package org.mpi.openmind.repository.utils;

import java.text.Normalizer;
import java.text.Normalizer.Form;
import java.util.ArrayList;
import java.util.List;
import java.util.regex.Pattern;

import org.apache.commons.lang.StringUtils;

/**
 * @author casties
 *
 * @see https://it-dev.mpiwg-berlin.mpg.de/tracs/OpenMind3/wiki/normalize_arabic_translit
 * 
 */
public class ArabicTranslitNormalizer {

    protected static List<ReplacementPattern> apostrophePatterns = new ArrayList<ReplacementPattern>();
    static {
        // `, ʿ, ʾ, ‘, ’ -> '
        //apostrophePatterns.put("'", Pattern.compile("\u0060|\u02BE|\u02BF|\u2018|\u2019"));
    	// remove apostrophes `, ʿ, ʾ, ‘, ’, '
        apostrophePatterns.add(new ReplacementPattern("", Pattern.compile("'|\u0060|\u02BE|\u02BF|\u2018|\u2019")));
    }    
    
    protected static List<ReplacementPattern> twoletterPatterns = new ArrayList<ReplacementPattern>();
    static {
        twoletterPatterns.add(new ReplacementPattern("j", Pattern.compile("ch")));
        twoletterPatterns.add(new ReplacementPattern("j", Pattern.compile("dj")));
        twoletterPatterns.add(new ReplacementPattern("t", Pattern.compile("th")));
        twoletterPatterns.add(new ReplacementPattern("h", Pattern.compile("kh")));
        twoletterPatterns.add(new ReplacementPattern("d", Pattern.compile("dh")));
        twoletterPatterns.add(new ReplacementPattern("s", Pattern.compile("sh")));
        twoletterPatterns.add(new ReplacementPattern("g", Pattern.compile("gh")));
    }

    protected static List<ReplacementPattern> wordpartPatterns = new ArrayList<ReplacementPattern>();
    static {
        // aẗ\b, at\b, ah\b -> a
        wordpartPatterns.add(new ReplacementPattern("a", Pattern.compile("a\u1E97\\b|at\\b|ah\\b")));
        // 'abd + space -> 'abd
        // now without apostrophe
        wordpartPatterns.add(new ReplacementPattern("abd", Pattern.compile("abd ")));
    }

    protected static List<ReplacementPattern> letterdiacritPatterns = new ArrayList<ReplacementPattern>();
    static {
        // ỳ -> a
        letterdiacritPatterns.add(new ReplacementPattern("a", Pattern.compile("\u1EF3")));
    }

    protected static List<ReplacementPattern> letterPatterns = new ArrayList<ReplacementPattern>();
    static {
        letterPatterns.add(new ReplacementPattern("j", Pattern.compile("g|c")));
    }
    
    protected static Pattern diacriticsPattern = Pattern.compile("\\p{InCombiningDiacriticalMarks}+");
    
    /**
     * Returns String normalized according to arabic transliteration rules.
     * 
     * @see https://it-dev.mpiwg-berlin.mpg.de/tracs/OpenMind3/wiki/normalize_arabic_translit
     * 
     * @param text
     * @return
     */
    public static String normalize(String text) {
        if (StringUtils.isEmpty(text)) {
            return text;
        }
        
        // everything is lowercase TODO: locale?
        text = text.toLowerCase();
        
        // replace "apostrophes"
        for (ReplacementPattern entry : apostrophePatterns) {
            text = entry.getPattern().matcher(text).replaceAll(entry.getReplacement());
        }
        
        // replace two-letter combinations
        for (ReplacementPattern entry : twoletterPatterns) {
            text = entry.getPattern().matcher(text).replaceAll(entry.getReplacement());
        }
        
        // replace word-parts
        for (ReplacementPattern entry : wordpartPatterns) {
            text = entry.getPattern().matcher(text).replaceAll(entry.getReplacement());
        }

        // replace letters with diacritics
        for (ReplacementPattern entry : letterdiacritPatterns) {
            text = entry.getPattern().matcher(text).replaceAll(entry.getReplacement());
        }

        // remove diacritics by de-composing and removing diacritical marks
        text = Normalizer.normalize(text, Form.NFD);
        text = diacriticsPattern.matcher(text).replaceAll("");
        
        // replace letters
        for (ReplacementPattern entry : letterPatterns) {
            text = entry.getPattern().matcher(text).replaceAll(entry.getReplacement());
        }

        return text;
    }
}