utf8.cpp

/*
  Copyright 2013 Marek "p2004a" Rusinowski
*/
#include <cstdio>
#include <cstdlib>
#include <cstring>
#include <ctime>

typedef unsigned short int uchar;

inline int u2c(char *dest, uchar c) {
  unsigned int e = 0;
  if (c < 0x0080) {
    dest[0] = (char) c;
    return 1;
  }
  int len = 1
    , mask = 0x40
    , last = 0x80;
  while (c > mask) {
    e |= 0x80 | (0x3F & c);
    e <<= 8;
    c >>= 6;
    last |= mask;
    mask >>= 1;
    ++len;
  }
  e |= last | c;
  ((unsigned int *) dest)[0] = e;
  return len;
}

inline int c2u(uchar *dest, char *src) {
  if (!(0x80 & src[0])) {
    *dest = (uchar) src[0];
    return 1;
  }
  int len = 0;
  unsigned long c = 0LL;
  int i, j, mask = 0x7F;
  for (i = 1, j = 0x40; src[len] & j; ++i, j >>= 1) {
    mask &= ~j;  
  }
  mask &= ~j;
  c |= mask & src[len];
  while (--i) {
    ++len;
    c <<= 6;
    c |= src[len] & 0x3F;
  }
  *dest = (uchar) c;
  return len + 1;
}

char *u2cs(uchar *str, int ulen, int &clen) {
  int size = ulen * 2;
  char *res = (char *) malloc(ulen * 3);
  for (int i = 0, &j = clen = 0; i < ulen; ++i) {
    j += u2c(res + j, str[i]);
  }
  return res;
}

uchar *c2us(char *str, int clen, int &ulen) {
  if (!strncmp(str, "\xEF\xBB\xBF", 3)) {
    fprintf(stderr, "WARNING: Passed utf8 string with BOM.\n");
    str += 3;
    clen -= 3;
  }
  uchar *res = (uchar *) malloc(sizeof(uchar) * clen);
  for (int i = 0, &j = ulen = 0; i < clen; ++j) {
    i += c2u(res + j, str + i);
  }
  return res;
}

void print_tdiff(struct timespec &p1, struct timespec &p2) {
  int s = p2.tv_sec - p1.tv_sec - 1;
  int ns = (1000000000 - p1.tv_nsec) + p2.tv_nsec;
  if (ns > 1000000000) {
    ns %= 1000000000;
    ++s;
  }
  fprintf(stderr, "time: %02d.%09d\n", s, ns);
}

int main(int argc, char *argv[]) {
  if (argc < 2) return 1;
  FILE *file = fopen(argv[1], "r");
  if (!file) return 2;
  fseek(file, 0, SEEK_END);
  int size = ftell(file);
  fseek(file, 0, SEEK_SET);
  char *content = (char *) malloc(size);
  fread(content, 1, size, file);
  fprintf(stderr, "Loaded file\n");
  fclose(file);
  int text_len;
  struct timespec p1, p2;
  clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &p1);
  uchar *text = c2us(content, size, text_len);
  clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &p2);
  print_tdiff(p1, p2);
  if (!text) return 3;
  int out_len;
  clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &p1);
  char *out = u2cs(text, text_len, out_len);
  clock_gettime(CLOCK_PROCESS_CPUTIME_ID, &p2);
  print_tdiff(p1, p2);
  fwrite(out, 1, out_len, stdout);
  free(content);
  free(text);
  free(out);
  return 0;
}