Function gdv_fn_initcap_utf8

cpp/src/gandiva/gdv_string_function_stubs.cc:515–612

For a given text, initialize the first letter after a word-separator and lowercase the others e.g: - "IT is a tEXt str" -> "It Is A Text Str"

// the others e.g:
//     - "IT is a tEXt str" -> "It Is A Text Str"
GANDIVA_EXPORT
const char* gdv_fn_initcap_utf8(int64_t context, const char* data, int32_t data_len,
                                int32_t* out_len) {
  if (data_len == 0) {
    *out_len = data_len;
    return "";
  }

  int32_t alloc_length = 0;
  if (ARROW_PREDICT_FALSE(!is_datalen_valid(context, data_len, &alloc_length, out_len))) {
    return "";
  }

  // If it is a single-byte character (ASCII), corresponding uppercase is always 1-byte
  // long; if it is >= 2 bytes long, uppercase can be at most 4 bytes long, so length of
  // the output can be at most twice the length of the input
  char* out = reinterpret_cast<char*>(gdv_fn_context_arena_malloc(context, alloc_length));
  if (out == nullptr) {
    gdv_fn_context_set_error_msg(context, "Could not allocate memory for output string");
    *out_len = 0;
    return "";
  }

  int32_t char_len = 0;
  int32_t out_char_len = 0;
  int32_t out_idx = 0;
  uint32_t char_codepoint;

  // Any character is considered as space, except if it is alphanumeric
  bool last_char_was_space = true;

  for (int32_t i = 0; i < data_len; i += char_len) {
    // An optimization for single byte characters:
    if (static_cast<signed char>(data[i]) >= 0) {  // 1-byte char (0x00 ~ 0x7F)
      char_len = 1;
      char cur = data[i];

      if (cur >= 0x61 && cur <= 0x7a && last_char_was_space) {
        // Check if the character is the first one of the word and it is
        // lowercase -> 'a' - 'z' : 0x61 - 0x7a.
        // Then turn it into uppercase -> 'A' - 'Z' : 0x41 - 0x5a
        out[out_idx++] = static_cast<char>(cur - 0x20);
        last_char_was_space = false;
      } else if (cur >= 0x41 && cur <= 0x5a && !last_char_was_space) {
        out[out_idx++] = static_cast<char>(cur + 0x20);
      } else {
        // Check if the ASCII character is not an alphanumeric character:
        // '0' - '9': 0x30 - 0x39
        // 'a' - 'z' : 0x61 - 0x7a
        // 'A' - 'Z' : 0x41 - 0x5a
        last_char_was_space = (cur < 0x30) || (cur > 0x39 && cur < 0x41) ||
                              (cur > 0x5a && cur < 0x61) || (cur > 0x7a);
        out[out_idx++] = cur;
      }
      continue;
    }

    char_len = gdv_fn_utf8_char_length(data[i]);