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("\u062A\u0631\u06CC\u0646")||
					word.substring(len-3, len+1).equals("\u0622\u0628\u0627\u062F")
					word.substring(len-3, len+1).equals("\u062A\u0631\u064A\u0646"))
			{
				word.delete(len - 3, len + 1);
				return;
			}
			if (word.substring(len-3, len+1).equals("\u06AF\u064A\u0631\u064A")||
					word.substring(len-3, len+1).equals("\u0647\u0627\u064A\u064A")||
					word.substring(len-3, len+1).equals("\u0647\u0627\u06CC\u06CC")||
					word.substring(len-3, len+1).equals("\u06AF\u06CC\u0631\u06CC")||
					word.substring(len-3, len+1).equals("\u0633\u0627\u0632\u064A")||
					word.substring(len-3, len+1).equals("\u0633\u0627\u0632\u06CC")||
					word.substring(len-3, len+1).equals("\u0631\u064A\u0632\u064A")||
					word.substring(len-3, len+1).equals("\u0631\u06CC\u0632\u06CC")||
					word.substring(len-3, len+1).equals("\u0628\u0646\u062F\u064A")||
					word.substring(len-3, len+1).equals("\u0628\u0646\u062F\u06CC")||
					word.substring(len-3, len+1).equals("\u0622\u0628\u0627\u062F")||
					word.substring(len-3, len+1).equals("\u0628\u0627\u0631\u0647"))
			{
				word.delete(len - 3, len + 1);
				return;
			}
		} /* end if len > 6 */
		if (len > 5) {
			if (word.substring(len-2, len+1).equals("\u0647\u0627\u064A")||
					word.substring(len-2, len+1).equals("\u0647\u0627\u06CC")||
					word.substring(len-2, len+1).equals("\u0627\u0646\u062F")||
					word.substring(len-2, len+1).equals("\u0627\u064A\u0645")||
					word.substring(len-2, len+1).equals("\u0627\u06CC\u0645")||
					word.substring(len-2, len+1).equals("\u0634\u0627\u0646"))
			{
				word.delete(len - 2, len + 1);
				return;
			}
		} /* end if len > 5 */		
		if (len > 4) {	
			if (word.substring( len-1, len+1).equals("\u0627\u0646")) 
			{
				word.delete(len - 1, len + 1);
				normalize_1(word);
				return;
			}
			if (word.substring( len-1, len+1).equals("\u0647\u0627")||
					word.substring( len-1, len+1).equals("\u06CC\u0646")||
					word.substring( len-1, len+1).equals("\u064A\u0646")||
					word.substring( len-1, len+1).equals("\u0627\u062A")||
					word.substring( len-1, len+1).equals("\u0647\u0621")||
					word.substring( len-1, len+1).equals("\u0627\u0634")||
					word.substring( len-1, len+1).equals("\u062A\u0631")||
					word.substring( len-1, len+1).equals("\u0631\u0627")||
					word.substring( len-1, len+1).equals("\u0648\u0646")||
					word.substring( len-1, len+1).equals("\u0627\u0645"))
			{
				word.delete(len - 1, len + 1);
				return;
			}						
		} /* end if len > 4 */
		if (len > 2) {			
			if (word.charAt(len)=='\u0647'||
					word.charAt(len)=='\u06CC'||
					word.charAt(len)=='\u064A'||
					word.charAt(len)=='\u0645'||
					word.charAt(len)=='\u062A'||
					word.charAt(len)=='\u0634') 
			{				
				word.deleteCharAt(len);
				return;
			}
		}
		return;
	}
	private void normalize_1(StringBuilder word) {
		int len = word.length() - 1;		
		if (len > 2) {			
			if (word.charAt(len)=='\u0649'||
					word.charAt(len)=='\u06AF'||
					word.charAt(len)=='\u0645'||
					word.charAt(len)=='\u062A'||
					word.charAt(len)=='\u0631'||
					word.charAt(len)=='\u0634') 
			{				
				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) == '\u0650' ) {
				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) == '\u06CC'||word.charAt(len) == '\u064A') 
			{
				word.deleteCharAt(len );
				return;
			}		
  	   	 }
	}
}