Access fields in a document with the field API

Use the field API to access document fields:

field('my_field').get(<default_value>)

This API fundamentally changes how you access documents in Painless. Previously, you had to access the doc map with the field name that you wanted to access:

doc['my_field'].value

Accessing document fields this way didn’t handle missing values or missing mappings, which meant that to write robust Painless scripts, you needed to include logic to check that both fields and values exist.

Instead, use the field API, which is the preferred approach to access documents in Painless. The field API handles missing values, and will evolve to abstract access to _source and doc_values.

The field API returns a Field object that iterates over fields with multiple values, providing access to the underlying value through the get(<default_value>) method, as well as type conversion and helper methods.

The field API returns the default value that you specify, regardless of whether the field exists or has any values for the current document. This means that the field API can handle missing values without requiring additional logic. For a reference type such as keyword, the default value can be null. For a primitive type such as boolean or long, the default value must be a matching primitive type, such as false or 1.

Instead of explicitly calling the field API with the get() method, you can include the $ shortcut. Just include the $ symbol, field name, and a default value, in case the field doesn’t have a value:

$(‘field’, <default_value>)

With these enhanced capabilities and simplified syntax, you can write scripts that are shorter, less complex, and easier to read. For example, the following script uses the outdated syntax to determine the difference in milliseconds between two complex datetime values from an indexed document:

if (doc.containsKey('start') && doc.containsKey('end')) {
   if (doc['start'].size() > 0 && doc['end'].size() > 0) {
       ZonedDateTime start = doc['start'].value;
       ZonedDateTime end = doc['end'].value;
       return ChronoUnit.MILLIS.between(start, end);
   } else {
       return -1;
   }
} else {
   return -1;
}

Using the field API, you can write this same script much more succinctly, without requiring additional logic to determine whether fields exist before operating on them:

ZonedDateTime start = field('start').get(null);
ZonedDateTime end = field('end').get(null);
return start == null || end == null ? -1 : ChronoUnit.MILLIS.between(start, end)

The following table indicates the mapped field types that the field API supports. For each supported type, values are listed that are returned by the field API (from the get and as<Type> methods) and the doc map (from the getValue and get methods).