/*    13 October 2003 */


/*
   Sample easy-to-use main to call the function string
     in order to search Tandem Repeats in a nucleotide sequence
     (contained in a file with extension .seq, written either
     in GenBank or in FASTA format).
   The user must provide the sequence file name without extension and
     optionally the score threshold value.
   In unix-like environment the user can provide them as two optional
     parameters (if both parameters are missing they are both requested,
     if only the first parameter is given, a default value is used for
     the second).
   The two output file names are automatically created by completing
     the file name with  extensions .ris and .tab .
*/


#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <ctype.h>


int  main (int argc, char** argv);
int  read_f (char*, unsigned long*, char**);
int  string (unsigned long, char*, FILE*, FILE*, int);


#define len_name 81

#define score_default 500

FILE *fi, *fo1, *fo2;



int main (int argc, char** argv)
{
  int err, sc;
  char name [len_name],
       name_s [len_name+4], /* sequence file name */
       name_r [len_name+4], /* detailed results file name */
       name_t [len_name+4]; /* tabled results file name */
  char *s;
  unsigned long bp_no;


  if (argc >= 2)
  {
    if (strlen (argv [1]) >= len_name)
    {
      printf ("FILE NAME TOO LONG (max 80)!\n");
      exit (2);
    }

    strcpy (name, argv [1]);
  }
  else
  {
    printf ("sequence name? ");
    scanf ("%80s", name);
  }

  strcpy (name_s, name);
  strcpy (name_r, name);
  strcpy (name_t, name);

  strcat (name_s, ".seq");
  strcat (name_r, ".ris");
  strcat (name_t, ".tab");


  sc = score_default;

  if (argc > 2)
    sc = (int) strtol (argv [2], (char**) NULL, 10);
  else
    if (argc < 2)
    {
      printf ("score? ");
      scanf ("%d", &sc);
    }

  if (!read_f (name_s, &bp_no, &s))
    exit (3);


  if (!bp_no)
  {
    printf ("WRONG FILE LENGTH!\n");
    exit (3);
  }


  if ((fo1 = fopen (name_r, "w")) == NULL)
  {
    printf ("IMPOSSIBLE TO OPEN %s!\n", name_r);
    exit (4);
  }

  if ((fo2 = fopen (name_t, "w")) == NULL)
  {
    printf ("IMPOSSIBLE TO OPEN %s!\n", name_t);
    exit (5);

  }


  fprintf (fo1, "sequence name: %s with threshold %d\n\n", 
           name_s, sc);

  printf ("\n--------sequence length %ld\n\n", bp_no);
  printf ("threshold %d\n", sc);
  printf ("%10s %10s %6s %7s %s\n","TR start", 
         "TR end", "length", "score", "word");


  err = string (bp_no, s, fo1, fo2, sc);


  free (s);

  if (!err)
  {
    printf ("DONE!\n");
    fprintf (fo1, "\n******* DONE!\n\n\n\n\n\n");
  }
  else
  {
    printf ("WARNING(S).........DONE!\n");
    fprintf (fo1, "\nWARNING(S).........DONE!\n\n\n\n\n\n");
  }
  fclose (fo1);
  fclose (fo2);
  return (err);
}




int read_f (char *name_s, unsigned long *bp_no, char** s)

/*
    unsophisticated function that:
      - reads from a file (name_s) a nucleotide sequence
          (either in GenBank or in FASTA format)
          filling the variable *s
      - sets the variable *bp_no to the length of the sequence
      - returns 1 (or 0) to flag the success (or failure)
*/

{
  char ch, row [300];
  int i;
  unsigned long num_bases = 0;


  *bp_no = 0;

  if ((fi = fopen (name_s, "r")) == NULL)
  {
    printf ("IMPOSSIBLE TO OPEN %s!\n", name_s);
    return 0;
  }


  while (1)
  {
    memset (row, 0, 300);
    i = -1;
    do
    {
      if (++i >= 300 || (fscanf (fi, "%c", &row [i]) != 1))
      {
        printf ("UNKNOWN FORMAT\n");
        return 0;
      }
    }
    while (isprint (row [i]));

    if (row [0] == '>')
      break;

    if (!memcmp (row, "ORIGIN", 6))
      break;
  }

  while (fscanf (fi, "%c", &ch) == 1)
    if (isalpha (ch))
      num_bases++;

  fclose (fi);

  if ((*s = (char*) malloc (num_bases)) == NULL)
  {
    printf ("MEMORY ALLOCATION FAILURE\n");
    return 0;
  }

  *bp_no = num_bases;

  num_bases = 0;

  fi = fopen (name_s, "r");
  while (1)
  {
    memset (row, 0, 300);
    i = -1;
    do
    {
      if (++i >= 300 || (fscanf (fi, "%c", &row [i]) != 1))
      {
        printf ("UNKNOWN FORMAT\n");
        return 0;
      }
    }
    while (isprint (row [i]));

    if (row [0] == '>')
      break;

    if (!memcmp (row, "ORIGIN", 6))
      break;
  }

  while (fscanf (fi, "%c", &ch) == 1)
    if (isalpha (ch))
      (*s) [num_bases++] = ch;

  fclose (fi);
  return 1;
}