Process Kafka streams with Clojure Transducers
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Boilerplate reduction for processing Kafka streams with Clojure Transducers


Clojars Project

    (:require [josef.consumer :as j]))

(defn msg->String
  (String. (.message m)))

(defn more-than-four
  (< 4 (count s)))

(defn -main
  [& args]
  (let [cnsmr (j/consumer "localhost:2181" "whatever") ;; get a Kafka consumer
        topic-pattern "test.*"                         ;; specify all of our test topics
        xf (comp (map msg->String)                     ;; compose our processing transducer xform
                 (filter more-than-four)
                 (map clojure.string/upper-case)
                 (partition-all 5))                     ;; and process 5 at a time
        rf println]                                     ;; and println out the results
    (j/process-stream! cnsmr topic-pattern xf rf)))


josef is a simple library for processing Kafka streams with Clojure transducers.

I had a bunch of similar but ultimately distinct filters that needed to process the same Kafka topic. Clojure transducers are a great way to compose various pieces and build transformations and filters, but I found myself writing the same boilerplate otherwise. After two, I broke it out into a separate, more general function. After a few more, I figured it was time for a little library.


Kafka streams are Iterable, which allows Clojure transducers to slice and dice them the same way you might use any other Clojure collection. josef wraps up some of the details and allows you to call a single function to create a long running Kafka processor with minimal effort. I've also included a simplistic Kafka Producer implementation as consuming and then producing some result is a fairly common paradigm. If you need more control over the producer, however, please see Paul Ingle's much more complete clj-kafka (without which josef would not have been possible). See the example application for a usage example of the included producer.

As a bonus, if for some reason you'd like access to the Kafka message streams themselves for other reasons, josef also exposes the functions used by process-stream! to access the raw Kafka message stream instances.

josef is named after Josef K. in Franz Kafka's "The Trial", or, it's original title in German, "Der Process" ... witty, no? Yeah, a bit of a stretch.


josef has been an idea for a few months now but has only been real code for a couple of weeks. There are almost certainly bugs, inconsistencies in the API, and spots where performance can be improved (type hints, etc). Use with caution. Suggestions and pull requests are welcome.


  • Tests


Copyright © 2015 Josh Rotenberg

Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.