// Program: text_count.cpp
// Author: Chuck Stewart
// Purpose: Count the lines, characters (non-whitespace), and words
// in a text file. Count the occurrences of each letter and the
// occurrences of each word. Ignore punctuation in the words.

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

unsigned int count_characters( const string& a_line );
void add_to_letter_counts( const string & a_line,
                           vector<int>& letter_counters );
vector<string> break_up_line( const string& a_line );
void add_to_word_counts( vector<string> const& line_words,
                         map<string,int> & word_counts );

int main( int argc, char* argv[] )
{
  if ( argc != 3 )
  {
    cerr << "Usage: " << argv[0] << " text-file results-file";
    return 1;
  }

  ifstream in_str( argv[1] );
  if ( !in_str )
  {
    cerr << "Couldn't open " << argv[1] << " to read.\n";
    return 1;
  }

  ofstream out_str( argv[2] );
  if ( !out_str )
  {
    cerr << "Couldn't open " << argv[2] << " to write the results.\n";
    return 1;
  }

  unsigned int character_count = 0;
  unsigned int line_count = 0;

  vector<int> letter_counters( 26, 0 ); // Counts for the individual letters
   // to update for letter 'g'
   //    'g' - 'a'

  int all_word_counts = 0; // Total number of words
  map<string,int> word_counts; // Occurrence count for each word

  // Handle one line at a time...
  string a_line;
  while ( getline( in_str, a_line ) )
  {
    line_count ++ ;
    character_count += count_characters( a_line );
    add_to_letter_counts( a_line, letter_counters );
    vector<string> words_in_line = break_up_line( a_line );
    add_to_word_counts( words_in_line, word_counts );
    all_word_counts += words_in_line.size();
  }

  // Output char, word and line counters
  out_str << "\nHere are the statistics on the input text file:\n"
          << " char count = " << character_count << "\n"
          << " word count = " << all_word_counts << "\n"
          << " line count = " << line_count << "\n";

  // Output the letter counts
  out_str << "\nHere are the letter counts:\n";
  for ( unsigned int i = 0; i < 26; ++ i ) {
    out_str << " " << char( 'a' + i ) << ": " << letter_counters[ i ] << "\n";
  }

  // Output word occurrences
  out_str << '\n' << "Here are the word occurrence counts\n";
  map<string,int>::iterator mp = word_counts.begin();
  while ( mp != word_counts.end() ) {
    out_str << mp->first << '\t' << mp->second << '\n';
    mp++;
  }
}

// Return the number of non-whitespace characters on the line
unsigned int
count_characters( const string& a_line )
{
  unsigned int r = 0;
  string::const_iterator i = a_line.begin();
  while ( i != a_line.end() )
  {
//    if ( isalpha( *i ) ) {
    if ( !isspace( *i ) ) {
      r++;
    }

    i++;
  }

  return r;
}

// For each letter seen add to the appropriate count in the vector.
void
add_to_letter_counts( const string & a_line,
                      vector<int>& letter_counters )
{
  string::const_iterator i = a_line.begin();
  while ( i != a_line.end() )
  {
    if ( isalpha( *i ) )
    {
      ++letter_counters[ *i - 'a' ];   // e.g. 'g'-'a' ==> 71-65 ==> 6
                                       //      abcdefg...
    }

    i++;
  }
}

// Break up a string storing a line of input into a vector of strings
// storing the words from the line
vector<string>
break_up_line( const string& a_line )
{
  // e.g. "...Hello from New York!  Good-bye...later"
  //                    ^
  //                    i
  //
  // TO DO: modify this to correctly include the "later" string

  bool in_a_word = false;
  vector<string> words;    // "Hello"  "from"
  string s;
  string::const_iterator i = a_line.begin();

  while ( i != a_line.end() )
  {
    if ( !in_a_word ) {
      // find the next alpha character....
      if ( isalpha( *i ) ) {
        in_a_word = true;
      }
    }

    if ( in_a_word ) {
      if ( isalpha( *i ) ) {
        s.push_back( *i );
      } else {
        in_a_word = false;
        words.push_back( s );
        s.clear();
      }
    }

    i++;
  }

  return words;
}

// For each word from the input line add to the map, increasing the
// word's count by 1 (implicitly starting from 0 if the word was not
// in the map).
void add_to_word_counts( vector<string> const& line_words,
                         map<string,int> & word_counts )
{
  vector<string>::const_iterator p = line_words.begin();
  while ( p != line_words.end() ) {
    word_counts[ *p ] ++ ;
    p++;
  }
}