package morphology.parser;

import java.io.*;
import java.text.*;
import java.util.*;

/**
 * An Experimental Morphological Analyzer of Latvian.
 * Limited to inflectional morphology of nouns.
 * Techniques used: two-level morphology and finite-state transducers.
 *
 * Project: NGSLT --> NLP --> Words --> Assignment #1
 * @author Normunds Grūzītis, Gunta Nešpore, Baiba Saulīte
 * @date February 2006
 */
public class Analyzer {

	private SortedMap stems;
	private SortedMap endings;
	private SortedMap paradigms;

	/**
	 * Reads configuration data from an external properties file.
	 * @param properties pattern configuration file.
	 * @return list of key-value pairs.
	 */
	private Properties readConfiguration(File properties) throws IOException {
		Properties config = new Properties();

		FileInputStream fis = new FileInputStream(properties);
		InputStreamReader isr = new InputStreamReader(fis, "Cp1257");
		BufferedReader reader = new BufferedReader(isr);
		String line = null;

		while ((line = reader.readLine()) != null) {
			line = line.trim();

			//Valid configuration entry satisfies pattern "KEY=VALUE".
			if (!line.startsWith("#") && line.indexOf("=") > 0 && line.indexOf("=") < line.length() - 1) {
				String key = line.substring(0, line.indexOf("="));
				String value = line.substring(line.indexOf("=") + 1);

				config.setProperty(key, value);
			}
		}

		reader.close();
		return config;
	}

	/**
	 * Reads table of noun stems into memory.
	 * @param lex lexicon data file.
	 */
	private void readNounStems(File lex) throws IOException {
		FileInputStream fis = new FileInputStream(lex);
		InputStreamReader isr = new InputStreamReader(fis, "Cp1257");
		BufferedReader reader = new BufferedReader(isr);

		String line = null;
		ArrayList features = null;

		while ((line = reader.readLine()) != null) {
			//entry[0] = stem; entry[1] = pos; entry[2] = paradigm.
			String[] entry = line.split("\t");

			if (!stems.containsKey(entry[0])) stems.put(entry[0], new ArrayList());
			features = (ArrayList)stems.get(entry[0]);
			features.add(new NounStemFeature(entry[1], entry[2]));
			stems.put(entry[0], features);
		}

		reader.close();
	}

	/**
	 * Reads table of noun endings into memory.
	 * @param lex lexicon data file.
	 */
	private void readNounEndings(File lex) throws IOException {
		FileInputStream fis = new FileInputStream(lex);
		InputStreamReader isr = new InputStreamReader(fis, "Cp1257");
		BufferedReader reader = new BufferedReader(isr);

		String line = null;
		ArrayList features = null;

		while ((line = reader.readLine()) != null) {
			//entry[0] = ending; entry[1] = paradigm; entry[2] = number; entry[3] = case.
			String[] entry = line.split("\t");

			if (!endings.containsKey(entry[0])) endings.put(entry[0], new ArrayList());
			features = (ArrayList)endings.get(entry[0]);
			features.add(new NounEndingFeature(entry[1], entry[2], entry[3]));
			endings.put(entry[0], features);
		}

		reader.close();
	}

	/**
	 * Reads table of noun declensions into memory.
	 * @param lex lexicon data file.
	 */
	private void readDeclensions(File lex) throws IOException {
		FileInputStream fis = new FileInputStream(lex);
		InputStreamReader isr = new InputStreamReader(fis, "Cp1257");
		BufferedReader reader = new BufferedReader(isr);
		String line = null;

		while ((line = reader.readLine()) != null) {
			//entry[0] = name; entry[1] = gender.
			String[] entry = line.split("\t");
			paradigms.put(entry[0], entry[1]);
		}
		reader.close();
	}

	/**
	 * Constructor.
	 * @param properties configuration file describing location of a database
	 */
	public Analyzer(File properties) throws IOException {
		Properties config = readConfiguration(properties);

		stems = new TreeMap(Collator.getInstance(new Locale("lv", "LV")));
		endings = new TreeMap(Collator.getInstance(new Locale("lv", "LV")));
		paradigms = new TreeMap();

		//Read database and create search tree.
		readNounStems(new File(config.getProperty("noun_stems")));
		readNounEndings(new File(config.getProperty("noun_endings")));
		readDeclensions(new File(config.getProperty("declensions")));

		//
		//Read FSTs - not implemented yet.
		//
	}

	/**
	 * Transforms given word form from surface level into lexical level representation.
	 * @param wf word form to parse.
	 * @return list of possible morphological representations of the given form.
	 */
	public ArrayList parse(String wf) {
		ArrayList output = null;
		String stem = "";

		//
		//Apply plugged in FSTs - not implemented yet.
		//

		//Searching for stem - hard-coded FST approach.
		for (int i = 0; i < wf.length(); i++) {
			//Read char by char from the surface type and look-up for corresponding stem in the stem tree.
			stem = stem + wf.charAt(i);

			if (stems.containsKey(stem)) {
				//Remaining part of the word form has to be legal ending for the paradigm(s),
				//which is/are associated with the stem in the lexicon.
				//Nevertheless search for longer stem if there is more than one char left on surface type.
				String ending = "";
				if (i < wf.length()) ending = wf.substring(i + 1);
				if (!endings.containsKey(ending)) continue;

				//Read morphological features assigned to the stem.
				ArrayList stem_features = (ArrayList)stems.get(stem);
				Iterator itSF = stem_features.iterator();
				while (itSF.hasNext()) {
					NounStemFeature nsf = (NounStemFeature)itSF.next();
					String pos = nsf.getPOS();
					String s_decl = nsf.getParadigm();

					//Read morphological features assigned to the ending.
					ArrayList ending_features = (ArrayList)endings.get(ending);
					Iterator itEF = ending_features.iterator();
					while (itEF.hasNext()) {
						NounEndingFeature nef = (NounEndingFeature)itEF.next();
						String e_decl = nef.getParadigm();
						String number = nef.getNumber();
						String e_case = nef.getCase();

						//Stem and ending feature mapping through joint paradigm.
						if (s_decl.equals(e_decl)) {
							//Word form accepted (recognized).
							//Map gender.
							String gender = (String)paradigms.get(s_decl);

							//Prepare output.
							if (output == null) output = new ArrayList();
							output.add(stem + " " + pos + " " + gender + " " + number + " " + e_case);
						}
					}
				}
			}
		}

		return output;
	}

}