[clang] [Clang][Preprocessor] Expand UCNs in macro concatenation (PR #145351)

[via cfe-commits]

https://github.com/yronglin created https://github.com/llvm/llvm-project/pull/145351

Fixs https://github.com/llvm/llvm-project/issues/145240.

>From 4892892f8a40c7fe3ebf58fcdb488a08b1eadd91 Mon Sep 17 00:00:00 2001
From: yronglin <yronglin777 at gmail.com>
Date: Tue, 24 Jun 2025 00:00:52 +0800
Subject: [PATCH] [Clang][Preprocessor] Expand UCNs in macro concatenation

Signed-off-by: yronglin <yronglin777 at gmail.com>
---
 clang/docs/ReleaseNotes.rst                          |  1 +
 clang/lib/Lex/TokenLexer.cpp                         | 11 +++++++++++
 clang/test/Preprocessor/macro_paste_identifier_ucn.c | 10 ++++++++++
 3 files changed, 22 insertions(+)
 create mode 100644 clang/test/Preprocessor/macro_paste_identifier_ucn.c

diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 96477ef6ddc9a..af107a2d51062 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -720,6 +720,7 @@ Bug Fixes in This Version
 - Fixed incorrect token location when emitting diagnostics for tokens expanded from macros. (#GH143216)
 - Fixed an infinite recursion when checking constexpr destructors. (#GH141789)
 - Fixed a crash when a malformed using declaration appears in a ``constexpr`` function. (#GH144264)
+- Fixed a bug when use unicode character name in macro concatenation. (#GH145240) 
 
 Bug Fixes to Compiler Builtins
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
diff --git a/clang/lib/Lex/TokenLexer.cpp b/clang/lib/Lex/TokenLexer.cpp
index 6e93416e01c0c..72f1ffa7ed06e 100644
--- a/clang/lib/Lex/TokenLexer.cpp
+++ b/clang/lib/Lex/TokenLexer.cpp
@@ -748,6 +748,7 @@ bool TokenLexer::pasteTokens(Token &LHSTok, ArrayRef<Token> TokenStream,
   const char *ResultTokStrPtr = nullptr;
   SourceLocation StartLoc = LHSTok.getLocation();
   SourceLocation PasteOpLoc;
+  bool HasUCNs = false;
 
   auto IsAtEnd = [&TokenStream, &CurIdx] {
     return TokenStream.size() == CurIdx;
@@ -885,6 +886,9 @@ bool TokenLexer::pasteTokens(Token &LHSTok, ArrayRef<Token> TokenStream,
 
     // Finally, replace LHS with the result, consume the RHS, and iterate.
     ++CurIdx;
+
+    // Set Token::HasUCN flag if LHS or RHS contains any UCNs.
+    HasUCNs = LHSTok.hasUCN() || RHS.hasUCN() || HasUCNs;
     LHSTok = Result;
   } while (!IsAtEnd() && TokenStream[CurIdx].is(tok::hashhash));
 
@@ -913,6 +917,13 @@ bool TokenLexer::pasteTokens(Token &LHSTok, ArrayRef<Token> TokenStream,
   // token pasting re-lexes the result token in raw mode, identifier information
   // isn't looked up.  As such, if the result is an identifier, look up id info.
   if (LHSTok.is(tok::raw_identifier)) {
+
+    // If there has any UNCs in concated token, we should mark this token
+    // with Token::HasUCN flag, then LookUpIdentifierInfo will expand UCNs in
+    // token.
+    if (HasUCNs)
+      LHSTok.setFlag(Token::HasUCN);
+
     // Look up the identifier info for the token.  We disabled identifier lookup
     // by saying we're skipping contents, so we need to do this manually.
     PP.LookUpIdentifierInfo(LHSTok);
diff --git a/clang/test/Preprocessor/macro_paste_identifier_ucn.c b/clang/test/Preprocessor/macro_paste_identifier_ucn.c
new file mode 100644
index 0000000000000..c9eb8190edfe8
--- /dev/null
+++ b/clang/test/Preprocessor/macro_paste_identifier_ucn.c
@@ -0,0 +1,10 @@
+// RUN: %clang_cc1 -fms-extensions %s -verify
+// RUN: %clang_cc1 -E -fms-extensions %s | FileCheck %s
+// expected-no-diagnostics
+
+#define CAT(a,b) a##b
+
+char foo\u00b5;
+char*p = &CAT(foo, \u00b5);
+// CHECK: char fooµ;
+// CHECK-NEXT: char*p = &fooµ;