Lucene 中的自定义排序的实现

最新推荐文章于 2025-05-05 09:18:07 发布

原创最新推荐文章于 2025-05-05 09:18:07 发布 · 2.4k 阅读

0 ·

CC 4.0 BY-SA版权

文章标签：

#lucene #import #sorting #parameters #exception #测试

JAVA 同时被 2 个专栏收录

258 篇文章

订阅专栏

杂谈

114 篇文章

订阅专栏

本文介绍如何在Lucene中实现自定义排序功能，包括实现SortComparatorSource和ScoreDocComparator接口的具体步骤，并提供了一个搜索附近餐馆的例子。

Lucene 中的自定义排序的实现使用Lucene来搜索内容,搜索结果的显示顺序当然是比较重要的.Lucene中Build-in的几个排序定义在大多数情况下是不适合我们使用的.要适合自己的应用程序的场景,就只能自定义排序功能,本节我们就来看看在Lucene中如何实现自定义排序功能.

Lucene中的自定义排序功能和Java集合中的自定义排序的实现方法差不多,都要实现一下比较接口. 在Java中只要实现Comparable接口就可以了.但是在Lucene中要实现SortComparatorSource接口和ScoreDocComparator接口.在了解具体实现方法之前先来看看这两个接口的定义吧.

SortComparatorSource接口的功能是返回一个用来排序ScoreDocs的comparator(Expert: returns a comparator for sorting ScoreDocs).该接口只定义了一个方法.如下:

public ScoreDocComparator newComparator(IndexReader reader,String fieldname) throws IOException

Creates a comparator for the field in the given index.

Parameters:

reader – Index to create comparator for.

fieldname – Field to create comparator for.

Returns:

Comparator of ScoreDoc objects.

Throws:



IOException

- If an error occurs reading the index.

该方法只是创造一个ScoreDocComparator 实例用来实现排序.所以我们还要实现ScoreDocComparator 接口.来看看ScoreDocComparator 接口.功能是比较来两个ScoreDoc 对象来排序(Compares two ScoreDoc objects for sorting) 里面定义了两个Lucene实现的静态实例.如下:

public static final ScoreDocComparator RELEVANCE

Special comparator for sorting hits according to computed relevance
(document score).

public static final ScoreDocComparator INDEXORDER

Special comparator for sorting hits according to index order (document
number).

有3个方法与排序相关,需要我们实现分别如下:

public int compare(ScoreDoc i,ScoreDoc j)

Compares two ScoreDoc objects and returns a result indicating their sort
order.

Parameters:

i – First ScoreDoc

j – Second ScoreDoc

Returns:

-1 if i should come before j

1 if i should come after j

0 if they are equal

public Comparable sortValue(ScoreDoc i)

Returns the value used to sort the given document. The object returned
must implement the java.io.Serializable interface. This is used by
multisearchers to determine how to collate results from their searchers.

Parameters:

i – Document

Returns:

Serializable object

public int sortType()

Returns the type of sort. Should return SortField.SCORE,
SortField.DOC, SortField.STRING,


SortField.INTEGER

, SortField.FLOAT or


SortField.CUSTOM

. It is not valid to return SortField.AUTO.
This is used by multisearchers to determine how to collate results from
their searchers.

Returns:

One of the constants in SortField.

看个例子吧!

该例子为Lucene in Action中的一个实现,用来搜索距你最近的餐馆的名字. 餐馆坐标用字符串"x,y"来存储.如下图:

Figure 6.1 Which Mexican restaurant is closest to home (at 0,0)
or work (at 10,10)?

此中情况下 Lucene中Build-in Sorting 实现就不可行了,看看如何自己实现吧.

01 package lia.extsearch.sorting; 02 03 import org.apache.lucene.search.SortComparatorSource; 04 import org.apache.lucene.search.ScoreDoc; 05 import org.apache.lucene.search.SortField; 06 import org.apache.lucene.search.ScoreDocComparator; 07 import org.apache.lucene.index.IndexReader; 08 import org.apache.lucene.index.TermEnum; 09 import org.apache.lucene.index.Term; 10 import org.apache.lucene.index.TermDocs; 11 12 import java.io.IOException; 13 // DistanceComparatorSource 实现了SortComparatorSource接口 14 public class DistanceComparatorSource implements SortComparatorSource { 15 // x y 用来保存坐标位置 16 private int x; 17 private int y; 18 19 public DistanceComparatorSource(int x, int y) { 20 this.x = x; 21 this.y = y; 22 } 23 // 返回ScoreDocComparator 用来实现排序功能 24 public ScoreDocComparator newComparator(IndexReader reader, String fieldname) 25 throws IOException { 26 return new DistanceScoreDocLookupComparator(reader, fieldname, x, y); 27 } 28 29 //DistanceScoreDocLookupComparator 实现了ScoreDocComparator 用来排序 30 private static class DistanceScoreDocLookupComparator implements 31 ScoreDocComparator { 32 private float[] distances; // 保存每个餐馆到指定点的距离 33 34 // 构造函数 , 构造函数在这里几乎完成所有的准备工作. 35 public DistanceScoreDocLookupComparator(IndexReader reader, 36 String fieldname, int x, int y) throws IOException { 37 38 final TermEnum enumerator = reader.terms(new Term(fieldname, "")); 39 distances = new float[reader.maxDoc()]; // 初始化distances 40 if (distances.length > 0) { 41 TermDocs termDocs = reader.termDocs(); 42 try { 43 if (enumerator.term() == null) { 44 throw new RuntimeException("no terms in field " 45 + fieldname); 46 } 47 int i = 0,j = 0; 48 do { 49 System.out.println("in do-while :" i +); 50 51 Term term = enumerator.term(); // 取出每一个Term 52 if (term.field() != fieldname) // 与给定的域不符合则比较下一个 53 break; 54 //Sets this to the data for the current term in a TermEnum. 55 //This may be optimized in some implementations. 56 termDocs.seek(enumerator); //参考TermDocs Doc 57 while (termDocs.next()) { 58 System.out.println(" in while :" j +); 59 System.out.println(" in while ,Term :" + term.toString()); 60 61 String[] xy = term.text().split(","); // 去处x y 62 int deltax = Integer.parseInt(xy[0]) - x; 63 int deltay = Integer.parseInt(xy[1]) - y; 64 // 计算距离 65 distances[termDocs.doc()] = (float) Math 66 .sqrt(deltax * deltax + deltay * deltay); 67 } 68 } while (enumerator.next()); 69 } finally { 70 termDocs.close(); 71 } 72 } 73 } 74 75 //有上面的构造函数的准备这里就比较简单了 76 public int compare(ScoreDoc i, ScoreDoc j) { 77 if (distances[i.doc] < distances[j.doc]) 78 return -1; 79 if (distances[i.doc] > distances[j.doc]) 80 return 1; 81 return 0; 82 } 83 84 // 返回距离 85 public Comparable sortValue(ScoreDoc i) { 86 return new Float(distances[i.doc]); 87 } 88 89 //指定SortType 90 public int sortType() { 91 return SortField.FLOAT; 92 } 93 } 94 95 public String toString() { 96 return "Distance from (" x "," y ")"; 97 } 98 99 }

这是一个实现了上面两个接口的两个类, 里面带有详细注释, 可以看出自定义排序并不是很难的.
该实现能否正确实现,我们来看看测试代码能否通过吧.

001 package lia.extsearch.sorting; 002 003 import junit.framework.TestCase; 004 import org.apache.lucene.analysis.WhitespaceAnalyzer; 005 import org.apache.lucene.document.Document; 006 import org.apache.lucene.document.Field; 007 import org.apache.lucene.index.IndexWriter; 008 import org.apache.lucene.index.Term; 009 import org.apache.lucene.search.FieldDoc; 010 import org.apache.lucene.search.Hits; 011 import org.apache.lucene.search.IndexSearcher; 012 import org.apache.lucene.search.Query; 013 import org.apache.lucene.search.ScoreDoc; 014 import org.apache.lucene.search.Sort; 015 import org.apache.lucene.search.SortField; 016 import org.apache.lucene.search.TermQuery; 017 import org.apache.lucene.search.TopFieldDocs; 018 import org.apache.lucene.store.RAMDirectory; 019 020 import java.io.IOException; 021 022 import lia.extsearch.sorting.DistanceComparatorSource; 023 // 测试自定义排序的实现 024 public class DistanceSortingTest extends TestCase { 025 private RAMDirectory directory; 026 027 private IndexSearcher searcher; 028 029 private Query query; 030 031 //建立测试环境 032 protected void setUp() throws Exception { 033 directory = new RAMDirectory(); 034 IndexWriter writer = new IndexWriter(directory, 035 new WhitespaceAnalyzer(), true); 036 addPoint(writer, "El Charro", "restaurant", 1, 2); 037 addPoint(writer, "Cafe Poca Cosa", "restaurant", 5, 9); 038 addPoint(writer, "Los Betos", "restaurant", 9, 6); 039 addPoint(writer, "Nico's Taco Shop", "restaurant", 3, 8); 040 041 writer.close(); 042 043 searcher = new IndexSearcher(directory); 044 045 query = new TermQuery(new Term("type", "restaurant")); 046 } 047 048 private void addPoint(IndexWriter writer, String name, String type, int x, 049 int y) throws IOException { 050 Document doc = new Document(); 051 doc.add(Field.Keyword("name", name)); 052 doc.add(Field.Keyword("type", type)); 053 doc.add(Field.Keyword("location", x + "," + y)); 054 writer.addDocument(doc); 055 } 056 057 public void testNearestRestaurantToHome() throws Exception { 058 //使用DistanceComparatorSource来构造一个SortField 059 Sort sort = new Sort(new SortField("location", 060 new DistanceComparatorSource(0, 0))); 061 062 Hits hits = searcher.search(query, sort); // 搜索 063 064 //测试 065 assertEquals("closest", "El Charro", hits.doc(0).get("name")); 066 assertEquals("furthest", "Los Betos", hits.doc(3).get("name")); 067 } 068 069 public void testNeareastRestaurantToWork() throws Exception { 070 Sort sort = new Sort(new SortField("location", 071 new DistanceComparatorSource(10, 10))); // 工作的坐标 10,10 072 073 //上面的测试实现了自定义排序,但是并不能访问自定义排序的更详细信息,利用 074 //TopFieldDocs 可以进一步访问相关信息 075 TopFieldDocs docs = searcher.search(query, null, 3, sort); 076 077 assertEquals(4, docs.totalHits); 078 assertEquals(3, docs.scoreDocs.length); 079 080 //取得FieldDoc 利用FieldDoc可以取得关于排序的更详细信息请查看FieldDoc Doc 081 FieldDoc fieldDoc = (FieldDoc) docs.scoreDocs[0]; 082 083 assertEquals("(10,10) -> (9,6) = sqrt(17)", new Float(Math.sqrt(17)), 084 fieldDoc.fields[0]); 085 086 Document document = searcher.doc(fieldDoc.doc); 087 assertEquals("Los Betos", document.get("name")); 088 089 dumpDocs(sort, docs); // 显示相关信息 090 } 091 092 // 显示有关排序的信息 093 private void dumpDocs(Sort sort, TopFieldDocs docs) throws IOException { 094 System.out.println("Sorted by: " + sort); 095 ScoreDoc[] scoreDocs = docs.scoreDocs; 096 for (int i = 0; i < scoreDocs.length; i++) { 097 FieldDoc fieldDoc = (FieldDoc) scoreDocs[i]; 098 Float distance = (Float) fieldDoc.fields[0]; 099 Document doc = searcher.doc(fieldDoc.doc); 100 System.out.println(" " doc.get( "name") " @ (" 101 doc.get( "location") ") -> " + distance); 102 } 103 } 104 }

完全通过测试,

输入信息如下:想进一步了解详细信息的可以研究一下:

in do-while :0

    in while :0

    in while ,Term :location:1,2

in do-while :1

    in while :1

    in while ,Term :location:3,8

in do-while :2

    in while :2

    in while ,Term :location:5,9

in do-while :3

    in while :3

    in while ,Term :location:9,6

in do-while :4

in do-while :0

    in while :0

    in while ,Term :location:1,2

in do-while :1

    in while :1

    in while ,Term :location:3,8

in do-while :2

    in while :2

    in while ,Term :location:5,9

in do-while :3

    in while :3

    in while ,Term :location:9,6

in do-while :4

Sorted by: