[clang-format] Add SpaceBeforeUnderscoreParens option for GNU gettext…

clang/docs/ClangFormatStyleOptions.rst

@@ -6771,6 +6771,18 @@ the configuration (without a prefix: ``Auto``).
      int a [5];                    vs.      int a[5];
      int a [5][5];                 vs.      int a[5][5];
 
+.. _SpaceBetweenUnderscoreParens:
+
+**SpaceBetweenUnderscoreParens** (``Boolean``) :versionbadge:`clang-format 19` :ref:`¶ <SpaceBetweenUnderscoreParens>`
+  If ``false``, spaces will be removed between underscore and an opening
+  parenthesis. This is specifically for the gettext macro ``_()`` commonly
+  used in GNU projects.
+
+  .. code-block:: c++
+
+     true:                                  false:
+     _ (message);                   vs.     _(message);
+
 .. _SpaceInEmptyBlock:
 
 **SpaceInEmptyBlock** (``Boolean``) :versionbadge:`clang-format 10` :ref:`¶ <SpaceInEmptyBlock>`

clang/include/clang/Format/Format.h

@@ -5030,6 +5030,16 @@ struct FormatStyle {
   /// \version 10
   bool SpaceBeforeSquareBrackets;
 
+  /// If ``false``, spaces will be removed between underscore and an opening
+  /// parenthesis. This is specifically for the gettext macro ``_()`` commonly
+  /// used in GNU projects.
+  /// \code
+  ///    true:                                  false:
+  ///    _ (message);                   vs.     _(message);
+  /// \endcode
+  /// \version 19
+  bool SpaceBetweenUnderscoreParens;
+
   /// If ``false``, spaces will be removed before range-based for loop
   /// colon.
   /// \code
@@ -5766,6 +5776,7 @@ struct FormatStyle {
            SpaceBeforeRangeBasedForLoopColon ==
                R.SpaceBeforeRangeBasedForLoopColon &&
            SpaceBeforeSquareBrackets == R.SpaceBeforeSquareBrackets &&
+           SpaceBetweenUnderscoreParens == R.SpaceBetweenUnderscoreParens &&
            SpaceInEmptyBraces == R.SpaceInEmptyBraces &&
            SpacesBeforeTrailingComments == R.SpacesBeforeTrailingComments &&
            SpacesInAngles == R.SpacesInAngles &&

clang/lib/Format/Format.cpp

@@ -1314,6 +1314,8 @@ template <> struct MappingTraits<FormatStyle> {
                    Style.SpaceBeforeRangeBasedForLoopColon);
     IO.mapOptional("SpaceBeforeSquareBrackets",
                    Style.SpaceBeforeSquareBrackets);
+    IO.mapOptional("SpaceBetweenUnderscoreParens",
+                   Style.SpaceBetweenUnderscoreParens);
     IO.mapOptional("SpaceInEmptyBraces", Style.SpaceInEmptyBraces);
     IO.mapOptional("SpacesBeforeTrailingComments",
                    Style.SpacesBeforeTrailingComments);
@@ -1821,6 +1823,7 @@ FormatStyle getLLVMStyle(FormatStyle::LanguageKind Language) {
   LLVMStyle.SpaceBeforeParensOptions.AfterIfMacros = true;
   LLVMStyle.SpaceBeforeRangeBasedForLoopColon = true;
   LLVMStyle.SpaceBeforeSquareBrackets = false;
+  LLVMStyle.SpaceBetweenUnderscoreParens = true;
   LLVMStyle.SpaceInEmptyBraces = FormatStyle::SIEB_Never;
   LLVMStyle.SpacesBeforeTrailingComments = 1;
   LLVMStyle.SpacesInAngles = FormatStyle::SIAS_Never;
@@ -2146,6 +2149,7 @@ FormatStyle getGNUStyle() {
   Style.FixNamespaceComments = false;
   Style.KeepFormFeed = true;
   Style.SpaceBeforeParens = FormatStyle::SBPO_Always;
+  Style.SpaceBetweenUnderscoreParens = false;
   return Style;
 }
 

clang/lib/Format/TokenAnnotator.cpp

@@ -4512,6 +4512,14 @@ unsigned TokenAnnotator::splitPenalty(const AnnotatedLine &Line,
 }
 
 bool TokenAnnotator::spaceRequiredBeforeParens(const FormatToken &Right) const {
+  // Handle underscore macro: _("string") or _(message)
+  // Special case when SpaceBetweenUnderscoreParens is false
+  const FormatToken *Left = Right.Previous;
+  if (Left && Left->is(tok::identifier) && Left->TokenText == "_" &&
+      !Style.SpaceBetweenUnderscoreParens) {
+    return false;
+  }
+
   if (Style.SpaceBeforeParens == FormatStyle::SBPO_Always)
     return true;
   if (Right.is(TT_OverloadedOperatorLParen) &&
@@ -4920,6 +4928,11 @@ bool TokenAnnotator::spaceRequiredBetween(const AnnotatedLine &Line,
     // Handle builtins like identifiers.
     if (Line.Type != LT_PreprocessorDirective &&
         (Left.Tok.getIdentifierInfo() || Left.is(tok::r_paren))) {
+      // Check for special case: single underscore token (gettext macro).
+      if (Left.Tok.getIdentifierInfo() && !Style.SpaceBetweenUnderscoreParens &&
+          Left.TokenText == "_") {
+        return false;
+      }
       return spaceRequiredBeforeParens(Right);
     }
     return false;

clang/unittests/Format/FormatTest.cpp

@@ -17877,6 +17877,37 @@ TEST_F(FormatTest, ConfigurableSpacesInParens) {
                Spaces);
 }
 
+TEST_F(FormatTest, SpaceBetweenUnderscoreParens) {
+  FormatStyle Style = getLLVMStyle();
+  Style.SpaceBeforeParens = FormatStyle::SBPO_Always;
+
+  EXPECT_TRUE(Style.SpaceBetweenUnderscoreParens);
+  verifyFormat("func (arg);", Style);
+  verifyFormat("my_func (arg);", Style);
+  verifyFormat("_ (message);", Style);
+  verifyFormat("underscore_ (param);", Style);
+
+  Style.SpaceBetweenUnderscoreParens = false;
+  verifyFormat("func (arg);", Style);
+  verifyFormat("my_func (arg);", Style);
+  verifyFormat("_(message);", Style);
+  verifyFormat("underscore_ (param);", Style);
+  verifyFormat("_private_func (data);", Style);
+
+  FormatStyle GNUStyle = getGNUStyle();
+  EXPECT_FALSE(GNUStyle.SpaceBetweenUnderscoreParens);
+  EXPECT_EQ(GNUStyle.SpaceBeforeParens, FormatStyle::SBPO_Always);
+  verifyFormat("func (arg);", GNUStyle);
+  verifyFormat("my_func (arg);", GNUStyle);
+  verifyFormat("_(message);", GNUStyle);
+  verifyFormat("_private_func (data);", GNUStyle);
+
+  verifyFormat("printf (_(\"Hello\"));\n"
+               "func (arg);\n"
+               "_(\"World\");",
+               GNUStyle);
+}
+
 TEST_F(FormatTest, ConfigurableSpacesInSquareBrackets) {
   verifyFormat("int a[5];");
   verifyFormat("a[3] += 42;");