超轻量级全文搜索框架的设计和实现

Lucene是Java领域最出色的全文搜索引擎，然而其API比较复杂，并且有严格的线程同步模型，直接使用不易。Compass则是封装了Lucene的一个OSEM：Object-SearchEngine Mapping，与Hibernate封装JDBC类似，然而过于复杂，支持的Lucene版本较低，在www.javaeedev.com发现雪峰开发一个类似Compass的简单封装Lucene的全文搜索框架，支持最新版本Lucene和Java 5泛型代码，用户通过简单的代码即可对自定义Bean进行搜索：

List<T> list = Searcher.search(Class<T>, String q, Page page);

下载地址：

http://code.google.com/p/lightweight-search/downloads/list

 

 

 

下面介绍下简单的用法：

 

1.自定义Bean，例如：Article

import com.javaeedev.lightweight.search.Index;
import com.javaeedev.lightweight.search.SearchableId;
import com.javaeedev.lightweight.search.SearchableProperty;
import com.javaeedev.lightweight.search.Store;

/**
 * Article bean that can be searched.
 *
 * @author Xuefeng
 */
public class Article {

    private String id;
    private String title;
    private String author;
    private String content;

    @SearchableId
    public String getId() {
        return id;
    }

    public void setId(String id) {
        this.id = id;
    }

    @SearchableProperty(store=Store.YES, index=Index.TOKENIZED, boost=3.0f)
    public String getTitle() {
        return title;
    }

    public void setTitle(String title) {
        this.title = title;
    }

    @SearchableProperty(store=Store.YES, index=Index.TOKENIZED, boost=2.0f)
    public String getAuthor() {
        return author;
    }

    public void setAuthor(String author) {
        this.author = author;
    }

    @SearchableProperty(store=Store.YES, index=Index.TOKENIZED)
    public String getContent() {
        return content;
    }

    public void setContent(String content) {
        this.content = content;
    }

其中特别的地方就是@SearchableId和@SearchableProperty

@SearchableId是描述lucene索引的ID

@SearchableProperty是描述lucene索引域的

以上二属性的详细信息请查询lucene，应为此框架的原理仅仅是对lucene的封装。
2.实现全文全文搜索

有两种方法：

        一 直接用行，所需参数在调用类中配置，例如：

import com.javaeedev.lightweight.common.Page;
import com.javaeedev.lightweight.search.Searcher;
import com.javaeedev.lightweight.search.SearcherProvider;

/**
 * Run directly.
 * 
 */
public class Run {

    @SuppressWarnings("unchecked")
    public static void main(String[] args) {
        // define classes that can be searched:
        Set<Class> searchableClasses = new HashSet<Class>();
        searchableClasses.add(Article.class);
        // init a provider:
        SearcherProvider provider = new SearcherProvider("./index-location", searchableClasses);
        provider.setAnalyzer(new StandardAnalyzer());
        provider.setHighlightPre("<b>");
        provider.setHighlightPost("</b>");
        // get a searcher:
        Searcher searcher = provider.get();
        // now do all search work with "searcher" object:
        // first add new Articles:
        Article a = new Article();
        a.setId(UUID.randomUUID().toString());
        a.setTitle("How to use lightweight search?");
        a.setAuthor("Liao Xuefeng");
        a.setContent("Lightweight search project is an open source project that using Lucene but simplify full text search.");
        searcher.index(a);
        // now search:
        Page page = new Page(1);
        List<Article> articles = searcher.search(Article.class, "lightweight", page);
        System.out.println("Results: " + page.getTotalCount());
        for(Article article : articles) {
            System.out.println(article.getTitle() + " by " + article.getAuthor());
        }
    }

}

（注意：请注意其代码中的注释部分）

       二 借助于Guice注入（相关Guice的资料）

    借助于Guice注入，首先要做的就是初始化Guice

import com.google.inject.Module;
import com.google.inject.name.Names;
import com.javaeedev.lightweight.search.Searcher;
import com.javaeedev.lightweight.search.SearcherProvider;

/**
 * Init Google Guice IoC Container.
 * 

 */
public class MyModule implements Module {

    @SuppressWarnings("unchecked")
    public void configure(Binder binder) {
        binder.bindConstant().annotatedWith(Names.named("HighlightPre")).to("<span class=\"highlight\">");
        binder.bindConstant().annotatedWith(Names.named("HighlightPost")).to("</span>");
        binder.bindConstant().annotatedWith(Names.named("IndexLocation")).to("./index-location");
        Set<Class> set = new HashSet<Class>();
        set.add(Article.class);
        binder.bind(new Key<Set<Class>>(Names.named("SearchableClasses")) {}).toInstance(set);
        binder.bind(Analyzer.class).to(StandardAnalyzer.class).asEagerSingleton();
        binder.bind(Searcher.class).toProvider(SearcherProvider.class).asEagerSingleton();
    }

}
其次创建调用

import java.util.List;
import java.util.UUID;

import com.google.inject.Guice;
import com.google.inject.Injector;
import com.javaeedev.lightweight.common.Page;
import com.javaeedev.lightweight.search.Searcher;

/**
 * Run demo with Google Guice as IoC container.
 *
 * @author Xuefeng
 */
public class RunWithGuice {

    public static void main(String[] args) {
        Injector injector = Guice.createInjector(new MyModule());
        Searcher searcher = injector.getInstance(Searcher.class);
        // now do all search work with "searcher" object:
        // first add new Articles:
        Article a = new Article();
        a.setId(UUID.randomUUID().toString());
        a.setTitle("How to use lightweight search?");
        a.setAuthor("Liao Xuefeng");
        a.setContent("Lightweight search project is an open source project that using Lucene but simplify full text search.");
        searcher.index(a);
        // now search:
        Page page = new Page(1);
        List<Article> articles = searcher.search(Article.class, "lightweight", page);
        System.out.println("Results: " + page.getTotalCount());
        for(Article article : articles) {
            System.out.println(article.getTitle() + " by " + article.getAuthor());
        }
    }

分享到：

一个完整的项目管理流程 | 通过dwr往session中设值

• 2008-08-31 14:17
• 浏览 2625
• 评论(1)
• 查看更多

评论

1 楼 ymgoodboy 2008-09-22  

死涛涛，哈哈