//File document.cpp

#include "document.h"
#include "chset.h"
//#include "concglob.h"
#include <iostream.h>
#include <string.h>
#include <stdlib.h>
#include <ctype.h>

chset others("!#$%^*()_-+=|\\{}[]:;\"'<,>.?/");  //does not include '@','&'

document::document(char* name):f() {
  f.open(name);
  if (!f) {
    cerr<<"file "<<name<<" not found!!\n";
    exit(1);
  }
  else {
    endofdocument=FALSE;
    linenumber=1;
    pagenumber=1;
  };
};

document::~document() {
  f.close();
};

void document::getwordandpage(char* word,int& page) {
  char ch;
  int len=0;
  f.get(ch);

  // a while loop to process the blank characters.
  while ((ch==' ' || ch=='\t' || ch=='\n' || others.in(ch)) && !f.eof()) {
    int tempLine = linenumber;                            
    if (ch=='\n')
      linenumber++;
    if ((((linenumber-1) % linePerPage) == 0) &&
        !(tempLine == linenumber) )  {
       pagenumber++;
       //cout << "In while loop, linenumber = " << linenumber << " pagenumber =="
       //     << pagenumber << " ch = "  << int (ch) << endl;
       //linenumber = 1; // reset linenumber
    }
    f.get(ch);
  };
  if (f.eof())
    endofdocument=TRUE;
  else {
    while (ch!=' ' && ch!='\t' && ch!='\n' && !f.eof() && !others.in(ch)) {
      s[len]=toupper(ch);
      len++;
      f.get(ch);
    };
  };
  s[len]='\0';
  //strupr(s);
  //line=linenumber;
  page=pagenumber;
    int tempLine = linenumber;
    if (ch=='\n')
      linenumber++;
    if ((((linenumber-1) % linePerPage) == 0) && !(tempLine == linenumber))  {
       pagenumber++;
       //cout << "In outside while loop, linenumber = " << linenumber
       //     << " pagenumber ==" << pagenumber << endl;
       //linenumber = 1; // reset linenumber
    }
    if (f.eof()) {
      endofdocument=TRUE;
      pagenumber--;
    }
  if (!f.eof())
     strcpy(word,s);
};