/*******************************************************
PROGRAM NAME - buildDict

FILE NAME = buildDict.cxx

PROGRAMMER NAME - Justin C. Miller

USAGE = type './buildDict' to run me

COMPILE = type 'make buildDict' to use my makefile

DATE - 10-10-2003

BUGS - I believe that everything works properly

DESCRIPTION - This program reads in any number of files
       from the user and creates a dictionary using the
       words in those files.  The dictionary is then 
       placed in a file called "words.dat".
   
       WARNING:  so if a file named "words.dat" already 
       exists in your directory it will be overwritten!
*******************************************************/

#include <iostream> // cout, cin
#include <fstream>  // ifstream, ofstream
#include <cstdlib>  // EXIT_SUCCES, EXIT_FAILURE
#include <cstring>  // tolower, isalpha

#include "apvector.h" // so I can create apvector objects
#include "wordList.h" // so I can create wordList objects

//#define DEBUG  // for debugging purposes, to turn of, comment out

using namespace std ; // so I don't have to type std::cout

// function prototypes (tell main what functions it can use)
void buildDictionary(wordList &, const string &);
void getFilenames(apvector<string> &) ;
void insertWord(wordList &, string &) ;
void writeToFile(const wordList &, const string &);

// main function, where my program starts
int main(int argc, char * argv[]){
  wordList dictionary ; // create an empty dictionary
  apvector<string> filenames ; // create a vector of filenames
  string out_filename = "words.dat" ; // file to write to

  getFilenames(filenames) ; // get filenames from user

  for(int i = 0 ; i < filenames.size() ; i++) // use each file
    buildDictionary(dictionary, filenames[i]) ; // to build dictionary

  writeToFile(dictionary, out_filename) ; // write to file

  return EXIT_SUCCESS ;
}//end of main function

// function definitions below...
// -----------------------------

// name : buildDictionary
// parameters :
//   dictionary - a word List to add words to from filename
//   filename   - the file that contains a bunch of words
// pre  : dictionary can be empty or non-empty, it doesn't matter
//        filename should exist and have something in it
// post : puts all the words in filename into the dictionary wordList
void buildDictionary(wordList & dictionary, const string & filename){
  ifstream in_file(filename.c_str(), ios::in) ; // open input file
  if(!in_file.is_open()) { cout << "\n" << filename << " couldn't be opened.\n" ; return ;}
  string word ; 
  while(in_file >> word) // read everything in the file
    insertWord(dictionary,word) ; // place each word into the dictionary
}//end of buildDictionary

// name : getFilenames
// parameters :
//   filenames - collection of filenames
// pre  : the collection of filenames could be empty or non-empty, doesn't matter
// post : gets filenames from user using stdin, places them into collection
void getFilenames(apvector<string> & filenames){
  string input ;
  while(true){ // read from user until they type 'continue'
    cout << "\nEnter a filename (or 'continue' to build the dictionary):" ;
    cin >> input ; // read from user
    if(input == "continue") break ; // they typed 'continue'
    filenames.push_back(input) ; // add to list of filenames
  }//while
}//end of getFilenames

// name : insertWord
// parameters :
//   dictionary - a collection of words
//   word       - a word to be inserted into dictionary
// pre  : dictionary could be empty or non-empty, doesn't matter
//        word could exist or not-exist in dictionary, doesn't matter
//        word does not have to be properly formatted
// post : word is properly formatted and if it doesn't exist
//          in the dictionary, then it is inserted, else nothing
void insertWord(wordList & dictionary, string & word){
  if(word.length() < 2) return ; // check for single letter words
  int start = 0, end = 0 ;
  string sub_word ;
#ifdef DEBUG
  cout << "original word: " << word << endl ;
#endif
  for(int i = 0 ; i < word.length() ; i++){
    for(; isalpha(word[i]) != 0 ; i++,end++) // skip thru characters
      word[i] = (char) tolower(word[i]) ; // 1st lowercase them
    sub_word = word.substr(start, end-start) ;
    start = end = i + 1 ;
    if(sub_word.length() < 2) continue ; // skip 0 or 1 letter sub words
#ifdef DEBUG
  cout << "\tword inserting: " << sub_word << endl ;
#endif
    if(dictionary.insert(sub_word) == false){ // insert sub word
#ifdef DEBUG
      cout << "\t--" << sub_word << " already in dictionary" << endl ;
#endif
    }//if
  }//for
}//end of insertWord

// name : writeToFile
// parameters :
//   dictionary   - collection of words to be written to out_filename
//   out_filename - name of file to write the dictionary to
// pre  : dictionary is assumed to have some words in it
//        out_filename is assumed to not exist, otherwise it will overwrite existing file
// post : all words in dictionary are written to out_filename
//          overwriting original out_filename if it did exist
void writeToFile(const wordList & dictionary, const string & out_filename){
  ofstream out_file(out_filename.c_str(), ios::out) ; // open file
  if(!out_file.is_open()) {cout << "\n" << out_filename << " couldn't be created.\n" ; exit(EXIT_FAILURE) ;}
  out_file << dictionary ; // write to file using overloaded <<
  out_file.close() ; // close file
  cout << "\n\n--Dictionary (with " << dictionary.numberWords() 
       << " words) written to " << out_filename << "--" << endl ;
}//end of writeToFile