Table of Contents

Transform Specs

Transform specs allow Apache Druid (incubating) to filter and transform input data during ingestion.


The syntax for the transformSpec is shown below:

"transformSpec": {
  "transforms: <List of transforms>,
  "filter": <filter>
property description required?
transforms A list of transforms to be applied to input rows. no
filter A filter that will be applied to input rows; only rows that pass the filter will be ingested. no


The transforms list allows the user to specify a set of column transformations to be performed on input data.

Transforms allow adding new fields to input rows. Each transform has a "name" (the name of the new field) which can be referred to by DimensionSpecs, AggregatorFactories, etc.

A transform behaves as a "row function", taking an entire row as input and outputting a column value.

If a transform has the same name as a field in an input row, then it will shadow the original field. Transforms that shadow fields may still refer to the fields they shadow. This can be used to transform a field "in-place".

Transforms do have some limitations. They can only refer to fields present in the actual input rows; in particular, they cannot refer to other transforms. And they cannot remove fields, only add them. However, they can shadow a field with another field containing all nulls, which will act similarly to removing the field.

Note that the transforms are applied before the filter.

Expression Transform

Druid currently supports one kind of transform, the expression transform.

An expression transform has the following syntax:

  "type": "expression",
  "name": <output field name>,
  "expression": <expr>
property description required?
name The output field name of the expression transform. yes
expression An expression that will be applied to input rows to produce a value for the transform's output field. no

For example, the following expression transform prepends "foo" to the values of a page column in the input data, and creates a fooPage column.

      "type": "expression",
      "name": "fooPage",
      "expression": "concat('foo' + page)"


The transformSpec allows Druid to filter out input rows during ingestion. A row that fails to pass the filter will not be ingested.

Any of Druid's standard filters can be used.

Note that the filtering takes place after the transforms, so filters will operate on transformed rows and not the raw input data if transforms are present.

For example, the following filter would ingest only input rows where a country column has the value "United States":

"filter": {
  "type": "selector",
  "dimension": "country",
  "value": "United States"