/*****************************************************************************
*** thorcommonsmi.c - how many & which smiles in common in dbs.
***                   Other primary ids not counted.
*** 
*** Author: Jeremy Yang
***   6 Feb 2001
*****************************************************************************/
#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include "dt_smiles.h"
#include "dt_thor.h"
#include "du_utils.h"

struct database {
  dt_Handle db;
  dt_Handle svr;
  char dbname[12];
  char dbpw[12];
  char host[12];
  char service[12];
  char user[12];
  char userpw[12];
  int size;
  dt_Handle smitype;
};

main(int argc,char *argv[])
{
  struct database *dbs;
  struct database *pdb;
  dt_Handle tdts,tdt,tdt2;
  char *prog=*argv;
  char *p,*q,*p2,*q2,*smi;
  char c[] = "|/-\\|/-\\";
  int i,j,ndb,is=0; /*** "is" = i-smallest ***/
  int printsmi=0;
  int tdtcount=0,len,isnew,hitcount=0,ci=0;
  
  /****************************************************************** 
  *** parse command line
  ******************************************************************/ 
  if (argc<3 || 0==strcmp(argv[1],"-help")) {
    printf("syntax: %s [-printsmi] db1 db2 [db3...]\n",prog);
    exit(1);
  }
  if (0==strcmp(argv[1],"-printsmi")) {
    printsmi=1; argv++; --argc;
  }
  dbs = (struct database *)malloc((argc-1)*sizeof(struct database));

  for (argv++,i=0;--argc;++argv,++i) {
    if (!du_parse_db(*argv,
                   dbs[i].dbname,
                   dbs[i].dbpw,
                   dbs[i].host,
                   dbs[i].service,
                   dbs[i].user,
                   dbs[i].userpw)) {
      fprintf(stderr,"ERROR: Can't parse db spec \"%s\"\n",*argv);
      exit(1);
    }
  }
  ndb=i;

  /****************************************************************** 
  *** open databases
  ******************************************************************/ 
  for (i=0;i<ndb;++i) {
    pdb=&dbs[i];
    pdb->svr = du_server(pdb->host,pdb->service,pdb->user,pdb->userpw);
    pdb->db = du_dbopen(pdb->svr,pdb->dbname,pdb->dbpw,"r");
    if (NULL_OB==pdb->svr || NULL_OB== pdb->db) {
      fprintf(stderr,"ERROR: Can't open \"%s@%s\"\n",pdb->dbname,pdb->host);
      exit(1);
    }
    pdb->size = dt_count(pdb->db,TYP_DATATREE);
    if (pdb->size < dbs[is].size) is = i;
    pdb->smitype = dt_thor_getdatatype(pdb->db,4,"$SMI");
    if (NULL_OB==pdb->smitype) {
      fprintf(stderr,"ERROR: No $SMI dtype in \"%s@%s\"\n",pdb->dbname,pdb->host);
      exit(1);
    }
  }
  fprintf(stderr,"NOTE: Smallest db is %s@%s with %d datatrees.\n",
    dbs[is].dbname,dbs[is].host,dbs[is].size);
  fprintf(stderr,"NOTE: Looking up all %s@%s smiles in other (%d) dbs...\n",
    dbs[is].dbname,dbs[is].host,ndb-1);

  /****************************************************************** 
  *** main loop - tdts of smallest database.
  ******************************************************************/ 
  tdts = dt_stream(dbs[is].db,TYP_DATATREE);
  for (tdtcount=0;NULL_OB!=(tdt=dt_next(tdts));++tdtcount) {
    smi=NULL;
    p = dt_thor_tdt2str(&len,tdt,0);
    if (strncmp(p,"$SMI<",5)) goto NEXTTDT;
    if (*(p+5)=='"') p2=p+6; else p2=p+5;
    q = du_find_character(len,p,">");
    if (*(q-1)=='"') --q;
    smi = du_strndup(p2,q-p2);

    for (i=0;i<ndb;++i) {
      if (i==is) continue;
      pdb=&dbs[i];

      tdt2 = dt_thor_tdtget(pdb->db,pdb->smitype,strlen(smi),smi,0,&isnew);
      if (NULL_OB==tdt2) goto NEXTTDT; /***Not a common smi.***/
      dt_dealloc(tdt2);
    }
    hitcount++;
    if (printsmi) printf("%s\n",smi);

NEXTTDT:
    if (!(tdtcount%100)) {
      fprintf(stderr,"%c%c",tdtcount?8:'\t',c[ci++%8]);
    }
    if (NULL_OB!=tdt) dt_dealloc(tdt);
    if (smi) free(smi);
  }
  dt_dealloc(tdts);
  fputc(8,stderr);
  fprintf(stderr,"\nRESULTS(%s): %d smiles shared by:\n",prog,hitcount);
  for (i=0;i<ndb;++i) {
    fprintf(stderr,"\t%s@%s\t%8d\n",dbs[i].dbname,dbs[i].host,dbs[i].size);
    dt_dealloc(dbs[i].db);
  }
}