Lucene3.6 之 排序篇

Lucene的默认排序是按照Document的得分进行排序的。当检索结果集中的两个Document的具有相同的得分时，默认按照Document的ID对结果进行排序。 

 

一、使用Sort、SortField类实现排序 

Lucene在查询的时候，可以通过以一个Sort作为参数构造一个检索器IndexSearcher，在构造Sort的时候，指定排序规则。 调用sIndexSearcher.search，例如： 
IndexSearcher.search(query, filter, n, sort);

关于Sort类，在其内部定义了3种构造方法： 

 

关于SortField类，其构造方法方法如下：

 

其中type对应的取值如下：

 

SortField. SCORE 按积分排序 
SortField. DOC 按文档排序 
SortField. AUTO 域的值为int、long、float都有效 
SortField.STRING 域按STRING排序 
SortField..FLOAT 
SortField.LONG 
SortField.DOUBLE 
SortField.SHORT 
SortField.CUSTOM 通过比较器排序 
SortField.BYTE 

示例代码

1、对单个字段进行排序

@Test   public void sortSingleField(){    try {     String path = "D://LuceneEx/day01";     String keyword = "android";     File file = new File(path);     Directory mdDirectory = FSDirectory.open(file);  //   Analyzer analyzer = new StandardAnalyzer(Version.LUCENE_36);     // 使用 商业分词器     Analyzer mAnalyzer = new IKAnalyzer();          IndexReader reader = IndexReader.open(mdDirectory);       IndexSearcher searcher = new IndexSearcher(reader);       String[] fields = {"title","category"};  // (在多个Filed中搜索)     QueryParser parser = new MultiFieldQueryParser(Version.LUCENE_36, fields, mAnalyzer);  //   String fieldName = "source";    //   QueryParser parser = new QueryParser(Version.LUCENE_36, fieldName, mAnalyzer);     Query query = parser.parse(keyword);       SortField field = new SortField("reputation", SortField.FLOAT);     Sort sort = new Sort(field );     TopDocs tops = searcher.search(query, 50, sort );          int count = tops.totalHits;          System.out.println("totalHits="+count);          ScoreDoc[] docs = tops.scoreDocs;          for(int i=0;i<docs.length;i++){      Document doc = searcher.doc(docs[i].doc);            int id = Integer.parseInt(doc.get("id"));      String title = doc.get("title");      String author = doc.get("author");      String publishTime = doc.get("publishTime");      String source = doc.get("source");      String category = doc.get("category");      float reputation = Float.parseFloat(doc.get("reputation"));            System.out.println(id+"\t"+title+"\t"+author+"\t"+publishTime+"\t"+source+"\t"+category+"\t"+reputation);     }          reader.close();     searcher.close();         } catch (CorruptIndexException e) {     e.printStackTrace();    } catch (IOException e) {     e.printStackTrace();    } catch (ParseException e) {     e.printStackTrace();    }   }

2、对多个字段进行排序

@Test   public void sortMultiField(){    try {     String path = "D://LuceneEx/day01";     String keyword = "Android";     File file = new File(path);     Directory mdDirectory = FSDirectory.open(file);  //   Analyzer analyzer = new StandardAnalyzer(Version.LUCENE_36);     // 使用 商业分词器     Analyzer mAnalyzer = new IKAnalyzer();          IndexReader reader = IndexReader.open(mdDirectory);       IndexSearcher searcher = new IndexSearcher(reader);       String[] fields = {"title","category"};  // (在多个Filed中搜索)     QueryParser parser = new MultiFieldQueryParser(Version.LUCENE_36, fields, mAnalyzer);  //   String fieldName = "source";    //   QueryParser parser = new QueryParser(Version.LUCENE_36, fieldName, mAnalyzer);     Query query = parser.parse(keyword);       SortField sortF1 =new SortField("reputation", SortField.FLOAT);     SortField sortF2 =new SortField("source", SortField.STRING);     Sort sort =new Sort(new SortField[]{sortF1 , sortF2});          TopDocs tops = searcher.search(query, null, 100, sort);     int count = tops.totalHits;          System.out.println("totalHits="+count);          ScoreDoc[] docs = tops.scoreDocs;          for(int i=0;i<docs.length;i++){      Document doc = searcher.doc(docs[i].doc);            int id = Integer.parseInt(doc.get("id"));      String title = doc.get("title");      String author = doc.get("author");      String publishTime = doc.get("publishTime");      String source = doc.get("source");      String category = doc.get("category");      float reputation = Float.parseFloat(doc.get("reputation"));            System.out.println(id+"\t"+title+"\t"+author+"\t"+publishTime+"\t"+source+"\t"+category+"\t"+reputation);     }          reader.close();     searcher.close();         } catch (CorruptIndexException e) {     e.printStackTrace();    } catch (IOException e) {     e.printStackTrace();    } catch (ParseException e) {     e.printStackTrace();    }   }

用到的两个工具方法代码

/**    * 创建文档对象的工具方法    * @param book    * @return    */   public Document createDocument(Book book){    Document doc = new Document();      Field id = new Field("id", book.getId() + "", Store.YES,      Index.ANALYZED);    Field title = new Field("title", book.getTitle(), Store.YES,      Index.ANALYZED);    Field author = new Field("author", book.getAuthor(), Store.YES,      Index.ANALYZED);    Field publishTime = new Field("publishTime", book.getPublishTime(),      Store.YES, Index.ANALYZED);    Field source = new Field("source", book.getSource(), Store.YES,      Index.ANALYZED);    Field category = new Field("category", book.getCategory(),      Store.YES, Index.ANALYZED);    Field reputation = new Field("reputation", book.getReputation()      + "", Store.YES, Index.ANALYZED);      doc.add(id);    doc.add(title);    doc.add(author);    doc.add(publishTime);    doc.add(source);    doc.add(category);    doc.add(reputation);        return doc;   }      /**    * 创建Book对象    * @param title    * @param author    * @param publishTime    * @param category    * @param reputation    * @return    */   public Book createBook(String title,String author,String publishTime,String category,float reputation){        Random r = new Random();    int id = r.nextInt(10000);        Book book = new Book();    book.setId(id);    book.setAuthor(author);    book.setTitle(title);    book.setCategory(category);    book.setPublishTime(publishTime);    book.setReputation(reputation);    book.setSource("清华大学出版社");        return book;   }

二、改变boost(激励因子) 

1、改变Document的boost(激励因子) 
改变boost的大小，会导致Document的得分的改变，从而按照Lucene默认的对检索结果集的排序方式，改变检索结果中Document的排序的提前或者靠后。在计算得分的时候，使用到了boost的值，默认boost的值为1.0，也就说默认情况下Document的得分与boost的无关的。一旦改变了默认的boost的值，也就从Document的得分与boost无关，变为相关了：boost值越大，Document的得分越高。 

2、改变Field的boost(激励因子) 
改变Field的boost值，和改变Document的boost值是一样的。因为Document的boost是通过添加到Docuemnt中Field体现的，所以改变Field的boost值，可以改变Document的boost值。 

示例代码

@Test   public void testBoost(){    try {     String path = "D://LuceneEx/day02";     String keyword = "android";     File file = new File(path);     Directory mdDirectory = FSDirectory.open(file);     // 使用 商业分词器     Analyzer mAnalyzer = new IKAnalyzer();       IndexReader reader = IndexReader.open(mdDirectory);       IndexSearcher searcher = new IndexSearcher(reader);       String[] fields = { "title", "category" }; // (在多个Filed中搜索)     QueryParser parser = new MultiFieldQueryParser(Version.LUCENE_36,       fields, mAnalyzer);     Query query = parser.parse(keyword);       TopDocs tops = searcher.search(query, null, 50);       int count = tops.totalHits;       System.out.println("totalHits=" + count);       ScoreDoc[] docs = tops.scoreDocs;       for (int i = 0; i < docs.length; i++) {            Document doc = searcher.doc(docs[i].doc);        float score = docs[i].score;            int id = Integer.parseInt(doc.get("id"));      String title = doc.get("title");      String author = doc.get("author");      String publishTime = doc.get("publishTime");      String source = doc.get("source");      String category = doc.get("category");      float reputation = Float.parseFloat(doc.get("reputation"));        System.out.println(id + "\t" + title + "\t" + author + "\t"        + publishTime + "\t" + source + "\t" + category + "\t"        + reputation+"\t"+score);     }       reader.close();     searcher.close();      } catch (CorruptIndexException e) {     e.printStackTrace();    } catch (IOException e) {     e.printStackTrace();    } catch (ParseException e) {     e.printStackTrace();    }   }      @Test   public void testAdd() {      try {     String path = "D://LuceneEx/day02";     File file = new File(path);     Directory mdDirectory = FSDirectory.open(file);       // 使用Lucene提供的分词器     // Analyzer mAnalyzer = new StandardAnalyzer(Version.LUCENE_36);     // 使用 商业分词器     Analyzer mAnalyzer = new IKAnalyzer();     IndexWriterConfig config = new IndexWriterConfig(Version.LUCENE_36,       mAnalyzer);       IndexWriter writer = new IndexWriter(mdDirectory, config);       Book book1 = createBook("Android内核揭秘", "ABC", "2010-07", "android 移动开发", 8.9f);     Document doc1 = createDocument(book1);     doc1.setBoost(2.0F); //boost：设置得分,2F在当前得分的基础上*2，使得分增高          Book book2 = createBook("Android多媒体开发", "BCD", "2011-07", "android 多媒体", 8.5f);     Document doc2 = createDocument(book2);     doc2.setBoost(2.5F); //boost：设置得分,2F在当前得分的基础上*2，使得分增高          Book book3 = createBook("Android企业应用开发", "QAB", "2012-05", "android 企业应用", 8.2f);     Document doc3 = createDocument(book3);     doc3.setBoost(1.5F); //boost：设置得分,2F在当前得分的基础上*2，使得分增高       writer.addDocument(doc1);     writer.addDocument(doc2);     writer.addDocument(doc3);       writer.close();      } catch (CorruptIndexException e) {     e.printStackTrace();    } catch (LockObtainFailedException e) {     e.printStackTrace();    } catch (IOException e) {     e.printStackTrace();    }   }

运行结果

totalHits=3
3383 Android多媒体开发BCD2011-07清华大学出版社android 多媒体8.51.259212
891 Android内核揭秘ABC2010-07清华大学出版社android 移动开发8.91.0073696
2919 Android企业应用开发QAB2012-05清华大学出版社android 企业应用8.20.75552726

三、自定义排序 
待完成。。。