lucene3.0_IndexSearcher的基础使用及注意事项

较之lucene2.4版本，lucene3.0在indexSearcher这块变动比较大，从api可以直观的了解到。

基类Searcher与搜索有关的方法：

void	search(Query query, Collector results)            Lower-level search API.
void	search(Query query, Filter filter, Collector results)            Lower-level search API.
TopDocs	search(Query query, Filter filter, int n)            Finds the top n hits for query, applying filter if non-null.
TopFieldDocs	search(Query query, Filter filter, int n, Sort sort)            Search implementation with arbitrary sorting.
TopDocs	search(Query query, int n)            Finds the top n hits for query.
abstract  void	search(Weight weight, Filter filter, Collector results)            Lower-level search API.
abstract  TopDocs	search(Weight weight, Filter filter, int n)            Expert: Low-level search implementation.
abstract  TopFieldDocs	search(Weight weight, Filter filter, int n, Sort sort)            Expert: Low-level search implementation with arbitrary sorting.

 

该文将讲解上表格中黄色高亮的3个方法：

1.search(Query query , int n)

2.search(Query query , Collector results)

3.search(Query query , Filter filter , int n , Sort sort)

--------------------------------------------------------------------------------

1.search(Query query , int n)示例

取出前n条目标结果。

 

publicvoid searcher(String queryString){
        try {
            FSDirectory dir = SimpleFSDirectory.open(new File("d:/20101015index"));
            //注意点1:创建IndexSearcher实例是传入IndexReader实例还是Directory实例呢？
            IndexReader reader = IndexReader.open(dir);
            IndexSearcher searcher =new IndexSearcher(reader);
            //使用：search(Query query, int n) 
            QueryParser parser =new QueryParser(Version.LUCENE_30, "f1", new StandardAnalyzer(Version.LUCENE_30));
            Query query = parser.parse(queryString);
            TopDocs tds = searcher.search(query, 5);
            ScoreDoc[] sd = tds.scoreDocs;
            for (int i =0; i < sd.length; i++) {
                System.out.println(reader.document(sd[i].doc));
                //注意点2：怎么查看每个文档的打分的详情。
                //explain(Weight weight, int doc) 
                //Expert: low-level implementation method Returns an Explanation that describes how doc scored against weight.
                System.out.println("Explanation:"+ (searcher.explain(query, sd[i].doc)));
            }
        } catch (IOException e) {
            e.printStackTrace();
        } catch (ParseException e) {
            e.printStackTrace();
        } 
    }

 

注意点：

 

1.创建IndexSearcher实例是传入IndexReader实例还是Directory实例呢？

详情请参见:

lucene问题_IndexSearcher初始化，IndexSearcher(Directory dir)和IndexSearcher(IndexReader reader)有什么区别？到底使用那个更合理？

2.查看每个目标文档的打分的详细情况。

在网上看到有朋友问到这个，故在此提出来。

3.真实的项目请不要像上面代码那样——将indexReader和IndexSearcher的实例在搜索方法中创建，为了节约系统开销、提高效率，

应该将这些实例作为单例模式。

这里做了反面教材了 = =！

 

2.search(Query query , Collector results)

在介绍这个方法之前，先来了解下Collector：

 Collectors are primarily meant to be used to gather raw results from a search, and implement sorting or custom result filtering, collation, etc.

先重点了解：

TopScoreDocCollector is a concrete subclass TopDocsCollector and sorts according to score + docID. 

This is used internally by the IndexSearcher search methods that do not take an explicitSort. It is likely the most frequently used collector.

他是最常用的collector子类，是默认相关度排序的。下面给出一个实例，使用TopScoreDocCollector进行结果的收集，并提供简单的分页功能。

 

publicvoid searcher(String queryString ,int start, int howMany){
        try {
            FSDirectory dir = SimpleFSDirectory.open(new File("d:/20101015index"));
            //注意点1:创建IndexSearcher实例是传入IndexReader实例还是Directory实例呢？
            IndexReader reader = IndexReader.open(dir);
            IndexSearcher searcher =new IndexSearcher(reader);
            //使用：search(Query query , Collector results)
            QueryParser parser =new QueryParser(Version.LUCENE_30, "f1", new StandardAnalyzer(Version.LUCENE_30));
            Query query = parser.parse(queryString);
            int hm = start+howMany ;
            TopScoreDocCollector res = TopScoreDocCollector.create(hm, false);
            System.out.println("total hits :"+res.getTotalHits());
            searcher.search(query, res);
            //注意点2:这里可以控制分页。
            TopDocs tds = res.topDocs(start, howMany);
            ScoreDoc[] sd = tds.scoreDocs;
            for (int i =0; i < sd.length; i++) {
                System.out.println(reader.document(sd[i].doc));
//                System.out.println("Explanation:" + (searcher.explain(query, sd[i].doc)));
            }
        } catch (IOException e) {
            e.printStackTrace();
        } catch (ParseException e) {
            e.printStackTrace();
        } 
    }

 

 

3.search(Query query , Filter filter , int n , Sort sort)

这种方式的排序很简单，直接给出实例代码，注意的问题其他文章有详细说明。

 

FSDirectory dir = SimpleFSDirectory.open(new File("d:/20101015index"));
            //注意点1:创建IndexSearcher实例是传入IndexReader实例还是Directory实例呢？
            IndexReader reader = IndexReader.open(dir);
            IndexSearcher searcher =new IndexSearcher(reader);
            //使用：search(Query query , Filter filter , int n , Sort sort)
            QueryParser parser =new QueryParser(Version.LUCENE_30, "f1", new StandardAnalyzer(Version.LUCENE_30));
            Query query = parser.parse(queryString);
            //sort
            SortField sf = new SortField("f1", SortField.INT);
            Sort sort = new Sort(sf);
            TopDocs tds = searcher.search(query, null, 5 , sort);
            ScoreDoc[] sd = tds.scoreDocs;
            for (int i =0; i < sd.length; i++) {
                System.out.println(reader.document(sd[i].doc));
                //注意点2：怎么查看每个文档的打分的详情。
                //explain(Weight weight, int doc) 
                //Expert: low-level implementation method Returns an Explanation that describes how doc scored against weight.
                System.out.println("Explanation:"+ (searcher.explain(query, sd[i].doc)));
            }

 

 

注意点：

lucene问题_检索结果怎么排序？对于不同类型（例如int型）的字段排序有什么区别吗？

lucene问题_怎么对多个字段进行排序？

 

---------------------------------------------------------

对于lucene3.0检索的基础使用就先介绍这些了，关于搜索应该重点关注：

1.collector的选用

2.分页，及分页效率问题

3.排序，及排序的效率问题

4.多索引搜索

5.实际项目中实时搜索和效率的保障