[PATCH 028/125] gccrs: ast: Add base nodes for FormatArgs

[Arthur Cohen]

This commit adds a base for creating AST FormatArgs nodes after expanding
invocations of `format_args!()`. These nodes will then be expanded to
the proper runtime function calls (to core::fmt::rt) during the AST
lowering.

gcc/rust/ChangeLog:

        * ast/rust-builtin-ast-nodes.h: New file.
        * ast/rust-ast-full-decls.h (class FormatArgs): Declare new class.
        * ast/rust-ast-collector.cc: Handle FormatArgs nodes properly.
        * ast/rust-ast-collector.h: Likewise.
        * ast/rust-ast-full.h: Likewise.
        * ast/rust-ast-visitor.cc: Likewise.
        * ast/rust-ast-visitor.h: Likewise.
        * ast/rust-ast.cc: Likewise.
        * ast/rust-ast.h: Likewise.
        * expand/rust-derive.h: Likewise.
        * hir/rust-ast-lower-base.cc: Likewise.
        * hir/rust-ast-lower-base.h: Likewise.
        * hir/rust-ast-lower-expr.cc: Likewise.
        * hir/rust-ast-lower-expr.h: Likewise.
        * resolve/rust-ast-resolve-base.cc: Likewise.
        * resolve/rust-ast-resolve-base.h: Likewise.
---
 gcc/rust/ast/rust-ast-collector.cc        |   8 ++
 gcc/rust/ast/rust-ast-collector.h         |   2 +
 gcc/rust/ast/rust-ast-full-decls.h        |   3 +
 gcc/rust/ast/rust-ast-full.h              |   1 +
 gcc/rust/ast/rust-ast-visitor.cc          |   6 +
 gcc/rust/ast/rust-ast-visitor.h           |   5 +
 gcc/rust/ast/rust-ast.cc                  |   6 +
 gcc/rust/ast/rust-ast.h                   |   1 +
 gcc/rust/ast/rust-builtin-ast-nodes.h     | 129 ++++++++++++++++++++++
 gcc/rust/expand/rust-derive.h             |   1 +
 gcc/rust/hir/rust-ast-lower-base.cc       |   5 +
 gcc/rust/hir/rust-ast-lower-base.h        |   3 +
 gcc/rust/hir/rust-ast-lower-expr.cc       |   8 ++
 gcc/rust/hir/rust-ast-lower-expr.h        |   4 +
 gcc/rust/resolve/rust-ast-resolve-base.cc |   6 +
 gcc/rust/resolve/rust-ast-resolve-base.h  |   3 +
 16 files changed, 191 insertions(+)
 create mode 100644 gcc/rust/ast/rust-builtin-ast-nodes.h

diff --git a/gcc/rust/ast/rust-ast-collector.cc 
b/gcc/rust/ast/rust-ast-collector.cc
index 181f1b100be..b8ec62367bc 100644
--- a/gcc/rust/ast/rust-ast-collector.cc
+++ b/gcc/rust/ast/rust-ast-collector.cc
@@ -16,6 +16,8 @@
 // along with GCC; see the file COPYING3.  If not see
 // <http://www.gnu.org/licenses/>.
 #include "rust-ast-collector.h"
+#include "rust-ast.h"
+#include "rust-diagnostics.h"
 #include "rust-item.h"
 #include "rust-keyword-values.h"
 
@@ -2805,5 +2807,11 @@ TokenCollector::visit (BareFunctionType &type)
     }
 }
 
+void
+TokenCollector::visit (AST::FormatArgs &fmt)
+{
+  rust_sorry_at (0, "unimplemented format_args!() visitor");
+}
+
 } // namespace AST
 } // namespace Rust
diff --git a/gcc/rust/ast/rust-ast-collector.h 
b/gcc/rust/ast/rust-ast-collector.h
index cf97c185a26..ec695ef5b31 100644
--- a/gcc/rust/ast/rust-ast-collector.h
+++ b/gcc/rust/ast/rust-ast-collector.h
@@ -398,6 +398,8 @@ public:
   void visit (SliceType &type);
   void visit (InferredType &type);
   void visit (BareFunctionType &type);
+
+  void visit (FormatArgs &fmt);
 };
 } // namespace AST
 
diff --git a/gcc/rust/ast/rust-ast-full-decls.h 
b/gcc/rust/ast/rust-ast-full-decls.h
index c96bbfb07d9..8d5c8dbc821 100644
--- a/gcc/rust/ast/rust-ast-full-decls.h
+++ b/gcc/rust/ast/rust-ast-full-decls.h
@@ -267,6 +267,9 @@ class SliceType;
 class InferredType;
 struct MaybeNamedParam;
 class BareFunctionType;
+
+// rust-builtin-ast-nodes.h
+class FormatArgs;
 } // namespace AST
 } // namespace Rust
 
diff --git a/gcc/rust/ast/rust-ast-full.h b/gcc/rust/ast/rust-ast-full.h
index f2152193a13..ebd38f2520c 100644
--- a/gcc/rust/ast/rust-ast-full.h
+++ b/gcc/rust/ast/rust-ast-full.h
@@ -28,5 +28,6 @@
 #include "rust-stmt.h"
 #include "rust-type.h"
 #include "rust-macro.h"
+#include "rust-builtin-ast-nodes.h"
 
 #endif
diff --git a/gcc/rust/ast/rust-ast-visitor.cc b/gcc/rust/ast/rust-ast-visitor.cc
index 27b9aa47d99..c72e2d72f6d 100644
--- a/gcc/rust/ast/rust-ast-visitor.cc
+++ b/gcc/rust/ast/rust-ast-visitor.cc
@@ -1395,6 +1395,12 @@ DefaultASTVisitor::visit (AST::BareFunctionType &type)
     visit (type.get_return_type ());
 }
 
+void
+DefaultASTVisitor::visit (AST::FormatArgs &)
+{
+  // FIXME: Do we have anything to do? any subnodes to visit? Probably, right?
+}
+
 void
 DefaultASTVisitor::visit (AST::VariadicParam &param)
 {
diff --git a/gcc/rust/ast/rust-ast-visitor.h b/gcc/rust/ast/rust-ast-visitor.h
index 6c9715eb077..c5c9a025ba6 100644
--- a/gcc/rust/ast/rust-ast-visitor.h
+++ b/gcc/rust/ast/rust-ast-visitor.h
@@ -229,6 +229,9 @@ public:
   virtual void visit (InferredType &type) = 0;
   virtual void visit (BareFunctionType &type) = 0;
 
+  // special AST nodes for certain builtin macros such as `asm!()`
+  virtual void visit (FormatArgs &fmt) = 0;
+
   // TODO: rust-cond-compilation.h visiting? not currently used
 };
 
@@ -390,6 +393,7 @@ protected:
   virtual void visit (AST::SelfParam &self) override;
   virtual void visit (AST::FunctionParam &param) override;
   virtual void visit (AST::VariadicParam &param) override;
+  virtual void visit (AST::FormatArgs &fmt) override;
 
   template <typename T> void visit (T &node) { node.accept_vis (*this); }
 
@@ -422,6 +426,7 @@ protected:
   virtual void visit (AST::MacroTranscriber &transcriber);
   virtual void visit (AST::StructPatternElements &spe);
   virtual void visit (AST::MaybeNamedParam &param);
+
   void visit (AST::Attribute &attribute) {}
 
   template <typename T> void visit_outer_attrs (T &node)
diff --git a/gcc/rust/ast/rust-ast.cc b/gcc/rust/ast/rust-ast.cc
index 90fe9269404..5d571b46622 100644
--- a/gcc/rust/ast/rust-ast.cc
+++ b/gcc/rust/ast/rust-ast.cc
@@ -5048,6 +5048,12 @@ MetaWord::accept_vis (ASTVisitor &vis)
   vis.visit (*this);
 }
 
+void
+FormatArgs::accept_vis (ASTVisitor &vis)
+{
+  vis.visit (*this);
+}
+
 } // namespace AST
 
 std::ostream &
diff --git a/gcc/rust/ast/rust-ast.h b/gcc/rust/ast/rust-ast.h
index 1422d77eade..c4d5858dd63 100644
--- a/gcc/rust/ast/rust-ast.h
+++ b/gcc/rust/ast/rust-ast.h
@@ -2029,6 +2029,7 @@ public:
 class PathExpr : public ExprWithoutBlock
 {
 };
+
 } // namespace AST
 } // namespace Rust
 
diff --git a/gcc/rust/ast/rust-builtin-ast-nodes.h 
b/gcc/rust/ast/rust-builtin-ast-nodes.h
new file mode 100644
index 00000000000..3998fbfb4a7
--- /dev/null
+++ b/gcc/rust/ast/rust-builtin-ast-nodes.h
@@ -0,0 +1,129 @@
+// Copyright (C) 2024 Free Software Foundation, Inc.
+
+// This file is part of GCC.
+
+// GCC is free software; you can redistribute it and/or modify it under
+// the terms of the GNU General Public License as published by the Free
+// Software Foundation; either version 3, or (at your option) any later
+// version.
+
+// GCC is distributed in the hope that it will be useful, but WITHOUT ANY
+// WARRANTY; without even the implied warranty of MERCHANTABILITY or
+// FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+// for more details.
+
+// You should have received a copy of the GNU General Public License
+// along with GCC; see the file COPYING3.  If not see
+// <http://www.gnu.org/licenses/>.
+
+#ifndef RUST_AST_BUILTIN_NODES_H
+#define RUST_AST_BUILTIN_NODES_H
+
+#include "rust-system.h"
+#include "line-map.h"
+#include "optional.h"
+#include "rust-ast.h"
+#include "rust-fmt.h"
+
+namespace Rust {
+namespace AST {
+
+// Definitions, from rustc's `FormatArgs` AST struct
+// 
https://github.com/rust-lang/rust/blob/1be468815c/compiler/rustc_ast/src/format.rs
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+// └──────────────────────────────────────────────┘
+//                     FormatArgs
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//                                     └─────────┘
+//                                      argument
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//              └───────────────────┘
+//                     template
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//               └────┘└─────────┘└┘
+//                      pieces
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//               └────┘           └┘
+//                   literal pieces
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//                     └─────────┘
+//                     placeholder
+//
+// format_args!("hello {abc:.xyz$}!!", abc="world");
+//                      └─┘  └─┘
+//                      positions (could be names, numbers, empty, or `*`)
+
+class FormatArgumentKind
+{
+public:
+  Identifier &get_ident ()
+  {
+    rust_assert (kind == Kind::Captured || kind == Kind::Named);
+
+    return ident.value ();
+  }
+
+private:
+  enum class Kind
+  {
+    Normal,
+    Named,
+    Captured,
+  } kind;
+
+  tl::optional<Identifier> ident;
+};
+
+class FormatArgument
+{
+  FormatArgumentKind kind;
+  std::unique_ptr<Expr> expr;
+};
+
+class FormatArguments
+{
+  std::vector<FormatArgument> args;
+};
+
+// TODO: Format documentation better
+// Having a separate AST node for `format_args!()` expansion allows some
+// important optimizations which help reduce generated code a lot. For example,
+// turning `format_args!("a {} {} {}", 15, "hey", 'a')` directly into
+// `format_args!("a 15 hey a")`, since all arguments are literals. Or,
+// flattening imbricated `format_args!()` calls: `format_args!("heyo {}",
+// format_args!("result: {}", some_result))` -> `format_args!("heyo result: 
{}",
+// some_result)`
+// FIXME: Move to rust-macro.h
+class FormatArgs : public Visitable
+{
+public:
+  enum class Newline
+  {
+    Yes,
+    No
+  };
+
+  FormatArgs (location_t loc, Fmt::PieceSlice template_str,
+             FormatArguments arguments)
+    : loc (loc), template_str (std::move (template_str)),
+      arguments (std::move (arguments))
+  {}
+
+  void accept_vis (AST::ASTVisitor &vis);
+
+private:
+  location_t loc;
+  Fmt::PieceSlice template_str;
+  FormatArguments arguments;
+};
+
+} // namespace AST
+} // namespace Rust
+
+#endif // ! RUST_AST_BUILTIN_NODES_H
diff --git a/gcc/rust/expand/rust-derive.h b/gcc/rust/expand/rust-derive.h
index 8fe29c0db75..f953c3decbf 100644
--- a/gcc/rust/expand/rust-derive.h
+++ b/gcc/rust/expand/rust-derive.h
@@ -221,6 +221,7 @@ private:
   virtual void visit (SelfParam &param) override final{};
   virtual void visit (FunctionParam &param) override final{};
   virtual void visit (VariadicParam &param) override final{};
+  virtual void visit (FormatArgs &param) override final{};
 };
 
 } // namespace AST
diff --git a/gcc/rust/hir/rust-ast-lower-base.cc 
b/gcc/rust/hir/rust-ast-lower-base.cc
index bcc4797df77..fa37d62d026 100644
--- a/gcc/rust/hir/rust-ast-lower-base.cc
+++ b/gcc/rust/hir/rust-ast-lower-base.cc
@@ -20,6 +20,7 @@
 #include "rust-ast-lower-type.h"
 #include "rust-ast-lower-pattern.h"
 #include "rust-ast-lower-extern.h"
+#include "rust-ast.h"
 #include "rust-attribute-values.h"
 #include "rust-item.h"
 #include "rust-system.h"
@@ -523,6 +524,10 @@ void
 ASTLoweringBase::visit (AST::SelfParam &param)
 {}
 
+void
+ASTLoweringBase::visit (AST::FormatArgs &fmt)
+{}
+
 HIR::Lifetime
 ASTLoweringBase::lower_lifetime (AST::Lifetime &lifetime,
                                 bool default_to_static_lifetime)
diff --git a/gcc/rust/hir/rust-ast-lower-base.h 
b/gcc/rust/hir/rust-ast-lower-base.h
index d7b94ac8a70..7b0ce375f83 100644
--- a/gcc/rust/hir/rust-ast-lower-base.h
+++ b/gcc/rust/hir/rust-ast-lower-base.h
@@ -19,6 +19,7 @@
 #ifndef RUST_AST_LOWER_BASE
 #define RUST_AST_LOWER_BASE
 
+#include "rust-ast.h"
 #include "rust-system.h"
 #include "rust-ast-full.h"
 #include "rust-ast-visitor.h"
@@ -253,6 +254,8 @@ public:
   virtual void visit (AST::VariadicParam &param);
   virtual void visit (AST::SelfParam &param);
 
+  virtual void visit (AST::FormatArgs &fmt);
+
 protected:
   ASTLoweringBase ()
     : mappings (Analysis::Mappings::get ()),
diff --git a/gcc/rust/hir/rust-ast-lower-expr.cc 
b/gcc/rust/hir/rust-ast-lower-expr.cc
index 35deb57abe6..8e07b19df4c 100644
--- a/gcc/rust/hir/rust-ast-lower-expr.cc
+++ b/gcc/rust/hir/rust-ast-lower-expr.cc
@@ -22,6 +22,8 @@
 #include "rust-ast-lower-struct-field-expr.h"
 #include "rust-ast-lower-pattern.h"
 #include "rust-ast-lower-type.h"
+#include "rust-ast.h"
+#include "rust-diagnostics.h"
 
 namespace Rust {
 namespace HIR {
@@ -828,5 +830,11 @@ ASTLoweringExpr::visit (AST::ClosureExprInnerTyped &expr)
                            expr.get_locus ());
 }
 
+void
+ASTLoweringExpr::visit (AST::FormatArgs &fmt)
+{
+  rust_sorry_at (0, "unimplemented format_args!() visitor");
+}
+
 } // namespace HIR
 } // namespace Rust
diff --git a/gcc/rust/hir/rust-ast-lower-expr.h 
b/gcc/rust/hir/rust-ast-lower-expr.h
index 56da9d1ab88..168dd014002 100644
--- a/gcc/rust/hir/rust-ast-lower-expr.h
+++ b/gcc/rust/hir/rust-ast-lower-expr.h
@@ -20,6 +20,7 @@
 #define RUST_AST_LOWER_EXPR
 
 #include "rust-ast-lower-base.h"
+#include "rust-ast.h"
 
 namespace Rust {
 namespace HIR {
@@ -121,6 +122,9 @@ public:
   void visit (AST::ClosureExprInner &expr) override;
   void visit (AST::ClosureExprInnerTyped &expr) override;
 
+  // Extra visitor for FormatArgs nodes
+  void visit (AST::FormatArgs &fmt) override;
+
 private:
   ASTLoweringExpr ();
 
diff --git a/gcc/rust/resolve/rust-ast-resolve-base.cc 
b/gcc/rust/resolve/rust-ast-resolve-base.cc
index c88bd58860a..04a0bb65ec2 100644
--- a/gcc/rust/resolve/rust-ast-resolve-base.cc
+++ b/gcc/rust/resolve/rust-ast-resolve-base.cc
@@ -650,5 +650,11 @@ void
 ResolverBase::visit (AST::FunctionParam &)
 {}
 
+void
+ResolverBase::visit (AST::FormatArgs &fmt)
+{
+  rust_sorry_at (0, "unimplemented format_args!() visitor");
+}
+
 } // namespace Resolver
 } // namespace Rust
diff --git a/gcc/rust/resolve/rust-ast-resolve-base.h 
b/gcc/rust/resolve/rust-ast-resolve-base.h
index 9bc64331f5a..3b4d28618e4 100644
--- a/gcc/rust/resolve/rust-ast-resolve-base.h
+++ b/gcc/rust/resolve/rust-ast-resolve-base.h
@@ -20,6 +20,7 @@
 #define RUST_AST_RESOLVE_BASE_H
 
 #include "rust-ast-visitor.h"
+#include "rust-ast.h"
 #include "rust-name-resolver.h"
 #include "rust-diagnostics.h"
 #include "rust-location.h"
@@ -198,6 +199,8 @@ public:
   void visit (AST::VariadicParam &param);
   void visit (AST::SelfParam &param);
 
+  void visit (AST::FormatArgs &fmt);
+
 protected:
   ResolverBase ()
     : resolver (Resolver::get ()), mappings (Analysis::Mappings::get ()),
-- 
2.45.2