Miscellany

/*
    File: Word.cc

    @authors:

    Antoine Waugh
    SID: 0309888

    @date: 02/06/05
*/

//System Includes

#include<string>
#include <iostream>
#include<vector>
#include<fstream>

//Other Includes

#include "Node.h"
#include "Word.h"
#include "Hit.h"

using namespace std;


// constructor that takes in the word to be searched
Word::Word(string searched_word) {

              //open word_cmp object file for searching
              word_cmp word_cmp_file("data/word.cmp");
              //open word_pos object file for linking word_cmp to word_idx
              word_pos word_pos_file("data/word.pos");
              //open word_idx object file to get required act,play and scene data
              word_idx word_idx_file("data/word.idx");
              //open speeches_idx object file to get required speech start and finish lines
              speeches_idx speeches_idx_file("data/speeches.idx");
              //open playfile_map object file to get required play name
              playfile_map playfile_map_file("data/playfile.map");

              //to be a vector<hits> once tested
              vector<Hit> returns;

                  //search through word_cmp for all records where word == searched_word

                  //iterate through all entries in word_cmp file
                  for(int i=0; i<word_cmp_file.size(); ++i) {

                          //check to see if current listing in file represents required word
                          if( searched_word.compare( trim(word_cmp_file[i].word) ) == 0) {
                                //find id from word_cmp to pass to word_pos
                                temp_id = word_cmp_file[i].id;


                                //create hit for found word

                                //use word_pos to link the identified index from word_cmp to word_idx
                                temp_position = word_pos_file[temp_id].position;
                                temp_extent   =  word_pos_file[temp_id].extent;

                                //this loops for each occurence of the word, ensuring all entries are returned (hits)
                                for(int x = 0; x<temp_extent; x++) {
                                    //increment temp_position for each entry found (extent)
                                    temp_position = temp_position + x;

                                    //use word_idx to gain play number, act, scene and speech position for further
                                    //querying
                                    play_num  = word_idx_file[temp_position].play;
          	                    act   = word_idx_file[temp_position].act;
          	                    scene = word_idx_file[temp_position].scene;
          	                    speech_pos = word_idx_file[temp_position].speech;
    
                                    //open value from speeches_idx for first and ending line numbers
                                    startLine = speeches_idx_file[speech_pos].startLine;
                                    endLine   = speeches_idx_file[speech_pos].endLine;
    
                                    play = playfile_map_file[play_num].getPlay();
                                    
                                    playfile_directory = "texts/";
                                    playfile_directory.append(playfile_map_file[play_num].second);
    
                                    ifstream fileIn(playfile_directory.c_str());
                                    unsigned int count =0;
                                    //account for first line
                                    string line;
                                    while(!fileIn.eof() && count <= endLine) {
                                            count++;
                                            getline(fileIn, line);
                                            if(count>startLine && count <=endLine+1) {
                                                    speech.append(line+"\n");
                                            }
                                    }
    
                                    //identify play, act and scene number from word_idx
                                    //identify play name
                                    Hit newHit(searched_word, play, act, scene, speech);
                                    returns.push_back(newHit);
                              }
                          }
                  }

                  for(int x = 0; x<returns.size(); ++x) {
                        cout << returns.at(x);
                  }

             //friend Hits operator()(const Context *context);

             //friend ostream& print(ostream& out);
}

/*

usage: to remove all whitespace from a given string
puropse: to be used in conjunction with the compare statement of the constructor
to ensure that any leading or trailing whitespace is not considered when evaluating
word_cmp's or the passed search_string.

*/
string
Word::trim(string x)
{
     //trim leading whitespace
     while(isWhitespace(x[0])) x.erase(0, 1);

     //generates iterator of type basic_string and assigns to beginning of string
     basic_string<char, std::char_traits<char> >::iterator i = x.begin();

     //iterate through string
     while(i != x.end())
     {
     	  //if iterator -> whitespace remove from string
          if(isWhitespace(*i)) i = x.erase(i);
          else i++;
     }

     return x;
}

bool 
Word::isWhitespace(const char& c)
{
     //add other whitespace chars as necessary and return
     return (c == ' ' || c == '\t');
}