#ifdef HAVE_CONFIG_H
#include <config.h>
#endif

#include <iostream>
#include <string>
#include <vector>
#include <cmath>
#include <boost/filesystem.hpp>
#include <boost/program_options.hpp>

#include "unicode.h"
#include "grammar.h"
#include "profile.h"
#include "derivReader.h"
#include "derivReader.cpp"

using namespace std;

enum TagType {LETYPE, LETYPEMORPH};

TagType tagtype;

void collectleaves(Grammar &g, tToken s, vector<tToken> &ancestors, 
	vector<tToken> &leaves);

void parse_options(int argc, char **argv, string *gfname, string *pname, 
	TagType *tagtype, tTokenFormat *format, 
	bool *numopt, int *lim, bool *goldonly, 
	int *parsenum, int *itemno);
bool is_infl(string rule);

int main (int argc, char **argv)
{
	// setting option variables
	string gfname, pname;
	tTokenFormat format; bool numopt, goldonly; 
	int lim, parsenum, itemno;
	parse_options(argc, argv, &gfname, &pname, 
		&tagtype, &format,	&numopt, &lim, 
		&goldonly, &parsenum, &itemno);

	// UTF-8 encoder
	initialize_encoding_converter("utf-8");

	Grammar g(gfname);
	Profile p(pname);
	vector<tToken> leaves; //used in collectleaves()
	DerivReader<vector<tToken> > reader(g, leaves, NULL, NULL, &collectleaves);

	pair<pair<tIid,int>,string> result = p.getResult();
	tIid context = -1; // context is item id
	int event = -1; 	// event is parse id
	vector<tToken> itokens;

	while (result.first.first >= 0) {//new item/context
		if (result.first.first != context) {//new item/context
			if (context != -1) {
				//finish last context
			}//finished last context
		}
		context = result.first.first;
		event = result.first.second;
		if (lim > 0 && p.getReadings(context) >= lim) { //skip items with 
			result = p.getResult();								// greater than lim
			continue;												// readings
		}
		if (itemno >= 0 && context != itemno) { //select single item
			result = p.getResult();
			continue;
		}
		if (goldonly && (p.numGold(context) > 1 || !p.isGold(context, event))) {
			result = p.getResult();
			continue;
		}
		if (parsenum >= 0 && event != parsenum) {
			result = p.getResult();
			continue;
		}
		//do stuff with result
		leaves.clear();
		cout << "ITEM: " << context << endl;
		reader.readDeriv(result.second);
		if (numopt)
			cout << context << " " << event << " ";
		for (int idx=0; idx < leaves.size(); ++idx) {
			//string tag("NULL\tNULL\tNULL");
			string tag("NULL");
			tag = leaves[idx].tag();
			UnicodeString us = Conv->convert(leaves[idx].surface());
			if (us.length() == 1 && u_ispunct(us.charAt(0))) {
				if ((u_charType(us.charAt(0)) == U_START_PUNCTUATION
					|| us.compare(Conv->convert(string("“"))) == 0
					|| us.compare(Conv->convert(string("‘"))) == 0) 
					) {
					tag = "O:PUNCT";
				} 
			}
			//leaves[idx].print(format, pair<string,double>(tag,1));
			leaves[idx].print(format);

			if (idx != (leaves.size()-1))
				cout << leaves[idx].separator(format);
		}
		cout << endl;
		result = p.getResult();
	}
	//finish last context

	return 0;
}

void parse_options(int argc, char **argv, string *gfname, string *pname, 
	TagType *tagtype, tTokenFormat *format,
	bool *numopt, int *lim,
	bool *goldonly, int *parsenum, int *itemno)
{
	namespace po = boost::program_options;
	po::options_description visible("Options");
	visible.add_options()
		("help,h", "This usage information.")
		("infl,i", "Tags include morphological inflection rules.")
		("format", po::value<string>()->default_value("TNT"),
			"token format:TNT,CANDC,CONLL,FSC (default: TNT)")
		("num,n", "Output item and parse number.")
		("goldonly,g", "Only extract tags from 'gold' trees")
		("analysis,a", po::value<int>(parsenum)->default_value(-1),
			"Select a specific analysis, default (-1): all.")
		("single,s", po::value<int>(itemno)->default_value(-1),
			"Select a specific item, default (-1): all.")
		("limit,l", po::value<int>(lim)->default_value(0),
			"Number of readings at which a context is ignored.  "
			"Set to nbest to negate the effect of using a model during parsing.")
	;
	po::options_description hidden("Hidden options");
	hidden.add_options()
		("grammar-file", po::value<string>(gfname), "grammar .tdl file")
		("profile", po::value<string>(pname), "profile")
	;

	po::options_description cmd_line ("Command line options");
	cmd_line.add(visible).add(hidden);

	po::positional_options_description p;
	p.add("grammar-file",1).add("profile",1);

	po::variables_map vm;
	po::store(po::command_line_parser(argc, argv).
		options(cmd_line).positional(p).run(), vm);
	notify(vm);

	if (vm.count("help")) {
		cout << "Usage: " << argv[0] << " [options] "
			<< "grammar-file profile" << endl;
		cout << visible << endl;
		exit(0);
	}

	if (!vm.count("grammar-file") || !vm.count("profile")) {
		cerr << "Insufficient arguments given." << endl;
		cerr << "Usage: " << argv[0] << " [options] "
			<< "grammar-file profile" << endl;
		cerr << visible << endl;
		exit(1);
	}

	if (vm.count("infl"))
		*tagtype = LETYPEMORPH;
	else
		*tagtype = LETYPE;

   string f(vm["format"].as<string>());
   if (f.compare("FSC") == 0)
      *format = FSC;
   else if (f.compare("CANDC") == 0)
      *format = CANDC;
   else if (f.compare("CONLL") == 0)
      *format = CONLL;
   else if (f.compare("TNT") == 0)
      *format = TNT;
   else {
      cerr << "Warning: Invalid format " << f << " given. "
         << "Setting format to TNT" << endl;
      *format = TNT;
   }

	if (vm.count("num"))
		*numopt = true;
	else
		*numopt = false;

	if (vm.count("goldonly"))
		*goldonly = true;
	else
		*goldonly = false;

}

bool is_infl(string rule)
{
	if (
			(rule.size() > 4 && rule.compare(rule.size()-4,4,"odlr") == 0) ||
			(rule.size() > 3 && 
				(rule.compare(rule.size()-3,3,"olr") == 0 ||
				rule.compare(rule.size()-3,3,"ilr") == 0)))
		return true;

	if (rule.size() > 9 && rule.compare(rule.size()-9,9,"infl-rule") == 0)
		return true;

	return false;
}

void collectleaves(Grammar &g, tToken s, vector<tToken> &ancestors, 
	vector<tToken> &leaves)
{
	leaves.push_back(s);
	string tag(ancestors.back().surface());
   if (tagtype == LETYPEMORPH) {
		for (int x=ancestors.size()-2; x >= 0 
			&& g.is_lexrule(ancestors[x].surface()); x--) {
			//&& is_infl(ancestors[x].surface()); x--) {
			string infl = ancestors[x].surface();
			if (infl.compare(infl.length()-3, 3,"plr") != 0
				&& (infl.length() < 5 || 
					(infl.length() >=5 && infl.compare(0,5,"punct") != 0))) {
				tag+=":";
				tag+=infl;
			}
		}
	}

	leaves.back().tag(tag);
}