[clang-format] Split line comments separated by backslashes (#147648)

Fixes #147341

Author: owenca

clang/docs/ClangFormatStyleOptions.rst

@@ -1554,9 +1554,9 @@ the configuration (without a prefix: ``Auto``).
 
     .. code-block:: c++
 
-      #define A                                                                      \
-        int aaaa;                                                                    \
-        int b;                                                                       \
+      #define A                                                            \
+        int aaaa;                                                          \
+        int b;                                                             \
         int dddddddddd;
 
 

clang/include/clang/Format/Format.h

@@ -513,9 +513,9 @@ struct FormatStyle {
     ENAS_LeftWithLastLine,
     /// Align escaped newlines in the right-most column.
     /// \code
-    ///   #define A                                                                      \
-    ///     int aaaa;                                                                    \
-    ///     int b;                                                                       \
+    ///   #define A                                                            \
+    ///     int aaaa;                                                          \
+    ///     int b;                                                             \
     ///     int dddddddddd;
     /// \endcode
     ENAS_Right,

clang/lib/Format/FormatTokenLexer.cpp

@@ -1329,23 +1329,26 @@ FormatToken *FormatTokenLexer::getNextToken() {
   if (FormatTok->is(tok::unknown))
     FormatTok->setType(TT_ImplicitStringLiteral);
 
+  const bool IsCpp = Style.isCpp();
+
   // JavaScript and Java do not allow to escape the end of the line with a
   // backslash. Backslashes are syntax errors in plain source, but can occur in
   // comments. When a single line comment ends with a \, it'll cause the next
   // line of code to be lexed as a comment, breaking formatting. The code below
   // finds comments that contain a backslash followed by a line break, truncates
   // the comment token at the backslash, and resets the lexer to restart behind
   // the backslash.
-  if ((Style.isJavaScript() || Style.isJava()) && FormatTok->is(tok::comment) &&
-      FormatTok->TokenText.starts_with("//")) {
-    size_t BackslashPos = FormatTok->TokenText.find('\\');
-    while (BackslashPos != StringRef::npos) {
-      if (BackslashPos + 1 < FormatTok->TokenText.size() &&
-          FormatTok->TokenText[BackslashPos + 1] == '\n') {
-        truncateToken(BackslashPos + 1);
+  if (const auto Text = FormatTok->TokenText;
+      Text.starts_with("//") &&
+      (IsCpp || Style.isJavaScript() || Style.isJava())) {
+    assert(FormatTok->is(tok::comment));
+    for (auto Pos = Text.find('\\'); Pos++ != StringRef::npos;
+         Pos = Text.find('\\', Pos)) {
+      if (Pos < Text.size() && Text[Pos] == '\n' &&
+          (!IsCpp || Text.substr(Pos + 1).ltrim().starts_with("//"))) {
+        truncateToken(Pos);
         break;
       }
-      BackslashPos = FormatTok->TokenText.find('\\', BackslashPos + 1);
     }
   }
 
@@ -1450,7 +1453,7 @@ FormatToken *FormatTokenLexer::getNextToken() {
     Column = FormatTok->LastLineColumnWidth;
   }
 
-  if (Style.isCpp()) {
+  if (IsCpp) {
     auto *Identifier = FormatTok->Tok.getIdentifierInfo();
     auto it = Macros.find(Identifier);
     if ((Tokens.empty() || !Tokens.back()->Tok.getIdentifierInfo() ||

clang/unittests/Format/FormatTestComments.cpp

@@ -747,16 +747,14 @@ TEST_F(FormatTestComments, DontSplitLineCommentsWithEscapedNewlines) {
                    "       // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
                    "       // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
                    getLLVMStyleWithColumns(50)));
-  // FIXME: One day we might want to implement adjustment of leading whitespace
-  // of the consecutive lines in this kind of comment:
-  EXPECT_EQ("double\n"
-            "    a; // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
-            "          // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
-            "          // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
-            format("double a; // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
-                   "          // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
-                   "          // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
-                   getLLVMStyleWithColumns(49)));
+  verifyFormat("double\n"
+               "    a; // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+               "       // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+               "       // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
+               "double a; // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+               "          // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\\\n"
+               "          // AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA",
+               getLLVMStyleWithColumns(49));
 }
 
 TEST_F(FormatTestComments, DontIntroduceMultilineComments) {

clang/unittests/Format/TokenAnnotatorTest.cpp

@@ -4126,6 +4126,13 @@ TEST_F(TokenAnnotatorTest, JsonCodeInRawString) {
   EXPECT_TOKEN(Tokens[6], tok::colon, TT_DictLiteral);
 }
 
+TEST_F(TokenAnnotatorTest, LineCommentTrailingBackslash) {
+  auto Tokens = annotate("// a \\\n"
+                         "// b");
+  ASSERT_EQ(Tokens.size(), 3u) << Tokens;
+  EXPECT_TOKEN(Tokens[1], tok::comment, TT_LineComment);
+}
+
 } // namespace
 } // namespace format
 } // namespace clang