#include "LinkGenerator.h"
//#include "../link41b/include/link-includes.h"

using namespace std;

LinkGenerator::LinkGenerator(void)
{
}

LinkGenerator::~LinkGenerator(void)
{
}

DependencyStructure LinkGenerator::Generate(vector<WordItem> article) {
	DependencyStructure q;
	return q;
}
DependencyStructure LinkGenerator::Generate(string filename) {
	/* For each sentence, it will
	- integrate the words with the DependencyStructure
	- Add dependencies based on the link generator */
	
	dependencies.open("dependencies.csv");
	if (!dependencies.is_open()) {
		cout << "Unable to open dependencies file" << endl;
		char* xxxx;
		cin >> xxxx;
	}
	dependencies << "\"Left Index\",\"Left Word\",\"Label\",\"Label Description\",\"Right Index\",\"Right Word\"" << endl;

	string article = string();
	string word = string();
	fstream articleFile;
	articleFile.open(filename.c_str());
	while (!articleFile.eof()) {
		articleFile >> word;
		article.append(word);
		article.append(" ");
	}

	Dictionary    dict = dictionary_create("../link41b/data/4.0.dict", "../link41b/data/4.0.knowledge", NULL, "../link41b/data/4.0.affix");
    Parse_Options opts = parse_options_create();
	parse_options_set_max_parse_time(opts, 30);
	
	s = DependencyStructure();
	int position = 0;
	int starting = 0;

	vector<int> location = vector<int>();
	string sentence = string();

	while (nextSentence(article,sentence)) {
		LinksToDependencies(sentence, dict, opts);
		sentence = string();
	}

	dependencies.close();
	return s;
}

bool LinkGenerator::nextSentence(string &article, string &sentence) {
	if (article.length() == 0)
		return false;
	int end = 0;
	for (int i = 0; i < article.length(); i++) {
		if (i > 2) {
		if (((article[i-1] == '.') || (article[i-1] == '?') || (article[i-1] == '!'))
		&&  ((article[i] == ' ') || (article[i] == '\n') || (article[i] == '\t'))){// it is punctuation with a space after
			if ((i > 4) 
				&& !(article[i-2] == '.')
				&& !((article[i-2] == 'm') && (article[i-1] == 'r'))
				&& !((article[i-2] == 'm') && (article[i-1] == 's'))
				&& !((article[i-3] == 'm') && (article[i-2] == 'r') && (article[i-1] == 's'))
				&& !((article[i-2] == 'd') && (article[i-1] == 'r'))
				) {
					end = i;
					break;
			}
		}
		}
	}
	int len = article.length();
	if (end == 0) {
		return false;
	}
	if (end == article.length()-1) {
		sentence = article;
		article = "";
		return true;
	}
	sentence = article.substr(0,end);
	article = article.substr(end,-1);

	return true;
}

/* GetNextSentence will use an article of words, and a position to begin within the article
   to generate a string of the next sentence
   and to generate a mapping of locations for words that represent the same subject. */
bool LinkGenerator::GetNextSentence(int &position, int &starting, string &sentence, vector<int> &location) {
	starting = position;
	vector<WordItem> article = a; // TODO: renaming moo


	while (position < article.size()) { // check the position
		// the first priority is to construct the sentence.
		sentence.append(" ");
		sentence.append(article[position].word);
		if (article[position].tag[0] == 'N') {
			// if the WordItem is a noun, try to find it first
			location.push_back(s.Find(article[position].word));
			if (location[position] == -1) // if it's not found, just add it
				location[position] = s.AddWord(article[position]);
		} else { // if it's not a noun, just add it
			location.push_back(s.AddWord(article[position]));
			if (article[position].word == ".")
				break;
			//else
			//	printf("|%s|", article[position].word);
		}
		position++;
	}
	position++;
	if (sentence != "")
		return true;
	else
		return false;
}


string LinkGenerator::linkLabelBeginning(string label) {
	for (int i = 1; i < label.size(); i++) {
		if ((label[i] < 'A') || (label[i] > 'Z'))
			string batty = label.substr(0,i);
			return label.substr(0,i);
	}
	return label;
}


void LinkGenerator::LinksToDependencies(string sentence, Dictionary dict, Parse_Options opts) {
	// Check to make sure this does what I think it does
	// Is there a more effective way?
	// Convert the stl::string to a non-const char*
	parse_options_reset_resources(opts);
	char* sentenceChars = new char[sentence.size()+1];
	for (int i = 0; i < sentence.size(); i++) {
		sentenceChars[i] = sentence[i];
	}
	sentenceChars[sentence.size()] = '\0';

	Sentence sent = sentence_create(sentenceChars, dict);
	int num_linkages = sentence_parse(sent, opts);
	int num_words = sentence_length(sent);

	int starting = s.Verticies.size()-1;
	for (int i = 1; i < sentence_length(sent)-1; i++) {
		s.AddWord(WordItem(sentence_get_word(sent,i),""));
	}

	if (VERBOSITY > 1)
		cout << num_linkages << " linkages in: " << sentence << endl;
	for (int q = 0; q < num_linkages; q += num_linkages) {
		Linkage linkage = linkage_create(q, sent, opts);

		linkage_compute_union(linkage);
		linkage_set_current_sublinkage(linkage, linkage_get_num_sublinkages(linkage)-1);

		int links = linkage_get_num_links(linkage);
		int l, r;
		string label;
		for (int j = 0; j < links; j++) {
			label = linkLabelBeginning(linkage_get_link_label(linkage,j));
			l = linkage_get_link_lword(linkage,j)+starting;
			r = linkage_get_link_rword(linkage,j)+starting;
			//printf("	%s --> %s\n", linkage_get_word(linkage,linkage_get_link_lword(linkage,j)), linkage_get_word(linkage,linkage_get_link_rword(linkage,j)));
			//printf("		%s --> %s\n", linkage,linkage_get_link_llabel(linkage,j), linkage_get_link_rlabel(linkage,j));
			if ((label == "A") // Adj->Noun
				|| (label == "AM") // "as much" etc..
				|| (label == "AN") // noun-modifiers->Nouns
				|| (label == "AZ") // As->Verb
				|| (label == "B") // rel. clauses
				|| (label == "BI") // "be" to some idiomatic expression
				|| (label == "BT") // fronted objects in time expressions
				|| (label == "BW") // What, whatever -> Verbs
				|| (label == "C") // Coordinating Conjunction -> Subject
				|| (label == "CC") // Clauses to Coordinating Conjunctions
				|| (label == "D") // Determiners -> Nouns
				|| (label == "DD") // Definitive Determiners -> #s or Adj. as Nouns
				|| (label == "DG") // Determiner -> Proper Noun
				|| (label == "DP") // Posessive Determiner -> Gerund
				|| (label == "DT") // Determiners -> Nouns in time expressions
				|| (label == "E") // verb-modifying adverbs
				|| (label == "EA") // Adverbs -> Adjectives
				|| (label == "EB") // Adverbs -> forms of "be"
				|| (label == "EC") // Adverbs -> comparative adjectives
				|| (label == "EE") // Adverbs -> Adverbs
				|| (label == "EF") // "Enough" -> Adj/Adverbs
				|| (label == "EI") // Adverbs (eg soon, shortly) -> after, before
				|| (label == "EL") // ????? ____ -> Else
				|| (label == "EN") // Adverbs -> Quantities
				|| (label == "EZ") // Adverbs -> "as"
				|| (label == "FL") // for -> long
				|| (label == "FM") // from -> preposition
				|| (label == "GN") // common noun -> proper noun
				|| (label == "I") // ????? verbs -> Infinitives
				|| (label == "IN") // in ->	idiomatic time expressions
				|| (label == "J") // prepositions -> Objects
				|| (label == "JG") // prepositions -> Proper nouns: "of science"
				|| (label == "JT") // conjunctions -> time expressions
				|| (label == "M") // nouns -> modifiers
				|| (label == "MG") // certain prepositions -> nouns
				|| (label == "MV") // verbs -> modifying phrases
				|| (label == "MX") // noun -> modifiers
				|| (label == "ND") // numerical determiner -> expressions
				|| (label == "NF") // fractional words -> of
				|| (label == "NI") // special idiomatic number phrase -> number
				|| (label == "NJ") // used with NF
				|| (label == "NN") // connects number words in a series
				|| (label == "NR") // fraction words -> superlatives
				|| (label == "NS") // for singular expressions: Numerical determiner -> expression
				|| (label == "NT") // not -> to
				|| (label == "NW") // numbers -> fractional numbers
				|| (label == "O") // transitive verbs -> objects
				|| (label == "OD") // rise/fall -> distances
				|| (label == "OF") // certain verb/adjectives -> of
				|| (label == "ON") // on -> time expressions
				|| (label == "OT") // verbs -> time expressions (as objects)
				|| (label == "ON") // on -> time expressions
				|| (label == "P") // be verbs -> complements
				|| (label == "PP") // have -> past participles
				|| (label == "R") // nouns -> relative clauses
				|| (label == "RS") // start of a relative clause -> subject of
				|| (label == "S") // subject nouns -> finite verbs
				|| (label == "SI") // verbs -> subjects in subject-verb inversion
				|| (label == "SX") // I -> was/am
				|| (label == "SXI") // I -> was/am in subject verb inversion
				|| (label == "TA") // adjectives -> month names
				|| (label == "TD") // day-of-the-week -> morning|afternoon|evening
				|| (label == "TA") // adjectives -> month names
				|| (label == "TH") // verbs -> that[clause]
				|| (label == "TI") // titles
				|| (label == "TS") // certain verbs -> conjunctive clauses
				|| (label == "TT") // "time" -> previous adjective
				|| (label == "TW") // days of the week -> month names
				|| (label == "U") // subject-object/determiner -> noun
				|| (label == "UN") // until|since -> time phrases
				|| (label == "WN") // when phrases -> time nouns
				|| (label == "Y") // idiomatic time + place expressions <-
				|| (label == "YS") // 's -> noun

				|| (label == "G") // proper noun in a series
				|| (label == "ID") // idiomatic strings
				|| (label == "TM") // month names -> day numbers
				) {
					if ((l < s.Verticies.size()) && (r < s.Verticies.size())) {
						char* linflected = linkage_get_word(linkage,l-starting);
						char* rinflected = linkage_get_word(linkage,r-starting);
						if ((strlen(linflected) >= 2) && (linflected[strlen(linflected)-2] == '.'))
							s.Verticies[l].tag = linflected[strlen(linflected)-1];
						if ((strlen(rinflected) >= 2) && (linflected[strlen(rinflected)-2] == '.'))
							s.Verticies[l].tag = linflected[strlen(rinflected)-1];
						s.Edges.push_back(Dependency(l,r,label));
						if (VERBOSITY > 3)
							printf("#%-4d%-10s -- %-40s -> #%-4d%-10s\n",l,sentence_get_word(sent,l-starting),linkLabelDescription(label).c_str(),r,sentence_get_word(sent,r-starting));
						dependencies << l << ",\"" << sentence_get_word(sent,l-starting) << "\"," << label << ",\"" << linkLabelDescription(label) << "\"," << r << ",\"" << sentence_get_word(sent,r-starting) << "\"" << endl;
					} 

			}
			/*	|| (label == "G") // proper noun in a series
				|| (label == "ID") // proper noun in a series
				|| (label == "TM") // month names -> day numbers
					 make them the same
					 */
		}

		linkage_delete(linkage);
	}

	sentence_delete(sent);

	// now review all the words in the sentence, looking for duplicate objects.
	int n;
	for (int i = starting; i < s.Verticies.size(); i++) {
		n = s.Find(s.Verticies[i]);
		if ((s.Verticies[i].tag.compare("n") == 0) && (n != i)) {
			// if it finds one, replace its references in the edges.
			for (int j = 0; j < s.Edges.size(); j++) {
				if (s.Edges[j].child == i)
					s.Edges[j].child = n;
				if (s.Edges[j].parent == i)
					s.Edges[j].parent = n;
			}
		}
	}
}

string LinkGenerator::linkLabelDescription(string label) {
	if (label == "A") return "Adj->Noun";
	else if (label == "AM") return "'as much' etc..";
	else if (label == "AN") return "noun-modifiers->Nouns";
	else if (label == "AZ") return "As->Verb";
	else if (label == "B") return "rel. clauses";
	else if (label == "BI") return "'be' to some idiomatic expression";
	else if (label == "BT") return "fronted objects in time expressions";
	else if (label == "BW") return "What, whatever -> Verbs";
	else if (label == "C") return "Coordinating Conjunction -> Subject";
	else if (label == "CC") return "Clauses to Coordinating Conjunctions";
	else if (label == "D") return "Determiners -> Nouns";
	else if (label == "DD") return "Definitive Determiners -> #s or Adj. as Nouns";
	else if (label == "DG") return "Determiner -> Proper Noun";
	else if (label == "DP") return "Posessive Determiner -> Gerund";
	else if (label == "DT") return "Determiners -> Nouns in time expressions";
	else if (label == "E") return "verb-modifying adverbs";
	else if (label == "EA") return "Adverbs -> Adjectives";
	else if (label == "EB") return "Adverbs -> forms of 'be'";
	else if (label == "EC") return "Adverbs -> comparative adjectives";
	else if (label == "EE") return "Adverbs -> Adverbs";
	else if (label == "EF") return "'Enough' -> Adj/Adverbs";
	else if (label == "EI") return "Adverbs (eg soon, shortly) -> after, before";
	else if (label == "EL") return "????? ____ -> Else";
	else if (label == "EN") return "Adverbs -> Quantities";
	else if (label == "EZ") return "Adverbs -> 'as'";
	else if (label == "FL") return "for -> long";
	else if (label == "FM") return "from -> preposition";
	else if (label == "GN") return "common noun -> proper noun";
	else if (label == "I") return "????? verbs -> Infinitives";
	else if (label == "IN") return "in ->	idiomatic time expressions";
	else if (label == "J") return "prepositions -> Objects";
	else if (label == "JG") return "prepositions -> Proper nouns: 'of science'";
	else if (label == "JT") return "conjunctions -> time expressions";
	else if (label == "M") return "nouns -> modifiers";
	else if (label == "MG") return "certain prepositions -> nouns";
	else if (label == "MV") return "verbs -> modifying phrases";
	else if (label == "MX") return "noun -> modifiers";
	else if (label == "ND") return "numerical determiner -> expressions";
	else if (label == "NF") return "fractional words -> of";
	else if (label == "NI") return "special idiomatic number phrase -> number";
	else if (label == "NJ") return "used with NF";
	else if (label == "NN") return "connects number words in a series";
	else if (label == "NR") return "fraction words -> superlatives";
	else if (label == "NS") return "singular expressions: Numerical determiner -> expression";
	else if (label == "NT") return "not -> to";
	else if (label == "NW") return "numbers -> fractional numbers";
	else if (label == "O") return "transitive verbs -> objects";
	else if (label == "OD") return "rise/fall -> distances";
	else if (label == "OF") return "certain verb/adjectives -> of";
	else if (label == "ON") return "on -> time expressions";
	else if (label == "OT") return "verbs -> time expressions (as objects)";
	else if (label == "ON") return "on -> time expressions";
	else if (label == "P") return "be verbs -> complements";
	else if (label == "PP") return "have -> past participles";
	else if (label == "R") return "nouns -> relative clauses";
	else if (label == "RS") return "start of a relative clause -> subject of";
	else if (label == "S") return "subject nouns -> finite verbs";
	else if (label == "SI") return "verbs -> subjects in subject-verb inversion";
	else if (label == "SX") return "I -> was/am";
	else if (label == "SXI") return "I -> was/am in subject verb inversion";
	else if (label == "TA") return "adjectives -> month names";
	else if (label == "TD") return "day-of-the-week -> morning|afternoon|evening";
	else if (label == "TA") return "adjectives -> month names";
	else if (label == "TH") return "verbs -> that[clause]";
	else if (label == "TI") return "titles";
	else if (label == "TS") return "certain verbs -> conjunctive clauses";
	else if (label == "TT") return "'time' -> previous adjective";
	else if (label == "TW") return "days of the week -> month names";
	else if (label == "U") return "subject-object/determiner -> noun";
	else if (label == "UN") return "until|since -> time phrases";
	else if (label == "WN") return "when phrases -> time nouns";
	else if (label == "Y") return "idiomatic time + place expressions <-";
	else if (label == "YS") return "'s -> noun";
	else if (label == "G") return "proper noun in a series";
	else if (label == "ID") return "proper noun in a series";
	else if (label == "TM") return "month names -> day numbers";
	else return "UNKNOWN";
}