我的网站集成ElasticSearch初体验

   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

using Core.SearchEngine.Client; using Microsoft.Extensions.DependencyInjection;  namespace Core.SearchEngine {     public static class ConfigureSearchEngine     {         public static void AddSearchEngine(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

using Core.SearchEngine.Attr; using System; using System.Collections.Generic; using System.Linq; using System.Text; using System.Threading.Tasks; using XianDan.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     public class Article_ES     {         public long Id { get; set; }         /// <summary>         /// 作者         /// </summary>         public string Author { get; set; }         /// <summary>         /// 标题                                                                                        /// </summary>         public string Title { get; set; }         /// <summary>         /// 标签         /// </summary>         public string Tag { get; set; }         /// <summary>         /// 简介                                                                                       /// </summary>         public string Description { get; set; }         /// <summary>         /// 内容         /// </summary>         public string ArticleContent { get; set; }         /// <summary>         /// 专栏         /// </summary>         public long ArticleCategoryId { get; set; }         /// <summary>         /// 是否原创         /// </summary>         public bool? IsOriginal { get; set; }         /// <summary>         /// 评论数         /// </summary>         public int? CommentCount { get; set; }         /// <summary>         /// 点赞数         /// </summary>         public int? PraiseCount { get; set; }         /// <summary>         /// 浏览次数         /// </summary>         public int? BrowserCount { get; set; }         /// <summary>         /// 收藏数量         /// </summary>         public int? 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

 private async 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);             await unitOfWork.GetRepository<TaskRecord>().InsertAsync(taskRecord);             await unitOfWork.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;                 await taskRecordRepository.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 =                 {                     new Field("title"),                     new Field("tag"),                     new Field("articleContent"),                     new Field("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(new Field("title"), new HighlightField()             {                 PreTags = new List<string> { "<em>" },                 PostTags = new List<string> { "</em>" },             });             highlightFields.Add(new Field("description"), new HighlightField()             {                 PreTags = new List<string> { "<em>" },                 PostTags = new List<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 = new string[] { keyword },                 Analyzer = analyzer             };             var analyzeResponse = await elasticsearchClient.Indices.AnalyzeAsync(analyzeIndexRequest);             if (analyzeResponse.Tokens == null)                 return new string[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

 public async Task<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 = new List<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 =                 {                     new Field("title"),                     new Field("tag"),                     new Field("articleContent"),                     new Field("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(new Field("title"), new HighlightField()             {                 PreTags = new List<string> { "<em>" },                 PostTags = new List<string> { "</em>" },             });             highlightFields.Add(new Field("description"), new HighlightField()             {                 PreTags = new List<string> { "<em>" },                 PostTags = new List<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 = await esArticleClient.GetClient().SearchAsync<Article_ES>(configureRequest);             foreach (var item in resp.Hits)             {                 if (item.Highlight == null)                     continue;                 foreach (var dict in 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();             return new Core.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。