iconv_speed.c

   1 #include <stdio.h>
   2 #include <iconv.h>
   3 #include <errno.h>
   4 #include <sys/time.h>
   5
   6 struct timeval tp1,tp2;
   7
   8 static void start_timer()
   9 {
  10         gettimeofday(&tp1,NULL);
  11 }
  12
  13 static double end_timer()
  14 {
  15         gettimeofday(&tp2,NULL);
  16         return((tp2.tv_sec - tp1.tv_sec) +
  17                (tp2.tv_usec - tp1.tv_usec)*1.0e-6);
  18 }
  19
  20 static size_t utf8_pull(char **inbuf, size_t *inbytesleft,
  21                          char **outbuf, size_t *outbytesleft)
  22 {
  23         while (*inbytesleft >= 1 && *outbytesleft >= 2) {
  24                 unsigned char *c = (unsigned char *)*inbuf;
  25                 unsigned char *uc = (unsigned char *)*outbuf;
  26                 int len = 1;
  27
  28                 if ((c[0] & 0x80) == 0) {
  29                         uc[0] = c[0];
  30                         uc[1] = 0;
  31                 } else if ((c[0] & 0xf0) == 0xe0) {
  32                         if (*inbytesleft < 3) {
  33                                 goto badseq;
  34                         }
  35                         uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF);
  36                         uc[0] = (c[1]<<6) | (c[2]&0x3f);
  37                         len = 3;
  38                 } else if ((c[0] & 0xe0) == 0xc0) {
  39                         if (*inbytesleft < 2) {
  40                                 goto badseq;
  41                         }
  42                         uc[1] = (c[0]>>2) & 0x7;
  43                         uc[0] = (c[0]<<6) | (c[1]&0x3f);
  44                         len = 2;
  45                 }
  46
  47                 (*inbuf)  += len;
  48                 (*inbytesleft)  -= len;
  49                 (*outbytesleft) -= 2;
  50                 (*outbuf) += 2;
  51         }
  52
  53         if (*inbytesleft > 0) {
  54                 errno = E2BIG;
  55                 return -1;
  56         }
  57
  58         return 0;
  59
  60 badseq:
  61         errno = EINVAL;
  62         return -1;
  63 }
  64
  65 static size_t utf8_push(char **inbuf, size_t *inbytesleft,
  66                          char **outbuf, size_t *outbytesleft)
  67 {
  68         while (*inbytesleft >= 2 && *outbytesleft >= 1) {
  69                 unsigned char *c = (unsigned char *)*outbuf;
  70                 unsigned char *uc = (unsigned char *)*inbuf;
  71                 int len=1;
  72
  73                 if ((uc[1] & 0xf8) == 0xd8) {
  74                         if (*outbytesleft < 3) {
  75                                 goto toobig;
  76                         }
  77                         c[0] = 0xed;
  78                         c[1] = 0x9f;
  79                         c[2] = 0xbf;
  80                         len = 3;
  81                 } else if (uc[1] & 0xf8) {
  82                         if (*outbytesleft < 3) {
  83                                 goto toobig;
  84                         }
  85                         c[0] = 0xe0 | (uc[1]>>4);
  86                         c[1] = 0x80 | ((uc[1]&0xF)<<2) | (uc[0]>>6);
  87                         c[2] = 0x80 | (uc[0]&0x3f);
  88                         len = 3;
  89                 } else if (uc[1] | (uc[0] & 0x80)) {
  90                         if (*outbytesleft < 2) {
  91                                 goto toobig;
  92                         }
  93                         c[0] = 0xc0 | (uc[1]<<2) | (uc[0]>>6);
  94                         c[1] = 0x80 | (uc[0]&0x3f);
  95                         len = 2;
  96                 } else {
  97                         c[0] = uc[0];
  98                 }
  99
 100
 101                 (*outbuf)[0] = (*inbuf)[0];
 102                 (*inbytesleft)  -= 2;
 103                 (*outbytesleft) -= len;
 104                 (*inbuf)  += 2;
 105                 (*outbuf) += len;
 106         }
 107
 108         if (*inbytesleft == 1) {
 109                 errno = EINVAL;
 110                 return -1;
 111         }
 112
 113         if (*inbytesleft > 1) {
 114                 errno = E2BIG;
 115                 return -1;
 116         }
 117
 118         return 0;
 119
 120 toobig:
 121         errno = E2BIG;
 122         return -1;
 123 }
 124
 125
 126
 127
 128 int main(int argc, char *argv[])
 129 {
 130         iconv_t cd;
 131         char *test_string, *buf1, *buf2;
 132         int buflen, string_len;
 133         int i;
 134         int len1, len2;
 135
 136         cd = iconv_open("UCS2", argv[1]);
 137         test_string = argv[2];
 138
 139         string_len = strlen(test_string);
 140         buflen = strlen(test_string)*5;
 141         buf1 = malloc(buflen);
 142         buf2 = malloc(buflen);
 143
 144         start_timer();
 145
 146         for (i=0; end_timer() < 1; i++) {
 147                 char *p = test_string;
 148                 char *q = buf1;
 149                 size_t inbytes = string_len;
 150                 size_t outbytes = buflen;
 151
 152                 iconv(cd, &p, &inbytes, &q, &outbytes);
 153                 len1 = buflen - outbytes;
 154
 155         }
 156
 157         printf("iconv: %g ops/sec\n", i/end_timer());
 158
 159         start_timer();
 160
 161         for (i=0; end_timer() < 1; i++) {
 162                 char *p = test_string;
 163                 char *q = buf2;
 164                 size_t inbytes = string_len;
 165                 size_t outbytes = buflen;
 166
 167                 utf8_pull(&p, &inbytes, &q, &outbytes);
 168                 len2 = buflen - outbytes;
 169         }
 170
 171         printf("internal: %g ops/sec\n", i/end_timer());
 172         printf("string_len=%d len1=%d len2=%d\n", string_len, len1, len2);
 173
 174         if (len1 != len2 || memcmp(buf1, buf2, len1) != 0) {
 175                 printf("results DIFFER! (%10.10s) (%10.10s)\n",
 176                        buf1, buf2);
 177         }
 178
 179         return 0;
 180 }