pyrkafka 0.2.1

Fast Python Kafka client powered by Rust and librdkafka — producer and consumer with native performance

pyrkafka: Fast Python Kafka Client Powered by Rust

pyrkafka is a fast, lightweight Python client for Apache Kafka, built on top of librdkafka via the Rust rdkafka crate. It provides a simple, Pythonic API for producing and consuming Kafka messages with native Rust performance through PyO3.

Features

• Rust-powered performance — wraps librdkafka through Rust, avoiding the overhead of pure-Python implementations
• GIL release during polling — consumer polling releases the Python GIL, allowing other threads to run while waiting for messages
• Simple Pythonic API — producer and consumer with familiar Python patterns (iterator protocol for consuming)
• Full librdkafka configuration — pass any librdkafka config option via a dict
• SSL and compression built-in — compiled with SSL (vendored OpenSSL) and zstd compression support
• Cross-platform wheels — pre-built for Linux (x86_64, aarch64), macOS (x86_64, Apple Silicon), and Windows

Installation

pip install pyrkafka

Requires Python 3.13+. Pre-built wheels are available for all major platforms — no Rust toolchain needed for installation.

Getting Started

Producing Messages

from pyrkafka import PyrKafkaProducer

producer = PyrKafkaProducer("localhost:9092")

# Send a message (partitioned round-robin)
producer.produce("my_topic", b"Hello, Kafka!")

# Send with a key (messages with the same key go to the same partition)
producer.produce_with_key("my_topic", b"Hello, Kafka!", "my_key")

# Flush pending messages (also happens automatically when the producer is dropped)
producer.flush()

Consuming Messages

from pyrkafka import PyrKafkaConsumer

consumer = PyrKafkaConsumer("localhost:9092", "my_topic", "my_group")

for message in consumer:
    print(message.decode())

Custom Configuration

Both producer and consumer accept an optional config dict for additional librdkafka configuration:

producer = PyrKafkaProducer("localhost:9092", config={
    "message.timeout.ms": "5000",
    "compression.type": "zstd",
})

consumer = PyrKafkaConsumer("localhost:9092", "my_topic", "my_group", config={
    "auto.offset.reset": "latest",
    "enable.auto.commit": "false",
})

License

MIT