Function gdv_fn_lower_utf8

cpp/src/gandiva/gdv_string_function_stubs.cc:217–286

Convert an utf8 string to its corresponding lowercase string

// Convert an utf8 string to its corresponding lowercase string
GANDIVA_EXPORT
const char* gdv_fn_lower_utf8(int64_t context, const char* data, int32_t data_len,
                              int32_t* out_len) {
  if (data_len == 0) {
    *out_len = 0;
    return "";
  }

  // If it is a single-byte character (ASCII), corresponding lowercase is always 1-byte
  // long; if it is >= 2 bytes long, lowercase 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, 2 * data_len));
  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, out_char_len, out_idx = 0;
  uint32_t char_codepoint;

  for (int32_t i = 0; i < data_len; i += char_len) {
    char_len = gdv_fn_utf8_char_length(data[i]);
    // For single byte characters:
    // If it is an uppercase ASCII character, set the output to its corresponding
    // lowercase character; else, set the output to the read character
    if (char_len == 1) {
      char cur = data[i];
      // 'A' - 'Z' : 0x41 - 0x5a
      // 'a' - 'z' : 0x61 - 0x7a
      if (cur >= 0x41 && cur <= 0x5a) {
        out[out_idx++] = static_cast<char>(cur + 0x20);
      } else {
        out[out_idx++] = cur;
      }
      continue;
    }

    // Control reaches here when we encounter a multibyte character
    const auto* in_char = (const uint8_t*)(data + i);

    // Decode the multibyte character
    bool is_valid_utf8_char =
        arrow::util::UTF8Decode((const uint8_t**)&in_char, &char_codepoint);

    // If it is an invalid utf8 character, UTF8Decode evaluates to false
    if (!is_valid_utf8_char) {
      gdv_fn_set_error_for_invalid_utf8(context, data[i]);
      *out_len = 0;
      return "";
    }

    // Convert the encoded codepoint to its lowercase codepoint
    int32_t lower_codepoint = utf8proc_tolower(char_codepoint);

    // UTF8Encode advances the pointer by the number of bytes present in the lowercase
    // character
    auto* out_char = (uint8_t*)(out + out_idx);