ElasticSearch系列学习
ElasticSearch第五步-.net平台下c#操作ElasticSearch详解
前面我们讲解了关于ElasticSearch的安装配置，以及CRUD
本章我将讲解怎么使用c#操作ElasticSearch。
首先你需要一定的技术储备，比如：asp.net webapi，mvc,jsonp,knockout。这些知识在这里不再讲解，请自行Google。
项目DEMO介绍
搜索和索引功能我是以服务(webapi项目)方式提供的，在客户端(mvc项目)中的view视图中，直接使用ajax(jsonp格式)方式调用webapi，然后使用knockout绑定到table上的。
项目结构如图：
引入驱动
工欲善其事必先利其器，首先我们需要在Supernova.Webapi层中引入操作ElasticSearch的驱动dll PlainElastic.Net。
如图：
封装操作ElasticSearch的ElasticSearchHelper
View Codedemo中涉及的实体对象模型
/// <summary>
/// ik分词结果对象
/// </summary>
public class ik
{
public List<tokens> tokens { get; set; }
}
public class tokens
{
public string token { get; set; }
public int start_offset { get; set; }
public int end_offset { get; set; }
public string type { get; set; }
public int position { get; set; }
}
/// <summary>
/// 测试数据对象
/// </summary>
public class personList
{
public personList()
{
this.list = new List<person>();
}
public int hits { get; set; }
public int took { get; set; }
public List<person> list { get; set; }
}
public class person
{
public string id { get; set; }
public string name { get; set; }
public int age { get; set; }
public bool sex { get; set; }
public DateTime birthday { get; set; }
public string intro { get; set; }
}
详细介绍ElasticSearchHelper里面的方法
1.索引文档（注意：索引文档之前先用配置filed对应的ik分词）：
public IndexResult Index(string indexName, string indexType, string id, string jsonDocument)
{
var serializer = new JsonNetSerializer();
string cmd = new IndexCommand(indexName, indexType, id);
OperationResult result = Client.Put(cmd, jsonDocument);
var indexResult = serializer.ToIndexResult(result.Result);
return indexResult;
}
public IndexResult Index(string indexName, string indexType, string id, object document)
{
var serializer = new JsonNetSerializer();
var jsonDocument = serializer.Serialize(document);
return Index(indexName, indexType, id, jsonDocument);
}
2.对单个字段的全文检索，字段intro 包含词组key中的任意一个单词。例如：词组（中国美好），只要每条数据的intro字段包含"中国"或者"美好"就返回。
public personList Search<person>(string indexName, string indexType, string key,int from ,int size)
{
string cmd = new SearchCommand(indexName, indexType);
string query = new QueryBuilder<person>()
//1 查询
.Query(b =>
b.Bool(m =>
//并且关系
m.Must(t =>
//分词的最小单位或关系查询
t.QueryString(t1=>t1.DefaultField("intro").Query(key))
//.QueryString(t1 => t1.DefaultField("name").Query(key))
// t .Terms(t2=>t2.Field("intro").Values("研究","方鸿渐"))
//范围查询
// .Range(r => r.Field("age").From("100").To("200") )
)
)
)
//分页
.From(from)
.Size(size)
//排序
// .Sort(c => c.Field("age", SortDirection.desc))
//添加高亮
.Highlight(h => h
.PreTags("<b>")
.PostTags("</b>")
.Fields(
f => f.FieldName("intro").Order(HighlightOrder.score),
f => f.FieldName("_all")
)
)
.Build();
string result = Client.Post(cmd, query);
var serializer = new JsonNetSerializer();
var list = serializer.ToSearchResult<Supernova.Webapi.DbHelper.person>(result);
personList datalist = new personList();
datalist.hits = list.hits.total;
datalist.took = list.took;
var personList= list.hits.hits.Select(c => new Supernova.Webapi.DbHelper.person() {
id=c._source.id,
age=c._source.age,
birthday =c._source.birthday,
intro=string.Join("",c.highlight["intro"]), //高亮显示的内容，一条记录中出现了几次
name=c._source.name,
sex=c._source.sex,
});
datalist.list.AddRange(personList);
return datalist;
}
3.字段intro 或者name 包含词组key中的所有单词。例如：词组（中国美好），只要每条数据的intro或者name字段包含"中国"并且包含"美好"就返回。
public personList SearchFullFileds<person>(string indexName, string indexType, string key, int from, int size)
{
MustQuery<person> mustNameQueryKeys = new MustQuery<person>();
MustQuery<person> mustIntroQueryKeys = new MustQuery<person>();
var arrKeys = GetIKTokenFromStr(key);
foreach (var item in arrKeys)
{
mustNameQueryKeys = mustNameQueryKeys.Term(t3 => t3.Field("name").Value(item)) as MustQuery<person>;
mustIntroQueryKeys = mustIntroQueryKeys.Term(t3 => t3.Field("intro").Value(item)) as MustQuery<person>;
}
string cmd = new SearchCommand(indexName, indexType);
string query = new QueryBuilder<person>()
//1 查询
.Query(b =>
b.Bool(m =>
m.Should(t =>
t.Bool(m1 =>
m1.Must(
t2 =>
//t2.Term(t3=>t3.Field("name").Value("研究"))
// .Term(t3=>t3.Field("name").Value("方鸿渐"))
mustNameQueryKeys
)
)
)
.Should(t =>
t.Bool(m1 =>
m1.Must(t2 =>
//t2.Term(t3 => t3.Field("intro").Value("研究"))
//.Term(t3 => t3.Field("intro").Value("方鸿渐"))
mustIntroQueryKeys
)
)
)
)
)
//分页
.From(from)
.Size(size)
//排序
// .Sort(c => c.Field("age", SortDirection.desc))
//添加高亮
.Highlight(h => h
.PreTags("<b>")
.PostTags("</b>")
.Fields(
f => f.FieldName("intro").Order(HighlightOrder.score),
f => f.FieldName("name").Order(HighlightOrder.score)
)
)
.Build();
string result = Client.Post(cmd, query);
var serializer = new JsonNetSerializer();
var list = serializer.ToSearchResult<Supernova.Webapi.DbHelper.person>(result);
personList datalist = new personList();
datalist.hits = list.hits.total;
datalist.took = list.took;
var personList = list.hits.hits.Select(c => new Supernova.Webapi.DbHelper.person()
{
id = c._source.id,
age = c._source.age,
birthday = c._source.birthday,
intro = c.highlight==null||!c.highlight.Keys.Contains("intro") ? c._source.intro : string.Join("", c.highlight["intro"]), //高亮显示的内容，一条记录中出现了几次
name = c.highlight==null||!c.highlight.Keys.Contains("name") ? c._source.name : string.Join("", c.highlight["name"]),
sex = c._source.sex
});
datalist.list.AddRange(personList);
return datalist;
}
3.搜索age在1到500之间，并且字段intro 或者name 包含词组key中的所有单词。
public personList SearchFullFiledss<person>(string indexName, string indexType, string key, int from, int size)
{
MustQuery<person> mustNameQueryKeys = new MustQuery<person>();
MustQuery<person> mustIntroQueryKeys = new MustQuery<person>();
var arrKeys = GetIKTokenFromStr(key);
foreach (var item in arrKeys)
{
mustNameQueryKeys = mustNameQueryKeys.Term(t3 => t3.Field("name").Value(item)) as MustQuery<person>;
mustIntroQueryKeys = mustIntroQueryKeys.Term(t3 => t3.Field("intro").Value(item)) as MustQuery<person>;
}
string cmd = new SearchCommand(indexName, indexType);
string query = new QueryBuilder<person>()
//1 查询
.Query(b =>
b.Bool(m =>
m.Must(t =>
t.Range(r => r.Field("age").From("1").To("500"))
.Bool(ms =>
ms.Should(ts =>
ts.Bool(m1 =>
m1.Must(
t2 =>
//t2.Term(t3=>t3.Field("name").Value("研究"))
// .Term(t3=>t3.Field("name").Value("方鸿渐"))
//
mustNameQueryKeys
)
)
)
.Should(ts =>
ts.Bool(m1 =>
m1.Must(t2 =>
//t2.Term(t3 => t3.Field("intro").Value("研究"))
//.Term(t3 => t3.Field("intro").Value("方鸿渐"))
//
mustIntroQueryKeys
)
)
)
)
)
)
)
//分页
.From(from)
.Size(size)
//排序
// .Sort(c => c.Field("age", SortDirection.desc))
//添加高亮
.Highlight(h => h
.PreTags("<b>")
.PostTags("</b>")
.Fields(
f => f.FieldName("intro").Order(HighlightOrder.score),
f => f.FieldName("name").Order(HighlightOrder.score)
)
)
.Build();
string result = Client.Post(cmd, query);
var serializer = new JsonNetSerializer();
var list = serializer.ToSearchResult<Supernova.Webapi.DbHelper.person>(result);
personList datalist = new personList();
datalist.hits = list.hits.total;
datalist.took = list.took;
var personList = list.hits.hits.Select(c => new Supernova.Webapi.DbHelper.person()
{
id = c._source.id,
age = c._source.age,
birthday = c._source.birthday,
intro = c.highlight==null||!c.highlight.Keys.Contains("intro") ? c._source.intro : string.Join("", c.highlight["intro"]), //高亮显示的内容，一条记录中出现了几次
name = c.highlight==null||!c.highlight.Keys.Contains("name") ? c._source.name : string.Join("", c.highlight["name"]),
sex = c._source.sex
});
datalist.list.AddRange(personList);
return datalist;
}
需要用到的方法：将语句用ik分词，返回分词结果的集合
private List<string> GetIKTokenFromStr(string key)
{
string s = "/db_test/_analyze?analyzer=ik";
var result = Client.Post(s, "{"+key+"}");
var serializer = new JsonNetSerializer();
var list = serializer.Deserialize(result, typeof(ik)) as ik;
return list.tokens.Select(c=>c.token).ToList();
}
ASP.NET WebApi 调用ElasticSearchHelper
1.首先我们添加一个基类ApiController
public class BaseApiController : ApiController
{
public MongoDatabase db;
public MongoCollection col = null;//用于直接返回查询的json
public BaseApiController() { }
public BaseApiController(string collectionName)
{
db = DbHelper.MongodbHelper.Instance.DB;
col = db.GetCollection(collectionName);
}
public string GetStringRequest(string paramter)
{
return HttpContext.Current.Request.QueryString[paramter] ?? "";
}
public int? GetIntRequest(string paramter)
{
string tmp = HttpContext.Current.Request.QueryString[paramter] ?? "";
int tag = 0;
if (!int.TryParse(tmp, out tag))
{
return null;
}
return tag;
}
}
2.操作ElasticSearch的apicontroller如下：
View Code3.索引数据的api如下：
/// <summary>
/// 索引数据
/// </summary>
/// <returns></returns>
[Route("estest/index")]
[HttpGet]
public object index()
{
int length = S.test.Length;
Random rd = new Random();
Random rdName = new Random();
ParallelOptions _po = new ParallelOptions();
_po.MaxDegreeOfParallelism = 4;
Parallel.For(0, 10000000, _po, c =>
{
var start = rd.Next(0, S.test.Length - 700);
var startName = rd.Next(0, S.test.Length - 30);
person p = new person() { age = DateTime.Now.Millisecond, birthday = DateTime.Now, id = Guid.NewGuid().ToString(), intro = S.test.Substring(start, 629) + c, name = S.test.Substring(startName, 29) + c, sex = true };
ElasticSearchHelper.Intance.Index("db_test", "person", Guid.NewGuid().ToString(), p);
});
return 1;
}
索引使用的测试数据如下：
View Code4.搜索api如下：
[Route("estest")]
[HttpGet]
public object Search()
{
//1 搜索数据
string key = GetStringRequest("Key");
int? from = GetIntRequest("from");
int? size = GetIntRequest("size");
return ElasticSearchHelper.Intance.Search<person>("db_test", "person", key ?? "方鸿渐", from == null ? 0 : from.Value, size == null ? 20 : size.Value);
}
[Route("estest/SearchFullFileds")]
[HttpGet]
public object SearchFullFileds()
{
//1 搜索数据
string key = GetStringRequest("Key");
int? from = GetIntRequest("from");
int? size = GetIntRequest("size");
return ElasticSearchHelper.Intance.SearchFullFileds<person>("db_test", "person", key ?? "方鸿渐", from == null ? 0 : from.Value, size == null ? 20 : size.Value);
}
[Route("estest/SearchFullFiledss")]
[HttpGet]
public object SearchFullFiledss()
{
//1 搜索数据
string key = GetStringRequest("Key");
int? from = GetIntRequest("from");
int? size = GetIntRequest("size");
return ElasticSearchHelper.Intance.SearchFullFiledss<person>("db_test", "person", key ?? "方鸿渐", from == null ? 0 : from.Value, size == null ? 20 : size.Value);
}
WebSite中的view视图调用webapi
说明：我是直接使用ajax(jsop格式)调用webapi，返回的数据直接用knockout绑定到table中的。
视图代码如下：
@{
ViewBag.Title = "ElasticSearch测试";
Layout = null;
}
<link href="~/Content/bootstrap.css" rel="stylesheet" />
<script src="~/Scripts/jquery-2.1.3.min.js"></script>
<script src="~/Scripts/knockout-3.3.0.js"></script>
<style>
b{
color:red;
}
</style>
<script>
$(function () {
function ViewModel() {
self = this;
self.getData = ko.observableArray();
//定义加载数据方法
self.loadData = function () {
$.get("http://192.168.0.230/api/estest//SearchFullFiledss", { key: "@Request.QueryString["key"]", from: "@Request.QueryString["from"]", size: "@Request.QueryString["size"]" }, function (data) {
// alert(data.hits);
// alert(data.list.length);
$("#count").text("符合条件的数据供：" + data.hits + " 共花费了：" + data.took + "毫秒");
$.each(data.list, function (i) {
var date=data.list[i].birthday;
});
self.getData(data.list);
}, "jsonp");
};
//调用定义方法
self.loadData();
}
ko.applyBindings(new ViewModel());
});
</script>
<div id="count"></div>
<div class="row">
<table class="table-bordered table-condensed table-hover table-striped">
<tbody data-bind="foreach: getData">
<tr>
<td data-bind="text: id" />
<td data-bind="html: name" />
<td data-bind="text: sex" />
<td data-bind="text: age" />
<td data-bind="text: birthday" />
<td data-bind="html: intro" />
</tr>
</tbody>
</table>
</div>
<script>
</script>
搜索结果测试如下（我是用"api/estest//SearchFullFiledss"这个api测试的，搜索age在1到500之间，并且字段intro 或者name 包含词组key中的所有单词。）：
1.首先我们看一看测试数据总共有多少：
我们可以看到总共db_test中总共有两千多万条数据。
搜索测试1
测试条件：key=上海方鸿渐&from=0&size=100，key指搜索关键短语，0是从第0条开始区数据，100是指取一百条数据，隐藏条件是age大于1小于500
我们可以看到，首次搜索时，两千多万条数据大约耗时3.5秒，这里还包括取100条数据的时间，如果把数据改为20条则会更快。这里的硬件配置还只限于我的本机测试(内存8G,处理器Intel i5-4590 3.3GHZ)。
搜索测试2
测试条件同测试1：key=上海方鸿渐&from=0&size=100，key指搜索关键短语，0是从第0条开始区数据，100是指取一百条数据，隐藏条件是age大于1小于500
我们可以看到，搜索耗时降到了大约1.5秒。这说明同一个关键词搜索越频繁，搜索速度越快，这是因为ElasticSearch会自动将搜索的内容缓存到内存中。
搜索测试3
测试条件：key=香烟德国&from=0&size=20，key指搜索关键短语，0是从第0条开始区数据，20是指取一百条数据，隐藏条件是age大于1小于500
搜索测试4
测试条件同测试3，二次相同条件搜索：key=香烟德国&from=0&size=20，key指搜索关键短语，0是从第0条开始区数据，20是指取一百条数据，隐藏条件是age大于1小于500
我们可以看到，在二次搜索条件相同，搜索数据降低到20条的时候，只耗时不到0.5秒。
本章完……
ElasticSearch系列学习
ElasticSearch第五步-.net平台下c#操作ElasticSearch详解
http://www.cnblogs.com/eggTwo/p/4425269.html