#include <fstream.h>
#include <fcl/vec.hh>
#include <matrix.hh>
#include <cluster.hh>

/* Program som använder Acquaintance-metoden för automatisk
   klassifikation av texter. Metoden baseras på de ideer som lanserades
   av Marc Damashek i artikeln "Gauging Similarity with n-grams: 
   Language-Independent Categorization of Text" som publicerades
   i Science 10 Feb. 1995, p 843ff.
   Smart inläsningsrutin som bara läser varje tecken en gång, oavsett
   vald längd på provträngarna (n-grammerna)

   Tillordning av en hashnyckel till varje teststräng.

   Inplockning av teststrängen i en hashtabell.

   Iteration genom texten.

   Presentation av textens koordinater - möjlighet till lagring och
   jämförelse med annan text.
   
   Aktuella n-grammer kan lagras i en sluten länkad lista. Varje 
   textinläsning gör att en n-gram blir färdig och kan hashas in.
   Samtidigt påbörjas en ny n-Gram. Specialfall erhålles i början 
   och slutet av texten; kanske kan dessa helt enkelt strykas.

   Hashtabellen sparas i en vektor.

   (C) This program and the subroutiner/headerfiles it uses are copyrighted
       by Jonas Gustavsson 1996.
*/;

const int n=5;

char ch;
double thr=0.07;
Vec<Cluster*> cl;
Vec<int> incl;

main()
{
  cl.insert(cl.end(),new Cluster("cluster of all texts"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/PraktiskSvenska/PRLeon"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/Busi/BETrans2.ascii"));
  cl[0]->hangon(new Hashtab(n,"../../../public_html/citat.e/kublakhan2.txt"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/appl2.txt"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/herrarne2.txt"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/kallocain2.txt"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/ucsdfreshman2.txt"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/Ling/art.txt"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/Ling/soc.txt"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/Ling/med.txt"));
  cl[0]->hangon(new Hashtab(n,"../../../maclada/texter/Ling/sci.txt"));

  bool placed;
  int focus=0;

  while (focus>=0)
  {
    int old_size=cl.size(); //Number of clusters so far.
    
    if (focus>0)
      cl[focus]->zoomin();  //"förstörande" inzoomning
    
    int cl_con=cl[focus]->ant();  //Number of texts in studied cluster
    incl.resize(cl_con);
    Matrix S(cl_con,cl_con);
    for (int i=0;i<cl_con;i++)
      incl[i]=-1;
    for (i=0;i<cl_con;i++)
    {
      S.put(1,i,i);
      Hashtab* texti=cl[focus]->medl(i);
      for (int j=i+1;j<cl_con;j++)
      {
	Hashtab* textj=cl[focus]->medl(j);
	cout<<"Similarity score ("<<texti->filn();
	cout<<" - "<<textj->filn()<<") : ";
	S.put(texti->comp(textj),j,i);
	S.put(S.get(j,i),i,j);
	cout<<S.get(i,j)<<'\n';
      }
    }
    //Texts with a similarity score above the threshold level are clustered
    //together.
    cl.insert(cl.end(),new Cluster("new cluster"));
    cl[cl.size()-1]->hangon(cl[focus]->medl(0));  //The first text starts off the first cluster.
    incl[0]=0;
    cout<<"Number of clusters: "<<cl.size()<<'\n';
    for (i=1;i<cl_con;i++)
    {
      placed=false;
      int j=0;
      while ((j<i) && (!placed))
      {
	cout<<"(i,j) S[i,j] ("<<i<<','<<j<<") "<<S.get(i,j)<<'\n';
	if (S.get(i,j)>thr)  //If text i is similar to text j, 
        {
	  cl[incl[j]]->hangon(cl[focus]->medl(i)); //it should be placed in
	  incl[i]=incl[j]; //the same cluster as that text.
	  placed=true;
	}
	j++;
      }
      if (!placed)
      {
	cout<<"No match for this text has been found! New cluster built!\n";
	cl.insert(cl.end(),new Cluster("new cluster"));
	cl[cl.size()-1]->hangon(cl[focus]->medl(i));
	incl[i]=cl.size()-1;
      }      
    }
    cout<<"Number of clusters: "<<cl.size()<<"  old_size="<<old_size<<'\n';
    for (i=old_size;i<cl.size();i++)
    {
      cout<<"Cluster no "<<i<<'\n';
      cl[i]->ready();  //detta tar lång tid...
      cl[i]->display();
      cl[i]->setname();
    }  
    cout<<"Similarity between clusters \n";
    
    for (i=old_size;i<cl.size()-1;i++)
    {
      for (int j=i+1;j<cl.size();j++)
      {
	cout<<"  "<<cl[i]->name()<<'-'<<cl[j]->name()<<" : "<<cl[i]->centroid()->comp(cl[j]->centroid())<<'\n';
      }
    }  
    cout<<"Please enter the number of the cluster you want to check out more in detail. \n";
    cout<<"(Enter -1 to exit program.)  ";
    cin>>focus;  
  }
  

  for (int i=0;i<cl.size();i++)
  {
    delete cl[i];
  }  
}