package morphology.parser;

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

/**
 * Service for acquisition of external information, e.g. list of stems and configuration data.
 *
 * Project: NGSLT --> NLP --> Words --> Assignment #1
 * @author Normunds Grūzītis, Gunta Nešpore, Baiba Saulīte
 * @version February-March 2006
 */
public class Utility {

	/**
	 * Reads configuration data from an external properties file.
	 * @param properties pattern configuration file.
	 * @return list of key-value pairs.
	 */
	public static 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.
	 * @return list of stems mapped by their morphological descriptions.
	 */
	public static SortedMap readNounStems(File lex) throws IOException {
		SortedMap stems = new TreeMap(Collator.getInstance(new Locale("lv", "LV")));

		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();
		return stems;
	}

	/**
	 * Reads table of morphotactics into memory.
	 * @param lex lexicon data file.
	 * @return list of rules.
	 */
	public static ArrayList readRegExRules(File lex) throws IOException {
		ArrayList rules = new ArrayList();

		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]=order; entry[1] = IF rule; entry[2] = THEN rule.
			String[] entry = line.split("\t");
			rules.add(new RegExRule(Integer.parseInt(entry[0]), entry[1], entry[2]));
		}

		reader.close();
		return rules;
	}

}