TermVector用法:相关搜索功能及提高高亮显示性能

转自：http://hi.baidu.com/z57354658/blog/item/b80f524b2c92e1fa82025cbd.html

1. public class TermVectorTest {  
2.       
3.     Analyzer analyzer = new SimpleAnalyzer();  
4.     Directory ramDir = new RAMDirectory();  
5.       
6.     public void createRamIndex() throws CorruptIndexException, LockObtainFailedException, IOException{  
7.           
8.         IndexWriter writer = new IndexWriter(ramDir,analyzer,IndexWriter.MaxFieldLength.LIMITED);  
9.           
10.         Document doc1 = new Document();  
11.         doc1.add(new Field("title","java",Store.YES,Index.ANALYZED));  
12.         doc1.add(new Field("author","callan",Store.YES,Index.ANALYZED));  
13.         doc1.add(new Field("subject","java一门编程语言,用java的人很多，编程语言也不少，但是java最流行",Store.YES,Index.ANALYZED,TermVector.WITH_POSITIONS_OFFSETS));  
14.           
15.         Document doc2 = new Document();  
16.         doc2.add(new Field("title","english",Store.YES,Index.ANALYZED));  
17.         doc2.add(new Field("author","wcq",Store.YES,Index.ANALYZED));  
18.         doc2.add(new Field("subject","英语用的人很多",Store.YES,Index.ANALYZED,TermVector.WITH_POSITIONS_OFFSETS));  
19.       
20.         Document doc3 = new Document();  
21.         doc3.add(new Field("title","asp",Store.YES,Index.ANALYZED));  
22.         doc3.add(new Field("author","ca",Store.YES,Index.ANALYZED));  
23.         doc3.add(new Field("subject","英语用的人很多",Store.YES,Index.ANALYZED,TermVector.WITH_POSITIONS_OFFSETS));  
24.           
25.         writer.addDocument(doc1);  
26.         writer.addDocument(doc2);  
27.         writer.addDocument(doc3);  
28.           
29.         writer.optimize();  
30.         writer.close();  
31.     }  
32.       
33.     public void search() throws CorruptIndexException, IOException{  
34.         IndexReader reader = IndexReader.open(ramDir);  
35.         IndexSearcher searcher = new IndexSearcher(reader);  
36.         Term term = new Term("title","java");   //在title里查询java词条  
37.         TermQuery query = new TermQuery(term);  
38.         Hits hits = searcher.search(query);  
39.         for (int i = 0; i < hits.length(); i++)  
40.         {  
41.             Document doc = hits.doc(i);  
42.             System.out.println(doc.get("title"));  
43.             System.out.println(doc.get("subject"));  
44.             System.out.println("moreLike search: ");  
45.               
46.             morelikeSearch(reader,hits.id(i));  
47.         }  
48.     }  
49.   
50.     private void morelikeSearch(IndexReader reader,int id) throws IOException  
51.     {  
52.         //根据这个document的id获取这个field的Term Vector 信息，就是这个field分词之后在这个field里的频率、位置、等信息  
53.         TermFreqVector vector = reader.getTermFreqVector(id, "subject");  
54.           
55.         BooleanQuery query = new BooleanQuery();    
56.           
57.         for (int i = 0; i < vector.size(); i++)  
58.         {  
59.              TermQuery tq = new TermQuery(new Term("subject",     
60.                         vector.getTerms()[i]));   //获取每个term保存的Token  
61.                      
62.                  query.add(tq, BooleanClause.Occur.SHOULD);     
63.   
64.         }  
65.           
66.         IndexSearcher searcher = new IndexSearcher(ramDir);     
67.              
68.         Hits hits = searcher.search(query);     
69.           
70.         //显示代码，略  
71.   
72.           
73.     }  
74.   
75. //Lucene使用TermVector提高高亮显示性能  
76.     public void highterLightSearch() throws CorruptIndexException, IOException{  
77.         IndexReader reader = IndexReader.open(ramDir);     
78.           
79.         IndexSearcher searcher = new IndexSearcher(reader);     
80.              
81.         TermQuery query = new TermQuery(new Term("subject","java"));     
82.              
83.         Hits hits = searcher.search(query);     
84.              
85.         //高亮显示设置     
86.         SimpleHTMLFormatter simpleHTMLFormatter = new SimpleHTMLFormatter("<font color='red'>","</font>");  
87.              
88.         Highlighter highlighter =new Highlighter(simpleHTMLFormatter,new QueryScorer(query));     
89.           
90.          // 这个100是指定关键字字符串的context的长度，你可以自己设定，因为不可能返回整篇正文内容     
91.         highlighter.setTextFragmenter(new SimpleFragmenter(100));     
92.     
93.         for(int i = 0; i < hits.length(); i++){     
94.                  
95.             Document doc = hits.doc(i);     
96.                  
97.             TermPositionVector termFreqVector = (TermPositionVector)reader.getTermFreqVector(hits.id(i), "subject");     
98.                
99.             TermFreqVector vector = reader.getTermFreqVector(hits.id(i), "subject");  
100.             TokenStream tokenStream = TokenSources.getTokenStream(termFreqVector);     
101.                  
102.             String result = highlighter.getBestFragment(tokenStream, doc.get("subject"));     
103.     
104.             System.out.println(doc.get("title"));     
105.                  
106.             System.out.println(result);     
107.                  
108.         }     
109.   
110.           
111.     }  
112.       
113.     public static void main(String[] args) throws CorruptIndexException, IOException  
114.     {  
115.         TermVectorTest  t = new TermVectorTest();  
116.         t.createRamIndex();  
117.         t.search();  
118.     }  
119.   
120. }