php - Elasticsearch仅按字母排序而不按数字排序

Question

我在使用PHP排序时遇到问题，这是我的映射:

{
  "jdbc": {
    "mappings": {
      "jdbc": {
        "properties": { 
          "admitted_date": {
            "type": "date",
            "format": "dateOptionalTime"
          },
          "et_tax": {
            "type": "string"
          },  
          "jt_tax": {
            "type": "string"
          }, 
          "loc_cityname": {
            "type": "string"
          }, 
          "location_countryname": {
            "type": "string"
          },
          "location_primary": {
            "type": "string"
          },  
          "pd_firstName": {
            "type": "string"
          } 
        }
      }
    }
  }
}

当我使用按结果对结果进行排序时，它将使用字母数字顺序对结果进行排序，它将首先加载数字形式的结果。我需要仅以字母开头来排序结果。现在它的命令是这样的:

http://localhost:9200/jdbc/_search?pretty=true&sort=pd_lawFirmName:asc

1. BM&A
2. Gomez-Acebo & Pombo
3. Addleshaw Goddard

如何排序这样的结果？

1. Addleshaw Goddard
2. BM&A
3. Gomez-Acebo & Pombo

这是我用于索引的查询

{
    "type" : "jdbc",
    "jdbc" : {
        "driver" : "com.mysql.jdbc.Driver",
        "url" : "jdbc:mysql://localhost:3306/dbname",
        "user" : "user",
        "password" : "pass",
        "sql" : "SQL QUERY",
        "poll" : "24h",
        "strategy" : "simple", 
        "scale" : 0,
        "autocommit" : true,
        "bulk_size" : 5000,
        "max_bulk_requests" : 30,
        "bulk_flush_interval" : "5s",
        "fetchsize" : 100,
        "max_rows" : 149669,
        "max_retries" : 3,
        "max_retries_wait" : "10s",
        "locale" : "in",
        "digesting" : true,
        "mappings": {
        "sorting": {
        "properties": { 
        "pd_lawFirmName": {
        "type": "string",
        "fields": {
          "raw": {
            "type": "string",
            "index": "not_analyzed"
          }
        }
      }
      }
    }
  }
  }
}

Answer 1

之所以这样，是因为Elasticsearch将使用默认分析器standard标记文本。例如，McDermott Will Amery的索引如下:

              "amery",
              "mcdermott",
              "will"

如果您想这样排序，我建议以如下方式更改pd_lawFirmName的映射:

  "pd_lawFirmName": {
    "type": "string",
    "fields": {
      "raw": {
        "type": "string",
        "index": "not_analyzed"
      }
    }
  }

并按raw子字段排序:

http://localhost:9200/jdbc/_search?pretty=true&sort=pd_lawFirmName.raw:asc