弹性搜索match_pharse查询无法正常工作

Question

Am trying to search the below document using match_phrase query in kibana but am not getting the response.

Please find the document below which is availabe in elastic search

    {  
       "took":7,
       "timed_out":false,
       "_shards":{  
          "total":5,
          "successful":5,
          "skipped":0,
          "failed":0
       },
       "hits":{  
          "total":2910,
          "max_score":1.0,
          "hits":[  
             {  
                "_index":"documents",
                "_type":"doc",
                "_id":"DmLD22MBFTg0XFZppYt8",
                "_score":1.0,
                "_source":{  
                   "doct_country":"DE",
                   "filename":"series_Accessories_v1_de-DE.pdf",


             }

          ]
       }
    }

Please find the query which am using to search this above document.

GET documents/_search
{
    "query": {
        "match_phrase" : {
            "message" : "Accessories_v1_de-DE.pdf"
        }
    }
}

For the above query am getting this response :

{
  "took": 0,
  "timed_out": false,
  "_shards": {
    "total": 5,
    "successful": 5,
    "skipped": 0,
    "failed": 0
  },
  "hits": {
    "total": 0,
    "max_score": null,
    "hits": []
  }
}

Answer 1

There are two issues. Presumably in your query you mean to use the filename field rather than message which is not present in you example document:

GET documents/_search
{
    "query": {
        "match_phrase" : {
            "filename" : "Accessories_v1_de-DE.pdf"
        }
    }
}

Second, you need Elasticsearch to know that the filename field should be indexed with _ treated as a split. This does not happen by default. One way to do this is to define your mapping as follows:

PUT /documents
{
    "mappings" : {
        "document" : {
            "properties" : {
                "filename" : { "type" : "text", "analyzer": "simple" }
            }
        }
    }
}

The simple analyzer will split on any non-letter, so _ and numbers will be treated as splits. Depending on your application, you may need finer grained control over tokenization. See the documentation .