/*  File: bin2hapmap.c
 *  Author: Luke Jostins (lj4@sanger.ac.uk)
 *  Copyright (C) Genome Research Limited, 2008
 *-------------------------------------------------------------------
 * Description:
 * Exported functions:
 * HISTORY:
 * Created: Mon June  22 16:52:22 2009 (lj4)
 *-------------------------------------------------------------------
 */

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#include <math.h>
#include "array.h"
#include "dict.h"
#include "tools.h"

// static int freadString (FILE *fil, char *s, int lenMax)
// { int len ;
//   fread (&len, sizeof(int), 1, fil) ;
//   if (len > lenMax)
//     { fprintf (stderr, "string too long (%d/%d)- abort\n",len,lenMax) ; return(0); /*exit (-1) ;*/ }
//   return fread (s, 1, len, fil) ;
// }

// static DICT *freadDict (FILE *fil)
// { char string[256] ;
//   DICT *dict ;
//   int n ;
//   if (!freadString (fil, string, 256) || strcmp (string, "DICT")
//       || !fread (&n, 1, sizeof(int), fil))
//     printf("failed!\n");//return 0 ;
//   dict = dictCreate (n) ;
//   while (n--)
//     if (!freadString (fil, string, 256) ||
// 	!dictAdd (dict, string, 0))
//       return 0 ;
//   return dict ;
// }

int main (int argc, char *argv[])
{
    char name2[256] ;
    FILE *genotypeFile = NULL;
    DICT *chromDict;		/* chromosome names */
    int i,n;
    int *gList;
    int list_sites = 0;
    int list_samples = 0;
    Array chrArray, posArray ;

    for (i=1; i<argc; i++)
    {
        if ( !strcmp(argv[i],"-l") || !strcmp(argv[i],"--list-sites") ) { list_sites=1; continue; }
        if ( !strcmp(argv[i],"-s") || !strcmp(argv[i],"--list-samples") ) { list_samples=1; continue; }
        if (!(genotypeFile = fopen(argv[i], "r")))
        { 
            fprintf(stderr, "Unknown argument or non-existent file: %s\n", argv[i]); 
            exit(-1); 
        }
    }
    if ( !genotypeFile )
    {
        fprintf(stderr,
            "Usage: bin2hapmap [OPTIONS] hapmap.bin\n"
            "Options:\n"
            "   -l, --list-sites        List only sites\n"
            "   -s, --list-samples      List only samples\n"
            "\n"
            );
        exit(-1);
    }


    /* first read the header of the genotypes file */

    chromDict = freadDict (genotypeFile) ;
    fread (&n, 1, sizeof(int), genotypeFile) ; /* n = number of sites */

    chrArray = arrayCreate (n, int) ;
    posArray = arrayCreate (n, int) ;
    fread (arrp(chrArray, 0, int), n, sizeof(int), genotypeFile) ;
    fread (arrp(posArray, 0, int), n, sizeof(int), genotypeFile) ;
    gList = (int*) malloc (n*sizeof(int)) ;

    static int map1[26];
    static int map2[26];

    map1[0] = map1[1] = map1[2] = map1[3] = 'A';

    map2[0] = 'A';
    map2[1] = 'C';
    map2[2] = 'G';
    map2[3] = 'T';

    map1[4] = map1[5] = map1[6] = 'C';
    map2[4] = 'C';
    map2[5] = 'G';
    map2[6] = 'T';

    map1[7] = map1[8] = 'G';

    map2[7] = 'G';
    map2[8] = 'T';

    map1[9] = map2[9] = 'T'; 
    map1[10] = map2[10] = 'N';

    /* read and check the samples from the genotypes file */

    while (freadString (genotypeFile, name2, 256))
    { 
        if ( list_samples )
        {
            printf("%s\n", name2);
            fseek(genotypeFile, n*sizeof(int), SEEK_CUR);
            continue;
        }

        if ( !list_sites )
            printf("# %s\n",name2);

        fread(gList, n, sizeof(int), genotypeFile) ;

        if (list_sites)
        {
            for (i = 0; i < n; i++) 
                printf("%s\t%d\n",dictName(chromDict,arr(chrArray,i,int)),arr(posArray,i,int));
            break;
        }
        else
        {
            for (i = 0; i < n; i++) 
                printf("%s\t%d\t%c%c\n",dictName(chromDict,arr(chrArray,i,int)),arr(posArray,i,int),map1[gList[i]],map2[gList[i]]);
        }
    }
    return 0;
}


/*************** enf of file *****************/