Function IsCasedCharacterUnicode

cpp/src/arrow/compute/kernels/scalar_string_utf8.cc:98–103

}

static inline bool IsCasedCharacterUnicode(uint32_t codepoint) {
  return HasAnyUnicodeGeneralCategory(codepoint, UTF8PROC_CATEGORY_LU,
                                      UTF8PROC_CATEGORY_LL, UTF8PROC_CATEGORY_LT) ||
         ((static_cast<uint32_t>(utf8proc_toupper(codepoint)) != codepoint) ||
          (static_cast<uint32_t>(utf8proc_tolower(codepoint)) != codepoint));
}

static inline bool IsLowerCaseCharacterUnicode(uint32_t codepoint) {
  // although this trick seems to work for upper case, this is not enough for lower case