#include "TaggedTreebank.h"

#include <iostream>
#include <fstream>

/* Set up a list of acceptable tags */
TaggedTreebank::TaggedTreebank(void)
{
	tags.push_back("CC"); // Coordinating conjunction
	tags.push_back("CD"); // Cardinal number
	tags.push_back("DT"); // Determiner
	tags.push_back("EX"); // Existential there
	tags.push_back("FW"); // Foreign WordItem
	tags.push_back("IN"); // Preposition or subordinating conjunction
	tags.push_back("JJ"); // Adjective
	tags.push_back("JJR"); // Adjective, comparative
	tags.push_back("JJS"); // Adjective, superlative
	tags.push_back("LS"); // List item marker
	tags.push_back("MD"); // Modal
	tags.push_back("NN"); // Noun, singular or mass
	tags.push_back("NNS"); // Noun, plural
	tags.push_back("NNP"); // Proper noun, singular
	tags.push_back("NNPS"); // Proper noun, plural
	tags.push_back("PDT"); // Predeterminer
	tags.push_back("POS"); // Possessive ending
	tags.push_back("PP"); // Personal pronoun
	tags.push_back("PPS"); // Possessive pronoun
	tags.push_back("RB"); // Adverb
	tags.push_back("RBR"); // Adverb, comparative
	tags.push_back("RBS"); // Adverb, superlative
	tags.push_back("RP"); // Particle
	tags.push_back("SYM"); // Symbol
	tags.push_back("TO"); // to
	tags.push_back("UH"); // Interjection
	tags.push_back("VB"); // Verb, base form
	tags.push_back("VBD"); // Verb, past tense
	tags.push_back("VBG"); // Verb, gerund or present participle
	tags.push_back("VBN"); // Verb, past participle
	tags.push_back("VBP"); // Verb, non-3rd person singular present
	tags.push_back("VBZ"); // Verb, 3rd person singular present
	tags.push_back("WDT"); // Wh-determiner
	tags.push_back("WP"); // Wh-pronoun
	tags.push_back("WPS"); // Possessive wh-pronoun
	tags.push_back("WRB"); // Wh-adverb
}

TaggedTreebank::~TaggedTreebank(void)
{
}

/* Pull the PoS from the treebank and assemble the list of words 
** Tags will always be right after a (, and right before a WordItem */
vector<WordItem> TaggedTreebank::Tag(string filename) {
	cout << endl << "Pulling PoS from a Tagged Treebank ... ";
	vector<WordItem> t;
	string line, word, tag;
	bool lookForTag = true;
	fstream Tagged;
	string outFile = filename;
	Tagged.open(outFile.c_str());
	while (!Tagged.eof()) {
		getline(Tagged,line);
		for (int i = 0; i < line.size(); i++) {
			if (line[i] == '(') {// a tag is coming soon
				lookForTag = true;
				tag = "";
			} else if (line[i] == ')') { // A tag and a WordItem have been read in. save them.
				if (word != "") {
					if (tag.find("#") != -1)
						tag.resize(tag.find("#"));
					if (isValidTag(tag)) {
						t.push_back(WordItem(word,tag));
					}
					tag = "";
					word = "";
				}
			} else if ((lookForTag) && (line[i] != ' ') && (line[i] != '\t')) // the tag has been read in and is done
				tag = tag + line[i];
			else if ((!lookForTag) && (line[i] != ' ') && (line[i] != '\t')) // the WordItem has been read in and is done
				word = word + line[i];
			else if ((lookForTag) && ((line[i] == ' ') || (line[i] == '\t'))) // stop looking for a tag
				lookForTag = false;
		}	
	}
	Tagged.close();
	return t;
}

/* Make sure that the tag is in the tagset */
bool TaggedTreebank::isValidTag(string tag) {
	bool result = false;
	for (int i = 0; i < tags.size(); i++) {
		if (tag == tags[i]) {
			result = true;
			break;
		}
	}
	return result;
}