/* cansmi.c */

    /**********************************************************************
    * This source code is public domain, and may be freely distributed,	  *
    * modified, and used for any purpose.  IT COMES WITH ABSOLUTELY NO    *
    * WARRANTY OF ANY KIND.						  *
    **********************************************************************/

/*****************************************************************************
*  cansmi.c -- a simple filter to uniquify SMILES
*/

#include <stdio.h>
#include <string.h>
#include "dt_smiles.h"
#include "du_utils.h"

int main(int argc, char *argv[])
{
  dt_Handle mol;
  char      *smi, *cansmi, *p;
  int       count, goodcount, modcount, lensmi, lencan, len;
  int       iso;

  /*** Make Isomeric SMILES if "-i" option.  ***/
  if (argc > 1 && !strcmp("-i", argv[1]))
    iso = 1;
  else
    iso = 0;

  count = goodcount = modcount = 0;
  while( NULL != (smi = du_fgetline(&len, stdin)) ) {

    /*** Permit space or tab between SMILES and name. ***/
    p = smi;
    while (*p && (*p!=' ') && (*p!='\t')) p++;
    lensmi = p - smi;
    mol = dt_smilin(lensmi, smi);

    if (NULL_OB == mol)
      fprintf(stderr, "ERROR: (dt_smilin) Can't parse \"%s\"\n", smi);
    else {
      cansmi = dt_cansmiles(&lencan, mol, iso);
      if (0 >= lencan)
        fprintf(stderr,
          "ERROR: (dt_cansmiles) Can't make unique SMILES for \"%s\"\n", smi);
      else {

        /*** Print name if present. ***/
        if (*p) printf("%.*s %s\n", lencan, cansmi, p+1);
        else    printf("%.*s\n", lencan, cansmi);
        goodcount++;
        if (strncmp(smi, cansmi, lencan)) modcount++;
      }
      dt_dealloc(mol);
    }
    du_printerrors(stderr, DX_ERR_NOTE);
    dt_errorclear();
    count++;
  }
  free(smi);
  fprintf(stderr, "cansmi results:\n\tin: %d\n\tout: %d\n\tchanged: %d\n",
          count, goodcount, modcount);
  return 0;
}