[clang] [clang-format] Recognize TableGen paste operator on separate line (PR #133722)

[via cfe-commits]

https://github.com/sstwcw created https://github.com/llvm/llvm-project/pull/133722

Formatting this piece of code made the program crash.

```
class TypedVecListRegOperand<RegisterClass Reg, int lanes, string eltsize>
    : RegisterOperand<Reg, "printTypedVectorList<" # lanes # ", '"
                                                   # eltsize # "'>">;
```

The line starting with the `#` was treated as a separate preprocessor directive line.  Then the code dereferenced a null pointer when it tried to continue parsing the first line that did not end in a semicolon.

Now the 2 problems are fixed.

>From b018450ed1a2901dd5b8cd5e146c1ce5d8613c86 Mon Sep 17 00:00:00 2001
From: sstwcw <su3e8a96kzlver at posteo.net>
Date: Mon, 31 Mar 2025 13:50:02 +0000
Subject: [PATCH] [clang-format] Recognize TableGen paste operator on separate
 line

Formatting this piece of code made the program crash.

```
class TypedVecListRegOperand<RegisterClass Reg, int lanes, string eltsize>
    : RegisterOperand<Reg, "printTypedVectorList<" # lanes # ", '"
                                                   # eltsize # "'>">;
```

The line starting with the `#` was treated as a separate preprocessor
directive line.  Then the code dereferenced a null pointer when it tried
to continue parsing the first line that did not end in a semicolon.

Now the 2 problems are fixed.
---
 clang/lib/Format/TokenAnnotator.cpp           |  2 ++
 clang/lib/Format/UnwrappedLineParser.cpp      | 11 +++++++++--
 clang/unittests/Format/FormatTestTableGen.cpp |  6 ++++++
 clang/unittests/Format/TokenAnnotatorTest.cpp | 17 +++++++++++++++++
 4 files changed, 34 insertions(+), 2 deletions(-)

diff --git a/clang/lib/Format/TokenAnnotator.cpp b/clang/lib/Format/TokenAnnotator.cpp
index d87b3a6088bd8..278355aa58586 100644
--- a/clang/lib/Format/TokenAnnotator.cpp
+++ b/clang/lib/Format/TokenAnnotator.cpp
@@ -949,6 +949,8 @@ class AnnotatingParser {
       HashTok->setType(TT_Unknown);
       if (!parseTableGenValue(ParseNameMode))
         return false;
+      if (!CurrentToken)
+        return true;
     }
     // In name mode, '{' is regarded as the end of the value.
     // See TGParser::ParseValue in TGParser.cpp
diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index f7712bea01c2c..aa0c372d5e15f 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -4853,9 +4853,16 @@ void UnwrappedLineParser::readToken(int LevelDifference) {
     PreviousWasComment = FormatTok->is(tok::comment);
 
     while (!Line->InPPDirective && FormatTok->is(tok::hash) &&
-           (!Style.isVerilog() ||
-            Keywords.isVerilogPPDirective(*Tokens->peekNextToken())) &&
            FirstNonCommentOnLine) {
+      // In Verilog, the backtick is used for macro invocations. In TableGen,
+      // the single hash is used for the paste operator.
+      const FormatToken *Next = Tokens->peekNextToken();
+      assert(Next); // There is an EOF token at the end.
+      if ((Style.isVerilog() && !Keywords.isVerilogPPDirective(*Next)) ||
+          (Style.isTableGen() &&
+           !Next->isOneOf(tok::pp_define, tok::pp_ifdef, tok::pp_ifndef))) {
+        break;
+      }
       distributeComments(Comments, FormatTok);
       Comments.clear();
       // If there is an unfinished unwrapped line, we flush the preprocessor
diff --git a/clang/unittests/Format/FormatTestTableGen.cpp b/clang/unittests/Format/FormatTestTableGen.cpp
index 92377c31f2e91..b78f79f20704f 100644
--- a/clang/unittests/Format/FormatTestTableGen.cpp
+++ b/clang/unittests/Format/FormatTestTableGen.cpp
@@ -218,6 +218,12 @@ TEST_F(FormatTestTableGen, PasteOperator) {
                "  string Z = [\"Traring\", \"Paste\", \"Traring\", \"Paste\",\n"
                "              \"Traring\", \"Paste\"]#;\n"
                "}");
+  verifyFormat("def x#x {}", "def x\n"
+                             "#x {}");
+  verifyFormat("def x#x {}", "def x\n"
+                             "#\n"
+                             "x {}");
+  verifyFormat("def x#x");
 }
 
 TEST_F(FormatTestTableGen, ClassDefinition) {
diff --git a/clang/unittests/Format/TokenAnnotatorTest.cpp b/clang/unittests/Format/TokenAnnotatorTest.cpp
index ac5e979aea071..fb8f5d30a669f 100644
--- a/clang/unittests/Format/TokenAnnotatorTest.cpp
+++ b/clang/unittests/Format/TokenAnnotatorTest.cpp
@@ -2832,6 +2832,23 @@ TEST_F(TokenAnnotatorTest, UnderstandTableGenTokens) {
   Tokens = Annotate("!cond");
   EXPECT_TOKEN(Tokens[0], tok::identifier, TT_TableGenCondOperator);
 
+  // The paste operator should not be treated as a preprocessor directive even
+  // if it is on a separate line.
+  Tokens = Annotate("def x\n"
+                    "#embed {}");
+  ASSERT_EQ(Tokens.size(), 7u) << Tokens;
+  EXPECT_TOKEN(Tokens[1], tok::identifier, TT_StartOfName);
+  EXPECT_TOKEN(Tokens[2], tok::hash, TT_Unknown);
+  EXPECT_EQ(Tokens[1]->Next, Tokens[2]);
+  Tokens = Annotate("def x\n"
+                    "#define x\n"
+                    "#embed {}");
+  ASSERT_EQ(Tokens.size(), 10u) << Tokens;
+  EXPECT_TOKEN(Tokens[1], tok::identifier, TT_StartOfName);
+  EXPECT_TOKEN(Tokens[2], tok::hash, TT_Unknown);
+  EXPECT_TOKEN(Tokens[5], tok::hash, TT_Unknown);
+  EXPECT_EQ(Tokens[1]->Next, Tokens[5]);
+
   auto AnnotateValue = [this, &Style](StringRef Code) {
     // Values are annotated only in specific context.
     auto Result = annotate(("def X { let V = " + Code + "; }").str(), Style);