landoop / kafka-connect-kcql-smt

Kafka-Connect SMT (Single Message Transformations) with SQL syntax (Using Apache Calcite for the SQL parsing)


Build Status

Kafka Connect Sql Single Message Transform

Use SQL to drive the transformation of the Kafka message(key or/and value) when using Kafka Connect. Before SMT you needed a KStream app to take the message from the source topic apply the transformation to a new topic. We have developed a KStreams library ( you can find on github) to make it easy expressing simple Kafka streams transformations.

However the extra topic is not required anymore for Kafka Connect!.


Sources or sinks might produce/deal-with data that is not in sync with what you want:

  • you have a kafka topic where you want to pick up specific fields for the sink
  • you might want to flatten the message structure
  • you might want to rename fields
  • (coming soon) might want to filter messages

And you want to express it with a simple syntax! This is where SQL SMT comes to help you!


Configuration Type Required Description
connect.transforms.sql.key String N Comma separated SQL targeting the key of a Kafka Message
connect.transforms.sql.value String N Comma separated SQL targeting the value of a Kafka Message

The SQL will define the mapping between the topic and the transformation to be applied. Each message on the specified topics will get the appropriate transformation.

Example configuration

connect.transforms.sql.value=SELECT as fieldName,ingredients.*, ingredients.sugar as fieldSugar FROM topic1 withstructure;SELECT name, as streetName, as streetName2 FROM topic2

Kafka Connect Payloads supported

In most cases the payload sent over Kafka is Avro. That might not always be the case for existing systems where in most cases the payload is json. As a result the transform is capable of handling more than just Avro. Supported payload type (applies to both key and value):

Schema Type Input Schema Output Output
Type.STRUCT Struct Type Struct Struct
Type.BYTES Json (byte[]) Schema.Bytes Json(byte[])
Type.STRING Json(string) Schema.STRING Json (string)
NULL Json (byte[]) NULL Json (byte[])
NULL Json (string) NULL Json (string)


We make use of Apache Calcite to handle the SQL parsing. The library support for SQL is quite large but for now we only handle simple SQL identifiers (nested structure is supported) with more to come like: WHERE condition and probably SQL operation(field concatenation for example) Syntax:


There are two modes for the SQL when it comes to Kafka Connect SMT

  • flatten the structure. General syntax is like this:SELECT ... FROM TOPIC_A
//rename and only pick fields on first level
SELECT calories as C ,vegan as V ,name as fieldName FROM topic

//Cherry pick fields on different levels in the structure
SELECT name, as streetName FROM topic

//Select and rename fields on nested level
SELECT name, address.street.*, as streetName2 FROM topic
  • retain structure. Syntax looks like SELECT ... FROM TOPIC_A WITHSTUCTURE. Notice the WITHSTRUCTRE keyword.
//you can select itself - obviously no real gain on this
SELECT * FROM topic withstructure 

//rename a field 
SELECT *, name as fieldName FROM topic withstructure

//rename a complex field
SELECT *, ingredients as stuff FROM topic withstructure

//select a single field
SELECT vegan FROM topic withstructure

//rename and only select nested fields
SELECT as fieldName, ingredients.sugar as fieldSugar, ingredients.* FROM topic withstructure

Not supported

Applying SQL to value to use the message key fields or metadata. Coming soon!

0.1 (2017-05-16)

  • first release


Requires gradle 3.4.1 to build.

To build

gradle compile

To test

gradle test

You can also use the gradle wrapper

./gradlew build

To view dependency trees

gradle dependencies #