[libc-commits] [libc] [libc] Fix character converter in C++20 (PR #177421)

[via libc-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-libc

Author: Michael Jones (michaelrj-google)

<details>
<summary>Changes</summary>

Internally character converter uses char8_t to represent a character.
Before C++20 we provide a typedef for it, but since C++20 it's a
keyword. The keyword version isn't listed in our is_integral, causing
countl to reject it as an invalid type. This patch just casts from
char8_t to int8_t to sidestep the issue, but in future we may want to
add char8_t, char16_t, and char32_t to our is_integral.


---
Full diff: https://github.com/llvm/llvm-project/pull/177421.diff


1 Files Affected:

- (modified) libc/src/__support/wchar/character_converter.h (+2-1) 


``````````diff
diff --git a/libc/src/__support/wchar/character_converter.h b/libc/src/__support/wchar/character_converter.h
index e7300166556ac..d0a942ecb3bb1 100644
--- a/libc/src/__support/wchar/character_converter.h
+++ b/libc/src/__support/wchar/character_converter.h
@@ -77,7 +77,8 @@ LIBC_INLINE bool CharacterConverter::isValidState() {
 }
 
 LIBC_INLINE int CharacterConverter::push(char8_t utf8_byte) {
-  uint8_t num_ones = static_cast<uint8_t>(cpp::countl_one(utf8_byte));
+  uint8_t num_ones =
+      static_cast<uint8_t>(cpp::countl_one(static_cast<uint8_t>(utf8_byte)));
   // Checking the first byte if first push
   if (isEmpty()) {
     // UTF-8 char has 1 byte total

``````````

</details>


https://github.com/llvm/llvm-project/pull/177421