// This file is a part of Framsticks SDK.  http://www.framsticks.com/
// Copyright (C) 1999-2015  Maciej Komosinski and Szymon Ulatowski.
// See LICENSE.txt for details.


#include <vector>
#include "frams/errmgr/stdouterr.h"
#include "frams/_demos/genotypeloader.h"
#include "frams/genetics/preconfigured.h"
#include "frams/virtfile/stdiofile.h"
#include "frams/model/similarity/simil_model.h"



using namespace std;

/** Computes a matrix of distances between all genotypes in the specified
    .gen file, using the matching and measure weights as specified in the
    command line.

    Command line parameters: [-names] <genotypesFile> <w_dP> <w_dDEG> <w_dNEU> <w_dGEO>

    Parameters:
      <genotypesFile> name of a file with genotypes (only f1 format)
      <w_dP> weight of the difference in the number of parts
      <w_dDEG> weight of the difference in degrees of matched parts
      <w_dNEU> weight of the difference in neurons of matched parts
      <w_dGEO> weight of the distance of matched parts

    Switches:
      -names specifies that the number and names of genotypes are to be printed to output
        before the distance matrix; by default the number and names are not printed

    Outputs a distance matrix in the format:
    <row_1> (columns in a row are separated by TABs)
    ...
    <row_n>

    Last modified: 2015-04-16
 */
int main(int argc, char *argv[])
{
    //StdoutErrorHandler err; //?TODO
    typedef double *pDouble;
    int iCurrParam = 0; // index of the currently processed parameter
    char *szCurrParam = NULL;
    ModelSimil M; // similarity computing object
    bool bPrintNames = false; // specifies if names of genotypes are to be printed
    int nResult = 0; // a temporary result

    if (argc < 6)
    {
        // too few parameters
        printf("Too few parameters!\n");
        printf("Command line: [-names] <genotypesFile> <matchType> <w_dP> <w_dDEG> <w_dNEU> <w_dGEO> <ifFUZZY>\n\n");
        printf("Parameters:\n");
        printf(" <genotypesFile> name of a file with genotypes (only f1 format is allowed)\n");
        printf(" <w_dP> weight of the difference in the number of parts\n");
        printf(" <w_dDEG> weight of the difference in degrees of matched parts\n");
        printf(" <w_dNEU> weight of the difference in neurons of matched parts\n");
        printf(" <w_dGEO> weight of the distance of matched parts\n\n");
        printf("Switches:\n");
        printf(" -names specifies that the number and names of genotypes are to be printed to output\n");
        printf("   before the distance matrix; by default the number and names are not printed\n\n");

        printf("Outputs a symmetric distance matrix in the format:\n");
        printf(" <row_1> (columns in a row are separated by TABs)\n");
        printf(" ...\n");
        printf(" <row_n>\n");

        return -1;
    }

    // prepare output parameters from .gen file
    vector<Geno *> *pvGenos = new vector<Geno *>();
    vector<char *> *pvNames = new vector<char *>();

    // check if there is a switch
    iCurrParam = 1;
    szCurrParam = argv[ iCurrParam ];
    if (strcmp(szCurrParam, "-names") == 0)
    {
        // switch "-names" was given; print names also
        bPrintNames = true;
        // pass to the next parameter
        iCurrParam++;
    }

    // check the parameters
    // get <genotypesFile> name from command line
    char *szFileName = argv[ iCurrParam ];

    // initially set measure components' weights to invalid values (negative)
    for (int i = 0; i < M.GetNOFactors(); i++)
    {
        M.m_adFactors[i] = -1.0;
    }

    const char *params[] = {"<w_dP>", "<w_dDEG>", "<w_dNEU>", "<w_dGEO>"};
    for (int i = 0; i < M.GetNOFactors(); i++)
    {
        iCurrParam++;
        szCurrParam = argv[ iCurrParam ];
        nResult = sscanf(szCurrParam, " %lf ", & M.m_adFactors[ i ]);
        if (nResult != 1)
        {
            // <w_dP> is not a number -- error
            printf("%s", params[i]);
            printf(" should be a number\n");
            return -1;
        }
        else
        {
            // <w_dP> is a number; check if nonnegative
            if (M.m_adFactors[ i ] < 0.0)
            {
                printf("%s", params[i]);
                printf(" should be a nonnegative number\n");
                return -1;
            }
        }
    }

    // read the input file
    // prepare loading of genotypes from a .gen file
    // create some basic genotype converters
    PreconfiguredGenetics genetics;
    StdioFileSystem_autoselect stdiofilesys;

    long count = 0, totalsize = 0;
    MiniGenotypeLoader loader(szFileName);
    MiniGenotype *loaded;
    while (loaded = loader.loadNextGenotype())
    {
        // while a valid genotype was loaded
        count++;
        totalsize += loaded->genotype.len();
        // create a Geno object based on the MiniGenotype
        Geno *pNextGenotype = new Geno(loaded->genotype);
        if ((pNextGenotype != NULL) && (pNextGenotype->isValid()))
        {
            pvGenos->push_back(pNextGenotype);
            char *szNewName = new char [ loaded->name.len() + 1];
            strcpy(szNewName, loaded->name.c_str());
            pvNames->push_back(szNewName);
        }
        else
        {
            printf("Genotype %2li is not valid\n", count);
        }
    }
    if (loader.getStatus() == MiniGenotypeLoader::OnError)
    {
        printf("Error: %s", loader.getError().c_str());
    }

    double dSimilarity = 0.0;
    double **aaSimil = NULL; // array of similarities

    // create the empty array of similarities
    aaSimil = new pDouble [pvGenos->size()];
    for (int k = 0; k < (int) pvGenos->size(); k++)
    {
        aaSimil[k] = new double [pvGenos->size()];
        for (int l = 0; l < (int) pvGenos->size(); l++)
            aaSimil[k][l] = 0.0;
    }

    // compute and remember similarities
    int i, j;
    for (i = 0; i < (int) pvGenos->size(); i++)
    {
        for (j = 0; j < (int) pvGenos->size(); j++)
        {
            dSimilarity = M.EvaluateDistance(pvGenos->operator[](i), pvGenos->operator[](j));
            aaSimil[i][j] = dSimilarity;
        }
    }

    if (bPrintNames)
    {
        // if "-names" switch was given,
        // print the number of genotypes and their names
        printf("%li\n", pvGenos->size());
        int iGen;
        for (iGen = 0; iGen < (int) pvNames->size(); iGen++)
        {
            printf("%s\n", pvNames->at(iGen));
        }
    } // if (bPrintNames)

    // print out the matrix of similarities
    for (i = 0; i < (int) pvGenos->size(); i++)
    {
        for (j = 0; j < (int) pvGenos->size(); j++)
        {
            printf("%.2lf\t", aaSimil[i][j]);
        }
        printf("\n");
    }

    // delete vectors and arrays
    for (i = 0; i < (int) pvGenos->size(); i++)
    {
        delete pvGenos->operator[](i);
        delete [] pvNames->operator[](i);
        delete [] aaSimil[i];
    }

    delete pvGenos;
    delete pvNames;
    delete [] aaSimil;

    return 0;
}