[clang] [clang-format] Handle templated elaborated type specifier in function… (PR #77013)

via cfe-commits cfe-commits at lists.llvm.org
Sat Jan 20 14:43:09 PST 2024


https://github.com/XDeme updated https://github.com/llvm/llvm-project/pull/77013

>From 45d01cbc7ec958518b1739daa9e9b0dc35c2d194 Mon Sep 17 00:00:00 2001
From: XDeme <fernando.tagawa.gamail.com at gmail.com>
Date: Thu, 4 Jan 2024 19:04:21 -0300
Subject: [PATCH 1/8] [clang-format] Handle templated elaborated type specifier
 in function return type. The behaviour now is consistent with the non
 templated version

---
 clang/lib/Format/UnwrappedLineParser.cpp | 8 ++++++++
 clang/unittests/Format/FormatTest.cpp    | 4 +---
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index 684609747a5513f..aaff6319dd45ef9 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -3914,7 +3914,15 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
   // (this would still leave us with an ambiguity between template function
   // and class declarations).
   if (FormatTok->isOneOf(tok::colon, tok::less)) {
+    int AngleNestingLevel = 0;
     do {
+      if (FormatTok->is(tok::less))
+        ++AngleNestingLevel;
+      else if (FormatTok->is(tok::greater))
+        --AngleNestingLevel;
+
+      if (AngleNestingLevel == 0 && FormatTok->is(tok::r_paren))
+        break;
       if (FormatTok->is(tok::l_brace)) {
         calculateBraceTypes(/*ExpectClassBody=*/true);
         if (!tryToParseBracedList())
diff --git a/clang/unittests/Format/FormatTest.cpp b/clang/unittests/Format/FormatTest.cpp
index 762fc8254bdfc9a..f304407d0ce2f46 100644
--- a/clang/unittests/Format/FormatTest.cpp
+++ b/clang/unittests/Format/FormatTest.cpp
@@ -14583,9 +14583,7 @@ TEST_F(FormatTest, UnderstandContextOfRecordTypeKeywords) {
   verifyFormat("template <> struct X < 15, i<3 && 42 < 50 && 33 < 28> {};");
   verifyFormat("int i = SomeFunction(a<b, a> b);");
 
-  // FIXME:
-  // This now gets parsed incorrectly as class definition.
-  // verifyFormat("class A<int> f() {\n}\nint n;");
+  verifyFormat("class A<int> f() {}\nint n;");
 
   // Elaborate types where incorrectly parsing the structural element would
   // break the indent.

>From 7a7d298c1e25940cc0607ba55eac5677b3b02f46 Mon Sep 17 00:00:00 2001
From: XDeme <fernando.tagawa.gamail.com at gmail.com>
Date: Thu, 4 Jan 2024 21:53:11 -0300
Subject: [PATCH 2/8] Fix edge case in template specialization

---
 clang/lib/Format/UnwrappedLineParser.cpp | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index aaff6319dd45ef9..dc6ece09059b5d7 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -3903,6 +3903,15 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
       parseParens();
   }
 
+  auto IsTemplate = [&] {
+    FormatToken *Tok = InitialToken.Previous;
+    while (Tok) {
+      if (Tok->is(tok::kw_template))
+        return true;
+      Tok = Tok->Previous;
+    }
+    return false;
+  };
   // Note that parsing away template declarations here leads to incorrectly
   // accepting function declarations as record declarations.
   // In general, we cannot solve this problem. Consider:
@@ -3921,8 +3930,10 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
       else if (FormatTok->is(tok::greater))
         --AngleNestingLevel;
 
-      if (AngleNestingLevel == 0 && FormatTok->is(tok::r_paren))
+      if (AngleNestingLevel == 0 && !IsTemplate() &&
+          FormatTok->is(tok::r_paren)) {
         break;
+      }
       if (FormatTok->is(tok::l_brace)) {
         calculateBraceTypes(/*ExpectClassBody=*/true);
         if (!tryToParseBracedList())

>From 019992a0e227755015bc46e637d406fee911b9b5 Mon Sep 17 00:00:00 2001
From: XDeme <fernando.tagawa.gamail.com at gmail.com>
Date: Thu, 4 Jan 2024 22:13:15 -0300
Subject: [PATCH 3/8] Add test

---
 clang/unittests/Format/FormatTest.cpp | 1 +
 1 file changed, 1 insertion(+)

diff --git a/clang/unittests/Format/FormatTest.cpp b/clang/unittests/Format/FormatTest.cpp
index f304407d0ce2f46..6ac290b64ca450a 100644
--- a/clang/unittests/Format/FormatTest.cpp
+++ b/clang/unittests/Format/FormatTest.cpp
@@ -14584,6 +14584,7 @@ TEST_F(FormatTest, UnderstandContextOfRecordTypeKeywords) {
   verifyFormat("int i = SomeFunction(a<b, a> b);");
 
   verifyFormat("class A<int> f() {}\nint n;");
+  verifyFormat("template <> class Foo<int> F() {\n} n;");
 
   // Elaborate types where incorrectly parsing the structural element would
   // break the indent.

>From e41076ed7799dc7867d8bb9854298da396307090 Mon Sep 17 00:00:00 2001
From: XDeme <fernando.tagawa.gamail.com at gmail.com>
Date: Sat, 6 Jan 2024 19:54:57 -0300
Subject: [PATCH 4/8] Add test

---
 clang/unittests/Format/TokenAnnotatorTest.cpp | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/clang/unittests/Format/TokenAnnotatorTest.cpp b/clang/unittests/Format/TokenAnnotatorTest.cpp
index 2cafc0438ffb46a..396ed1c003e975a 100644
--- a/clang/unittests/Format/TokenAnnotatorTest.cpp
+++ b/clang/unittests/Format/TokenAnnotatorTest.cpp
@@ -2453,6 +2453,13 @@ TEST_F(TokenAnnotatorTest, BraceKind) {
   EXPECT_BRACE_KIND(Tokens[4], BK_Block);
   EXPECT_BRACE_KIND(Tokens[5], BK_Block);
 
+  Tokens = annotate("class Foo<int> f() {}");
+  ASSERT_EQ(Tokens.size(), 11u) << Tokens;
+  EXPECT_TOKEN(Tokens[5], tok::identifier, TT_FunctionDeclarationName);
+  EXPECT_TOKEN(Tokens[8], tok::l_brace, TT_FunctionLBrace);
+  EXPECT_BRACE_KIND(Tokens[8], BK_Block);
+  EXPECT_BRACE_KIND(Tokens[9], BK_Block);
+
   Tokens = annotate("void f() override {};");
   ASSERT_EQ(Tokens.size(), 9u) << Tokens;
   EXPECT_TOKEN(Tokens[1], tok::identifier, TT_FunctionDeclarationName);

>From 8de0863118d818e3e82342d475082b38b51f145e Mon Sep 17 00:00:00 2001
From: XDeme <fernando.tagawa.gamail.com at gmail.com>
Date: Wed, 10 Jan 2024 18:07:29 -0300
Subject: [PATCH 5/8] Comment and update

---
 clang/lib/Format/UnwrappedLineParser.cpp | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index dc6ece09059b5d7..2d887d719d11c68 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -3915,7 +3915,8 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
   // Note that parsing away template declarations here leads to incorrectly
   // accepting function declarations as record declarations.
   // In general, we cannot solve this problem. Consider:
-  // class A<int> B() {}
+  // template<typename T>
+  // class A<T> B() {}
   // which can be a function definition or a class definition when B() is a
   // macro. If we find enough real-world cases where this is a problem, we
   // can parse for the 'template' keyword in the beginning of the statement,
@@ -3931,7 +3932,7 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
         --AngleNestingLevel;
 
       if (AngleNestingLevel == 0 && !IsTemplate() &&
-          FormatTok->is(tok::r_paren)) {
+          FormatTok->is(tok::l_paren)) {
         break;
       }
       if (FormatTok->is(tok::l_brace)) {

>From a814099281d2cc88901cc6a1c0ff57a0d0a685e1 Mon Sep 17 00:00:00 2001
From: XDeme <fernando.tagawa.gamail.com at gmail.com>
Date: Thu, 11 Jan 2024 20:34:02 -0300
Subject: [PATCH 6/8] Update to verify macro

---
 clang/lib/Format/UnwrappedLineParser.cpp      | 34 +++++--------------
 clang/unittests/Format/FormatTest.cpp         |  2 ++
 clang/unittests/Format/TokenAnnotatorTest.cpp |  7 ++++
 3 files changed, 17 insertions(+), 26 deletions(-)

diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index 2d887d719d11c68..b21d3287c6daf84 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -3873,6 +3873,9 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
   const FormatToken &InitialToken = *FormatTok;
   nextToken();
 
+  auto IsNonMacroIdentifier = [](FormatToken *Tok) {
+    return Tok->is(tok::identifier) && Tok->TokenText != Tok->TokenText.upper();
+  };
   // The actual identifier can be a nested name specifier, and in macros
   // it is often token-pasted.
   // An [[attribute]] can be before the identifier.
@@ -3894,35 +3897,14 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
     }
     if (FormatTok->is(tok::l_square) && handleCppAttributes())
       continue;
-    bool IsNonMacroIdentifier =
-        FormatTok->is(tok::identifier) &&
-        FormatTok->TokenText != FormatTok->TokenText.upper();
     nextToken();
     // We can have macros in between 'class' and the class name.
-    if (!IsNonMacroIdentifier && FormatTok->is(tok::l_paren))
+    if (!IsNonMacroIdentifier(FormatTok->Previous) &&
+        FormatTok->is(tok::l_paren)) {
       parseParens();
+    }
   }
 
-  auto IsTemplate = [&] {
-    FormatToken *Tok = InitialToken.Previous;
-    while (Tok) {
-      if (Tok->is(tok::kw_template))
-        return true;
-      Tok = Tok->Previous;
-    }
-    return false;
-  };
-  // Note that parsing away template declarations here leads to incorrectly
-  // accepting function declarations as record declarations.
-  // In general, we cannot solve this problem. Consider:
-  // template<typename T>
-  // class A<T> B() {}
-  // which can be a function definition or a class definition when B() is a
-  // macro. If we find enough real-world cases where this is a problem, we
-  // can parse for the 'template' keyword in the beginning of the statement,
-  // and thus rule out the record production in case there is no template
-  // (this would still leave us with an ambiguity between template function
-  // and class declarations).
   if (FormatTok->isOneOf(tok::colon, tok::less)) {
     int AngleNestingLevel = 0;
     do {
@@ -3931,8 +3913,8 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
       else if (FormatTok->is(tok::greater))
         --AngleNestingLevel;
 
-      if (AngleNestingLevel == 0 && !IsTemplate() &&
-          FormatTok->is(tok::l_paren)) {
+      if (AngleNestingLevel == 0 && FormatTok->is(tok::l_paren) &&
+          IsNonMacroIdentifier(FormatTok->Previous)) {
         break;
       }
       if (FormatTok->is(tok::l_brace)) {
diff --git a/clang/unittests/Format/FormatTest.cpp b/clang/unittests/Format/FormatTest.cpp
index 6ac290b64ca450a..0a1105cd52db0e9 100644
--- a/clang/unittests/Format/FormatTest.cpp
+++ b/clang/unittests/Format/FormatTest.cpp
@@ -14584,6 +14584,8 @@ TEST_F(FormatTest, UnderstandContextOfRecordTypeKeywords) {
   verifyFormat("int i = SomeFunction(a<b, a> b);");
 
   verifyFormat("class A<int> f() {}\nint n;");
+  verifyFormat("template <typename T> class A<T> f() {}\nint n;");
+
   verifyFormat("template <> class Foo<int> F() {\n} n;");
 
   // Elaborate types where incorrectly parsing the structural element would
diff --git a/clang/unittests/Format/TokenAnnotatorTest.cpp b/clang/unittests/Format/TokenAnnotatorTest.cpp
index 396ed1c003e975a..a8cc522de3095c1 100644
--- a/clang/unittests/Format/TokenAnnotatorTest.cpp
+++ b/clang/unittests/Format/TokenAnnotatorTest.cpp
@@ -2460,6 +2460,13 @@ TEST_F(TokenAnnotatorTest, BraceKind) {
   EXPECT_BRACE_KIND(Tokens[8], BK_Block);
   EXPECT_BRACE_KIND(Tokens[9], BK_Block);
 
+  Tokens = annotate("template <typename T> class Foo<T> f() {}");
+  ASSERT_EQ(Tokens.size(), 16u) << Tokens;
+  EXPECT_TOKEN(Tokens[10], tok::identifier, TT_FunctionDeclarationName);
+  EXPECT_TOKEN(Tokens[13], tok::l_brace, TT_FunctionLBrace);
+  EXPECT_BRACE_KIND(Tokens[13], BK_Block);
+  EXPECT_BRACE_KIND(Tokens[14], BK_Block);
+
   Tokens = annotate("void f() override {};");
   ASSERT_EQ(Tokens.size(), 9u) << Tokens;
   EXPECT_TOKEN(Tokens[1], tok::identifier, TT_FunctionDeclarationName);

>From be0d0029b3ceb99060858b197af2de52c1a4076b Mon Sep 17 00:00:00 2001
From: XDeme <fernando.tagawa.gamail.com at gmail.com>
Date: Sat, 20 Jan 2024 19:29:13 -0300
Subject: [PATCH 7/8] Wrap new after \n

---
 clang/unittests/Format/FormatTest.cpp | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/clang/unittests/Format/FormatTest.cpp b/clang/unittests/Format/FormatTest.cpp
index 0a1105cd52db0e9..692ef8c83168b2b 100644
--- a/clang/unittests/Format/FormatTest.cpp
+++ b/clang/unittests/Format/FormatTest.cpp
@@ -14583,10 +14583,13 @@ TEST_F(FormatTest, UnderstandContextOfRecordTypeKeywords) {
   verifyFormat("template <> struct X < 15, i<3 && 42 < 50 && 33 < 28> {};");
   verifyFormat("int i = SomeFunction(a<b, a> b);");
 
-  verifyFormat("class A<int> f() {}\nint n;");
-  verifyFormat("template <typename T> class A<T> f() {}\nint n;");
+  verifyFormat("class A<int> f() {}\n"
+               "int n;");
+  verifyFormat("template <typename T> class A<T> f() {}\n"
+               "int n;");
 
-  verifyFormat("template <> class Foo<int> F() {\n} n;");
+  verifyFormat("template <> class Foo<int> F() {\n"
+               "} n;");
 
   // Elaborate types where incorrectly parsing the structural element would
   // break the indent.

>From 223fec615222a12f5b8146a712694270bf05b1f7 Mon Sep 17 00:00:00 2001
From: XDeme <66138117+XDeme at users.noreply.github.com>
Date: Sat, 20 Jan 2024 19:42:59 -0300
Subject: [PATCH 8/8] Apply suggestions from code review

Co-authored-by: Owen Pan <owenpiano at gmail.com>
---
 clang/lib/Format/UnwrappedLineParser.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
index b21d3287c6daf84..027907c7ba8f71e 100644
--- a/clang/lib/Format/UnwrappedLineParser.cpp
+++ b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -3873,7 +3873,7 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
   const FormatToken &InitialToken = *FormatTok;
   nextToken();
 
-  auto IsNonMacroIdentifier = [](FormatToken *Tok) {
+  auto IsNonMacroIdentifier = [](const FormatToken *Tok) {
     return Tok->is(tok::identifier) && Tok->TokenText != Tok->TokenText.upper();
   };
   // The actual identifier can be a nested name specifier, and in macros



More information about the cfe-commits mailing list