CODEWITHSUNDEEP
elasticsearchkibana
eran
eran

Reputation: 15136

Prefix auto-generated field in elasticsearch

I have a document with some uid. I would likt to create a field which is indexed as not_analyzed that is auto-generated as the 2-letter prefix (or suffix) of the uid.

Is there a way to create such a template that will auto-compute that field?

The use case is for showing sampled-down (filter is prefix ='00' for example) statistics on Kibana (so a unique count aggregation will take much less time).

I've used it succesfully, but I create the prefix field on the client before writing the document to the server.

Upvotes: 0

Views: 268

Answers (2)

Nir
Nir

Reputation: 26

I used the edgeNGram tokenizer, seems like it provides the same result.

{
  "settings": {
    "analysis": {
      "analyzer": {
        "edge_ngram_analyzer": {
          "tokenizer": "edge_ngram_tokenizer"
        }
      },
      "tokenizer": {
        "edge_ngram_tokenizer": {
          "type": "edgeNGram",
          "min_gram": "2",
          "max_gram": "2"
        }
      }
    }
  }
}

Upvotes: 1

imotov
imotov

Reputation: 30163

Indexing prefix can be done using a custom analyzer build with keyword tokenizer and truncate token filter. Here is an example of how to index a field test_prefix containing first 2 characters of the field test: 

curl -XPUT localhost:9200/test-idx -d '{
    "settings": {
        "index": {
            "number_of_shards": 1,
            "number_of_replicas": 0
        },
        "analysis": {
            "analyzer": {
                "prefix": {
                    "tokenizer": "keyword",
                    "filter": "prefix"
                }
            },
            "filter": {
                "prefix": {
                    "type": "truncate",
                    "length": 2
                }
            }
        }
    },
    "mappings": {
        "doc": {
            "properties": {
                "test": {
                    "type": "string",
                    "copy_to": "test_prefix"
                },
                "test_prefix": {
                    "type": "string",
                    "analyzer": "prefix"
                }

            }
        }
    }
}'

Upvotes: 1

Related Questions