ES按照整个字段统计

在使用terms facets时,经常想统计一个整个字段的值出现了多少次.如果用下面的消息发送请求:
{
  "query": {
    "bool": {
      "must": [
        {
          "query_string": {
            "default_field": "content",
            "query": "风暴"
          }
        }
      ],
      "must_not": [],
      "should": []
    }
  },
  "from": 0,
  "size": 50,
  "sort": [],
  "facets": {
    "sd3560531facet": {
      "terms": {
        "field": "siteName",
        "all_terms": true
      }
    }
  }
}
得出的统计结果是:
facets: {
province: {
_type: terms
missing: 0
total: 18
other: 0
terms: [
{
term: 区
count: 4
}
{
term: 论
count: 3
}
{
term: 讨
count: 3
}
{
term: 战
count: 2
}
{
term: 场
count: 2
}
{
term: 风
count: 1
}
{
term: 综
count: 1
}
{
term: 纪
count: 1
}
{
term: 合
count: 1
}
{
term: 本
count: 0
}
]
}
默认统计的是每个字出现的次数.不符合实际应用要求.
重新修改消息请求如下:
{
  "query": {
    "bool": {
      "must": [
        {
          "query_string": {
            "default_field": "content",
            "query": "风暴"
          }
        }
      ],
      "must_not": [],
      "should": []
    }
  },
  "from": 0,
  "size": 50,
  "sort": [],
  "facets": {
    "sd3560531facet": {
      "terms": {
        "script_field": "_source.siteName",
        "all_terms": true
      }
    }
  }
}
得出的结果为:
facets: {
province: {
_type: terms
missing: 0
total: 4
other: 0
terms: [
{
term: 战场讨论区
count: 2
}
{
term: 风纪区
count: 1
}
{
term: 综合讨论区
count: 1
}
]
}
}
表示符合条件的4条记录中,siteName字段里,战场讨论区有两条,风纪区有1条,综合讨论区有一条.