Elasticsearch项目实战（elasticsearch 详解）

目录

• Elasticsearch+kibana
• 环境搭建
• windows 10环境配置
• 安装Elasticsearch
• head安装(非必需)
• 安装kibana
• DSL的基本使用
• 增加
• 修改
• 查询
• 删除
• Elasticsearch .Net
• Low level client基本使用
• 项目实战
• 总结
• 参考

Elasticsearch是一个基于Apache Lucene(TM)的开源搜索引擎。无论在开源还是专有领域，Lucene可以被认为是迄今为止最 先进、性能最好的、功能最全的搜索引擎库。

一说到全文搜索,lucene久负盛名。早年间,因为项目需要,接触过一个叫盘古分词的开源项目,借助其中的分词实现了分词搜索的功能。而盘古分词就是lucence的.NET版本。据说这个开源项目已经恢复更新并支持. NET Core,有兴趣的童鞋可以去围观一下(https://github.com/LonghronShen/Lucene.Net.Analysis.PanGu/tree/netcore2.0)。

我想很多童鞋都听过ELK,ELK是Elasticsearch、Logstash、Kibana。正好公司运维同事引入了这样一套体系,用于建立集中式日志收集系统，将所有节点上的日志统一收集，管理，访问。虽然能够从一定程度上解决基本的问题,但是原生的kibana界面和查询方式都不够友好,很难推向广大的开发人员。于是我在想，我们是否可以利用这个开源的库集成到运维自动化平台当中,让这把利剑发挥出更大的价值。

一、环境搭建

本文是基于windows 10操作系统的es环境的搭建。

1. java环境安装

由于es是java语言开发的,所以这里要安装java环境。

jdk下载:

https://www.oracle.com/technetwork/java/javase/downloads/jdk8-downloads-2133151.html

安装完成之后就是配置环境变量:

查看是否安装成功:

2.安装Elasticsearch

Elasticsearch版本已经比较多,初学者可能比较懵。特别是在安装head和Kibana的时候,如果版本不匹配,往往会导致无法使用。这里使用的是elasticsearch-5.6.11版本。

elasticsearch-5.6.11下载:

https://www.elastic.co/downloads/past-releases/elasticsearch-5-6-11

解压到C:\ELk 备用。

3.head安装(非必需)

es 4.x 版本安装head很简单,只需下载head插件解压到指定目录即可。es 5.x+需要借助node安装。

head下载:

https://github.com/mobz/elasticsearch-head

解压到C:\ELk\elasticsearch-5.6.11

node下载:

https://nodejs.org/dist/v8.12.0/node-v8.12.0-win-x64.zip

安装node

检查node和npm是否安装成功

path环境变量末尾 会自动增加 C:\Program Files\nodejs\

安装 phantomjs 官网:http://phantomjs.org/下载【配置环境变量】

安装grunt npm install -g grunt-cli

执行C:\ELk\elasticsearch-5.6.11\bin\elasticsearch.bat

执行命令启动 head

浏览器访问:http://localhost:9100/

4.安装kibana

导致为止,其实elasticsearch自身已经安装完成。通过Head就能很方便的操作es,但是kibana集成了head类似功能,并提供了更加友好的访问界面。

kibana-5.6.9-windows-x86下载:

https://www.elastic.co/downloads/past-releases/kibana-5-6-9

下载之后,解压到C:\ELk\kibana-5.6.9-windows-x86

执行C:\ELk\kibana-5.6.9-windows-x86\bin\kibana.bat

浏览器访问:http://localhost:5601

二、DSL的基本使用

elasticsearch也像mysql一样提供了专门的语法来操作数据。Elasticsearch provides a full Query DSL (Domain Specific Language) based on JSON to define queries.

• 创建文档

PUT people/person/1?op_type=create
{
 "user" : "kimchy",
 "post_date" : "2009-11-15T14:12:12",
 "message" : "trying out Elasticsearch"
}

• 修改

POST /user/guest/20/_update
{
 "doc": {
 "RealName":"LukyHuu20"
 }
}

• 查询

GET /user/guest/_search
{
 "query": {
 "match": {
 "Id":22
 }
 }
}

• 删除

DELETE /user/guest/15
{
 
}

三、Elasticsearch .Net

elasticsearch是以restfulAPI方式对外提供接口,并提供客户端给多种语言使用。Elasticsearch uses standard RESTful APIs and JSON. We also build and maintain clients in many languages such as Java, Python, .NET, SQL, and PHP. Plus, our community has contributed many more. They’re easy to work with, feel natural to use, and, just like Elasticsearch, don't limit what you might want to do with them. 参考(https://www.elastic.co/products/elasticsearch)

1.Low level client基本使用

本文是介绍ES的.NET客户端,Elasticsearch .Net - Low level client[5.x]

通过引入对应的版本的客户端,便可通过C#操作ES。参考(https://www.elastic.co/guide/en/elasticsearch/client/net-api/5.x/elasticsearch-net.html)

连接

var settings = new ConnectionConfiguration(new Uri("http://example.com:9200"))
 .RequestTimeout(TimeSpan.FromMinutes(2));
var lowlevelClient = new ElasticLowLevelClient(settings);

插入文档

 var indexResponse = lowlevelClient.Index<byte[]>("user", "guest", user.Id.ToString(), user);
 byte[] responseBytes = indexResponse.Body;

更新文档

var searchResponse = lowlevelClient.Update<string>("user", "guest", id.ToString(), new
 {
 doc = new
 {
 RealName = realname,
 Description = description
 }
 });
bool successful = searchResponse.Success;

查询

var searchResponse = lowlevelClient.Search<string>("user", "guest", new
 {
 query = new
 {
 match = new
 {
 Id = id
 }
 }
 });
bool successful = searchResponse.Success;

删除

var searchResponse = lowlevelClient.Delete<string>("user", "guest", id.ToString());
 bool successful = searchResponse.Success;

2.项目实战

前面大致介绍了ES的安装和基本使用。那么,如何在项目中落地呢？

使用nuget安装Elasticsearch.Net 5.6.4

Install-Package Elasticsearch.Net -Version 5.6.4

安装完后,

基本的增删该查在项目中的实现上面已经有所介绍,这里重点讲一下查询:

笔者使用的.NET MVC5 Web框架,对于返回的结果笔者做了一个简单封装：

public class ESearchRoot<T>
 {
 /// <summary>
 /// 
 /// </summary>
 public int took { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string timed_out { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public _shards _shards { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public Hits<T> hits { get; set; }
 }
 public class _shards
 {
 /// <summary>
 /// 
 /// </summary>
 public int total { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public int successful { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public int skipped { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public int failed { get; set; }
 }
 public class HitsItem<T>
 {
 /// <summary>
 /// 
 /// </summary>
 public string _index { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string _type { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string _id { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string _score { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public T _source { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public List<int> sort { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public Highlight highlight { get; set; }
 }
 public class Hits<T>
 {
 /// <summary>
 /// 
 /// </summary>
 public int total { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string max_score { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public List<HitsItem<T>> hits { get; set; }
 }
 public class Highlight
 {
 /// <summary>
 /// 
 /// </summary>
 public List<string> Description { get; set; }
 }

因为soure返回的对象是不定的,所以使用了泛型。 本项目soure对应的类,user:

 ///<summary>
 /// 
 /// </summary>
 public class User
 {
 /// <summary>
 /// 
 /// </summary>
 public string Account { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string Phone { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string Email { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string RealName { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string CanReview { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string CanExcute { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string Avatar { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string IsUse { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public int Id { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string Name { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public string Description { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public DateTime CreateTime { get; set; }
 /// <summary>
 /// 
 /// </summary>
 public DateTime ModifyTime { get; set; }
 }
 

项目使用了带条件的分页查询:

public List<AdminUser> GetBySomeWhere(string keyword, int limit, int pageSize, out int total)
 {
 List<AdminUser> users = new List<AdminUser>();
 total = 0;
 try
 {
 var settings = new ConnectionConfiguration(new Uri("http://localhost:9200/"))
 .RequestTimeout(TimeSpan.FromMinutes(2));
 var lowlevelClient = new ElasticLowLevelClient(settings);
 //根据不同的参数 来构建不同的查询条件
 var request = new object();
 if (!String.IsNullOrEmpty(keyword))
 {
 request = new
 {
 from = limit,
 size = pageSize,
 query = new
 {
 match = new
 {
 Description = keyword
 }
 },
 highlight = new
 {
 fields = new
 {
 Description = new { }
 }
 },
 sort = new
 {
 Id = new
 {
 order = "desc"
 }
 }
 };
 }
 else
 {
 request = new
 {
 from = limit,
 size = pageSize,
 query = new
 {
 match_all = new
 {
 }
 },
 highlight = new
 {
 fields = new
 {
 Description = new { }
 }
 },
 sort = new
 {
 Id = new
 {
 order = "desc"
 }
 }
 };
 }
 var searchResponse = lowlevelClient.Search<string>("user", "guest", request);
 bool successful = searchResponse.Success;
 var responseJson = searchResponse.Body;
 if (!successful)
 {
 return users;
 }
 ESearchRoot<User> root = JsonHelper.JSONStringObject<ESearchRoot<User>>(responseJson);
 if (root != null)
 {
 total = root.hits.total;
 foreach (HitsItem<User> item in root.hits.hits)
 {
 if (item._source != null)
 {
 string highlightDescription = String.Empty;
 StringBuilder sbDs = new StringBuilder();
 if (item.highlight != null && item.highlight.Description.Count > 0)
 {
 //ighlightDescription = item.highlight.Description[0];
 foreach (var d in item.highlight.Description)
 {
 sbDs.Append(d);
 }
 highlightDescription = sbDs.ToString();
 }
 AdminUser user = new AdminUser
 {
 Id = item._source.Id,
 RealName = item._source.RealName,
 Account = item._source.Account,
 Email = item._source.Email,
 Phone = item._source.Phone,
 //IsUse=item._source.IsUse,
 Avatar = item._source.Avatar,
 Description = item._source.Description,
 HighlightDescription = highlightDescription,
 CreateTime = item._source.CreateTime,
 ModifyTime = item._source.ModifyTime
 };
 users.Add(user);
 }
 }
 }
 return users;
 }
 catch (ElasticsearchClientException ex)
 {
 //Log4Helper.Error
 }
 return users;
 }

项目最终的效果如下:

四、总结

elasticsearch是很强大的开源工具,在实现全文搜索上有其独到之处,也是大数据的分析方面利器,值得大家深入去研究和实践。

五、参考

• Elasticsearch权威指南
• Elasticsearch.Net 5.x
• Elasticsearch Reference [5.6] ? Document APIs
• elasticsearch-net-example
• ElasticSearch安装及HEAD插件配置

个人微信公众号:码农商业参谋