#include <iostream.h>
#include <fstream.h>
#include "apstring.h"
#include "strutils.h"
#include "apvector.h"
#include "apset3.h"
#include "systimer.h"

void ReadWords(const apstring & filename,
	           apset<apstring> & wordset)
// postcondition: all unique words in input stored in wordset    
{
    ifstream input;
    input.open(filename.c_str());
    apstring word;
    int count = 0;
    
    cout << "reading " << filename << endl;
    
    wordset.makeEmpty();
    while (input >> word)              // reading word succeeded
    {
	ToLower(word);
        StripPunc(word);		
	count++;
	wordset.add(word);
    }
    cout << "read " << count << " words, unique = " << wordset.size() << endl;    
}

int main()
{
    apstring basedir = "d:\\data\\shakespeare\\";
    const int NUM_PLAYS = 6;
    apvector<apstring> plays(6);
    plays[0] = "romeo.txt";
    plays[1] = "hamlet.txt";
    plays[2] = "errors.txt";
    plays[3] = "macbeth.txt";
    plays[4] = "errors.txt";
    plays[5] = "allswell.txt";        

    int k;
    apset<apstring> words,setunion,setintersection;
    SysTimer timer;

    timer.start();
    ReadWords(basedir+plays[0],words);
    timer.stop();
    cout << " first read = " << timer.elapsedTime() << endl;
    setunion = setintersection = words;

    for(k=1; k < NUM_PLAYS; k++)
    {
	timer.start();
	ReadWords(basedir+plays[k],words);
	timer.stop();
	cout << " first read = " << timer.elapsedTime() << endl;
		
	timer.start();
	setunion += words;
	timer.stop();
	cout << "union time = " << timer.elapsedTime() << endl;
		
	timer.start();
	setintersection *= words;
	timer.stop();
	cout << "intersect time = " << timer.elapsedTime() << endl;
	cout << "union size: " << setunion.size() << endl;
	cout << "intersection size:" << setintersection.size() << endl;
    }
         
    return 0;    
}