package?com.chenlb.mmseg4j;

import?java.util.ArrayList;
import?java.util.HashMap;
import?java.util.Map;

/**
?*?所有詞都記錄在第一個字的結點下.
?*?
?*?@author?chenlb?2009-2-20?下午11:30:14
?*/
public?class?CharNode?{

	private?int?freq?=?-1;	//Degree?of?Morphemic?Freedom?of?One-Character?單字才需要
	private?int?maxLen?=?0;	//wordTail的最長

	private?KeyTree?ktWordTails?=?new?KeyTree（）;
	private?int?wordNum?=?0;
	
	public?CharNode（）?{
		
	}
	
	public?void?addWordTail（char[]?wordTail）?{
		ktWordTails.add（wordTail）;
		wordNum++;
		if（wordTail.length?>?maxLen）?{
			maxLen?=?wordTail.length;
		}
	}
	public?int?getFreq（）?{
		return?freq;
	}
	
	public?void?setFreq（int?freq）?{
		this.freq?=?freq;
	}
	
	public?int?wordNum（）?{
		return?wordNum;
	}
	
	/**
	?*?@param?sen?句子?一串文本.
	?*?@param?offset?詞在句子中的位置
	?*?@param?tailLen?詞尾的長度?實際是去掉詞的長度.
	?*?@author?chenlb?2009-4-8?下午11:10:30
	?*/
	public?int?indexOf（char[]?sen?int?offset?int?tailLen）?{
		//return?binarySearch（wordTails?sen?offset+1?tailLen?casc）;
		return?ktWordTails.match（sen?offset+1?tailLen）???1?:?-1;
	}
	
	/**
	?*?@param?sen?句子?一串文本.
	?*?@param?wordTailOffset?詞在句子中的位置?實際是?offset?后面的開始找.
	?*?@return?返回詞尾長?沒有就是?0
	?*?@author?chenlb?2009-4-10?下午10:45:51
	?*/
	public?int?maxMatch（char[]?sen?int?wordTailOffset）?{
		return?ktWordTails.maxMatch（sen?wordTailOffset）;
	}
	
	/**
	?*?
	?*?@return?至少返回一個包括?0的int
	?*?@author?chenlb?2009-4-12?上午10:01:35
	?*/
	public?ArrayList?maxMatch（ArrayList?tailLens?char[]?sen?int?wordTailOffset）?{
		return?ktWordTails.maxMatch（tailLens?sen?wordTailOffset）;
	}
	
	public?int?getMaxLen（）?{
		return?maxLen;
	}
	public?void?setMaxLen（int?maxLen）?{
		this.maxLen?=?maxLen;
	}
	
	public?static?class?KeyTree?{
		TreeNode?head?=?new?TreeNode（‘?‘）;
		
		public?void?add（char[]?w）?{
			if（w.length?				return;
			}
			TreeNode?p?=?head;
			for（int?i=0;?i				TreeNode?n?=?p.subNode（w[i]）;
				if（n?==?null）?{
					n?=?new?TreeNode（w[i]）;
					p.born（w[i]?n）;
				}
				p?=?n;
			}
			p.alsoLeaf?=?true;
		}
		
		/**
		?*?@return?返回匹配最長詞的長度?沒有找到返回?0.
		?*/
		public?int?maxMatch（char[]?sen?int?offset）?{
			int?idx?=?offset?-?1;
			TreeNode?node?=?head;
			for（int?i=offset;?i				node?=?node.subNode（sen[i]）;
				if（node?!=?null）?{
					if（node.isAlsoLeaf（））?{
						idx?=?i;?
					}
				}?else?{
					break;
				}
			}
			return?idx?-?offset?+?1;
		}
		
		public?ArrayList?maxMatch（ArrayList?tailLens?char[]?sen?int?offset）?{
			TreeNode?node?=?head;
			for（int?i=offset;?i				node?=?node.subNode（sen[i]）;
				if（node?!=?null）?{
					if（node.isAlsoLeaf（））?{
						tailLens.add（i-offset+1）;?
					}
				}?else?{
					break;
				}
			}
			return?tailLens;
		}
		
		public?boolean?match（char[]?sen?int?offset?int?len）?{
			TreeNode?node?=?head;
			for（int?i=0;?i		

?屬性????????????大小?????日期????時間???名稱
-----------?---------??----------?-----??----
?????目錄???????????0??2015-11-10?17:44??com\
?????目錄???????????0??2015-11-10?17:44??com\chenlb\
?????目錄???????????0??2015-11-10?17:44??com\chenlb\mmseg4j\
?????目錄???????????0??2015-11-10?17:44??com\chenlb\mmseg4j\analysis\
?????文件?????????566??2015-11-10?17:26??com\chenlb\mmseg4j\analysis\ComplexAnalyzer.java
?????文件????????4526??2015-11-10?17:26??com\chenlb\mmseg4j\analysis\CutLetterDigitFilter.java
?????文件?????????571??2015-11-10?17:26??com\chenlb\mmseg4j\analysis\MaxWordAnalyzer.java
?????文件????????1381??2015-11-10?17:26??com\chenlb\mmseg4j\analysis\MMSegAnalyzer.java
?????文件????????2617??2015-11-10?17:26??com\chenlb\mmseg4j\analysis\MMSegTokenizer.java
?????文件?????????572??2015-11-10?17:26??com\chenlb\mmseg4j\analysis\SimpleAnalyzer.java
?????文件????????1731??2015-11-10?17:26??com\chenlb\mmseg4j\analysis\TokenUtils.java
?????文件????????3802??2015-11-10?17:28??com\chenlb\mmseg4j\CharNode.java
?????文件????????2681??2015-11-10?17:28??com\chenlb\mmseg4j\Chunk.java
?????文件????????4097??2015-11-10?17:28??com\chenlb\mmseg4j\ComplexSeg.java
?????文件???????12847??2015-11-10?17:28??com\chenlb\mmseg4j\Dictionary.java
?????文件????????1439??2015-11-10?17:28??com\chenlb\mmseg4j\MaxWordSeg.java
?????文件???????10355??2015-11-10?17:28??com\chenlb\mmseg4j\MMSeg.java
?????目錄???????????0??2015-11-10?17:44??com\chenlb\mmseg4j\rule\
?????文件?????????696??2015-11-10?17:28??com\chenlb\mmseg4j\rule\LargestAvgLenRule.java
?????文件?????????813??2015-11-10?17:28??com\chenlb\mmseg4j\rule\LargestSumDegreeFreedomRule.java
?????文件?????????622??2015-11-10?17:28??com\chenlb\mmseg4j\rule\MaxMatchRule.java
?????文件????????1147??2015-11-10?17:28??com\chenlb\mmseg4j\rule\Rule.java
?????文件?????????761??2015-11-10?17:28??com\chenlb\mmseg4j\rule\SmallestVarianceRule.java
?????文件????????2345??2015-11-10?17:28??com\chenlb\mmseg4j\Seg.java
?????文件????????1173??2015-11-10?17:28??com\chenlb\mmseg4j\Sentence.java
?????文件?????????695??2015-11-10?17:28??com\chenlb\mmseg4j\SimpleSeg.java
?????文件????????2465??2015-11-10?17:28??com\chenlb\mmseg4j\Word.java
?????文件???????43234??2015-11-10?17:43??mmseg4j-1.9.2.jar