/************************************************************************ 
 **** Nicholas Erho **** OWNAGE.CA ****** C++ **** December 14, 2004 ****
 ************************************************************************
 *                             Anagram Finder                           *
 ************************************************************************
 * This program is designed to load a word list file compare every word *
 * in the word list file with every other word and determine the        *
 * anagrams of the word and outout them to a file.                      *
 * NOTE: For long word lists such as 81000 words this program can take  *
 * a considerable amount of time to complete.                           *  
 ************************************************************************ 
 * If you make any improvements which I am sure their are lots; I would *
 * love to see what you have done. So email me @ nicholaserho@ownage.ca *
 ************************************************************************/

#include <iostream> // General input and output.
#include <cctype> // This is going to be used to capitalize chars with the 'toupper' function. 
#include <string> // Casing to stings.
#include <fstream> // Used to get access to the filestream.

using namespace std;

// This function determines if string 1 and string 2 conatain the exact same characters elements, where order does not matter.
bool isanagram(string str1, string str2){
  register string tempstr1 = str1; // Fuction works on temp strings of 1 and 2 so that origional strings are left untouched.
  register string tempstr2 = str2; 
  if (tempstr1 == tempstr2){ // False because exactly the same word which violates the definition of an anagram.
    return false;
  }
  int strlen1 = str1.length();
  int strlen2 = str2.length(); 
  if (strlen1 != strlen2){ //False because of wrong string length and therefor can not contain the exact same elements.
    return false;
  }
  else{
    for (int i = 0; i <= strlen1-1; i++){ // runs though every char in the string1 array. 
      int j = 0;
      bool kill_loop = false;
      do{
	if (j > (strlen2-1)){ // False because missing common character.
	 return false;
	}
	if ((toupper(tempstr1[i]) == toupper(tempstr2[j])) && ((tempstr1[i] != '*') || (tempstr2[i] != '*'))){
	  tempstr1[i] = '*'; // insure that duplicate letters are not repeated
	  tempstr2[j] = '*';
	  kill_loop = true; // Exits the infinte loop if a match is found.
	}
	j++;
      }while(!kill_loop); 
    }
    return true; // If all of the above is not false, then the strings must be anagrams.
  }
}


int main(void){
  const int listsize = 81000; // Size of word list. 
  int number = 0; // number of anagram collections writen to the file.
  string wordlist[listsize]; // An array holding the entire list of words. NOTE: This is a BIG array and the stack may have to be increased.
  ifstream infile("wordlist.txt", ios::in); // Opens the wordlist text file for read.
  if (!infile){ // Error if file could not be opened.
    cerr << "This file could not be opened for read access." << endl;
    exit(1); // Kill program cause there is no point going on.
  }
  for (int i = 0; i <= listsize; i++){ // Run though the entire word list, this could probably be done better with dynamic means.
    if (!(infile >> wordlist[i])){
      wordlist[i] = ""; // Since 81000 is just a bit bigger than the word list it fills in the empty words with blanks. See above comment.
    }
  }
  infile.close(); // Since all the words are stored in the array, no need for this file to be open.
  ofstream outfile("anagramfile.txt", ios::out); // Open the output file.
  if (!outfile){ // Again if the file could not be opened display error message.
    cerr << "This file could not be opened for write access." << endl;
    exit(2); // Kill program cause if the file is not open there is no point going on.
  }
  // Display a header in the text file.
  outfile << "************************************************************************\n" 
             "**** Nicholas Erho **** OWNAGE.CA ****** C++ **** December 14, 2004 ****\n"
             "************************************************************************\n"
             "*                             Anagram Finder                           *\n"
             "************************************************************************\n"
             "* This file contains a list of anagrams generated from a wordlist file *\n"
	     "* using the Anagram Finder program.                                    *\n"  
             "************************************************************************" << endl;
  cout << "Percent done... " << endl;     
  for (register int j = 0; j <= listsize; j++){
    if ((j % 4050) == 0){ // Used 4050 just because that is exactly 5% of 81000 any number could be used.
      cout << (float(j)/listsize)*100 << "%... " << endl; // Prints the precent completed.
    }
    int anagramnumber = 0; // Number of anagrams in each anagram collection.
    for (register int k = 0; k <= listsize; k++){ // Runs though the word list looking for anagrams of the given word in wordlist[j].
      if (isanagram(wordlist[j], wordlist[k])){ // If new anagram found.
        anagramnumber++;
        if (anagramnumber == 1){ // if first anagram found.
          outfile << number++ << ".  " << wordlist[j] << "   " << wordlist[k];
        }
        else{ // If not first anagram found.
          outfile << "   " << wordlist[k];
        }
	wordlist[k] = ""; // Deletes the word from the word list so no duplicate anagrams are printed.
      } 
    }
    if (anagramnumber > 0){
      outfile << endl;  
    }
    wordlist[j] = ""; // Deletes the intial word from the word list so again no duplicate anagrams are printed.
  } 
  outfile.close(); // Closes the file output file.
  return 0; // Kills the program. 
}