#include "FreeLingTagger.h"

#include <iostream>
#include <fstream>

FreeLingTagger::FreeLingTagger(void)
{
}

FreeLingTagger::~FreeLingTagger(void)
{
}

/* Tag takes in a filename of a text file containing sentences
** and returns a list of the tagged words, using FreeLing's Tagger */
vector<WordItem> FreeLingTagger::Tag(string filename) {
	// Setup
	cout << endl << "Running FreeLing's Tagger ... ";
	vector<WordItem> t;
	string word, morph, tag;
	fstream Tagged;
	// Run FreeLing; it saves output as a file
	string outFile = CallFreeLing(filename);
	// Open the file and read the words into an array.
	Tagged.open(outFile.c_str());
	while (!Tagged.eof()) {
		t.push_back(WordItem(word,ConvertTag(word,tag)));
		Tagged >> word;
		Tagged >> morph;
		Tagged >> tag;
	}
	Tagged.close();
	return t;
}

/* This function issues a command line call to the FreeLing utility
** it returns the file that FreeLing saves it's tagged information in */
string FreeLingTagger::CallFreeLing(string filename) {
	string inFile = filename;
	string outFile = "Tagged.txt";
	string command = "cd FreeLing && analyzer.exe -f analyzer.cfg > ../" + outFile + " < ../" + inFile;
	system(command.c_str());
	return outFile;
}

/* ConvertTag takes in a WordItem and a tag from FreeLing's tagger
** it converts that tag into the subset of the Penn Treebank Tagset
** that MaltParser uses */
string FreeLingTagger::ConvertTag(string WordItem, string inTag) {
	if (inTag.compare("Z") == 0) // Z is a Cardinal Number
		return "CD";
	else if (inTag.compare("PP$") == 0) // Posessive Pronoun
		return "PRP$";
	else if (inTag.compare("NP") == 0) // Proper Noun, Singular
		return "NNP";
	else if ((WordItem.compare("to") == 0) && (inTag.compare("IN") == 0)) // To
		return "TO";
	else if (inTag[0] == 'F') { // Punctuation
		if ((WordItem[0] == '(') || (WordItem[0] == '{') || (WordItem[0] == '['))
			return "-LRB-";
		else if ((WordItem[0] == ')') || (WordItem[0] == '}') || (WordItem[0] == ']'))
			return "-RRB-";
		else if (WordItem[0] == '`')
			return "``";
		else if (WordItem[0] == '\'')
			return "\"";
		else if (WordItem[0] == '-')
			return ",";
		else if ((WordItem[0] == '?') || (WordItem[0] == '!'))
			return ".";
		else if ((WordItem[0] == ';') || (WordItem.compare("...") == 0))
			return ":";
		else 
			return WordItem;
	} else 
		return inTag;
}