package com.gftech.ictclas4j.segment;

import java.util.ArrayList;

import com.gftech.ictclas4j.bean.Sentence;
import com.gftech.ictclas4j.utility.Utility;
import com.gftech.util.GFString;

public class SentenceSeg {
	private String src;
	private ArrayList<Sentence> sens;
	
	public SentenceSeg(String src){
		this.src=src;
		sens=sentenceSplit();
	}
	/**
	 * 進行句子分隔
	 * 
	 * @param src
	 * @return
	 */
	private ArrayList<Sentence> sentenceSplit( ) {
		ArrayList<Sentence> result = null;

		if (src != null) {
			result = new ArrayList<Sentence>();
			String s1 = Utility.SENTENCE_BEGIN;
			String[] ss = GFString.atomSplit(src);

			for (int i = 0; i < ss.length; i++) {
				// 如果是分隔符，比如回車換行/逗號等
				if (Utility.SEPERATOR_C_SENTENCE.indexOf(ss[i]) != -1
						|| Utility.SEPERATOR_LINK.indexOf(ss[i]) != -1
						|| Utility.SEPERATOR_C_SUB_SENTENCE.indexOf(ss[i]) != -1
						|| Utility.SEPERATOR_E_SUB_SENTENCE.indexOf(ss[i]) != -1) {
					// 如果不是回車換行和空格
					if (Utility.SEPERATOR_LINK.indexOf(ss[i]) == -1)
						s1 += ss[i];
					// 斷句
					if (s1.length() > 0 && !Utility.SENTENCE_BEGIN.equals(s1)) {
						if (Utility.SEPERATOR_C_SUB_SENTENCE.indexOf(ss[i]) == -1
								&& Utility.SEPERATOR_E_SUB_SENTENCE
										.indexOf(ss[i]) == -1)
							s1 += Utility.SENTENCE_END;

						result.add(new Sentence(s1, true));
						s1 = "";
					}

					// 是回車換行符或空格，則不需要進行分析處理
					if (Utility.SEPERATOR_LINK.indexOf(ss[i]) != -1) {
						result.add(new Sentence(ss[i]));
						s1 = Utility.SENTENCE_BEGIN;

					} else if (Utility.SEPERATOR_C_SENTENCE.indexOf(ss[i]) != -1
							|| Utility.SEPERATOR_E_SENTENCE.indexOf(ss[i]) != -1)
						s1 = Utility.SENTENCE_BEGIN;
					else
						s1 = ss[i];

				} else
					s1 += ss[i];
			}

			if (s1.length() > 0 && !Utility.SENTENCE_BEGIN.equals(s1)) {
				s1 += Utility.SENTENCE_END;
				result.add(new Sentence(s1, true));
			}
		}
		return result;
	}
	public ArrayList<Sentence> getSens() {
		return sens;
	}
	
	
}