Performance of ICU Locale parsing vs. unic-locale proposal

add_likely_subtags.cpp

#include <stdio.h>
#include <chrono>
#include <vector>
#include <cstring>

#include "unicode/uloc.h"

    std::vector<std::string> ids;
    ids.push_back("en-US");
    ids.push_back("en-GB");
    ids.push_back("es-AR");
    ids.push_back("it");
    ids.push_back("zh-Hans-CN");
    ids.push_back("de-AT");
    ids.push_back("pl");
    ids.push_back("fr-FR");
    ids.push_back("de-AT");
    ids.push_back("sr-Cyrl-SR");
    ids.push_back("nb-NO");
    ids.push_back("fr-FR");
    ids.push_back("mk");
    ids.push_back("uk");
    ids.push_back("und-PL");
    ids.push_back("und-Latn-AM");
    ids.push_back("ug-Cyrl");
    ids.push_back("sr-ME");
    ids.push_back("mn-Mong");
    ids.push_back("lif-Limb");
    ids.push_back("gan");
    ids.push_back("zh-Hant");
    ids.push_back("yue-Hans");
    ids.push_back("unr");
    ids.push_back("unr-Deva");
    ids.push_back("und-Thai-CN");
    ids.push_back("ug-Cyrl");
    ids.push_back("en-Latn-DE");
    ids.push_back("pl-FR");
    ids.push_back("de-CH");
    ids.push_back("tuq");
    ids.push_back("sr-ME");
    ids.push_back("ng");
    ids.push_back("klx");
    ids.push_back("kk-Arab");
    ids.push_back("en-Cyrl");
    ids.push_back("und-Cyrl-UK");
    ids.push_back("und-Arab");
    ids.push_back("und-Arab-FO");
		auto start = std::chrono::steady_clock::now();

    for (auto& id : ids) {
      const int32_t kLocaleMax = 160;
      char maxLocale[kLocaleMax];

      UErrorCode status = U_ZERO_ERROR;
      uloc_addLikelySubtags(id.c_str(), maxLocale, kLocaleMax, &status);

      nsDependentCString maxLocStr(maxLocale);
      // printf("For string %s, result is %s.\n", id.c_str(), maxLocale);
    }

		auto end = std::chrono::steady_clock::now();
		auto diff = end - start;
		auto diff_sec = std::chrono::duration_cast<std::chrono::nanoseconds>(diff);
    printf("Create Locale from str time: %d ns\n", diff_sec);

locale.cpp

#include <unicode/ures.h>
#include <unicode/putil.h>
#include <unicode/ustdio.h>
#include <unicode/locid.h>
#include <unicode/localebuilder.h>
#include <stdio.h>
#include <chrono>
#include <vector>
#include <cstring>

#ifndef _UASSERT
#define _UASSERT

#define ASSERT_OK(x) if(U_FAILURE(x)) { fprintf(stderr,"%s:%d: ICU Failure %s\n", __FILE__, __LINE__, u_errorName(x)); }

#endif


static void show(void);

void show(void)
{
  UErrorCode status = U_ZERO_ERROR;

  {
    std::vector<std::string> ids;
    ids.push_back("en-US");
    ids.push_back("en-GB");
    ids.push_back("es-AR");
    ids.push_back("it");
    ids.push_back("zh-Hans-CN");
    ids.push_back("de-AT");
    ids.push_back("pl");
    ids.push_back("fr-FR");
    ids.push_back("de-AT");
    ids.push_back("sr-Cyrl-SR");
    ids.push_back("nb-NO");
    ids.push_back("fr-FR");
    ids.push_back("mk");
    ids.push_back("uk");
		auto start = std::chrono::steady_clock::now();

    for (auto& id : ids) {
      icu_64::Locale aLocale = icu_64::Locale(id.c_str());
    }
   
		auto end = std::chrono::steady_clock::now();
		if(U_FAILURE(status)) {
			printf("Can't open resource bundle. Error is %s\n", u_errorName(status));
			return;
		}

		auto diff = end - start;
		auto diff_sec = std::chrono::duration_cast<std::chrono::nanoseconds>(diff);
    printf("Create Locale from str time: %d ns\n", diff_sec);
  }
}


int main() {
  show();
  return 0;
}

results.txt

On Thinkpad Carbon X1 gen 5, Arch Linux:

C++ code based on `language_identifier_from_str` benchmark in `unic-langid` crate.

Create Locale from str time:

C++ (gcc 8.3.): 68605 ns
Rust (Rust 1.34): 1268 ns