//////////////////////////////////////////////////////////////////
//
//	slowmapwordcounter.cpp
//
//	Counts and reports the number of times each word in
//	the given input file occurs, and puts the report in
// 	the given output file.
//
//	Uses an elegant word-counting loop, but it's slower
//	than the ugly "insert"-based loop.
//
//	REVISION HISTORY:
//		2/10/01 (Jeff Ondich) Wrote the original version.
//		2/14/01 (JO) Modified the counting loop.
//
//////////////////////////////////////////////////////////////////

#include	<fstream>
#include	<map>
#include	<string>


void CleanWord( string& theWord );

int main( int argc, char *argv[] )
{
	// Deal with command line and open the files.
	if( argc != 3 )
	{
		cerr << "Usage: " << argv[0] << " inputFile outputFile" << endl;
		cerr << argv[0] << " collects word frequency data from inputFile" << endl;
		cerr << " and places the results in outputFile." << endl;
		exit( 1 );
	}

	ifstream	in( argv[1] );
	if( !in.is_open() )
	{
		cerr << "Cannot open " << argv[1] << endl;
		exit( 1 );
	}
		
	ofstream	out( argv[2] );
	if( !out.is_open() )
	{
		cerr << "Cannot open " << argv[2] << endl;
		exit( 1 );
	}
	

	// Get the words out of the input file and count them.
	map<string,int>							wordMap;
	string									word;

	while( in >> word )
	{
		CleanWord( word );
		if( wordMap.find( word ) == wordMap.end() )
			wordMap[word] = 1;
		else
			wordMap[word]++;
	}


	// Send the results to the output file.
	map<string,int>::const_iterator		wordIterator = wordMap.begin();
	while( wordIterator != wordMap.end() )
	{
		out << (*wordIterator).first << '\t' << (*wordIterator).second << endl;
		wordIterator++;
	}

	out.close();
	in.close();

	return 0;
}


//////////////////////////////////////////////////////////////////
//
//	CleanWord removes all non-alphabetic characters
//	from the given string (except for hyphens and
//	apostrophes), and reduces all upper case letters
//	to lower case.  Thus, once CleanWord is done, theWord
//	consists entirely of lower case letters, hyphens,
//	and apostrophes.
//
//////////////////////////////////////////////////////////////////

void CleanWord( string& theWord )
{
	string	s = "";
	int		originalLength = theWord.length();
	
	for( int i=0; i < originalLength; i++ )
	{
		char	c = theWord[i];
		if( isalpha(c)  ||  c == '\''  ||  c == '-' )
			s += tolower(c);
	}

	theWord = s;
}