Elasticsearch 8.7.x中keyword_repeat导致bool should查询结果异常

在Elasticsearch 8.7.0及8.7.1版本中，当索引配置使用keyword_repeat过滤器时，执行包含should子句的bool查询会出现异常：仅匹配should列表中的第一个字段，调换字段顺序后才能正常找到目标文档。移除keyword_repeat过滤器后，查询恢复正常；而8.6.2、8.8.0/8.8.1版本无此问题。

重现步骤 #

1. 创建索引 #

curl -X PUT "elasticsearch:9200/my-test-index?pretty" -H 'Content-Type: application/json' -d'
{
  "settings": {
    "analysis": {
      "analyzer": {
        "default": {          
          "tokenizer": "default_tokenizer",
          "filter": [
            "lowercase",
            "keyword_repeat",
            "default_stemmer"
          ]
        }
      },
      "tokenizer": {
        "default_tokenizer": {
          "type": "standard"
        }
      },
      "filter": {
        "default_stemmer": {
          "type": "stemmer",
          "language": "english"
        },
        "unique_stem": {
          "type": "unique",
          "only_on_same_position": true
        }
      }
    }
  },
  "mappings": {
    "properties": {
      "field1": {
        "type": "text"
      },
      "field2": {
        "type": "text"
      }
    }
  }
}
'

2. 插入文档 #

curl -X POST "elasticsearch:9200/my-test-index/_doc/1?pretty" -H 'Content-Type: application/json' -d'
{
"field1": "running man",
"field2": "other text"
}
'

3. 异常查询（无结果返回） #

curl -X GET "elasticsearch:9200/my-test-index/_search?pretty" -H 'Content-Type: application/json' -d'
{
  "query": {
    "bool": {
      "should": [
        { "match": { "field2":  "running" }},
        { "match": { "field1": "running" }}
      ]
    }
  }
}
'

4. 正常查询（调换字段顺序后返回目标文档） #

curl -X GET "elasticsearch:9200/my-test-index/_search?pretty" -H 'Content-Type: application/json' -d'
{
  "query": {
    "bool": {
      "should": [
        { "match": { "field1":  "running" }},
        { "match": { "field2": "running" }}
      ]
    }
  }
}
'

现象分析 #

通过_termvectors查看索引词项，field1包含"running"和"run"等有效词项，但异常查询无法匹配到该字段，仅当调换should子句中的字段顺序时，才能命中目标文档。

内容的提问来源于stack exchange，提问作者Ilya