Dynamic templates

edit

Dynamic templates allow you greater control of how Elasticsearch maps your data beyond the default dynamic field mapping rules. You enable dynamic mapping by setting the dynamic parameter to true or runtime. You can then use dynamic templates to define custom mappings that can be applied to dynamically added fields based on the matching condition:

Use the {name} and {dynamic_type} template variables in the mapping specification as placeholders.

Dynamic field mappings are only added when a field contains a concrete value. Elasticsearch doesn’t add a dynamic field mapping when the field contains null or an empty array. If the null_value option is used in a dynamic_template, it will only be applied after the first document with a concrete value for the field has been indexed.

Dynamic templates are specified as an array of named objects:

  "dynamic_templates": [
    {
      "my_template_name": { 
        ... match conditions ... 
        "mapping": { ... } 
      }
    },
    ...
  ]

The template name can be any string value.

The match conditions can include any of : match_mapping_type, match, match_pattern, unmatch, path_match, path_unmatch.

The mapping that the matched field should use.

Validating dynamic templates

edit

If a provided mapping contains an invalid mapping snippet, a validation error is returned. Validation occurs when applying the dynamic template at index time, and, in most cases, when the dynamic template is updated. Providing an invalid mapping snippet may cause the update or validation of a dynamic template to fail under certain conditions:

  • If no match_mapping_type has been specified but the template is valid for at least one predefined mapping type, the mapping snippet is considered valid. However, a validation error is returned at index time if a field matching the template is indexed as a different type. For example, configuring a dynamic template with no match_mapping_type is considered valid as string type, but if a field matching the dynamic template is indexed as a long, a validation error is returned at index time. It is recommended to configure the match_mapping_type to the expected JSON type or configure the desired type in the mapping snippet.
  • If the {name} placeholder is used in the mapping snippet, validation is skipped when updating the dynamic template. This is because the field name is unknown at that time. Instead, validation occurs when the template is applied at index time.

Templates are processed in order — the first matching template wins. When putting new dynamic templates through the update mapping API, all existing templates are overwritten. This allows for dynamic templates to be reordered or deleted after they were initially added.

Mapping runtime fields in a dynamic template

edit

If you want Elasticsearch to dynamically map new fields of a certain type as runtime fields, set "dynamic":"runtime" in the index mappings. These fields are not indexed, and are loaded from _source at query time.

Alternatively, you can use the default dynamic mapping rules and then create dynamic templates to map specific fields as runtime fields. You set "dynamic":"true" in your index mapping, and then create a dynamic template to map new fields of a certain type as runtime fields.

Let’s say you have data where each of the fields start with ip_. Based on the dynamic mapping rules, Elasticsearch maps any string that passes numeric detection as a float or long. However, you can create a dynamic template that maps new strings as runtime fields of type ip.

The following request defines a dynamic template named strings_as_ip. When Elasticsearch detects new string fields matching the ip* pattern, it maps those fields as runtime fields of type ip. Because ip fields aren’t mapped dynamically, you can use this template with either "dynamic":"true" or "dynamic":"runtime".

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          strings_as_ip: {
            match_mapping_type: 'string',
            match: 'ip*',
            runtime: {
              type: 'ip'
            }
          }
        }
      ]
    }
  }
)
puts response
PUT my-index-000001/
{
  "mappings": {
    "dynamic_templates": [
      {
        "strings_as_ip": {
          "match_mapping_type": "string",
          "match": "ip*",
          "runtime": {
            "type": "ip"
          }
        }
      }
    ]
  }
}

See this example for how to use dynamic templates to map string fields as either indexed fields or runtime fields.

match_mapping_type and unmatch_mapping_type

edit

The match_mapping_type parameter matches fields by the data type detected by the JSON parser, while unmatch_mapping_type excludes fields based on the data type.

Because JSON doesn’t distinguish a long from an integer or a double from a float, any parsed floating point number is considered a double JSON data type, while any parsed integer number is considered a long.

With dynamic mappings, Elasticsearch will always choose the wider data type. The one exception is float, which requires less storage space than double and is precise enough for most applications. Runtime fields do not support float, which is why "dynamic":"runtime" uses double.

Elasticsearch automatically detects the following data types:

Elasticsearch data type

JSON data type

"dynamic":"true"

"dynamic":"runtime"

null

No field added

No field added

true or false

boolean

boolean

double

float

double

long

long

long

object

object

No field added

array

Depends on the first non-null value in the array

Depends on the first non-null value in the array

string that passes date detection

date

date

string that passes numeric detection

float or long

double or long

string that doesn’t pass date detection or numeric detection

text with a .keyword sub-field

keyword

You can specify either a single data type or a list of data types for either the match_mapping_type or unmatch_mapping_type parameters. You can also use a wildcard (*) for the match_mapping_type parameter to match all data types.

For example, if we wanted to map all integer fields as integer instead of long, and all string fields as both text and keyword, we could use the following template:

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          numeric_counts: {
            match_mapping_type: [
              'long',
              'double'
            ],
            match: 'count',
            mapping: {
              type: '{dynamic_type}',
              index: false
            }
          }
        },
        {
          integers: {
            match_mapping_type: 'long',
            mapping: {
              type: 'integer'
            }
          }
        },
        {
          strings: {
            match_mapping_type: 'string',
            mapping: {
              type: 'text',
              fields: {
                raw: {
                  type: 'keyword',
                  ignore_above: 256
                }
              }
            }
          }
        },
        {
          non_objects_keyword: {
            match_mapping_type: '*',
            unmatch_mapping_type: 'object',
            mapping: {
              type: 'keyword'
            }
          }
        }
      ]
    }
  }
)
puts response

response = client.index(
  index: 'my-index-000001',
  id: 1,
  body: {
    my_integer: 5,
    my_string: 'Some string',
    my_boolean: 'false',
    field: {
      count: 4
    }
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "numeric_counts": {
          "match_mapping_type": ["long", "double"],
          "match": "count",
          "mapping": {
            "type": "{dynamic_type}",
            "index": false
          }
        }
      },
      {
        "integers": {
          "match_mapping_type": "long",
          "mapping": {
            "type": "integer"
          }
        }
      },
      {
        "strings": {
          "match_mapping_type": "string",
          "mapping": {
            "type": "text",
            "fields": {
              "raw": {
                "type":  "keyword",
                "ignore_above": 256
              }
            }
          }
        }
      },
      {
        "non_objects_keyword": {
          "match_mapping_type": "*",
          "unmatch_mapping_type": "object",
          "mapping": {
            "type": "keyword"
          }
        }
      }
    ]
  }
}

PUT my-index-000001/_doc/1
{
  "my_integer": 5, 
  "my_string": "Some string", 
  "my_boolean": "false", 
  "field": {"count": 4} 
}

The my_integer field is mapped as an integer.

The my_string field is mapped as a text, with a keyword multi-field.

The my_boolean field is mapped as a keyword.

The field.count field is mapped as a long.

match and unmatch

edit

The match parameter uses one or more patterns to match on the field name, while unmatch uses one or more patterns to exclude fields matched by match.

The match_pattern parameter adjusts the behavior of the match parameter to support full Java regular expressions matching on the field name instead of simple wildcards. For example:

  "match_pattern": "regex",
  "match": "^profit_\d+$"

The following example matches all string fields whose name starts with long_ (except for those which end with _text) and maps them as long fields:

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          longs_as_strings: {
            match_mapping_type: 'string',
            match: 'long_*',
            unmatch: '*_text',
            mapping: {
              type: 'long'
            }
          }
        }
      ]
    }
  }
)
puts response

response = client.index(
  index: 'my-index-000001',
  id: 1,
  body: {
    long_num: '5',
    long_text: 'foo'
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "longs_as_strings": {
          "match_mapping_type": "string",
          "match":   "long_*",
          "unmatch": "*_text",
          "mapping": {
            "type": "long"
          }
        }
      }
    ]
  }
}

PUT my-index-000001/_doc/1
{
  "long_num": "5", 
  "long_text": "foo" 
}

The long_num field is mapped as a long.

The long_text field uses the default string mapping.

You can specify a list of patterns using a JSON array for either the match or unmatch fields.

The next example matches all fields whose name starts with ip_ or ends with _ip, except for fields which start with one or end with two and maps them as ip fields:

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          ip_fields: {
            match: [
              'ip_*',
              '*_ip'
            ],
            unmatch: [
              'one*',
              '*two'
            ],
            mapping: {
              type: 'ip'
            }
          }
        }
      ]
    }
  }
)
puts response

response = client.index(
  index: 'my-index',
  id: 1,
  body: {
    one_ip: 'will not match',
    ip_two: 'will not match',
    three_ip: '12.12.12.12',
    ip_four: '13.13.13.13'
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "ip_fields": {
          "match":   ["ip_*", "*_ip"],
          "unmatch": ["one*", "*two"],
          "mapping": {
            "type": "ip"
          }
        }
      }
    ]
  }
}

PUT my-index/_doc/1
{
  "one_ip":   "will not match", 
  "ip_two":   "will not match", 
  "three_ip": "12.12.12.12", 
  "ip_four":  "13.13.13.13" 
}

The one_ip field is unmatched, so uses the default mapping of text.

The ip_two field is unmatched, so uses the default mapping of text.

The three_ip field is mapped as type ip.

The ip_four field is mapped as type ip.

path_match and path_unmatch

edit

The path_match and path_unmatch parameters work in the same way as match and unmatch, but operate on the full dotted path to the field, not just the final name, e.g. some_object.*.some_field.

This example copies the values of any fields in the name object to the top-level full_name field, except for the middle field:

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          full_name: {
            path_match: 'name.*',
            path_unmatch: '*.middle',
            mapping: {
              type: 'text',
              copy_to: 'full_name'
            }
          }
        }
      ]
    }
  }
)
puts response

response = client.index(
  index: 'my-index-000001',
  id: 1,
  body: {
    name: {
      first: 'John',
      middle: 'Winston',
      last: 'Lennon'
    }
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "full_name": {
          "path_match":   "name.*",
          "path_unmatch": "*.middle",
          "mapping": {
            "type":       "text",
            "copy_to":    "full_name"
          }
        }
      }
    ]
  }
}

PUT my-index-000001/_doc/1
{
  "name": {
    "first":  "John",
    "middle": "Winston",
    "last":   "Lennon"
  }
}

And the following example uses an array of patterns for both path_match and path_unmatch.

The values of any fields in the name object or the user.name object are copied to the top-level full_name field, except for the middle and midinitial fields:

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          full_name: {
            path_match: [
              'name.*',
              'user.name.*'
            ],
            path_unmatch: [
              '*.middle',
              '*.midinitial'
            ],
            mapping: {
              type: 'text',
              copy_to: 'full_name'
            }
          }
        }
      ]
    }
  }
)
puts response

response = client.index(
  index: 'my-index-000001',
  id: 1,
  body: {
    name: {
      first: 'John',
      middle: 'Winston',
      last: 'Lennon'
    }
  }
)
puts response

response = client.index(
  index: 'my-index-000001',
  id: 2,
  body: {
    user: {
      name: {
        first: 'Jane',
        midinitial: 'M',
        last: 'Salazar'
      }
    }
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "full_name": {
          "path_match":   ["name.*", "user.name.*"],
          "path_unmatch": ["*.middle", "*.midinitial"],
          "mapping": {
            "type":       "text",
            "copy_to":    "full_name"
          }
        }
      }
    ]
  }
}

PUT my-index-000001/_doc/1
{
  "name": {
    "first":  "John",
    "middle": "Winston",
    "last":   "Lennon"
  }
}

PUT my-index-000001/_doc/2
{
  "user": {
    "name": {
      "first":      "Jane",
      "midinitial": "M",
      "last":       "Salazar"
    }
  }
}

Note that the path_match and path_unmatch parameters match on object paths in addition to leaf fields. As an example, indexing the following document will result in an error because the path_match setting also matches the object field name.title, which can’t be mapped as text:

response = client.index(
  index: 'my-index-000001',
  id: 2,
  body: {
    name: {
      first: 'Paul',
      last: 'McCartney',
      title: {
        value: 'Sir',
        category: 'order of chivalry'
      }
    }
  }
)
puts response
PUT my-index-000001/_doc/2
{
  "name": {
    "first":  "Paul",
    "last":   "McCartney",
    "title": {
      "value": "Sir",
      "category": "order of chivalry"
    }
  }
}

Template variables

edit

The {name} and {dynamic_type} placeholders are replaced in the mapping with the field name and detected dynamic type. The following example sets all string fields to use an analyzer with the same name as the field, and disables doc_values for all non-string fields:

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          named_analyzers: {
            match_mapping_type: 'string',
            match: '*',
            mapping: {
              type: 'text',
              analyzer: '{name}'
            }
          }
        },
        {
          no_doc_values: {
            match_mapping_type: '*',
            mapping: {
              type: '{dynamic_type}',
              doc_values: false
            }
          }
        }
      ]
    }
  }
)
puts response

response = client.index(
  index: 'my-index-000001',
  id: 1,
  body: {
    english: 'Some English text',
    count: 5
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "named_analyzers": {
          "match_mapping_type": "string",
          "match": "*",
          "mapping": {
            "type": "text",
            "analyzer": "{name}"
          }
        }
      },
      {
        "no_doc_values": {
          "match_mapping_type":"*",
          "mapping": {
            "type": "{dynamic_type}",
            "doc_values": false
          }
        }
      }
    ]
  }
}

PUT my-index-000001/_doc/1
{
  "english": "Some English text", 
  "count":   5 
}

The english field is mapped as a string field with the english analyzer.

The count field is mapped as a long field with doc_values disabled.

Dynamic template examples

edit

Here are some examples of potentially useful dynamic templates:

Structured search

edit

When you set "dynamic":"true", Elasticsearch will map string fields as a text field with a keyword subfield. If you are only indexing structured content and not interested in full text search, you can make Elasticsearch map your fields only as keyword fields. However, you must search on the exact same value that was indexed to search those fields.

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          strings_as_keywords: {
            match_mapping_type: 'string',
            mapping: {
              type: 'keyword'
            }
          }
        }
      ]
    }
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "strings_as_keywords": {
          "match_mapping_type": "string",
          "mapping": {
            "type": "keyword"
          }
        }
      }
    ]
  }
}

text-only mappings for strings

edit

Contrary to the previous example, if you only care about full-text search on string fields and don’t plan on running aggregations, sorting, or exact searches, you could tell instruct Elasticsearch to map strings as text:

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          strings_as_text: {
            match_mapping_type: 'string',
            mapping: {
              type: 'text'
            }
          }
        }
      ]
    }
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "strings_as_text": {
          "match_mapping_type": "string",
          "mapping": {
            "type": "text"
          }
        }
      }
    ]
  }
}

Alternatively, you can create a dynamic template to map your string fields as keyword fields in the runtime section of the mapping. When Elasticsearch detects new fields of type string, those fields will be created as runtime fields of type keyword.

Although your string fields won’t be indexed, their values are stored in _source and can be used in search requests, aggregations, filtering, and sorting.

For example, the following request creates a dynamic template to map string fields as runtime fields of type keyword. Although the runtime definition is blank, new string fields will be mapped as keyword runtime fields based on the dynamic mapping rules that Elasticsearch uses for adding field types to the mapping. Any string that doesn’t pass date detection or numeric detection is automatically mapped as a keyword:

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          strings_as_keywords: {
            match_mapping_type: 'string',
            runtime: {}
          }
        }
      ]
    }
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "strings_as_keywords": {
          "match_mapping_type": "string",
          "runtime": {}
        }
      }
    ]
  }
}

You index a simple document:

response = client.index(
  index: 'my-index-000001',
  id: 1,
  body: {
    english: 'Some English text',
    count: 5
  }
)
puts response
PUT my-index-000001/_doc/1
{
  "english": "Some English text",
  "count":   5
}

When you view the mapping, you’ll see that the english field is a runtime field of type keyword:

response = client.indices.get_mapping(
  index: 'my-index-000001'
)
puts response
GET my-index-000001/_mapping
{
  "my-index-000001" : {
    "mappings" : {
      "dynamic_templates" : [
        {
          "strings_as_keywords" : {
            "match_mapping_type" : "string",
            "runtime" : { }
          }
        }
      ],
      "runtime" : {
        "english" : {
          "type" : "keyword"
        }
      },
      "properties" : {
        "count" : {
          "type" : "long"
        }
      }
    }
  }
}

Disabled norms

edit

Norms are index-time scoring factors. If you do not care about scoring, which would be the case for instance if you never sort documents by score, you could disable the storage of these scoring factors in the index and save some space.

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          strings_as_keywords: {
            match_mapping_type: 'string',
            mapping: {
              type: 'text',
              norms: false,
              fields: {
                keyword: {
                  type: 'keyword',
                  ignore_above: 256
                }
              }
            }
          }
        }
      ]
    }
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "strings_as_keywords": {
          "match_mapping_type": "string",
          "mapping": {
            "type": "text",
            "norms": false,
            "fields": {
              "keyword": {
                "type": "keyword",
                "ignore_above": 256
              }
            }
          }
        }
      }
    ]
  }
}

The sub keyword field appears in this template to be consistent with the default rules of dynamic mappings. Of course if you do not need them because you don’t need to perform exact search or aggregate on this field, you could remove it as described in the previous section.

Time series

edit

When doing time series analysis with Elasticsearch, it is common to have many numeric fields that you will often aggregate on but never filter on. In such a case, you could disable indexing on those fields to save disk space and also maybe gain some indexing speed:

response = client.indices.create(
  index: 'my-index-000001',
  body: {
    mappings: {
      dynamic_templates: [
        {
          unindexed_longs: {
            match_mapping_type: 'long',
            mapping: {
              type: 'long',
              index: false
            }
          }
        },
        {
          unindexed_doubles: {
            match_mapping_type: 'double',
            mapping: {
              type: 'float',
              index: false
            }
          }
        }
      ]
    }
  }
)
puts response
PUT my-index-000001
{
  "mappings": {
    "dynamic_templates": [
      {
        "unindexed_longs": {
          "match_mapping_type": "long",
          "mapping": {
            "type": "long",
            "index": false
          }
        }
      },
      {
        "unindexed_doubles": {
          "match_mapping_type": "double",
          "mapping": {
            "type": "float", 
            "index": false
          }
        }
      }
    ]
  }
}

Like the default dynamic mapping rules, doubles are mapped as floats, which are usually accurate enough, yet require half the disk space.