#include "MaltParserGenerator.h"

#include <iostream>
#include <fstream>

MaltParserGenerator::MaltParserGenerator(void)
{
}

MaltParserGenerator::~MaltParserGenerator(void)
{
}

/* Generate dependencies on a set of tagged sentences. */
DependencyStructure MaltParserGenerator::Generate(vector<WordItem> article) {
	cout << endl << "Running MaltParser ... ";
	// Save the list of tagged words in MaltTab
	DependencyStructure s;
	SaveAsMaltTab(article);
	// Run MaltParser on the file
	CallMaltParser();
	cout << "... ";
	// Read the dependency structure back from the MaltTab file
	vector<int> location;
	location.push_back(0);
	int i = 0;
	string word, tag, label;
	int parent;
	fstream Tagged;
	Tagged.open("DependencyTagged.txt");
	ofstream OriginalArticle;
	OriginalArticle.open("Original.txt", ios::out);
	// the malt tab file references parents by their original location
	// but we know that some of the nouns refer to the same thing
	// so let's create a new listing of what words go to what locations.
	while (!Tagged.eof()) { // for each WordItem
		i++;
		Tagged >> word;
		OriginalArticle << word << " ";
		Tagged >> tag;
		Tagged >> parent;
		Tagged >> label;
		if ((tag == "NP") || (tag == "NNP") || (tag == "NN")|| (tag == "NNS")) {
			// if the WordItem is a noun, try to find it first
			location.push_back(s.Find(word));
			if (location[i] == -1) // if it's not found, just add it
				location[i] = s.AddWord(WordItem(word,tag));
		} else { // if it's not a noun, just add it
			location.push_back(s.AddWord(WordItem(word,tag)));
		}
	}
	OriginalArticle.close();
	Tagged.close();
	fstream Dependencies;
	// Once we have that listing of what words go to what locations,
	// we can record the dependencies in the Edges array
	Dependencies.open("DependencyTagged.txt");
	i = 0;
	while (!Dependencies.eof()) {
		i++;
		Dependencies >> word;
		Dependencies >> tag;
		Dependencies >> parent;
		Dependencies >> label;
		s.Edges.push_back(Dependency(location[parent],location[i],label));
	}
	Dependencies.close();
	return s;
}

/* This saves the list of tagged words in MaltTab format
** MaltTab has one WordItem per line, with the WordItem and its tag separated by a tab */
void MaltParserGenerator::SaveAsMaltTab(vector<WordItem> article) {
	ofstream MaltTabArticle;
	MaltTabArticle.open("MaltTab.txt", ios::out);
	for (int i = 0; i < article.size(); i++) {
		if (article[i].word != "")
			MaltTabArticle << article[i].word << '\t' << article[i].tag << endl;		
	}
	MaltTabArticle.close();
}

/* Issues a systme call to the MaltParser program, which takes in the
** MaltTab data file and generates dependencies from it */
void MaltParserGenerator::CallMaltParser() {
	string command = "cd MaltParser && maltparser.exe -f eng/option.dat";
	system(command.c_str());
}