utf8 length from utf32 benchmark

utf8_length_from_utf32_bench.cpp

#include <cstddef>
#include <cstdint>
#include <cuchar>

size_t utf8_length_from_utf32(const char32_t* buf, size_t len) {
  // We are not BOM aware.
  const uint32_t* p = reinterpret_cast<const uint32_t*>(buf);
  size_t counter{0};
  for (size_t i = 0; i < len; i++) {
    /** ASCII **/
    if (p[i] <= 0x7F) {
      counter++;
    }
    /** two-byte **/
    else if (p[i] <= 0x7FF) {
      counter += 2;
    }
    /** three-byte **/
    else if (p[i] <= 0xFFFF) {
      counter += 3;
    }
    /** four-bytes **/
    else {
      counter += 4;
    }
  }
  return counter;
}

size_t utf8_length_from_utf32v(const char32_t* buf, size_t len) {
  // We are not BOM aware.
  const uint32_t* p = reinterpret_cast<const uint32_t*>(buf);
  size_t counter{0};
  for (size_t i = 0; i < len; i++) {
    ++counter; /** ASCII **/
    counter += static_cast<size_t>(p[i] > 0x7F); /** two-byte **/
    counter += static_cast<size_t>(p[i] > 0x7FF); /** three-byte **/
    counter += static_cast<size_t>(p[i] > 0xFFFF); /** four-bytes **/
  }
  return counter;
}

const char32_t text[] = U"eckwd4c7cu47r2wfeckwd4c7cu47r2wfeckwd4c7cu47r2wfeckwd4c7cu47r2wfeckwd4c7cu47r2wfeckwd4c7cu47r2wfeckwd4c7cu47r2wfeckwd4c7cu47r2wf";
// const char32_t text[] = U"MajiでKoiする5秒前MajiでKoiする5秒前MajiでKoiする5秒前MajiでKoiする5秒前MajiでKoiする5秒前MajiでKoiする5秒前MajiでKoiする5秒前MajiでKoiする5秒前MajiでKoiする5秒前";

static void BH_length(benchmark::State& state) {
  const auto text = (const char32_t *)state.range(0);
  for (auto _ : state) {
    benchmark::DoNotOptimize(utf8_length_from_utf32(text, 129));
  }
}
BENCHMARK(BH_length)->Arg((int64_t)text);

static void BH_lengthVec(benchmark::State& state) {
  const auto text = (const char32_t *)state.range(0);
  for (auto _ : state) {
    benchmark::DoNotOptimize(utf8_length_from_utf32v(text, 129));
  }
}
BENCHMARK(BH_lengthVec)->Arg((int64_t)text);