package com.dlj.ir.analysis.pe

import java.util.Map;
import java.util.WeakHashMap;

/**
 * Persian stemmer light-  
 * Done by Ljiljana Doalmic(University of Neuchatel, www.unine.ch/info/clef/)
 *         
 * @author Ljiljana Dolamic. Email: ljiljana.dolamic@unine.ch 
 */
public class PersianStemmerLight {
	/**
	 * A cache of words and their stems
	 */
	static private Map<String, String> cache = new WeakHashMap<String, String>();
	/**
	 * A buffer of the current word being stemmed
	 */
	private StringBuilder sb = new StringBuilder();
	/**
	 * Default constructor
	 */
	public PersianStemmerLight() {
	}

	public String stem(String word) {
		String result = cache.get(word);

		if (result != null)
			return result;

		sb.delete(0, sb.length());

		sb.append(word);

		remove_kasra(sb);
		
		remove_suffix(sb);		
		
		remove_kasra(sb);

		result = sb.toString();
		cache.put(word, result);

		return result;
	}

	private void remove_suffix(StringBuilder word) {
		int len = word.length() - 1;
		if (len > 6) {		
			if (word.substring(len-3, len+1).equals("ترین")||
					word.substring(len-3, len+1.equals("آباد")||
					word.substring(len-3, len+1).equals("ترين"))
			{
				word.delete(len - 3, len + 1);
				return;
			}
			if (word.substring(len-3, len+1).equals("گيري ")||
					word.substring(len-3, len+1).equals("هايي")||
					word.substring(len-3, len+1).equals("هایی")||
					word.substring(len-3, len+1).equals("گیری")||
					word.substring(len-3, len+1).equals("سازي")||
					word.substring(len-3, len+1).equals("سازی")||
					word.substring(len-3, len+1).equals("ريزي")||
					word.substring(len-3, len+1).equals("ریزی")||
					word.substring(len-3, len+1).equals("بندي")||
					word.substring(len-3, len+1).equals("بندی")||
					word.substring(len-3, len+1).equals("آباد ")||
					word.substring(len-3, len+1).equals("باره"))
			{
				word.delete(len - 3, len + 1);
				return;
			}
		} /* end if len > 6 */
		if (len > 5) {			
			if (word.substring(len-2, len+1).equals("هاي")||
					word.substring(len-2, len+1).equals("های")||
					word.substring(len-2, len+1).equals("اند")||
					word.substring(len-2, len+1).equals("ايم")||
					word.substring(len-2, len+1).equals("ایم")||
					word.substring(len-2, len+1).equals("شان"))
			{
				word.delete(len - 2, len + 1);
				return;
			}
		} /* end if len > 5 */		
		if (len > 4) {	
			if (word.substring( len-1, len+1).equals("ان"))
			 {
				 word.delete(len - 1, len + 1);
				 normalize_1(word);
				 return;
			}
			if (word.substring( len-1, len+1).equals("ها")||
					word.substring( len-1, len+1).equals("ين")||
					word.substring( len-1, len+1).equals("ين")||
					word.substring( len-1, len+1).equals("ات")||
					word.substring( len-1, len+1).equals(" هء")||
					word.substring( len-1, len+1).equals("اش")||
					word.substring( len-1, len+1).equals("تر")||
					word.substring( len-1, len+1).equals("را")||
					word.substring( len-1, len+1).equals("ون")||
					word.substring( len-1, len+1).equals("ام"))
			{
				word.delete(len - 1, len + 1);
				return;
			}						
		} /* end if len > 4 */
		if (len > 2) {			
			if (word.charAt(len)=='ه'||
					word.charAt(len)=='ی'||  // Farsi yeh \u06CC
					word.charAt(len)=='ي'||  // Arabic yeh \u064A
					word.charAt(len)=='م'||
					word.charAt(len)=='ت'||
					word.charAt(len)=='ش') 
			{				
				word.deleteCharAt(len);
				return;
			}
		}
		return;
	}

	private void normalize_1(StringBuilder word) {
		int len = word.length() - 1;		
		if (len > 2) {			
			if (word.charAt(len)=='ى'|| // Alef maksura \u0649
					word.charAt(len)=='گ'||
					word.charAt(len)=='م'||
					word.charAt(len)=='ت'||
					word.charAt(len)=='ر'||
					word.charAt(len)=='ش') 
			{				
				word.deleteCharAt(len);
				normalize_2(word);
			}
		}            
		return;
	}
	private void remove_kasra(StringBuilder word) {
		int len = word.length() - 1;
		if (len > 3) {			
			if (word.charAt(len) == '  ِ' )
			 {
				word.deleteCharAt(len);		
				return;
			}			
		} /* end if len > 3 */		
	}
	private void normalize_2(StringBuilder word){
		int len = word.length() - 1;  
		if(len > 2){		
			if (word.charAt(len) == 'ی' || word.charAt(len) == 'ي')
			{
				word.deleteCharAt(len );
				return;
			}		
     	 	}
	}
}