rmoff's random ramblings
about talks

๐ŸŽ„ Twelve Days of SMT ๐ŸŽ„ - Day 12: Community Transformations

Published Dec 23, 2020 by in Kafka Connect, Single Message Transform, TwelveDaysOfSMT at https://rmoff.net/2020/12/23/twelve-days-of-smt-day-12-community-transformations/

Apache Kafka ships with many Single Message Transformations (SMT) included - but the great thing about it being an open API is that people can, and do, write their own transformations. Many of these are shared with the wider community, and in this final installment of the series Iโ€™m going to look at some of the transformations written by Jeremy Custenborder and available in kafka-connect-transform-common which can be downloaded and installed from Confluent Hub (or built from source, if you like that kind of thing). Also check out the XML transformation by the same author, which Iโ€™ve written about previously.

Other cool community SMT

Here are some other Single Message Transform that I discovered after making this video, and are well worth a look:

  • kafka-connect-transform-kryptonite - field-level encryption/decryption of records. Uses authenticated encryption with associated data (AEAD) and in particular applies AES in GCM mode.

  • kafka-connect-transform-tojsonstring - takes the entire key or value record and transforms it to a new record which contains exactly one field with a JSON representation of the origin record.

  • kafka-connect-jmespath - check record keys or values against a JMESPath query expressions. Can be used to filter records or transform them conditionally.

  • kafka-connect-transform-grok - for parsing unstructured data text into a data Struct from the entire key or value using a Grok expression.

Change the topic case

๐Ÿ‘‰ Reference

curl -i -X PUT -H "Accept:application/json" \
  -H  "Content-Type:application/json" http://localhost:8083/connectors/sink-jdbc-mysql-day12-00/config \
  -d '{
      "connector.class"          : "io.confluent.connect.jdbc.JdbcSinkConnector",
      "connection.url"           : "jdbc:mysql://mysql:3306/demo",
      "connection.user"          : "mysqluser",
      "connection.password"      : "mysqlpw",
      "topics"                   : "day12-sys01",
      "tasks.max"                : "4",
      "auto.create"              : "true",
      "auto.evolve"              : "true",

      "transforms"               : "topicCase",
      "transforms.topicCase.type": "com.github.jcustenborder.kafka.connect.transform.common.ChangeTopicCase",
      "transforms.topicCase.from": "LOWER_HYPHEN",
      "transforms.topicCase.to"  : "UPPER_CAMEL"
      }'

The source topic name of day12-sys01 gets modified to Day12Sys01:

mysql> show tables;
+----------------+
| Tables_in_demo |
+----------------+
| Day12Sys01     |

Add the timestamp of a field to the topic name

A nice little triumvirate of transformations here, which use the timestamp in a field of a message to modify the topic name.

The three steps are:

  1. TimestampConverter to transform the field from a string to a Timestamp (not necessary if it already is). Ships with Apache Kafka.

  2. ExtractTimestamp to set the timestamp of the Kafka message to the value of the specified field. Custom SMT from Jeremy Custenborder.

  3. TimestampRouter to modify the topic name to include the timestamp components required. Ships with Apache Kafka.

curl -i -X PUT -H "Accept:application/json" \
  -H  "Content-Type:application/json" http://localhost:8083/connectors/sink-jdbc-mysql-day12-01/config \
  -d '{
      "connector.class"                        : "io.confluent.connect.jdbc.JdbcSinkConnector",
      "connection.url"                         : "jdbc:mysql://mysql:3306/demo",
      "connection.user"                        : "mysqluser",
      "connection.password"                    : "mysqlpw",
      "topics"                                 : "day12-sys01",
      "tasks.max"                              : "4",
      "auto.create"                            : "true",
      "auto.evolve"                            : "true",

      "transforms"                              : "convertTS,extractTS,setTopicName",
      "transforms.convertTS.type"               : "org.apache.kafka.connect.transforms.TimestampConverter$Value",
      "transforms.convertTS.field"              : "txn_date",
      "transforms.convertTS.format"             : "EEE MMM dd HH:mm:ss zzz yyyy",
      "transforms.convertTS.target.type"        : "Timestamp",
      "transforms.extractTS.type"               : "com.github.jcustenborder.kafka.connect.transform.common.ExtractTimestamp$Value",
      "transforms.extractTS.field.name"         : "txn_date",
      "transforms.setTopicName.type"            : "org.apache.kafka.connect.transforms.TimestampRouter",
      "transforms.setTopicName.topic.format"    : "${topic}_${timestamp}",
      "transforms.setTopicName.timestamp.format": "YYYY-MM-dd"
      }'

Resulting topic takes the date from the message field txn_date and generates table names accordingly:

mysql> show tables;
+------------------------+
| Tables_in_demo         |
+------------------------+
| day12-sys01_2020-12-07 |
| day12-sys01_2020-12-08 |
| day12-sys01_2020-12-09 |
| day12-sys01_2020-12-10 |
| day12-sys01_2020-12-11 |
| day12-sys01_2020-12-12 |
| day12-sys01_2020-12-13 |
| day12-sys01_2020-12-14 |
| day12-sys01_2020-12-15 |
| day12-sys01_2020-12-16 |
+------------------------+
12 rows in set (0.01 sec)

Add the current timestamp to the message payload

๐Ÿ‘‰ Reference

curl -i -X PUT -H "Accept:application/json" \
  -H  "Content-Type:application/json" http://localhost:8083/connectors/sink-jdbc-mysql-day12-02/config \
  -d '{
      "connector.class"          : "io.confluent.connect.jdbc.JdbcSinkConnector",
      "connection.url"           : "jdbc:mysql://mysql:3306/demo",
      "connection.user"          : "mysqluser",
      "connection.password"      : "mysqlpw",
      "topics"                   : "day12-sys01",
      "tasks.max"                : "4",
      "auto.create"              : "true",
      "auto.evolve"              : "true",

      "transforms"                : "addTSNow",
      "transforms.addTSNow.type"  : "com.github.jcustenborder.kafka.connect.transform.common.TimestampNowField$Value",
      "transforms.addTSNow.fields": "processingTS"
      }'
mysql> select product, amount, txn_date, processingTS from `day12-sys01` ORDER BY units  LIMIT 5;
+------------------------------+--------+------------------------------+-------------------------+
| product                      | amount | txn_date                     | processingTS            |
+------------------------------+--------+------------------------------+-------------------------+
| Sublimely Self-Righteous Ale | 61.25  | Mon Dec 14 09:12:03 GMT 2020 | 2020-12-17 00:43:02.550 |
| Arrogant Bastard Ale         | 88.65  | Wed Dec 09 18:05:02 GMT 2020 | 2020-12-17 00:43:02.559 |
| Sublimely Self-Righteous Ale | 30.81  | Fri Dec 11 14:49:14 GMT 2020 | 2020-12-17 00:43:02.551 |
| Arrogant Bastard Ale         | 20.45  | Tue Dec 08 10:30:21 GMT 2020 | 2020-12-17 00:43:02.223 |
| Sublimely Self-Righteous Ale | 56.95  | Wed Dec 16 23:12:23 GMT 2020 | 2020-12-17 00:43:02.233 |
+------------------------------+--------+------------------------------+-------------------------+
5 rows in set (0.00 sec)

Using SimulatorSinkConnector (and Single Message Transform TRACE logging)

Not a transformation as such, but a useful tip for examining the output of Transforms without needing to route the data to an actual target:

curl -i -X PUT -H  "Content-Type:application/json" \
    http://localhost:8083/connectors/sink-simulator-day12-02/config \
    -d '{
        "connector.class"           : "com.github.jcustenborder.kafka.connect.simulator.SimulatorSinkConnector",
        "topics"                    : "day12-sys01",
        "log.entries"               : "true",
        "transforms"                : "addTSNow",
        "transforms.addTSNow.type"  : "com.github.jcustenborder.kafka.connect.transform.common.TimestampNowField$Value",
        "transforms.addTSNow.fields": "processingTS"
    }'

You can see the message after itโ€™s been processed by the transform(s) in the Kafka Connect worker log:

[2020-12-18 00:29:59,651] INFO [sink-simulator-day12-02|task-0] record.value=Struct{units=39,product=Delirium Tremens,amount=32.60,txn_date=Wed Dec 16 07:27:19 GMT 2020,source=SYS01,processingTS=Fri Dec 18 00:29:59 GMT 2020} (com.github.jcustenborder.kafka.connect.simulator.SimulatorSinkTask:50)

You can also get the Kafka Connect runtime to log TRACE messages that show the source messages before a transformation (c.f. Changing the Logging Level for Kafka Connect Dynamically):

curl -s -X PUT -H "Content-Type:application/json" \
    http://localhost:8083/admin/loggers/org.apache.kafka.connect.runtime.TransformationChain \
    -d '{"level": "TRACE"}' \
    | jq '.'

With that set the Kafka Connect worker then logs the record before it is transformed, and then from the SimulatorSink its state after transform:

[2020-12-18 00:31:54,572] TRACE [sink-simulator-day12-02|task-0] Applying transformation
com.github.jcustenborder.kafka.connect.transform.common.TimestampNowField$Value to
SinkRecord{kafkaOffset=121, timestampType=CreateTime} ConnectRecord{topic='day12-sys01',
kafkaPartition=0, key=fd403528-90c3-45a1-a1c5-3f9ebe2799be, keySchema=Schema{STRING},
value=Struct{units=6,product=Nugget Nectar,amount=91.30,txn_date=Thu Dec 10 06:51:22 GMT
2020,source=SYS01}, valueSchema=Schema{io.mdrogalis.Gen0:STRUCT}, timestamp=1608251514568,
headers=ConnectHeaders(headers=)} (org.apache.kafka.connect.runtime.TransformationChain:47)

[2020-12-18 00:31:54,572] INFO [sink-simulator-day12-02|task-0]
record.value=Struct{units=6,product=Nugget Nectar,amount=91.30,txn_date=Thu Dec 10 06:51:22 GMT
2020,source=SYS01,processingTS=Fri Dec 18 00:31:54 GMT 2020}
(com.github.jcustenborder.kafka.connect.simulator.SimulatorSinkTask:50)

Try it out!

You can find the full code for trying this outโ€”including a Docker Compose so you can spin it up on your local machineโ€” ๐Ÿ‘พ here


Robin Moffatt

Robin Moffatt is a Principal DevEx Engineer at LakeFS. He likes writing about himself in the third person, eating good breakfasts, and drinking good beer.

Story logo

© 2023