py-queue-bus 0.1.1

Python port of queue-bus compatible with node-queue-bus and resque-bus

py-queue-bus

Python implementation of Queue-Bus semantics compatible with node-queue-bus and Ruby resque-bus. Uses redis-py to publish bus events and a rider worker to fan out events to subscribed queues.

Install

pip install py-queue-bus

Project on PyPI: https://pypi.org/project/py-queue-bus/

Basic usage

Publish Events (no rider needed in the publishing process):

from py_queue_bus import Bus

connection = {"host": "127.0.0.1", "port": 6379, "db": 0, "namespace": "resque"}
# Or use a Redis URL (takes precedence over host/port/db if both are provided):
# connection = {"url": "redis://:password@127.0.0.1:6379/0", "namespace": "resque"}

bus = Bus(connection=connection)
bus.connect()
bus.publish("order_created", {"order_id": 1, "total": 10.0})

Subscribe Events:

from py_queue_bus import Bus, Rider

connection = {"host": "127.0.0.1", "port": 6379, "db": 0, "namespace": "resque"}
# Or use a Redis URL (takes precedence over host/port/db if both are provided):
# connection = {"url": "redis://:password@127.0.0.1:6379/0", "namespace": "resque"}
app_key = "example_service"
priority = "default"
queue = f"{app_key}_{priority}"

bus = Bus(connection=connection)
bus.connect()

# Define handlers
def order_created_handler(payload):
    print("order_created_handler received:", payload)

def heartbeat_handler(payload):
    print("heartbeat_handler received:", payload)

jobs = {
    "order_created_job": order_created_handler,
    "heartbeat_job": heartbeat_handler,
}

# Subscribe
# Note: job name can differ from handler function name and event name; it must exist in the jobs dict
bus.subscribe(app_key, priority, "order_created_job", {"bus_event_type": "order_created"})
bus.subscribe(app_key, priority, "heartbeat_job", {"bus_event_type": "heartbeat_minutes"}) # See Heartbeat section below.

rider = Rider(connection=connection, jobs=jobs, queues=[queue], to_drive=True)
rider.connect()
rider.start()  # blocking worker

Scheduling (RQ)

• publish_at / publish_in enqueue scheduled publishes using RQ. Run a worker:

  rq worker --with-scheduler queue_bus_schedule # defaults to localhost
  # if you use a Redis URL, pass it explicitly:
  export REDIS_URL=redis://:password@host:port/db
  rq worker --with-scheduler --url "$REDIS_URL" queue_bus_schedule
  

• Use the official RQ CLI (as above) under a supervisor (systemd/docker/pm2) so it restarts if Redis drops the connection.
• Note: Scheduling is Python-native via RQ. Node/Ruby schedulers (resque-scheduler/node-resque) won’t see Python-scheduled jobs; they only see fired jobs after RQ publishes them.

Heartbeat

• publish_heartbeat() emits a QueueBus::Heartbeat job; riders emit heartbeat_minutes once per minute (with Redis locking) for cron-like tasks.

Examples

• Subscriber + Rider: py-queue-bus/examples/subscriber_service.py
• Publisher: py-queue-bus/examples/publisher_service.py
• RQ worker stub: py-queue-bus/examples/rq_worker.py
• Tip: In production, keep handlers in a folder and import the jobs dict into your rider entrypoint, e.g.:

  app/
    subscribers/
      __init__.py  # exports jobs = {"my_job": handler}
      order_handlers.py
    worker.py      # from subscribers import jobs; Rider(..., jobs=jobs)
  

Tests

• Unit tests (pytest): pytest py-queue-bus/test
• Microservice harnesses (manual cross-language checks):
  • Python service scripts: py-queue-bus/test/python_service
  • Node service scripts (npm node-queue-bus): py-queue-bus/test/node_service
  • Payloads include ts and explicit logs for debugging.

Compatibility notes

• Redis schema matches node/resque (queues, subscription hashes/sets).
• Payload metadata matches node/resque-bus, so Node/Ruby workers can consume events published here and vice versa.