            }
         }
     }
     // drain the elements added in this instance of expand_subtree
diff --git a/src/tools/rust-analyzer/crates/mbe/src/lib.rs b/src/tools/rust-analyzer/crates/mbe/src/lib.rs
index f7b6e8f54b5..dd71e46db36 100644
--- a/src/tools/rust-analyzer/crates/mbe/src/lib.rs
+++ b/src/tools/rust-analyzer/crates/mbe/src/lib.rs
@@ -6,6 +6,11 @@
 //! The tests for this functionality live in another crate:
 //! `hir_def::macro_expansion_tests::mbe`.
 
+#[cfg(not(feature = "in-rust-tree"))]
+extern crate ra_ap_rustc_lexer as rustc_lexer;
+#[cfg(feature = "in-rust-tree")]
+extern crate rustc_lexer;
+
 mod expander;
 mod parser;
 
diff --git a/src/tools/rust-analyzer/crates/mbe/src/parser.rs b/src/tools/rust-analyzer/crates/mbe/src/parser.rs
index 2efe318f614..b55edf4a5e0 100644
--- a/src/tools/rust-analyzer/crates/mbe/src/parser.rs
+++ b/src/tools/rust-analyzer/crates/mbe/src/parser.rs
@@ -84,6 +84,10 @@ pub(crate) enum Op {
         // FIXME: `usize`` once we drop support for 1.76
         depth: Option<usize>,
     },
+    Concat {
+        elements: Box<[ConcatMetaVarExprElem]>,
+        span: Span,
+    },
     Repeat {
         tokens: MetaTemplate,
         kind: RepeatKind,
@@ -98,6 +102,18 @@ pub(crate) enum Op {
     Ident(tt::Ident<Span>),
 }
 
+#[derive(Clone, Debug, PartialEq, Eq)]
+pub(crate) enum ConcatMetaVarExprElem {
+    /// There is NO preceding dollar sign, which means that this identifier should be interpreted
+    /// as a literal.
+    Ident(tt::Ident<Span>),
+    /// There is a preceding dollar sign, which means that this identifier should be expanded
+    /// and interpreted as a variable.
+    Var(tt::Ident<Span>),
+    /// For example, a number or a string.
+    Literal(tt::Literal<Span>),
+}
+
 #[derive(Copy, Clone, Debug, PartialEq, Eq)]
 pub(crate) enum RepeatKind {
     ZeroOrMore,
@@ -384,6 +400,32 @@ fn parse_metavar_expr(src: &mut TtIter<'_, Span>) -> Result<Op, ()> {
             let depth = if try_eat_comma(&mut args) { Some(parse_depth(&mut args)?) } else { None };
             Op::Count { name: ident.sym.clone(), depth }
         }
+        s if sym::concat == *s => {
+            let mut elements = Vec::new();
+            while let Some(next) = args.peek_n(0) {
+                let element = if let tt::TokenTree::Leaf(tt::Leaf::Literal(lit)) = next {
+                    args.next().expect("already peeked");
+                    ConcatMetaVarExprElem::Literal(lit.clone())
+                } else {
+                    let is_var = try_eat_dollar(&mut args);
+                    let ident = args.expect_ident_or_underscore()?.clone();
+
+                    if is_var {
+                        ConcatMetaVarExprElem::Var(ident)
+                    } else {
+                        ConcatMetaVarExprElem::Ident(ident)
+                    }
+                };
+                elements.push(element);
+                if args.peek_n(0).is_some() {
+                    args.expect_comma()?;
+                }
+            }
+            if elements.len() < 2 {
+                return Err(());
+            }
+            Op::Concat { elements: elements.into_boxed_slice(), span: func.span }
+        }
         _ => return Err(()),
     };
 
@@ -414,3 +456,11 @@ fn try_eat_comma(src: &mut TtIter<'_, Span>) -> bool {
     }
     false
 }
+
+fn try_eat_dollar(src: &mut TtIter<'_, Span>) -> bool {
+    if let Some(tt::TokenTree::Leaf(tt::Leaf::Punct(tt::Punct { char: '$', .. }))) = src.peek_n(0) {
+        let _ = src.next();
+        return true;
+    }
+    false
+}
diff --git a/src/tools/rust-analyzer/crates/tt/src/iter.rs b/src/tools/rust-analyzer/crates/tt/src/iter.rs
index e96bed0319e..587b903aa97 100644
--- a/src/tools/rust-analyzer/crates/tt/src/iter.rs
+++ b/src/tools/rust-analyzer/crates/tt/src/iter.rs
@@ -57,6 +57,13 @@ impl<'a, S: Copy> TtIter<'a, S> {
         }
     }
 
+    pub fn expect_comma(&mut self) -> Result<(), ()> {
+        match self.expect_leaf()? {
+            Leaf::Punct(Punct { char: ',', .. }) => Ok(()),
+            _ => Err(()),
+        }
+    }
+
     pub fn expect_ident(&mut self) -> Result<&'a Ident<S>, ()> {
         match self.expect_leaf()? {
             Leaf::Ident(it) if it.sym != sym::underscore => Ok(it),