#include "postag.h"
#include <cstdlib>
#include <string>
#include <cstring>
#include <sstream>
#include <iostream>
#include <stdio.h>
#include <fstream>

using namespace std;

void tnttag(vector<tToken> *tokens, string model)
{
	string tntcall("tnt -v0 -z100 ");
	string tntm(model);
	if (tntm.empty()) { // use default wsj model from the LOGON tree
		char *logonroot = getenv("LOGONROOT");
		if (!logonroot) {
			cerr << "Please set LOGONROOT so the TnT model can be found." << endl;
			exit(1);
		}
		tntm = string(logonroot)+"/coli/tnt/models/wsj.tnt";
	}

	string tmpdir("/tmp");
	char *tmpvar = getenv("TMPDIR");
	if (tmpvar)
		tmpdir = string(tmpvar);
	string ppidstr;
	stringstream pid;
	pid << getpid();

	char *tntinname = strdup(string(tmpdir+"/tntin."+pid.str()).c_str());
	ofstream tntinput(tntinname);
	if (!tntinput.is_open()) {
		cerr << "Couldn't open temp file " << tntinname << endl;
		exit(1);
	}
	//one token per line
	for (vector<tToken>::iterator iter = tokens->begin();
		iter != tokens->end(); ++iter)
		tntinput << iter->surface() << endl;
	tntinput.close();

	char *tntoutname = strdup(string(tmpdir+"/tntout."+pid.str()).c_str());
	tntcall = tntcall + tntm + " " + tntinname + " >" + tntoutname;
	int tntret = system(tntcall.c_str());
	if (tntret != 0) {
		perror(string(string("Error in TnT call: ")+tntcall).c_str());
		exit(1);
	}
	remove(tntinname);

	ifstream tntout(tntoutname);
	if (!tntout.is_open()) {
		cerr << "Couldn't open TnT output in " << tntoutname << endl;
		exit(1);
	}
	string line;
	int line_count = 0;
	getline(tntout, line);
	while (!tntout.eof()) {
		istringstream tagline(line);
		string tok, tag, prob;
		tagline >> tok;
		if (tok.compare((*tokens)[line_count].surface()) != 0) {
			cerr << "Token mismatch " << tok << " vs " 
				<< (*tokens)[line_count].surface() << endl;
			exit(1);
		}
		while (!tagline.eof()) {
			tagline >> tag >> prob;
			if (tagline.fail()) {
				cerr << "Malformed TnT ouput: " << line << endl;
				exit(1);
			}
			(*tokens)[line_count].tag(tag);
			(*tokens)[line_count].prob(prob);
		}

		line_count++;
		getline(tntout, line);
	}
	remove(tntoutname);
}

void chasentag(vector<tToken> *tokens, string chasenrc)
{
	string chasencall("chasen -i w");
	string chasenrcfile(chasenrc);
	if (chasenrcfile.empty()) { // use default .chasenrc from the LOGON tree
		char *logonroot = getenv("LOGONROOT");
		if (!logonroot) {
			cerr << "Please set LOGONROOT so .chasenrc file can be found." << endl;
			exit(1);
		}
		chasenrcfile = string(logonroot)+"/.chasenrc";
	}
	//this format needs to match how the output is read below
	string cformat("'%m  %P-+%Tn-%Fn\\n'");

	string input;
	for (vector<tToken>::iterator iter = tokens->begin();
		iter != tokens->end(); ++iter) {
		 input += iter->surface();
		 if (iter + 1 != tokens->end()) 
		 	input += " ";
	}

	string tmpdir("/tmp");
	char *tmpvar = getenv("TMPDIR");
	if (tmpvar)
		tmpdir = string(tmpvar);
	string ppidstr;
	stringstream pid;
	pid << getpid();
	char *chasenoutname = strdup(string(tmpdir+"/chasenout."+pid.str()).c_str());
	chasencall = "echo \"" +input + "\" |"+ chasencall + " -r " + chasenrcfile 
		+ " -F " + cformat + " > " + chasenoutname;
	int chasenret = system(chasencall.c_str());
	if (chasenret != 0) {
		perror(string(string("Error in chasen call: ")+chasencall).c_str());
		exit(1);
	}

	ifstream chasenout(chasenoutname);
	if (!chasenout.is_open()) {
		cerr << "Couldn't open chasen output in " << chasenoutname << endl;
		exit(1);
	}
	string line;
	int line_count = 0;
	getline(chasenout, line);
	while (!chasenout.eof()) {
		istringstream tagline(line);
		string tok, tag;
		tagline >> tok;
		if (tokens->size() > line_count) {	
			if (tok.compare((*tokens)[line_count].surface()) != 0) {
				cerr << "Token mismatch " << tok << " vs " 
					<< (*tokens)[line_count].surface() << endl;
				exit(1);
			}
			while (!tagline.eof()) {
				tagline >> tag;
				if (tagline.fail()) {
					cerr << "Malformed chasen ouput: " << line << endl;
					exit(1);
				}
				(*tokens)[line_count].tag(tag);
				(*tokens)[line_count].prob("1.00");
			}
		}

		line_count++;
		getline(chasenout, line);
	}
	remove(chasenoutname);
}