[clang-format] Fix a bug in indenting lambda trailing arrows

[c8ef]

Closes #94181

[llvmbot]

@llvm/pr-subscribers-clang-format

Author: None (c8ef)

Changes

close: #94181

---

Full diff: https://github.com/llvm/llvm-project/pull/94560.diff

2 Files Affected:

• (modified) clang/lib/Format/ContinuationIndenter.cpp (+5)
• (modified) clang/unittests/Format/FormatTest.cpp (+25)

diff --git a/clang/lib/Format/ContinuationIndenter.cpp b/clang/lib/Format/ContinuationIndenter.cpp
index 6b9fbfe0ebf53..630a4ebff9843 100644
--- a/clang/lib/Format/ContinuationIndenter.cpp
+++ b/clang/lib/Format/ContinuationIndenter.cpp
@@ -1457,6 +1457,11 @@ unsigned ContinuationIndenter::getNewLineColumn(const LineState &State) {
       !Current.isOneOf(tok::colon, tok::comment)) {
     return ContinuationIndent;
   }
+  if (Style.isCpp() && Current.is(tok::arrow) &&
+      Previous.isOneOf(tok::kw_noexcept, tok::kw_mutable, tok::kw_constexpr,
+                       tok::kw_consteval, tok::kw_static)) {
+    return ContinuationIndent;
+  }
   if (Current.is(TT_ProtoExtensionLSquare))
     return CurrentState.Indent;
   if (Current.isBinaryOperator() && CurrentState.UnindentOperator) {
diff --git a/clang/unittests/Format/FormatTest.cpp b/clang/unittests/Format/FormatTest.cpp
index 4e427268fb82a..d117efc06c2a7 100644
--- a/clang/unittests/Format/FormatTest.cpp
+++ b/clang/unittests/Format/FormatTest.cpp
@@ -22858,6 +22858,31 @@ TEST_F(FormatTest, FormatsLambdas) {
       "      //\n"
       "    });");
 
+  verifyFormat("int main() {\n"
+               "  very_long_function_name_yes_it_is_really_long(\n"
+               "      [](auto n)\n"
+               "          -> std::unordered_map<very_long_type_name_A, "
+               "very_long_type_name_B> {\n"
+               "        really_do_something();\n"
+               "      });\n"
+               "}");
+  verifyFormat("int main() {\n"
+               "  very_long_function_name_yes_it_is_really_long(\n"
+               "      [](auto n) noexcept\n"
+               "          -> std::unordered_map<very_long_type_name_A, "
+               "very_long_type_name_B> {\n"
+               "        really_do_something();\n"
+               "      });\n"
+               "}");
+  verifyFormat("int main() {\n"
+               "  very_long_function_name_yes_it_is_really_long(\n"
+               "      [](auto n) constexpr\n"
+               "          -> std::unordered_map<very_long_type_name_A, "
+               "very_long_type_name_B> {\n"
+               "        really_do_something();\n"
+               "      });\n"
+               "}");
+
   FormatStyle DoNotMerge = getLLVMStyle();
   DoNotMerge.AllowShortLambdasOnASingleLine = FormatStyle::SLS_None;
   verifyFormat("auto c = []() {\n"

[HazardyKnusperkeks]

Suggested change

	if (Style.isCpp() && Current.is(tok::arrow) &&
	if (Style.isCpp() && Current.is(TT_TrailingReturnArrow) &&

This is what you need, right?

[c8ef]

Exactly, thanks for your suggestions!

[HazardyKnusperkeks]

I also think you can drop the isCpp.

[c8ef]

Done.

[rymiel]

Technically also applies to attributes applied to the return type of the lambda, i.e.

int main() {
  very_long_function_name_yes_it_is_really_long(
      // also happens with constexpr specifier
      [](auto n) [[attribute]]
      -> std::enable_if_t<
          std::is_arithmetic<decltype(n)>::value &&
          !std::is_same<std::remove_cv_t<decltype(n)>, bool>::value> {
        do_something(n * 2);
      });
}

...but currently at least no valid C++ attribute can go in that spot

[owenca]

Technically also applies to attributes applied to the return type of the lambda, i.e.

int main() {
  very_long_function_name_yes_it_is_really_long(
      // also happens with constexpr specifier
      [](auto n) [[attribute]]
      -> std::enable_if_t<
          std::is_arithmetic<decltype(n)>::value &&
          !std::is_same<std::remove_cv_t<decltype(n)>, bool>::value> {
        do_something(n * 2);
      });
}

...but currently at least no valid C++ attribute can go in that spot

We can easily take care of it though. See #94560 (comment).

[c8ef]

Dear reviewers, is there anything else I need to do for this PR? If not, could you please assist me in merging it? Thank you!