utils/sc2wngram.c

   1 #include <stdio.h>
   2 #include <string.h>
   3
   4 int main(int argc,char **argv)
   5 {
   6         char w1[750],w2[750],w3[750];
   7         char ow1[750],ow2[750],ow3[750];
   8         double count,total_count;
   9         total_count = 0;
  10         int n_digit = 2;
  11         long long int base;
  12
  13         if (argc == 2)
  14                 sscanf(argv[1],"%d",&n_digit);
  15         for (base = 1;n_digit > 0; n_digit --) base *= 10;
  16
  17         while (scanf("%s %s %lf",w1,w2,&count) == 3) {
  18                 if (!strcmp(w1,ow1) && !strcmp(w2,ow2))
  19                         total_count += count;
  20                 else {
  21                         if (total_count > 0)
  22                                 printf("%s %s %Ld\n",ow1, ow2, (long long int)(total_count*base) > 0 ? (long long int)(total_count*base) : (long long int)1);
  23                         strcpy(ow1,w1);
  24                         strcpy(ow2,w2);
  25                         total_count = count;
  26                 }
  27         }
  28
  29         /*
  30           Force to long long int because <s> <digit>/<opaque>/<punct>
  31           can easily run over int limit
  32         */
  33         if ((long long int)(total_count*base) > 0)
  34                 printf("%s %s %Ld\n",ow1, ow2, (long long int)(total_count*base) > 0 ? (long long int)(total_count*base) : (long long int)1);
  35         return 0;
  36 }