/* interval.c
   Compile with: gcc -g interval.c -lm -o interval

   Example run:  interval hs2_DEVIATION.X.l6 13

   Given a "DEVIATION" file (agree false positives & false negatives data)
   and the number of characters in the denomination of the runs preceding
   the length indicator, it outputs the intervals [start,end] of invariance
   for the percent identity, with the following format:
      start end   #fp   #fn   #fn+#fp

   Example:
   The denomination 'agree.hs2.X.l4.p37.list' in the DEVIATION file
   specifies an agree run for the hs2 region with the parameter values
   4 for minimum length of reported region, and 37% for the required
   percent identity, respectively. Gaps are not allowed (X). The
   <chars_to_len> distance for the file is then 13 (strlen("agree.hs2.X.l")).

*/

#include <stdio.h>
#include <stdlib.h>
#include <strings.h>
#include <ctype.h>
#include <sys/types.h>
#include <sys/param.h>

#define  EPSILON    1

#define Usage "%s <input file> <chars-to-percent>"

void fatal(char *message);
void fatalf(char *msg, char *val);
FILE *ckopen(char *name, char *mode);
int get_percent(char *);

int chars_to_length;

int main(int argc, char *argv[])
{
    char   buffer[200], key[50], *s;
    int    percent, ppercent;
    int    i, min, fp, fn, pfp, pfn, count;
    int    From[1000], To[1000];
    int    FP[1000], FN[1000];
    FILE  *file;
    
    if (argc!=3) fatalf(Usage, argv[0]);
    chars_to_length = atoi(argv[2]);
    
    file = ckopen(argv[1],"r");
    min = 100000;
    pfp = pfn = -1; count = -1;
    ppercent = -1;

    while (fgets(buffer,200,file)!=NULL) {
       if (buffer[0]=='#') continue;
       s = buffer; 
       i = 0;
       while (!isspace(*s)) { key[i] = *s; s++; i++; }
       key[i] = '\0';
       percent = get_percent(key);

       while (isspace(*s)) s++;     /* go past blanks */
       while (!isspace(*s)) s++;    /* go past fp: label */
       while (isspace(*s)) s++;     /* go past blanks */  
       fp = 0;
       while (isdigit(*s)) {         /* go past fp values */
          fp = fp*10 + *s-'0';  
          s++;
       }
       while (isspace(*s)) s++;     /* go past blanks */
       while (!isspace(*s)) s++;    /* go past fn: label */
       while (isspace(*s)) s++;     /* go past blanks */  
       fn = 0;
       while (isdigit(*s)) {        /* go past fn values */
          fn = fn*10 + *s-'0';
          s++;
       }  

       if ((fn!=pfn) || (fp!=pfp)) {
           /* new interval */
           if (count>=0) To[count] = percent-EPSILON; 
           From[++count] = percent;        
           To[count] = percent; 
           FP[count] = fp; FN[count] = fn;
           pfp = fp; pfn = fn;
       } else {
           /* same interval */
           To[count] = percent;
       }
    }
    fclose(file);
   
    /* print out the values */
    for (i= count; i>=0; i--)
         printf("%3d  %3d  %3d  %3d    %3d\n", From[count-i], To[count-i], 
                 FP[count-i], FN[count-i], FP[count-i]+FN[count-i]); 
 
    return 0;
}

void fatal(char *msg)
{                
        (void)fprintf(stderr, "%s\n", msg);
        exit(1); 
}    
     
void fatalf(char *msg, char *val)
{    
        (void)fprintf(stderr, msg, val);  
        (void)putc('\n', stderr);
        exit(1);
}            
          
FILE *ckopen(char *name, char *mode)
{    
        FILE *fp;
     
        if ((fp = fopen(name, mode)) == NULL)
                fatalf("Cannot open %s.", name);
        return fp;
}

/* prune key */
int get_percent(char *key)
{
     int   value, decs, sign = 1; 
     char *s;
     
     s = key;
     s = s+chars_to_length;
     while (isdigit(*s)) s++;    /* go past length field */
     s++; s++;                   /* go past '.' and the 'p' */
     
     if (*s=='-') { sign = -1; s++; }
     value = 0;
     while (isdigit(*s)) {
        value = value*10+*s-'0'; 
        s++;
     }
     s++;   /* replace '.' and skip it */
/* 
     decs = 0;
     while (isdigit(*s)) {      
        value = value*10+*s-'0';
        s++; decs++;
     }  
     rval = value;
     while (--decs>=0) rval = rval/10;
*/
      
     return sign*value;
}