#include "opencv2/highgui/highgui.hpp"
#include "opencv2/core/core.hpp"
#include <iostream>

using namespace cv;
using namespace std;

struct pts
{
    float dot,high,space,num;
    bool IsSpam;
    pts()
    {
        dot=high=space=num=0;
    }
};

pts Str2Point(char str[])
{
    pts ret;
    int i;

    if(str[0]=='h')
    {
        i=4;
        ret.IsSpam=0;
    }
    else if(str[0]=='s')
    {
        i=5;
        ret.IsSpam=1;
    }
    else
    {
        printf("Error str!\n");
        return ret;
    }

    for(;str[i]!=0;i++)
    {
        if((str[i]>=65) && (str[i]<91))
            ret.high++;
        if((str[i]=='.') || (str[i]==',') || (str[i]=='?') || (str[i]=='!') || (str[i]=='-') || (str[i]==':') || (str[i]==';'))
            ret.dot++;
        if(str[i]==' ')
            ret.space++;
        if((str[i]>=48) && (str[i]<58))
            ret.num++;
    }
    //ret.high/=(float)i;
    ret.high=(float)i/*/905.0*/;
    //ret.dot/=(float)i;
    //ret.num/=(float)i;
    //ret.space/=(float)i;
    ret.space=ret.high/*/(float)i*/;

    //printf("dot=%f\nspace=%f\nnum=%f\nhigh=%f\ni=%d\n\n",ret.dot,ret.space,ret.num,ret.high,i);
    //system("pause>nul");
    return ret;
}

pts* FRead(char* Filename, int* SCount)
{
    FILE* f=fopen(Filename,"r");

    char tmp[1024];
    int i=0;

    while(!feof(f))
    {
        fgets(tmp,1023,f);
        *SCount=*SCount+1;
    }

    pts* P = new pts[*SCount];
    fseek(f,0,SEEK_SET);

    while(i<*SCount)
    {
        fgets(tmp,1023,f);
        P[i] = Str2Point(tmp);
        i++;
    }

    return P;
}

int main()
{
    int CLUSTERS;
    int SCount=0;

    printf("Enter number of clusters: ");
    scanf("%d",&CLUSTERS);

    pts* p=FRead("F:\\Second\\Second\\SMSSpamCollection",&SCount);

    Mat points(SCount, 1, CV_32FC4), labels(SCount, 1, CV_32FC4);
    Mat centers(CLUSTERS, 1, CV_32FC4/*points.type()*/);

    for(int j=0;j<SCount;j++)
    {
        points.at<float>(j,0) = p[j].high;
        points.at<float>(j,1) = p[j].num;
        points.at<float>(j,2) = p[j].space;
        points.at<float>(j,3) = p[j].dot;
    }

    kmeans(points, CLUSTERS, labels,
    TermCriteria( CV_TERMCRIT_EPS+CV_TERMCRIT_ITER, 20, 0.01),
    5, KMEANS_RANDOM_CENTERS, centers);

    int* spam=new int[CLUSTERS], *ham=new int[CLUSTERS];
    for(int i=0;i<CLUSTERS;i++)
    {
        spam[i]=ham[i]=0;
        for(int j=0;j<SCount;j++)
        if(labels.at<int>(j)==i)
        {
            if(p[j].IsSpam)
            spam[i]++;
            else
            ham[i]++;
        }
    }

    for(int i=0;i<CLUSTERS;i++)
    printf("CLUSTER #%d: \n\tSPAM: %d\n\tHAM: %d\n\n",1+i,spam[i],ham[i]);

    system("pause>nul");
    return 0;
}