Quix Streams

3.23.6 verified Fri May 01 auth: no python

Quix Streams is a Python library for building stream processing applications with Apache Kafka. Version 3.23.6 supports stateful operations, windowing, connectors, and Quix Cloud integration. Requires Python >=3.9, <4. Release cadence is active with frequent minor releases.

pip install quixstreams

Common errors

error ModuleNotFoundError: No module named 'confluent_kafka' ↓

cause Missing Kafka C library or Python bindings.

fix

Install confluent-kafka: pip install quixstreams[kafka] or separately: pip install confluent-kafka

error AttributeError: 'Application' object has no attribute 'run' ↓

cause Using a very old version (<0.5) where the method was named differently.

fix

Upgrade to latest: pip install --upgrade quixstreams

error ValueError: Topic 'my-topic' does not exist and auto_create_topics is False ↓

cause Kafka topic must be created beforehand or auto-creation must be enabled.

fix

Set auto_create_topics=True in Application() or create the topic manually via Kafka admin.

error TypeError: 'NoneType' object is not callable in StreamingDataFrame.apply() ↓

cause The apply function returned None instead of a value.

fix

Ensure your apply function returns a value (can be the input row modified).

Warnings

breaking In v3.22.0, the default for `quix_portal_api` URL was removed. You must explicitly set it when connecting to Quix Cloud. ↓

fix Set the `quix_portal_api` parameter in Application or use the corresponding environment variable.

deprecated The `lookup.quix_configuration_service` function is deprecated; use `app.quix_config_service` instead. ↓

fix Replace function-based configuration lookup with the method on the Application instance.

gotcha When using `StreamingDataFrame.to_topic()`, the callable must return a `Topic` object, not a topic name string. ↓

fix Ensure your routing callable returns `app.topic('destination-topic')`.

gotcha State stores are local and not automatically replicated; replaying a stream from a different consumer group will start with empty state. ↓

fix Design state to be idempotent or use external storage for persistence.

Install

pip install quixstreams[kafka]

Imports

Application

wrong

from quixstreams.app import Application

correct

from quixstreams import Application

Application is a top-level export since v3.0

StreamingDataFrame

wrong

from quixstreams.dataframe import StreamingDataFrame

correct

from quixstreams import StreamingDataFrame

Correct import is from quixstreams directly

Topic

wrong

from quixstreams.topic import Topic

correct

from quixstreams import Topic

Topic is re-exported at package level

Quickstart

Minimal stream processing app that increments a counter for each message.

from quixstreams import Application
import os

app = Application(broker_address=os.environ.get('KAFKA_BROKER', 'localhost:9092'))
topic = app.topic('my-topic', value_deserializer='json')

@app.dataframe(topic)
def process(df):
    df = df.apply(lambda row: {'count': row.get('count', 0) + 1})
    return df

if __name__ == '__main__':
    app.run()