/* spellchk.cpp - Matt Mahoney, mmahoney@cs.fit.edu

This program takes 2 file name arguments.  It prints any word
that occurs in the first file but not in the second, along with
the line number.  If the second file is a dictionary, then the
program acts as a spell checker.

A word is any sequence of letters and no other characters.  For
example, "It's" is 2 words, "it" and "s".  Comparison
is not case sensitive.  The first line is 1.  For example, if
file1 and file2 are as follows:

file1:

  This is
  file 1.

file2:

  THIS is the
  OTHER-file...

The program runs as follows:

  spellchk file1 file2    (no output)

  spellchk file2 file1
  1: the
  2: OTHER

  spellchk file1 file1    (no output)

The program prints an error message if there are not exactly 2 file name
arguments or if either file cannot be read.

*/

#include <iostream>
#include <fstream>
#include <string>
#include <map>
#include <cctype>
using namespace std;

// Return string s with all upper case letters converted to lower case
string lowercase(string s) {
  for (string::iterator p=s.begin(); p!=s.end(); ++p) {
    if (isupper(*p))
      *p = tolower(*p);
  }
  return s;
}

// Print words in file argv[1] that are not in argv[2]
int main(int argc, char** argv) {

  // Check for 2 file names
  if (argc!=3) {
    cout << "Usage: spellchk test_file dictionary_file\n";
    return 1;
  }

  // Open files, exit on failure
  ifstream f1(argv[1]), f2(argv[2]);
  if (!f1)
    cout << "File " << argv[1] << " not found\n";
  if (!f2)
    cout << "File " << argv[2] << " not found\n";
  if (!f1 || !f2)
    return 1;

  // Store lower case words from file f2 in dict
  map<string, bool> dict;  // dict[word] is true if word is read from f2
  string s;  // current word
  char c;
  while (f2.get(c)) {
    if (isalpha(c))
      s+=c;
    else if (s!="") {
      dict[lowercase(s)]=true;
      s="";
    }
  }

  // Test words in file f1.  Print "line: word" if not in dict.
  s="";
  int line=1;  // Line number in f1
  while (f1.get(c)) {
    if (isalpha(c))
      s+=c;
    else if (s!="") {
      if (!dict[lowercase(s)])
        cout << line << ": " << s << "\n";
      s="";
    }
    if (c=='\n')
      ++line;
  }
  return 0;
}