我的网站集成ElasticSearch初体验

编程知识1372024-09-22评论

最近，我给我的网站(https://www.xiandanplay.com/)尝试集成了一下es来实现我的一个搜索功能，因为这个是我第一次了解运用elastic，所以如果有不对的地方，大家可以指出来，话不多说，先看看我的一个大致流程

这里我采用的sdk的版本是Elastic.Clients.Elasticsearch, Version=8.0.0.0，官方的网址Installation | Elasticsearch .NET Client [8.0] | Elastic

我的es最开始打算和我的应用程序一起部署到ubuntu上面，结果最后安装kibana的时候，各种问题，虽好无奈，只好和我的SqlServer一起安装到windows上面，对于一个2G内容的服务器来说，属实有点遭罪了。

1、配置es

在es里面，我开启了密码认证。下面是我的配置

"Search": {
    "IsEnable": "true",
    "Uri": "http://127.0.0.1:9200/",
    "User": "123",
    "Password": "123"
  }
然后新增一个程序集

然后再ElasticsearchClient里面去写一个构造函数去配置es

using Core.Common;
using Core.CPlatform;
using Core.SearchEngine.Attr;
using Elastic.Clients.Elasticsearch;
using Elastic.Clients.Elasticsearch.IndexManagement;
using Elastic.Transport;

namespace Core.SearchEngine.Client
{
    public class ElasticSearchClient : IElasticSearchClient
    {
        private ElasticsearchClient elasticsearchClient;
        public ElasticSearchClient()
        {
            string uri = ConfigureProvider.configuration.GetSection("Search:Uri").Value;
            string username = ConfigureProvider.configuration.GetSection("Search:User").Value;
            string password = ConfigureProvider.configuration.GetSection("Search:Password").Value;
            var settings = new ElasticsearchClientSettings(new Uri(uri))
                          .Authentication(new BasicAuthentication(username, password)).DisableDirectStreaming();
            elasticsearchClient = new ElasticsearchClient(settings);
        }
        public ElasticsearchClient GetClient()
        {
            return elasticsearchClient;
        }
    }
}

　　然后，我们看skd的官网有这个这个提示

客户端应用程序应创建一个该实例，该实例在整个应用程序中用于整个应用程序辈子。在内部，客户端管理和维护与节点的 HTTP 连接，重复使用它们以优化性能。如果您使用依赖项注入容器中，客户端实例应注册到单例生存期

所以我直接给它来一个AddSingleton

usingCore.SearchEngine.Client;usingMicrosoft.Extensions.DependencyInjection;namespaceCore.SearchEngine{publicstaticclass ConfigureSearchEngine { publicstaticvoidAddSearchEngine(this IServiceCollection services) { services.AddSingleton<IElasticSearchClient, ElasticSearchClient>(); } }}

2、提交文章并且同步到es

然后就是同步文章到es了，我是先写入数据库，再同步到rabbitmq，通过事件总线（基于事件总线EventBus实现邮件推送功能）写入到es

先定义一个es模型

usingCore.SearchEngine.Attr;usingSystem;usingSystem.Collections.Generic;usingSystem.Linq;usingSystem.Text;usingSystem.Threading.Tasks;usingXianDan.Model.BizEnum;namespace XianDan.Domain.Article{ [ElasticsearchIndex(IndexName ="t_article")]//自定义的特性，sdk并不包含这个特性publicclass Article_ES { publiclong Id { get;set; } ///<summary>///作者///</summary>publicstring Author { get;set; } ///<summary>///标题///</summary>publicstring Title { get;set; } ///<summary>///标签///</summary>publicstring Tag { get;set; } ///<summary>///简介///</summary>publicstring Description { get;set; } ///<summary>///内容///</summary>publicstring ArticleContent { get;set; } ///<summary>///专栏///</summary>publiclong ArticleCategoryId { get;set; } ///<summary>///是否原创///</summary>publicbool? IsOriginal { get;set; } ///<summary>///评论数///</summary>publicint? CommentCount { get;set; } ///<summary>///点赞数///</summary>publicint? PraiseCount { get;set; } ///<summary>///浏览次数///</summary>publicint? BrowserCount { get;set; } ///<summary>///收藏数量///</summary>publicint? CollectCount { get;set; } ///<summary>///创建时间///</summary>public DateTime CreateTime { get;set; } }}

然后创建索引

string index = esArticleClient.GetIndexName(typeof(Article_ES));await esArticleClient.GetClient().Indices.CreateAsync<Article_ES>(index, s => s.Mappings( x => x.Properties( t => t.LongNumber(l => l.Id) .Text(l=>l.Title,z=>z.Analyzer(ik_max_word)) .Keyword(l=>l.Author) .Text(l=>l.Tag,z=>z.Analyzer(ik_max_word)) .Text(l=>l.Description,z=>z.Analyzer(ik_max_word)) .Text(l=>l.ArticleContent,z=>z.Analyzer(ik_max_word)) .LongNumber(l=>l.ArticleCategoryId) .Boolean(l=>l.IsOriginal) .IntegerNumber(l=>l.BrowserCount) .IntegerNumber(l=>l.PraiseCount) .IntegerNumber(l=>l.PraiseCount) .IntegerNumber(l=>l.CollectCount) .IntegerNumber(l=>l.CommentCount) .Date(l=>l.CreateTime) ) ) );

然后每次增删改文章的时候写入到mq，例如

privateasync Task SendToMq(Article article, Operation operation) { ArticleEventData articleEventData =new ArticleEventData(); articleEventData.Operation = operation; articleEventData.Article_ES = MapperUtil.Map<Article, Article_ES>(article); TaskRecord taskRecord =new TaskRecord(); taskRecord.Id = CreateEntityId(); taskRecord.TaskType = TaskRecordType.MQ; taskRecord.TaskName ="发送文章"; taskRecord.TaskStartTime = DateTime.Now; taskRecord.TaskStatu = (int)MqMessageStatu.New; articleEventData.Unique = taskRecord.Id.ToString(); taskRecord.TaskValue =JsonConvert.SerializeObject(articleEventData);awaitunitOfWork.GetRepository<TaskRecord>().InsertAsync(taskRecord);awaitunitOfWork.CommitAsync();try { eventBus.Publish(GetMqExchangeName(), ExchangeType.Direct, BizKey.ArticleQueueName, articleEventData); } catch (Exception ex) { var taskRecordRepository = unitOfWork.GetRepository<TaskRecord>(); TaskRecord update =await taskRecordRepository.SelectByIdAsync(taskRecord.Id); update.TaskStatu = (int)MqMessageStatu.Fail; update.LastUpdateTime = DateTime.Now; update.TaskResult ="发送失败"; update.AdditionalData =ex.Message;awaittaskRecordRepository.UpdateAsync(update);await unitOfWork.CommitAsync(); } }

mq订阅之后写入es，具体的增删改的方法就不写了吧

3、开始查询es

等待写入文章之后，开始查询文章，这里sdk提供的查询的方法比较复杂，全都是通过lmbda一个个链式去拼接的，但是我又没有找到更好的方法，所以就先这样吧

先创建一个集合存放查询的表达式

List<Action<QueryDescriptor<Article_ES>>> querys = new List<Action<QueryDescriptor<Article_ES>>>();

然后定义一个几个需要查询的字段

我这里使用MultiMatch来实现多个字段匹配同一个查询条件，并且指定使用ik_smart分词

Field[] fields ={newField("title"),newField("tag"),newField("articleContent"),newField("description") }; querys.Add(s => s.MultiMatch(y => y.Fields(Fields.FromFields(fields)).Analyzer(ik_smart).Query(keyword).Type(TextQueryType.MostFields)));

定义查询结果高亮，给查询出来的匹配到的分词的字段添加标签，同时前端需要对这个样式处理，

:deep(.search-words) em {

color: #ee0f29;

font-style: initial;

}

 Dictionary<Field, HighlightField> highlightFields = new Dictionary<Field, HighlightField>(); highlightFields.Add(newField("title"),new HighlightField() { PreTags =newList<string> { "<em>" }, PostTags =newList<string> { "</em>" }, }); highlightFields.Add(newField("description"),new HighlightField() { PreTags =newList<string> { "<em>" }, PostTags =newList<string> { "</em>" }, }); Highlight highlight =new Highlight() { Fields = highlightFields };

为了提高查询的效率，我只查部分的字段

 SourceFilter sourceFilter = new SourceFilter(); sourceFilter.Includes = Fields.FromFields(new Field[] { "title","id","author","description","createTime","browserCount","commentCount" }); SourceConfig sourceConfig =new SourceConfig(sourceFilter); Action<SearchRequestDescriptor<Article_ES>> configureRequest = s => s.Index(index) .From((homeArticleCondition.CurrentPage -1) * homeArticleCondition.PageSize) .Size(homeArticleCondition.PageSize) .Query(x => x.Bool(y => y.Must(querys.ToArray()))) .Source(sourceConfig) .Sort(y => y.Field(ht => ht.CreateTime, new FieldSort() { Order=SortOrder.Desc}))

获取查询的分词结果

var analyzeIndexRequest = new AnalyzeIndexRequest { Text =newstring[] { keyword }, Analyzer = analyzer }; var analyzeResponse = awaitelasticsearchClient.Indices.AnalyzeAsync(analyzeIndexRequest);if (analyzeResponse.Tokens == null)returnnewstring[0];return analyzeResponse.Tokens.Select(s => s.Token).ToArray();

到此，这个就是大致的查询结果，完整的如下

publicasyncTask<Core.SearchEngine.Response.SearchResponse<Article_ES>> SelectArticle(HomeArticleCondition homeArticleCondition) { string keyword =homeArticleCondition.Keyword.Trim();bool isNumber = Regex.IsMatch(keyword, RegexPattern.IsNumberPattern); List<Action<QueryDescriptor<Article_ES>>> querys = newList<Action<QueryDescriptor<Article_ES>>>();if (isNumber) { querys.Add(s => s.Bool(x => x.Should( should => should.Term(f => f.Field(z => z.Title).Value(keyword)) , should => should.Term(f => f.Field(z => z.Tag).Value(keyword)) , should => should.Term(f => f.Field(z => z.ArticleContent).Value(keyword)) ))); } else { Field[] fields ={newField("title"),newField("tag"),newField("articleContent"),newField("description") }; querys.Add(s => s.MultiMatch(y => y.Fields(Fields.FromFields(fields)).Analyzer(ik_smart).Query(keyword).Type(TextQueryType.MostFields))); } if (homeArticleCondition.ArticleCategoryId.HasValue) { querys.Add(s => s.Term(t => t.Field(f => f.ArticleCategoryId).Value(FieldValue.Long(homeArticleCondition.ArticleCategoryId.Value)))); } string index = esArticleClient.GetIndexName(typeof(Article_ES)); Dictionary<Field, HighlightField> highlightFields = new Dictionary<Field, HighlightField>(); highlightFields.Add(newField("title"),new HighlightField() { PreTags =newList<string> { "<em>" }, PostTags =newList<string> { "</em>" }, }); highlightFields.Add(newField("description"),new HighlightField() { PreTags =newList<string> { "<em>" }, PostTags =newList<string> { "</em>" }, }); Highlight highlight =new Highlight() { Fields = highlightFields }; SourceFilter sourceFilter =new SourceFilter(); sourceFilter.Includes = Fields.FromFields(new Field[] { "title","id","author","description","createTime","browserCount","commentCount" }); SourceConfig sourceConfig =new SourceConfig(sourceFilter); Action<SearchRequestDescriptor<Article_ES>> configureRequest = s => s.Index(index) .From((homeArticleCondition.CurrentPage -1) * homeArticleCondition.PageSize) .Size(homeArticleCondition.PageSize) .Query(x => x.Bool(y => y.Must(querys.ToArray()))) .Source(sourceConfig) .Sort(y => y.Field(ht => ht.CreateTime, new FieldSort() { Order=SortOrder.Desc})).Highlight(highlight);var resp = awaitesArticleClient.GetClient().SearchAsync<Article_ES>(configureRequest);foreach(varitemin resp.Hits) { if (item.Highlight == null)continue;foreach(vardictin item.Highlight) { switch (dict.Key) { case"title": item.Source.Title =string.Join("...", dict.Value); break;case"description": item.Source.Description =string.Join("...", dict.Value); break; } } } string[] analyzeWords = await esArticleClient.AnalyzeAsync(homeArticleCondition.Keyword); List<Article_ES> articles =resp.Documents.ToList();returnnewCore.SearchEngine.Response.SearchResponse<Article_ES>(articles, analyzeWords); }

4、演示效果

搞完之后，发布部署，看看效果，分词这里要想做的像百度那样，估计目前来看非常有难度的

那么这里我也向大家求教一下，如何使用SearchRequest封装多个查询条件，如下

SearchRequest searchRequest = new SearchRequest();
searchRequest.From = 0;
searchRequest.Size = 10;
searchRequest.Query=多个查询条件

因为我觉得这样代码读起来比lambda可读性高些，能更好的动态封装。

博

博客园

这个人很懒...

用户评论 (0)

发表评论

您的昵称:

评分

评论内容:

验证码:

我的网站集成ElasticSearch初体验

博客园

用户评论 (0)

发表评论

相关文章