kafka-connect-protobuf-converter

Converter plugin for Kafka Connect. A converter controls the format of the data that will be written to Kafka for source connectors or read from Kafka for sink connectors.

Compatibility

The 2.x release series is compatible with Kafka Connect 5.x. and up (older releases have been validated to work all the way back to Kafka Connect 3.2.0, although we highly recommend updating to the latest version).

N.B. This converter only works with proto3 protocol buffer schemas.

Usage

Copy the kafka-connect-protobuf-converter jar and the jar containing your compiled protocol buffers to /usr/share/java/kafka-serde-tools on your Kafka Connect instance and restart Kafka Connect.

Converters can be specified on a per-connector basis.

To use the protobuf converter in Kafka Connect, specify the converter as your key and value converter and specify the protocol buffer class you want to use to deserialize the message (ex: com.google.protobuf.Int32Value).

Note: Nested classes must be specified using the $ notation, for example com.blueapron.connect.protobuf.NestedTestProtoOuterClass$NestedTestProto

Example Kafka Connect JDBC source:

connector.class=io.confluent.connect.jdbc.JdbcSourceConnector
mode=bulk
topic.prefix=blueapron-protos-int
tasks.max=1
query=select * from int32value
name=blueapron-protos-source-int
connection.url=jdbc:postgresql://192.168.99.100:5432/test?user=postgres&password=mysecretpassword
value.converter=com.blueapron.connect.protobuf.ProtobufConverter
value.converter.protoClassName=com.google.protobuf.Int32Value
key.converter=com.blueapron.connect.protobuf.ProtobufConverter

Example Kafka Connect JDBC sink:

connector.class=io.confluent.connect.jdbc.JdbcSinkConnector
topics=blueapron-protos-int
topics.to.tables=blueapron-protos-int=int32output
auto.create=true
tasks.max=1
name=blueapron-protos-sink-int
connection.url=jdbc:postgresql://192.168.99.100:5432/test?user=postgres&password=mysecretpassword
value.converter=com.blueapron.connect.protobuf.ProtobufConverter
value.converter.protoClassName=com.google.protobuf.Int32Value
key.converter=com.blueapron.connect.protobuf.ProtobufConverter

Supported Conversions

List of proto3 types

From Connect

Connect Schema Type	Protobuf Type
INT32	int32
INT64	int64
FLOAT32	float
FLOAT64	double
BOOLEAN	bool
STRING	string
BYTES	bytes

Connect Logical Type	Protobuf Type
Timestamp	Timestamp
Date	Date

To Connect

Protobuf Type	Connect Schema Type
int32	INT32
int64	INT64
sint32	INT32
sint64	INT64
uint32	INT64
float	FLOAT32
double	FLOAT64
bool	BOOLEAN
string	STRING
bytes	BYTES
repeated	ARRAY
map	ARRAY of STRUCT*

Protobuf Type	Connect Logical Type
Timestamp	Timestamp
Date	Date
uint64	Decimal

* Protobuf map type fields are converted by default to an Array of Struct, but the converter supports also converting these fields into Connect Schema Map type. To enable this conversion mode, set the configuration field protoMapConversionType to map:

value.converter.protoMapConversionType=map

Handling field renames and deletes

Renaming and removing fields is supported by the proto IDL, but certain output formats (for example, BigQuery) do not support renaming or removal. In order to support these output formats, we use a custom field option to specify the original name and keep the Kafka Connect schema consistent.

You can specify the name for this field option using the legacyName configuration item. By default, the field option used is legacy_name

Example: value.converter.legacyName=blueapron.connect.protobuf.legacy_name

This annotation provides a hint that allows renamed fields to be mapped to correctly to their output schema names.

// Before rename:
message TestMessage {
  string test_string = 1;
}

// After first rename:
import "path/to/LegacyName.proto";

message TestMessage {
  string renamed_string = 1 [(blueapron.connect.protobuf.legacy_name) = "test_string"];
}

// After subsequent rename:
import "path/to/LegacyName.proto";

message TestMessage {
  string yet_another_named_string = 1 [(blueapron.connect.protobuf.legacy_name) = "test_string"];
}

We can also use the same syntax to support removing fields. You can treat deleted fields as renamed fields, prefixing the deprecated field name with OBSOLETE_ and including a legacy_name field option as previously detailed.

// Before deprecation:
message TestMessage {
  string test_string = 1;
  int32 test_count = 2;
}

// After deprecation:
import "path/to/LegacyName.proto";

message TestMessage {
  string OBSOLETE_test_string = 1 [(blueapron.connect.protobuf.legacy_name) = "test_string"];
  int32 test_count = 2;
}

Contributing

Contributions are warmly welcomed. If fixing a bug, please open an issue on the project before opening a PR. This helps with documentation for other users and will make code review go quicker 🙌

Development

Run tests:

mvn test

Create the JAR:

mvn clean package

Copy the JAR with dependencies (kafka-connect-protobuf-converter-*-jar-with-dependencies.jar) to /usr/share/java/kafka-serde-tools on your local Kafka Connect instance to make the converter available in Kafka Connect.

Deploy to Maven Central:

Note: only internal Blue Apron employees can deploy to Maven Central

1. Update the version to include -SNAPSHOT
2. Prepare your local environment for the deploy. If you already have a custom ~/.m2/settings.xml, you may want to back up or merge your existing file rather than overwrite it.

   cp settings.xml ~/.m2/settings.xml
   export GPG_PASSPHRASE=
   export OSSRH_JIRA_USERNAME=
   export OSSRH_JIRA_PASSWORD=
   

3. Confirm the tests run

   mvn test
   mvn release:clean
   

4. Prepare the release

   mvn release:prepare
   

5. Perform the release

   mvn release:perform
   

Generate protos:

Run the below command from the repository root.
```
protoc -I=src/test/java/com/blueapron/connect/protobuf/protos --java_out=src/test/java/ src/test/java/com/blueapron/connect/protobuf/protos/*.proto
```