This provides a library called thoth-messaging used in project Thoth. It is a basic module to encapsule all messaging (here it is Kafka via Faust) primitives.
This module uses pydantic for type hinting and enforcing a regular schema in messaging. If you are using mypy to check your code please add the following to your mypy configuration file:
[mypy]
plugins = pydantic.mypy
If you are creating an instance of a pydantic model in your own module you should directly use the MessageContents
class within the associate message file. Using foo_bar_message.model
will only type hint for BaseMessageContents.
For development and testing it is very useful to have a local instance of Kafka running on your machine
We provide a docker-compose
file to get you up and running quickly with a basic Kafka server; this file is based on
Single Zookeeper/Multiple Kafka.
In order to start Zookeeper as well as the Kafka Servers simply run $ podman-compose up or $ docker-compose up, choose the appropriate option based on the system which you are using.
Once you have Kafka up and running you should be ready to begin coding your own messaging producers and consumers. The interface between Kafka and Python is handled by a library called Confluent Kafka. Faust's documentation will be extremely helpful to you when you are developing your own applications. If you would like examples of producers and consumers from Team Thoth, look at the following two repositories, investigator and package-update.
You may find it useful to use console producers and consumers while testing your, to create one simply attach a bash shell to one of your Kafka Servers by running: $ podman exec -it messaging_kafka1_1 bash, your container names should be the same as given here, if not, run $ podman ps and choose the correct container. These containers have all Kafka binaries in appropriate places so you can simply run $ kafka-console-consumer, $ kafka-console-producer, or any other kafka command that you may find useful.
example:
kafka-console-consumer --bootstrap-server localhost:9092 --topic test --from-beginning
You can test sending a message via the CLI using a file as -
example:
pipenv shell
python cli.py --message-file messages_to_be_sent.json
Note Data is not persistent. Once pods are deleted so is the data associated with them.
Some schemas in this library are defined as Dict[str, Any]. This usually does not accurately reflect the actual schema required. These schemas can be purposefully vague because they are defined elsewhere and importing them from the requisite libraries couples thoth-messaging version too closely to other components in the best case, and introduces circular dependencies in the worst case. In short, thoth-messaging is naive and can only enforce schemas which it has defined.