#include "tRepp.h"
#include <boost/filesystem.hpp>
#include <boost/program_options.hpp>
#include <iostream>
#include <fstream>
#include "postag.h"

using namespace std;
void parse_options(int ac, char **argv, string *tokfname, ifstream *inputfile, 
	vector<string> *calllist, tTokenFormat *format, bool *tnt, bool *numopt,
	string *tntmodel, bool *quiet);

int main(int argc, char **argv)
{
	//option handling
	tTokenFormat format;
	vector<string> calllist;
	string tokfname;
	string tntmodel;
	ifstream inputfile;
	bool tnt,numopt,quiet;
	parse_options(argc, argv, &tokfname, &inputfile, &calllist, &format, &tnt,
		&numopt, &tntmodel, &quiet);
	//tokeniser
	boost::filesystem::path tokenizer(tokfname);
	#if BOOST_FILESYSTEM_VERSION >= 3
	tRepp repp(tokenizer.parent_path().string(), 
		tokenizer.stem().string(), quiet);
	#else
	tRepp repp(tokenizer.parent_path().string(),
		tokenizer.stem(), quiet);
	#endif


	//take input from file if given, else stdin
	istream &input = (inputfile.is_open()?inputfile:cin);

	string line;
	getline(input, line);
	while (!input.eof()) {
		string itemno = "UNKNOWN";
		if (numopt) {
			//input file can have itemno in first column, separated by @
			int at = line.find_first_of('@');
			if (at != string::npos) {
				itemno = line.substr(0,at);
				line.erase(0,at+1);
			}
		}
		vector<tToken> tokens(repp.tokenize(line, calllist));
		if (tnt) {
			tnttag(&tokens, tntmodel);
		}
		if (format == FSC || format == TNT)
			print_header(format, itemno, line, tokens.size());
		for (vector<tToken>::iterator iter = tokens.begin(); 
			iter != tokens.end(); ++iter) {
			iter->print(format);
			if (iter+1 != tokens.end()) {
				cout << iter->separator(format);
			}
		}
		if(numopt)
			cout << "#" << itemno;
		print_footer(format);
		getline(input, line);
	}
	return 0;
}

void parse_options(int ac, char **av, string *tokfname, ifstream *inputfile, 
	vector<string> *calls, tTokenFormat *format, bool *tnt, bool *numopt,
	string *tntmodel, bool *quiet)
{
	namespace po = boost::program_options;
	po::options_description opts("Options");
	opts.add_options()
		("help,h,?", "this usage information")
		("call,c", po::value< vector<string> >(calls)->composing(),
			"rpp calls, multiple options valid.")
		("format", po::value<string>()->default_value("STRING"), 
			"token format:STRING,PET,FSC,CANDC,TNT (default: STRING)")
		("tnt,t", "Use TnT POS tags")
		("num,n", "Output item and parse number.")
		("model,m", po::value<string>(tntmodel),
			"TnT model file (defaults to wsj model in LOGON tree)")
		("quiet,q", "Suppress informational messages.")
	;
	po::options_description hidden("Hidden options");
	hidden.add_options()
		("tokenizer-file", po::value<string>(tokfname), "tokenizer .rpp file")
		("input-file", po::value<string>(), "input file")
	;
	po::options_description cmd_line("Command line options");
	cmd_line.add(opts).add(hidden);
	po::positional_options_description p;
	p.add("tokenizer-file", 1).add("input-file", 1);

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

	if (vm.count("help")) {
		cout << "Usage: " << av[0] << " [options] tokenizer-file [input-file]"
			<< endl;
		cout << opts << endl;
		exit(0);
	}
	if (!vm.count("tokenizer-file")) {
		cerr << "Usage: " << av[0] << " [options] tokenizer-file [input-file]"
			<< endl;
		cerr << opts << endl;
		exit(1);
	}
	if (vm.count("input-file")) {
		inputfile->open(vm["input-file"].as<string>().c_str());
		if (!inputfile->is_open()) {
			cerr << "Couldn't open " << vm["input-file"].as<string>() << endl;
			exit(1);
		}
	}
	if (!vm.count("model"))
		*tntmodel = "";
	else {
		if (vm.count("tnt")) 
			cerr << "Warning: TnT model given, but TnT is not requested." 
				<< endl;
	}
	
	string f(vm["format"].as<string>());
	if (f.compare("STRING") == 0)
		*format = STRING;
	else if (f.compare("PET") == 0)
		*format = PET;
	else if (f.compare("FSC") == 0)
		*format = FSC;
	else if (f.compare("CANDC") == 0)
		*format = CANDC;
	else if (f.compare("TNT") == 0)
		*format = TNT;
	else {
		cerr << "Warning: invalid format " << f << " given. " 
			<< "Setting format to STRING" << endl;
		*format = STRING;
	}

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

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