ElasticSearch Notes

Posted on | In |

ElasticSearch可以說是一個高效的搜索引擎 同時也可以說是一個數據庫

ES的特性：

Base On

  • 開源 基於lucene

Store Data

  • schemea free–不需要定義檔案格式
  • document oriented–基於文檔儲存

Communicate

  • RESTful Api
  • 返回 Json 格式

Scale

  • 分佈式
  • 多租戶 – 多用戶環境下保持資料隔離性
  • 良好衝突管理機制
  • 高可用性

Search

  • 支持full text search 全文搜索
  • 實時搜索/分析

安裝和配置

步驟：

  1. 下載 最新版 elasticsearch : https://www.elastic.co/downloads/elasticsearch
  2. 安裝kibana和sense
    mac 建議用homebrew下載安裝,能省下不少時間
  3. 安裝分詞插件
    elasticsearch 自帶的分詞插件對中文支持十分不好,需要安裝分詞插件，目前搜索到ik和jieba
    jieba：https://github.com/huaban/elasticsearch-analysis-jieba

  4. 到這一步基本上安裝完畢，到ES目錄下的./bin,啟動ES
    之後將會用RESTful api與其通訊

    1
    2
    3
    4
    5
    6
    7
    Action : GET,PUT,DELETE,POST
+
URI : Unique Resource
+
Content : head +json body
+
Status Code : 200,400
  5. 設cluster和mapping 
    1
    2
    3
    4
    5
    6
    7
    8
    9
    10
    11
    12
    13
    14
    15
    16
    PUT--localhost:9200/index
{
    "settings" : {
        "number_of_shards" : 1,
        "number_of_replicas" : 0
    
    },
    "mappings" : {
        "type" : {
            "_all" : { "enabled" : false },
            "properties" : {
                "field" : { "type" : "string", "analyzer" : "name" }
            }
        }
    }
}

  6. 加入數據

    1
    2
    3
    4
     PUT http://localhost:9200 /index/type/id
 {
     "field" : "value"
 }
  7. 檢查 是否加入成功 和 analyzer有否正常工作
    搜尋所有結果： 
    1
     GET http://localhost:9200/index/type/_search

    analyzer分詞測試:

    1
    2
    3
    4
     PUT http://localhost:9200/index/document/number
 json:{
     content
 }

ElasticSearch於傳統數據庫結構上的區別

Relational DB ： Databases -> Tables -> Rows -> Columns
Elastic search ： Index -> Types -> Documents -> Fields

Full architecture of ES :
Cluster ->Node ->Index(Shard|Replication) -> Documents -> Fields

ElasticSearch-Jieba

Install elasticsearch and install jieba plugin:
https://github.com/huaban/elasticsearch-analysis-jieba

After install it set the mapping:
remember to set the index document and field analyzer：

  • index 主要用于索引分词，分词粒度较细
  • search 主要用于查询分词，分词粒度较粗
  • other 全角转半角、大写转小写、字符分词 
    1
    2
    3
    4
    5
    6
    7
    8
    9
    10
    11
    12
    13
    14
    15
    16
    PUT--localhost:9200/index
{
"settings" : {
    "number_of_shards" : 1,
    "number_of_replicas" : 0

},
"mappings" : {
    "document" : {
        "_all" : { "enabled" : false },
        "properties" : {
            "field" : { "type" : "string", "analyzer" : "jieba_index", "search_analyzer" : "jieba_search" }
        }
    }
}
}

To update the jieba dict go to elasticsearch/plugins/jieba/dic

  • sougou.dict –courpus from sougou
  • stopwords.txt –stopwords will take out from the text
  • user.dict –the dict can append after edit it , need to restart elasticsearch

    jieba support Traditional chinese formatting syntax is to make it as readable to upload data used restful api

    1
    2
    3
    4
    PUT http://localhost:9200/index/document/number
json:{
content
}

To test upload success or not, can try to make a query:

1
http://localhost:9200/index/document/_search?q=content:value

To test jieba

1
2
3
4
post
http://localhost:9200/test/_analyze?analyzer=jieba_search
body:
{text}

about how es counting the score:http://www.cnblogs.com/richaaaard/p/5254988.html