Elasticsearch search on phone numbers

Question

I have postgres array column which I wanted to be indexed and then use it in search. Here is example below,

phones = [ "+175 (2) 123-25-32", "123456789", "+12 111-111-11" ]

I have analyzed the tokens using analyze api, elasticsearch is tokenizing the field into multiple fields as follow

curl -XGET 'localhost:9200/_analyze' -d '
{
  "analyzer" : "standard",
  "text" : [ "+175 (2) 123-25-32", "123456789", "+12 111-111-11" ]
}'


{
  "tokens": [
    {
      "token": "analyzer",
      "start_offset": 6,
      "end_offset": 14,
      "type": "<ALPHANUM>",
      "position": 1
    },
    {
      "token": "standard",
      "start_offset": 19,
      "end_offset": 27,
      "type": "<ALPHANUM>",
      "position": 2
    },
    {
      "token": "text",
      "start_offset": 33,
      "end_offset": 37,
      "type": "<ALPHANUM>",
      "position": 3
    },
    {
      "token": "175",
      "start_offset": 45,
      "end_offset": 48,
      "type": "<NUM>",
      "position": 4
    },
    {
      "token": "2",
      "start_offset": 50,
      "end_offset": 51,
      "type": "<NUM>",
      "position": 5
    },
    {
      "token": "123",
      "start_offset": 53,
      "end_offset": 56,
      "type": "<NUM>",
      "position": 6
    },
    {
      "token": "25",
      "start_offset": 57,
      "end_offset": 59,
      "type": "<NUM>",
      "position": 7
    },
    {
      "token": "32",
      "start_offset": 60,
      "end_offset": 62,
      "type": "<NUM>",
      "position": 8
    },
    {
      "token": "123456789",
      "start_offset": 66,
      "end_offset": 75,
      "type": "<NUM>",
      "position": 9
    },
    {
      "token": "12",
      "start_offset": 80,
      "end_offset": 82,
      "type": "<NUM>",
      "position": 10
    },
    {
      "token": "111",
      "start_offset": 83,
      "end_offset": 86,
      "type": "<NUM>",
      "position": 11
    },
    {
      "token": "111",
      "start_offset": 87,
      "end_offset": 90,
      "type": "<NUM>",
      "position": 12
    },
    {
      "token": "11",
      "start_offset": 91,
      "end_offset": 93,
      "type": "<NUM>",
      "position": 13
    }
  ]
}

I wanted elasticsearch either to not do the tokenization and store the numbers without special characters e.g "+175 (2) 123-25-32" to be converted into "+17521232532" OR simply index the number as it is so that It would be available in search result.

My mapping is as below,

{ :id => { :type => "string"}, :secondary_phones => { :type => "string" } }

Here is how I am trying todo the query

      settings = {
        query: {
          filtered: {
            filter: {
              bool: {
                should: [
                  { terms: { phones: [ "+175 (2) 123-25-32", "123456789", "+12 111-111-11" ] } },
                ]
              }
            }
          }
        },
        size: 100,
      }

P.S I have also tried by removing the special characters but no luck.

I am sure it is achievable and I am missing something. Suggestions please.

Thanks.

Answer 1

If you just want to perform exact matches against your data, as in your terms query example, the best approach is to simply set the index mapping parameter in your mapping to be not_analyzed. Have a look at the documentation here.

This disables analysis (or tokenization) of the value entirely and treats the contents of the field (each item in your array) as a single token/keyword.