CODEWITHSUNDEEP
elasticsearch
Vivek
Vivek

Reputation: 67

Dynamic Mappings for nested field in elasticsearch

Example document:

{
   "first_name": "Uttam",
   "last_name": "Kini",
   "metadata": [
        {
           "field_type": "small_text",
           "field_value": "bla bla"
        }, 
        {
           "field_type": "large_text",
           "field_value": "bla bla bla bla bla bla bla bla"
        }, 
     ]
}

How do I dynamically set the mapping of metadata.field_value based on metadata.field_type?

For example, when metadata.field_type is small_text I would like to use an analyzer with keyword tokenizer and when metadata.field_type is large_text I would like to use a different analyzer.

Upvotes: 3

Views: 258

Answers (1)

Artur Nowak
Artur Nowak

Reputation: 5354

I think it is not possible to do this directly. The important reason it is not supported is that it is not possible to provide good quality results out of field that contains products of different analyzers. Think about parsing the query: which analyzer would you use on query to match terms produced by two different analyzers?

The solution to this is to use multi-fields. You basically create two related fields for field_value, applying different mapping to each of them. Then, at query time you construct a query with two branches: 

(field_type:small_text AND field_value.small_text:abc)
  OR (field_type:large_text AND field_value.large_text:abc)

Note: I've used query string here for readability, you probably had better use filtered query with or and and filters

Upvotes: 1

Related Questions