Pattern Replace Character Filter

Pattern Replace Character Filter 使用正则表达式来匹配应替换为指定替换字符串的字符。替换字符串可以引用正则表达式中的捕获组。

小心"病态"正则表达式

Pattern Replace Character Filter使用Java正则表达式

一个“病态”的正则表达式可能会运行得非常慢,甚至会抛出一个StackOverflowError,还会导致其运行的节点突然退出。

更多信息请查看 pathological regular expressions and how to avoid them.

配置

Pattern Replace Character Filter 会接收以下参数:

参数名称 参数说明
pattern 一个Java正则表达式. 必填
replacement 替换字符串,可以使用$ 1 .. $ 9语法来引用捕获组,可以参考这里
flags Java正则表达式标志。 标志应该用“|”进行分割,例如“CASE_INSENSITIVE | COMMENTS”。

在下面例子中,我们使用Pattern Replace Character Filter实现用下划线替代任何嵌入的破折号,即123-456-789→123_456_789:

案例

PUT my_index
{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_analyzer": {
          "tokenizer": "standard",
          "char_filter": [
            "my_char_filter"
          ]
        }
      },
      "char_filter": {
        "my_char_filter": {
          "type": "pattern_replace",
          "pattern": "(\\d+)-(?=\\d)",
          "replacement": "$1_"
        }
      }
    }
  }
}

POST my_index/_analyze
{
  "analyzer": "my_analyzer",
  "text": "My credit card is 123-456-789"
}

上面案例将返回如下结果:

[ My, credit, card, is 123_456_789 ]

出于搜索目的使用替换字符串,会引起原始文本长度的更改,进而导致不正确的高亮显示。案例如下所示。

这个案例在遇到小写字母后跟大写字母(即fooBarBaz→foo Bar Baz)时插入空格,允许单独查询camelCase字词:

案例

PUT my_index
{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_analyzer": {
          "tokenizer": "standard",
          "char_filter": [
            "my_char_filter"
          ],
          "filter": [
            "lowercase"
          ]
        }
      },
      "char_filter": {
        "my_char_filter": {
          "type": "pattern_replace",
          "pattern": "(?<=\\p{Lower})(?=\\p{Upper})",
          "replacement": " "
        }
      }
    }
  },
  "mappings": {
    "my_type": {
      "properties": {
        "text": {
          "type": "text",
          "analyzer": "my_analyzer"
        }
      }
    }
  }
}

POST my_index/_analyze
{
  "analyzer": "my_analyzer",
  "text": "The fooBarBaz method"
}

案例返回的结果如下:

[ the, foo, bar, baz, method ]

查询 bar 可以正确找到文档,但突出显示结果将产生不正确的高光,因为我们的字符过滤器更改了原始文本的长度:

PUT my_index/my_doc/1?refresh
{
  "text": "The fooBarBaz method"
}

GET my_index/_search
{
  "query": {
    "match": {
      "text": "bar"
    }
  },
  "highlight": {
    "fields": {
      "text": {}
    }
  }
}

以上的输出结果是:

{
  "timed_out": false,
  "took": $body.took,
  "_shards": {
    "total": 5,
    "successful": 5,
    "failed": 0
  },
  "hits": {
    "total": 1,
    "max_score": 0.2824934,
    "hits": [
      {
        "_index": "my_index",
        "_type": "my_doc",
        "_id": "1",
        "_score": 0.2824934,
        "_source": {
          "text": "The fooBarBaz method"
        },
        "highlight": {
          "text": [
            "The foo<em>Ba</em>rBaz method" 
          ]
        }
      }
    ]
  }
}