#include "derivReader.h"
#include <iostream>
#include <sstream>

using namespace std;

template <typename Data>
DerivReader<Data>::DerivReader(Grammar const &g, Data &d,
	void (*start_int_node)(Grammar &, string, vector<tToken> &, int, double, 
		int, int, Data &),
	void (*end_int_node)(Grammar &, vector<tToken> &, vector<tToken> &, Data &),
	void (*leaf)(Grammar &, tToken, vector<tToken> &, Data &))
	:_g(g), _d(d), _start_int_node_funct(start_int_node),
	_end_int_node_funct(end_int_node), _leaf_funct(leaf)
{
	rootnode.assign("^\\s*\\(([^\\s\"]+)\\s*\\(");
	endnode.assign("^\\s*\\)\\s*");
	startnode.assign("^\\s*\\(\\s*");
	intnode.assign( "^\\s*\\((\\d+) (\\S+) (\\S+) (\\d+) (\\d+)\\s*\\(");
	leafnode.assign( "^\\s*\\(\"(([^\"]*(\\\\\")*)*[^\"]*)\"(\\s*\\d+\\s*)?");
	tokenre.assign("^\\s*\"token (\\[.*?\\])\"");
}

template <typename Data>
DerivReader<Data>::~DerivReader()
{
}

template <typename Data>
void DerivReader<Data>::readDeriv(string treestring)
{
	string rest = treestring;
   vector<tToken> ancestors;

	readNode(rest, ancestors);
}

template <typename Data>
string DerivReader<Data>::readNode(const string &rest, vector<tToken> &ancestors)
{
	boost::smatch res;
	string next = rest;
	if (boost::regex_search(next, res, rootnode)) {
		ancestors.push_back(tToken(res[1].str()));
		next = next.substr(string(res[0].first, res[1].second).length());
		next = readNode(next, ancestors);
		if (!boost::regex_match(next, res, endnode)) {
			cerr << "Unbalanced tree at: " << next << endl;
			exit(1);
		}
		else {
			// finished node's children: function
			if (_end_int_node_funct != NULL) {
				vector<tToken> children;
				_end_int_node_funct(_g, children, ancestors, _d);
			}
		}
	} else if (boost::regex_search(next, res, intnode)) {
		string s1 = res[2].str();
      // record lexical type, not lex entry
      string let = _g.letype(s1);
      if (!let.empty()) {
         s1 = let;
         //modified to grab lemma, lex entry _and_ lex type
         //string lex = _g.lexeme(s1);
         //s1 = string(lex+"\t"+s1+"\t"+let);
      }
      for (unsigned int i=0;i<s1.length();i++) { // lower case all rules
         s1[i] = tolower(s1[i]);
      }

		// start and end of these internal 'tokens' are the vertix numbers, not
		// character positions
		int edge, start, end;
		double score;
		istringstream(res[1].str()) >> edge;
		istringstream(res[3].str()) >> score;
		istringstream(res[4].str()) >> start;
		istringstream(res[5].str()) >> end;

		// funct call to deal with internal nodes
		if (_start_int_node_funct != NULL)
			_start_int_node_funct(_g, s1, ancestors, edge, score, start, end, _d);
		// end funct call to deal with internal nodes

		ancestors.push_back(tToken(s1));
		ancestors.back().start(start);
		ancestors.back().end(end);
		next = next.substr(string(res[0].first,res[5].second).length());
		while (boost::regex_search(next, res, startnode)) {
			next = readNode(next, ancestors);
		}
		if (boost::regex_search(next, res, endnode)) {
			next = next.substr(string(res[0].first,res[0].second).length());
			ancestors.pop_back();
			// finished node's children: function
			if (_end_int_node_funct != NULL) {
				vector<tToken> children;
				_end_int_node_funct(_g, children, ancestors, _d);
			}
		} else {
			cerr << "Unbalanced node at: " << next << endl;
			exit(1);
		}
	} else if (boost::regex_search(next, res, leafnode)) {
		string s1 = res[1].str();	
		tToken newtok = tToken(s1);
		next = next.substr(string(res[0].first,res[0].second).length());
		while (next.substr(0,8) == "\"token [") {
			map<string,string> tokfeatures;
			next = readTokenFS(next, tokfeatures);
			string listfeat("+ID.LIST.FIRST");
			while (tokfeatures.count(listfeat) == 1 && 
				isdigit(tokfeatures[listfeat].at(0))) {
				istringstream idstr(tokfeatures[listfeat]);
				int tid;
				idstr >> tid;
				newtok.tid(tid);
				listfeat.erase(listfeat.length()-5);
				listfeat += "REST.FIRST";
			}
			if (tokfeatures.count("+CARG") == 1) {
				string carg = newtok.carg();
				if (!carg.empty()) 
					carg += "_";
				carg += tokfeatures["+CARG"];
				newtok.carg(carg);
			}
			if (tokfeatures.count("+FROM") == 1 && newtok.start() < 0) {
				istringstream numstr(tokfeatures["+FROM"]);
				int from;
				numstr >> from;
				newtok.start(from);
			}
			if (tokfeatures.count("+TO") == 1) {
				istringstream numstr(tokfeatures["+TO"]);
				int to;
				numstr >> to;
				newtok.end(to);
			}
			listfeat = "+STAG.+STAGS.FIRST";
			string listfeat2("+STAG.+SPRBS.FIRST");
			while (tokfeatures.count(listfeat) == 1) {
				string stag = tokfeatures[listfeat];
				string sprob = "0";
				if (tokfeatures.count(listfeat2) == 1) {
					sprob = tokfeatures[listfeat2];
				}
				newtok.supertag(stag);
				newtok.superprob(sprob);
				listfeat.erase(listfeat.length()-5);
				listfeat += "REST.FIRST";
				listfeat2.erase(listfeat2.length()-5);
				listfeat2 += "REST.FIRST";
			}
		}
		if (boost::regex_search(next, res, endnode)) {
			next = next.substr(string(res[0].first,res[0].second).length());
		} else {
			cerr << "Unbalanced node at: " << next << endl;
			exit(1);
		}
		
      // function to deal with finding a leaf
      if (_leaf_funct != NULL) 
         _leaf_funct(_g, newtok, ancestors, _d);
      // end function to deal with leaf

//      ancestors.push_back(tToken("\""+s1+"\""));

	} else {
		cerr << "Ill-formed derivation tree at: " << next << endl;
		exit(1);
	}
	return next;
}

template <typename Data>
string DerivReader<Data>::readTokenFS(string &rest, 
	map<string,string> &tokfeatures) {
	boost::smatch res;
	string tstring;
	map<int,string> reentrancies;

	if (boost::regex_search(rest, res, tokenre)) {
		tstring = string(res[1].first, res[1].second);
		rest = rest.substr(string(res[0].first,res[0].second).length());
		readFS(tstring, string(), tokfeatures, reentrancies);
		if (!tstring.empty() && tstring.at(0) == '"') {
			tstring.erase(0,1);
		}
	} else {
		cerr << "no token FS found at: " << rest << endl;
		exit(1);
	}
	removeWhitespace(rest);
	while (!rest.empty() && isdigit(rest.at(0))) { //another token
		rest.erase(0,1);
	}
	removeWhitespace(rest);
	return rest;
}

template <typename Data>
void DerivReader<Data>::readFS(string &tok, string featstr,
	map<string,string> &tokfeatures, map<int,string> &reentrancies) {

	string feat, val;
	pair<int,string> value;

	if (tok.at(0) != '[') {
		cerr << "not a FS at: " << tok << endl;
		exit(1);
	}
	tok = tok.substr(1);
	while (!tok.empty() && tok.at(0) != ']') {
		removeWhitespace(tok);
		feat = readFeat(tok);
		string fullfeat = featstr;
		if (!fullfeat.empty())
			fullfeat += ".";
		fullfeat += feat;
		removeWhitespace(tok);
		value = readVal(tok);
		if (value.first >= 0) {
			if (value.second.empty()) { //look up index
				val = tokfeatures[reentrancies[value.first]];
			} else { //first defn of index
				reentrancies.insert(pair<int,string>(value.first, fullfeat));
				val = value.second;
			}
		} else {
			val = value.second;
		}
		tokfeatures.insert(pair<string,string>(fullfeat, val));

		if (!tok.empty() && tok.at(0) == '[') {
			readFS(tok, fullfeat, tokfeatures, reentrancies);
		}
		removeWhitespace(tok);
	}
	if (tok.empty()) {
		cerr << "unterminated FS" << endl;
		exit(1);
	}
	if (tok.at(0) == ']') {
		tok = tok.erase(0,1);
	}
}


template <typename Data>
string DerivReader<Data>::readFeat(string &tok) {
	string feat;
	feat = parseSymbol(tok);
	removeWhitespace(tok);
	return feat;
}

template <typename Data>
pair<int,string> DerivReader<Data>::readVal(string &tok) {
	int index = -1;
	string val;
	if (tok.substr(0,3).compare("\\\\\"") == 0) {//parse string
		val = parseString(tok);
	} else { //parse symbol
		if (tok.at(0) == '#') { //index
			tok.erase(0,1);
			string indexstr;
			while (!tok.empty() && isdigit(tok.at(0))) {
				indexstr += tok.at(0);
				tok.erase(0,1);
			}
			istringstream indexstream(indexstr);
			indexstream >> index;
			if (tok.at(0) == '=') {//index and val
				tok.erase(0,1);
				if (tok.substr(0,3).compare("\\\\\"") == 0) {
					val = parseString(tok);
				} else {
					val = parseSymbol(tok);
				}
			}
		} else { //symbol
			val = parseSymbol(tok);
		}
	}
	removeWhitespace(tok);
	return pair<int,string>(index,val);
}

template <typename Data>
string DerivReader<Data>::parseSymbol(string &tok) {
	string symbol;
	while (!tok.empty() && !isspace(tok.at(0))) {
		symbol += tok.at(0);
		tok.erase(0,1);
	}
	return symbol;
}

template <typename Data>
string DerivReader<Data>::parseString(string &tok) {
	string val;
	tok.erase(0,3);
	while (!tok.empty()) {
		if (tok.length() >= 4 && tok.at(3) == '"') {
			if (tok.substr(0,3) != "\\\\\\") {
				if (tok.substr(1,2) == "\\\\") {
					val += tok.at(0);
					tok.erase(0,1);
					break;
				} else {
					cerr << "quoting wrong at: " << tok << endl;
					exit(1);
				}
			}
		}
		val += tok.at(0);
		tok.erase(0,1);
	}
	if (tok.length() >= 3 && tok.substr(0,3) == "\\\\\"") {
		tok.erase(0,3);
	} else {
		cerr << "Unterminated quoted string at: " << tok << endl;
		exit(1);
	}
	return val;
}

template <typename Data>
void DerivReader<Data>::removeWhitespace(std::string &rest) {
  while (!rest.empty() && isspace(rest.at(0)))
    rest.erase(0,1);
}