[clang] [Clang][Comments] Support for parsing headers in Doxygen \par commands (PR #91100)

via cfe-commits cfe-commits at lists.llvm.org
Sat Jun 15 00:34:16 PDT 2024


https://github.com/hdoc updated https://github.com/llvm/llvm-project/pull/91100

>From 926abee4c34bc23a92e0c71551348bf74e39ff56 Mon Sep 17 00:00:00 2001
From: hdoc <github at hdoc.io>
Date: Sat, 4 May 2024 18:50:16 -0700
Subject: [PATCH 1/3] Support for parsing headers in Doxygen \par commands

---
 .../include/clang/AST/CommentCommandTraits.h  |   4 +
 clang/include/clang/AST/CommentCommands.td    |   3 +-
 clang/include/clang/AST/CommentParser.h       |   4 +-
 clang/lib/AST/CommentParser.cpp               |  77 ++++++++++
 clang/test/Index/comment-misc-tags.m          |   8 +-
 clang/unittests/AST/CommentParser.cpp         | 137 ++++++++++++++++++
 .../ClangCommentCommandInfoEmitter.cpp        |   1 +
 7 files changed, 227 insertions(+), 7 deletions(-)

diff --git a/clang/include/clang/AST/CommentCommandTraits.h b/clang/include/clang/AST/CommentCommandTraits.h
index 0c3254d84eb00..78c484fff3aed 100644
--- a/clang/include/clang/AST/CommentCommandTraits.h
+++ b/clang/include/clang/AST/CommentCommandTraits.h
@@ -88,6 +88,10 @@ struct CommandInfo {
   LLVM_PREFERRED_TYPE(bool)
   unsigned IsHeaderfileCommand : 1;
 
+  /// True if this is a \\par command.
+  LLVM_PREFERRED_TYPE(bool)
+  unsigned IsParCommand : 1;
+
   /// True if we don't want to warn about this command being passed an empty
   /// paragraph.  Meaningful only for block commands.
   LLVM_PREFERRED_TYPE(bool)
diff --git a/clang/include/clang/AST/CommentCommands.td b/clang/include/clang/AST/CommentCommands.td
index 06b2fa9b5531c..a410cd4039bee 100644
--- a/clang/include/clang/AST/CommentCommands.td
+++ b/clang/include/clang/AST/CommentCommands.td
@@ -18,6 +18,7 @@ class Command<string name> {
   bit IsThrowsCommand = 0;
   bit IsDeprecatedCommand = 0;
   bit IsHeaderfileCommand = 0;
+  bit IsParCommand = 0;
 
   bit IsEmptyParagraphAllowed = 0;
 
@@ -156,7 +157,7 @@ def Date       : BlockCommand<"date">;
 def Invariant  : BlockCommand<"invariant">;
 def Li         : BlockCommand<"li">;
 def Note       : BlockCommand<"note">;
-def Par        : BlockCommand<"par">;
+def Par        : BlockCommand<"par"> { let IsParCommand = 1; let NumArgs = 1; }
 def Post       : BlockCommand<"post">;
 def Pre        : BlockCommand<"pre">;
 def Remark     : BlockCommand<"remark">;
diff --git a/clang/include/clang/AST/CommentParser.h b/clang/include/clang/AST/CommentParser.h
index a2d0e30835e2c..289f0b2c066b9 100644
--- a/clang/include/clang/AST/CommentParser.h
+++ b/clang/include/clang/AST/CommentParser.h
@@ -105,6 +105,9 @@ class Parser {
   ArrayRef<Comment::Argument>
   parseThrowCommandArgs(TextTokenRetokenizer &Retokenizer, unsigned NumArgs);
 
+  ArrayRef<Comment::Argument>
+  parseParCommandArgs(TextTokenRetokenizer &Retokenizer, unsigned NumArgs);
+
   BlockCommandComment *parseBlockCommand();
   InlineCommandComment *parseInlineCommand();
 
@@ -123,4 +126,3 @@ class Parser {
 } // end namespace clang
 
 #endif
-
diff --git a/clang/lib/AST/CommentParser.cpp b/clang/lib/AST/CommentParser.cpp
index 5baf81a509fb6..bbe93ebc37d13 100644
--- a/clang/lib/AST/CommentParser.cpp
+++ b/clang/lib/AST/CommentParser.cpp
@@ -222,6 +222,63 @@ class TextTokenRetokenizer {
     return true;
   }
 
+  /// Check if this line starts with @par or \par
+  bool startsWithParCommand() {
+    unsigned Offset = 1;
+
+    /// Skip all whitespace characters at the beginning.
+    /// This needs to backtrack because Pos has already advanced past the
+    /// actual \par or @par command by the time this function is called.
+    while (isWhitespace(*(Pos.BufferPtr - Offset)))
+      Offset++;
+
+    /// Check if next four characters are \par or @par
+    llvm::StringRef LineStart(Pos.BufferPtr - 5, 4);
+    return LineStart.starts_with("\\par") || LineStart.starts_with("@par");
+  }
+
+  /// Extract a par command argument-header.
+  bool lexParHeading(Token &Tok) {
+    if (isEnd())
+      return false;
+
+    Position SavedPos = Pos;
+
+    consumeWhitespace();
+    SmallString<32> WordText;
+    const char *WordBegin = Pos.BufferPtr;
+    SourceLocation Loc = getSourceLocation();
+
+    if (!startsWithParCommand())
+      return false;
+
+    // Read until the end of this token, which is effectively the end of the
+    // line This gets us the content of the par header, if there is one.
+    while (!isEnd()) {
+      WordText.push_back(peek());
+      if (Pos.BufferPtr + 1 == Pos.BufferEnd) {
+        consumeChar();
+        break;
+      } else {
+        consumeChar();
+      }
+    }
+
+    const unsigned Length = WordText.size();
+    if (Length == 0) {
+      Pos = SavedPos;
+      return false;
+    }
+
+    char *TextPtr = Allocator.Allocate<char>(Length + 1);
+
+    memcpy(TextPtr, WordText.c_str(), Length + 1);
+    StringRef Text = StringRef(TextPtr, Length);
+
+    formTokenWithChars(Tok, Loc, WordBegin, Length, Text);
+    return true;
+  }
+
   /// Extract a word -- sequence of non-whitespace characters.
   bool lexWord(Token &Tok) {
     if (isEnd())
@@ -394,6 +451,23 @@ Parser::parseThrowCommandArgs(TextTokenRetokenizer &Retokenizer,
   return llvm::ArrayRef(Args, ParsedArgs);
 }
 
+ArrayRef<Comment::Argument>
+Parser::parseParCommandArgs(TextTokenRetokenizer &Retokenizer,
+                            unsigned NumArgs) {
+  auto *Args = new (Allocator.Allocate<Comment::Argument>(NumArgs))
+      Comment::Argument[NumArgs];
+  unsigned ParsedArgs = 0;
+  Token Arg;
+
+  while (ParsedArgs < NumArgs && Retokenizer.lexParHeading(Arg)) {
+    Args[ParsedArgs] = Comment::Argument{
+        SourceRange(Arg.getLocation(), Arg.getEndLocation()), Arg.getText()};
+    ParsedArgs++;
+  }
+
+  return llvm::ArrayRef(Args, ParsedArgs);
+}
+
 BlockCommandComment *Parser::parseBlockCommand() {
   assert(Tok.is(tok::backslash_command) || Tok.is(tok::at_command));
 
@@ -449,6 +523,9 @@ BlockCommandComment *Parser::parseBlockCommand() {
     else if (Info->IsThrowsCommand)
       S.actOnBlockCommandArgs(
           BC, parseThrowCommandArgs(Retokenizer, Info->NumArgs));
+    else if (Info->IsParCommand)
+      S.actOnBlockCommandArgs(BC,
+                              parseParCommandArgs(Retokenizer, Info->NumArgs));
     else
       S.actOnBlockCommandArgs(BC, parseCommandArgs(Retokenizer, Info->NumArgs));
 
diff --git a/clang/test/Index/comment-misc-tags.m b/clang/test/Index/comment-misc-tags.m
index 47ee9d9aa392a..6d018dbfcf193 100644
--- a/clang/test/Index/comment-misc-tags.m
+++ b/clang/test/Index/comment-misc-tags.m
@@ -91,18 +91,16 @@ @interface IOCommandGate
   
 struct Test {int filler;};
 
-// CHECK:       (CXComment_BlockCommand CommandName=[par]
+// CHECK:       (CXComment_BlockCommand CommandName=[par] Arg[0]=User defined paragraph:
 // CHECK-NEXT:     (CXComment_Paragraph
-// CHECK-NEXT:        (CXComment_Text Text=[ User defined paragraph:] HasTrailingNewline)
 // CHECK-NEXT:        (CXComment_Text Text=[ Contents of the paragraph.])))
 // CHECK:       (CXComment_BlockCommand CommandName=[par]
 // CHECK-NEXT:     (CXComment_Paragraph
-// CHECK-NEXT:        (CXComment_Text Text=[ New paragraph under the same heading.])))
+// CHECK-NEXT:        (CXComment_Text Text=[New paragraph under the same heading.])))
 // CHECK:       (CXComment_BlockCommand CommandName=[note]
 // CHECK-NEXT:     (CXComment_Paragraph
 // CHECK-NEXT:        (CXComment_Text Text=[ This note consists of two paragraphs.] HasTrailingNewline)
 // CHECK-NEXT:        (CXComment_Text Text=[ This is the first paragraph.])))
 // CHECK:       (CXComment_BlockCommand CommandName=[par]
 // CHECK-NEXT:     (CXComment_Paragraph
-// CHECK-NEXT:     (CXComment_Text Text=[ And this is the second paragraph.])))
-
+// CHECK-NEXT:     (CXComment_Text Text=[And this is the second paragraph.])))
diff --git a/clang/unittests/AST/CommentParser.cpp b/clang/unittests/AST/CommentParser.cpp
index 1c57c899f9074..e0df182d430c3 100644
--- a/clang/unittests/AST/CommentParser.cpp
+++ b/clang/unittests/AST/CommentParser.cpp
@@ -1639,6 +1639,143 @@ TEST_F(CommentParserTest, ThrowsCommandHasArg9) {
   }
 }
 
+TEST_F(CommentParserTest, ParCommandHasArg1) {
+  const char *Sources[] = {
+      "/// @par Paragraph header:",     "/// @par Paragraph header:\n",
+      "/// @par Paragraph header:\r\n", "/// @par Paragraph header:\n\r",
+      "/** @par Paragraph header:*/",
+  };
+
+  for (size_t i = 0, e = std::size(Sources); i != e; i++) {
+    FullComment *FC = parseString(Sources[i]);
+    ASSERT_TRUE(HasChildCount(FC, 2));
+
+    ASSERT_TRUE(HasParagraphCommentAt(FC, 0, " "));
+    {
+      BlockCommandComment *BCC;
+      ParagraphComment *PC;
+      ASSERT_TRUE(HasBlockCommandAt(FC, Traits, 1, BCC, "par", PC));
+      ASSERT_TRUE(HasChildCount(PC, 0));
+      ASSERT_TRUE(BCC->getNumArgs() == 1);
+      ASSERT_TRUE(BCC->getArgText(0) == "Paragraph header:");
+    }
+  }
+}
+
+TEST_F(CommentParserTest, ParCommandHasArg2) {
+  const char *Sources[] = {
+      "/// @par Paragraph header: ",     "/// @par Paragraph header: \n",
+      "/// @par Paragraph header: \r\n", "/// @par Paragraph header: \n\r",
+      "/** @par Paragraph header: */",
+  };
+
+  for (size_t i = 0, e = std::size(Sources); i != e; i++) {
+    FullComment *FC = parseString(Sources[i]);
+    ASSERT_TRUE(HasChildCount(FC, 2));
+
+    ASSERT_TRUE(HasParagraphCommentAt(FC, 0, " "));
+    {
+      BlockCommandComment *BCC;
+      ParagraphComment *PC;
+      ASSERT_TRUE(HasBlockCommandAt(FC, Traits, 1, BCC, "par", PC));
+      ASSERT_TRUE(HasChildCount(PC, 0));
+      ASSERT_TRUE(BCC->getNumArgs() == 1);
+      ASSERT_TRUE(BCC->getArgText(0) == "Paragraph header: ");
+    }
+  }
+}
+
+TEST_F(CommentParserTest, ParCommandHasArg3) {
+  const char *Sources[] = {
+      ("/// @par Paragraph header:\n"
+       "/// Paragraph body"),
+      ("/// @par Paragraph header:\r\n"
+       "/// Paragraph body"),
+      ("/// @par Paragraph header:\n\r"
+       "/// Paragraph body"),
+  };
+
+  for (size_t i = 0, e = std::size(Sources); i != e; i++) {
+    FullComment *FC = parseString(Sources[i]);
+    ASSERT_TRUE(HasChildCount(FC, 2));
+
+    ASSERT_TRUE(HasParagraphCommentAt(FC, 0, " "));
+    {
+      BlockCommandComment *BCC;
+      ParagraphComment *PC;
+      TextComment *TC;
+      ASSERT_TRUE(HasBlockCommandAt(FC, Traits, 1, BCC, "par", PC));
+      ASSERT_TRUE(HasChildCount(PC, 1));
+      ASSERT_TRUE(BCC->getNumArgs() == 1);
+      ASSERT_TRUE(BCC->getArgText(0) == "Paragraph header:");
+      ASSERT_TRUE(GetChildAt(PC, 0, TC));
+      ASSERT_TRUE(TC->getText() == " Paragraph body");
+    }
+  }
+}
+
+TEST_F(CommentParserTest, ParCommandHasArg4) {
+  const char *Sources[] = {
+      ("/// @par Paragraph header:\n"
+       "/// Paragraph body1\n"
+       "/// Paragraph body2"),
+      ("/// @par Paragraph header:\r\n"
+       "/// Paragraph body1\n"
+       "/// Paragraph body2"),
+      ("/// @par Paragraph header:\n\r"
+       "/// Paragraph body1\n"
+       "/// Paragraph body2"),
+  };
+
+  for (size_t i = 0, e = std::size(Sources); i != e; i++) {
+    FullComment *FC = parseString(Sources[i]);
+    ASSERT_TRUE(HasChildCount(FC, 2));
+
+    ASSERT_TRUE(HasParagraphCommentAt(FC, 0, " "));
+    {
+      BlockCommandComment *BCC;
+      ParagraphComment *PC;
+      TextComment *TC;
+      ASSERT_TRUE(HasBlockCommandAt(FC, Traits, 1, BCC, "par", PC));
+      ASSERT_TRUE(HasChildCount(PC, 2));
+      ASSERT_TRUE(BCC->getNumArgs() == 1);
+      ASSERT_TRUE(BCC->getArgText(0) == "Paragraph header:");
+      ASSERT_TRUE(GetChildAt(PC, 0, TC));
+      ASSERT_TRUE(TC->getText() == " Paragraph body1");
+      ASSERT_TRUE(GetChildAt(PC, 1, TC));
+      ASSERT_TRUE(TC->getText() == " Paragraph body2");
+    }
+  }
+}
+
+TEST_F(CommentParserTest, ParCommandHasArg5) {
+  const char *Sources[] = {
+      ("/// @par \n"
+       "/// Paragraphs with no text before newline have no heading"),
+      ("/// @par \r\n"
+       "/// Paragraphs with no text before newline have no heading"),
+      ("/// @par \n\r"
+       "/// Paragraphs with no text before newline have no heading"),
+  };
+
+  for (size_t i = 0, e = std::size(Sources); i != e; i++) {
+    FullComment *FC = parseString(Sources[i]);
+    ASSERT_TRUE(HasChildCount(FC, 2));
+
+    ASSERT_TRUE(HasParagraphCommentAt(FC, 0, " "));
+    {
+      BlockCommandComment *BCC;
+      ParagraphComment *PC;
+      TextComment *TC;
+      ASSERT_TRUE(HasBlockCommandAt(FC, Traits, 1, BCC, "par", PC));
+      ASSERT_TRUE(HasChildCount(PC, 1));
+      ASSERT_TRUE(BCC->getNumArgs() == 0);
+      ASSERT_TRUE(GetChildAt(PC, 0, TC));
+      ASSERT_TRUE(TC->getText() ==
+                  "Paragraphs with no text before newline have no heading");
+    }
+  }
+}
 
 } // unnamed namespace
 
diff --git a/clang/utils/TableGen/ClangCommentCommandInfoEmitter.cpp b/clang/utils/TableGen/ClangCommentCommandInfoEmitter.cpp
index a113b02e19995..07b26dc2f6b8b 100644
--- a/clang/utils/TableGen/ClangCommentCommandInfoEmitter.cpp
+++ b/clang/utils/TableGen/ClangCommentCommandInfoEmitter.cpp
@@ -44,6 +44,7 @@ void clang::EmitClangCommentCommandInfo(RecordKeeper &Records,
        << Tag.getValueAsBit("IsThrowsCommand") << ", "
        << Tag.getValueAsBit("IsDeprecatedCommand") << ", "
        << Tag.getValueAsBit("IsHeaderfileCommand") << ", "
+       << Tag.getValueAsBit("IsParCommand") << ", "
        << Tag.getValueAsBit("IsEmptyParagraphAllowed") << ", "
        << Tag.getValueAsBit("IsVerbatimBlockCommand") << ", "
        << Tag.getValueAsBit("IsVerbatimBlockEndCommand") << ", "

>From b7f01740533794d9322df50871105f728135d92b Mon Sep 17 00:00:00 2001
From: hdoc <github at hdoc.io>
Date: Mon, 6 May 2024 13:57:42 -0700
Subject: [PATCH 2/3] Address review feedback

---
 clang/lib/AST/CommentParser.cpp | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/clang/lib/AST/CommentParser.cpp b/clang/lib/AST/CommentParser.cpp
index bbe93ebc37d13..1158df610d560 100644
--- a/clang/lib/AST/CommentParser.cpp
+++ b/clang/lib/AST/CommentParser.cpp
@@ -259,12 +259,11 @@ class TextTokenRetokenizer {
       if (Pos.BufferPtr + 1 == Pos.BufferEnd) {
         consumeChar();
         break;
-      } else {
-        consumeChar();
       }
+      consumeChar();
     }
 
-    const unsigned Length = WordText.size();
+    unsigned Length = WordText.size();
     if (Length == 0) {
       Pos = SavedPos;
       return false;
@@ -454,6 +453,7 @@ Parser::parseThrowCommandArgs(TextTokenRetokenizer &Retokenizer,
 ArrayRef<Comment::Argument>
 Parser::parseParCommandArgs(TextTokenRetokenizer &Retokenizer,
                             unsigned NumArgs) {
+  assert(NumArgs > 0);
   auto *Args = new (Allocator.Allocate<Comment::Argument>(NumArgs))
       Comment::Argument[NumArgs];
   unsigned ParsedArgs = 0;

>From 362080b5ab4e0401b24e85a9c28ece3a9b0e0286 Mon Sep 17 00:00:00 2001
From: hdoc <github at hdoc.io>
Date: Mon, 3 Jun 2024 14:31:40 -0700
Subject: [PATCH 3/3] Small fix to par heading check and comments

---
 clang/lib/AST/CommentParser.cpp | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/clang/lib/AST/CommentParser.cpp b/clang/lib/AST/CommentParser.cpp
index 1158df610d560..24aa1ab8fc189 100644
--- a/clang/lib/AST/CommentParser.cpp
+++ b/clang/lib/AST/CommentParser.cpp
@@ -222,18 +222,19 @@ class TextTokenRetokenizer {
     return true;
   }
 
-  /// Check if this line starts with @par or \par
+  // Check if this line starts with @par or \par
   bool startsWithParCommand() {
     unsigned Offset = 1;
 
-    /// Skip all whitespace characters at the beginning.
-    /// This needs to backtrack because Pos has already advanced past the
-    /// actual \par or @par command by the time this function is called.
+    // Skip all whitespace characters at the beginning.
+    // This needs to backtrack because Pos has already advanced past the
+    // actual \par or @par command by the time this function is called.
     while (isWhitespace(*(Pos.BufferPtr - Offset)))
       Offset++;
 
-    /// Check if next four characters are \par or @par
-    llvm::StringRef LineStart(Pos.BufferPtr - 5, 4);
+    // Once we've reached the whitespace, backtrack and check if the previous four
+    // characters are \par or @par.
+    llvm::StringRef LineStart(Pos.BufferPtr - Offset - 3, 4);
     return LineStart.starts_with("\\par") || LineStart.starts_with("@par");
   }
 
@@ -253,7 +254,7 @@ class TextTokenRetokenizer {
       return false;
 
     // Read until the end of this token, which is effectively the end of the
-    // line This gets us the content of the par header, if there is one.
+    // line. This gets us the content of the par header, if there is one.
     while (!isEnd()) {
       WordText.push_back(peek());
       if (Pos.BufferPtr + 1 == Pos.BufferEnd) {



More information about the cfe-commits mailing list