Pasted

/*
 * Copyright 2015 Florian Philipp
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
 
#include <wchar.h>
/* using wcsrtombs, wcrtomb, wcstrlen */
#include <stdlib.h>
/* using malloc, realloc, free */
#include <string.h>
/* using memset, memcpy */
#include <stdio.h>
 /* using printf, perror */
#include <locale.h>
/* using setlocale */
#include <errno.h>
/* using errno */
#include <time.h>
/* using clock_gettime, difftime */
 
 
/**
 * Utility to preserve errno when calling free
 */
static void _free_keep_errno(void* buf)
{
  int tmp_errno;
  tmp_errno = errno;
  free(buf);
  errno = tmp_errno;
}
 
/**
 * Converts a wide character string to a multi-byte character string
 *
 * Operates by measuring the size and allocating the resulting string once
 *
 * \param ws a null-terminated wide character string
 * \return the corresponding multi-byte character string or NULL on error
 */ 
static char* s_wstring_to_string_meas(const wchar_t* ws)
{
  const wchar_t* count_ws;
  mbstate_t mbstate;
  size_t len;
  char* buf;
  count_ws = ws;
  memset(&mbstate, 0, sizeof(mbstate));
  if((len = wcsrtombs(NULL, &count_ws, 0, &mbstate)) == (size_t) -1)
    goto err_rtrn;
  len += 1;
  if(! (buf = malloc(len)))
    goto err_rtrn;
  if(wcsrtombs(buf, &ws, len, &mbstate) == (size_t) -1)
    goto err_free;
  return buf;
 err_free:
  _free_keep_errno(buf);
 err_rtrn:
  return NULL;
}
 
/**
 * Converts a wide character string to a multi-byte character string
 *
 * Operates by allocating enough memory if each wchar_t translates into
 * MB_CUR_MAX bytes, then shrinking the string in the end
 *
 * \param ws a null-terminated wide character string
 * \return the corresponding multi-byte character string or NULL on error
 */ 
static char* s_wstring_to_string_max(const wchar_t* ws)
{
  size_t len;
  char* buf, *shrunk;
  mbstate_t mbstate;
  /* make a worst case estimate. Note that MB_CUR_MAX is a runtime variable */
  len = MB_CUR_MAX * wcslen(ws) + 1;
  if(! (buf = malloc(len)))
    goto err_rtrn;
  memset(&mbstate, 0, sizeof(mbstate));
  if((len = wcsrtombs(buf, &ws, len, &mbstate)) == (size_t) -1)
    goto err_free;
  if(! (shrunk = realloc(buf, len + 1)))
    goto err_free;
  return shrunk;
 err_free:
  _free_keep_errno(buf);
 err_rtrn:
  return NULL;
}
 
/**
 * Converts a wide character string to a multi-byte character string
 *
 * Helper function for different high-level implementations.
 * Uses an initial guess for the size.
 * If the guessed size is insufficient, larger memory is allocated
 * exponentially.
 *
 * \param ws a null-terminated wide character string
 * \param buflen initial size estimate
 * \return the corresponding multi-byte character string or NULL on error
 */
static char* _s_wstring_to_string_realloc(const wchar_t* ws, size_t buflen)
{
  char* buffer;
  char* resized;
  size_t bufpos ;
  mbstate_t mbstate;
  memset(&mbstate, 0, sizeof(mbstate));
  /* to avoid parsing the string twice, we allocate memory speculatively
   * with exponential growth. Then we shrink it at the end
   */
  for(buffer = NULL, bufpos = 0; ws; buflen *= 2) {
    size_t converted;
    if(! (resized = realloc(buffer, buflen)))
      goto err;
    buffer = resized;
    if((converted = wcsrtombs(buffer + bufpos, &ws, buflen - bufpos, &mbstate))
       == (size_t) -1)
      goto err;
    bufpos += converted;
  }
  /* shrink buffer to actually required size */
  if(! (resized = realloc(buffer, bufpos + 1)))
    goto err;
  return resized;
 err:
  _free_keep_errno(buffer); /* buffer may be NULL */
  return NULL;
}
 
 
/**
 * Converts a wide character string to a multi-byte character string
 *
 * Operates by allocating the one byte per wide character and then extending it
 * exponentially
 *
 * \param ws a null-terminated wide character string
 * \return the corresponding multi-byte character string or NULL on error
 */ 
static char* s_wstring_to_string_min(const wchar_t* ws)
{
  return _s_wstring_to_string_realloc(ws, wcslen(ws) + 1);
}
 
/**
 * Converts a wide character string to a multi-byte character string
 *
 * Operates by making an uneducated guess on the memory size and then extending
 * it exponentially.
 * Currently, the initial guess is 64, i.e. one cache line
 *
 * \param ws a null-terminated wide character string
 * \return the corresponding multi-byte character string or NULL on error
 */ 
static char* s_wstring_to_string_guess(const wchar_t* ws)
{
  return _s_wstring_to_string_realloc(ws, 64);
}
 
/**
 * Converts a wide character string to a multi-byte character string
 *
 * Operates by allocating exponentially larger strings. Then concatenates them
 * in the end
 *
 * \param ws a null-terminated wide character string
 * \return the corresponding multi-byte character string or NULL on error
 */ 
static char* s_wstring_to_string_rope(const wchar_t* ws)
{
  struct _String
  {
    size_t len;
    char* str;
  };
  /* Array of exponentially growing strings */
  struct _String* rope;
  size_t ropelen = 0, ropecap = 64 / sizeof(struct _String);
  /* allocation size. Initial size is a cache line */
  size_t lastreserved = 64;
  size_t i;
  size_t outlen = 1;
  char* outbuf;
  char* outpos;
  int errno_cpy;
  mbstate_t mbstate;
  memset(&mbstate, 0, sizeof(mbstate));
  if(! (rope = malloc(ropecap * sizeof(struct _String))))
    goto err_rtrn;
  /* Convert the string into increasingly large substrings */
  for(i = 0; ws; ++i, lastreserved *= 2) {
    if(i == ropecap) {
      struct _String* newrope;
      ropecap *= 2;
      if(! (newrope = realloc(rope, ropecap * sizeof(struct _String))))
    goto err_free;
      rope = newrope;
    }
    if(! (rope[i].str = malloc(lastreserved)))
      goto err_free;
    ropelen = i + 1;
    if((rope[i].len = wcsrtombs(rope[i].str, &ws, lastreserved, &mbstate))
       == (size_t) -1)
      goto err_free;
    outlen += rope[i].len;
  };
  /* allocate the final string. reallocing the first strings saves us the
   * hassle of copying its content and freeing it.
   * It also deals with the common case that the first string was big enough
   * and just needs to be shrunk
   */
  if(! (outbuf = realloc(rope[0].str, outlen)))
    goto err_free;
  /* append the remaining strings */
  outpos = outbuf + rope[0].len;
  for(i = 1; i < ropelen; ++i) {
    outpos = memcpy(outpos, rope[i].str, rope[i].len) + rope[i].len;
    free(rope[i].str);
  }
  *outpos = '\0';
  free(rope);
  return outbuf;
 err_free:
  /* free all entries. The last entry may be NULL if allocation failed.
   * free may call munmap so we preserve errno before calling it
   */
  errno_cpy = errno;
  for(i = 0; i < ropelen; ++i)
    free(rope[i].str);
  free(rope);
  errno = errno_cpy;
 err_rtrn:
  return NULL;
}
 
/**
 * Creates n random valid wchar_ts
 */
static void make_wchars(wchar_t* ws, size_t n)
{
  mbstate_t lastvalid, cur;
  size_t i;
  memset(&lastvalid, 0, sizeof(lastvalid));
  for(i = 0; i < n; ++i) {
    char buf[MB_CUR_MAX];
    do {
      memcpy(&cur, &lastvalid, sizeof(cur));
      ws[i] = (wchar_t) (rand() + 1);
    }
    while(wcrtomb(buf, ws[i], &cur) == (size_t) -1);
    memcpy(&lastvalid, &cur, sizeof(lastvalid));
  }
}
 
int main(void)
{
  typedef char* (*algorithm_t)(const wchar_t*);
  static const algorithm_t algorithms[] = {
    s_wstring_to_string_meas,
    s_wstring_to_string_max,
    s_wstring_to_string_min,
    s_wstring_to_string_guess,
    s_wstring_to_string_rope
  };
  static const char* const names[] = {
    "measure", "max", "min", "guess", "rope"
  };
  const size_t len = 4096, algo_n = sizeof(names) / sizeof(names[0]);
  wchar_t* wstr;
  size_t curlen;
  const char* operation;
  setlocale(LC_ALL, "");
  if(! (wstr = malloc(len * sizeof(wchar_t)))) {
    operation = "Allocating wide characters";
    goto err_rtrn;
  }
  make_wchars(wstr, len - 1);
  wstr[len - 1] = L'\0';
  for(curlen = 1; curlen <= len; curlen *= 2) {
    const wchar_t* cur_wstr;
    size_t repetitions;
    size_t algo_i;
    cur_wstr = wstr + len - curlen;
    repetitions = 4096 * 1024 * 8 / curlen;
    for(algo_i = 0; algo_i < algo_n; ++algo_i) {
      struct timespec starttime, endtime;
      size_t repetition_i;
      double spenttime, time_per_wchar;
      if(clock_gettime(CLOCK_MONOTONIC, &starttime)) {
    operation = "Getting time";
    goto err_free;
      }
      for(repetition_i = 0; repetition_i < repetitions; ++repetition_i) {
    char* mbstr;
    if(! (mbstr = algorithms[algo_i](cur_wstr))) {
      operation = names[algo_i];
      goto err_free;
    }
    free(mbstr);
      }
      if(clock_gettime(CLOCK_MONOTONIC, &endtime)) {
    operation = "Getting time";
    goto err_free;
      }
      spenttime = difftime(endtime.tv_sec, starttime.tv_sec)
    + (endtime.tv_nsec - starttime.tv_nsec) * 1e-9;
      time_per_wchar = spenttime / (repetitions * curlen);
      if(printf("%zu\t%s\t%g\n", curlen, names[algo_i], time_per_wchar) < 0) {
    operation = "Printing results";
    goto err_free;
      }
    }
  }
  free(wstr);
  return 0;
 err_free:
  _free_keep_errno(wstr);
 err_rtrn:
  perror(operation);
  return 1;
}