brewblox-service 0.10.1.dev10

No project description provided

# Scaffolding for BrewBlox service applications

In order to reduce code duplication between services, generic functionality is implemented here.

For an example on how to implement your own service based on `brewblox-service`, see <https://github.com/brewblox/brewblox-boilerplate>.

`brewblox-service` can technically be launched as a standalone application, but will not be very useful.

## [brewblox_service](./brewblox_service/__init__.py)

Small generic tools are defined here.

`brewblox_logger` can be used for creating module-specific loggers. It is not required, but will play a bit nicer with default formatting of the log.

Example:

```python
from brewblox_service import brewblox_logger

LOGGER = brewblox_logger(__name__)
LOGGER.info('hello')
```

## [service.py](./brewblox_service/service.py)

Parses commandline arguments, creates an `aiohttp` app, and runs it.

The shortest implementation is:

```python
app = service.create_app(default_name='my_service')
service.furnish(app)
service.run(app)
```

This will get you a working web application, but it will only support the `/_service/status` health check endpoint.

Applications can configure their own features, and add new commandline arguments.

Example:

```python
# Separately creating the parser allows adding arguments to the default set
parser = service.create_parser(default_name='my_service')
parser.add_argument('--my-arg')

# Now create the app
app = service.create_app(parser=create_parser())

# Add features for this service
device.setup(app)
api.setup(app)

# Furnish and run
service.furnish(app)
service.run(app)
```

## [features.py](./brewblox_service/features.py)

Many service features are application-scoped. Their lifecycle should span multiple requests, either because they are not request-driven, or because they manage asynchronous I/O operations (such as listening to AMQP messages).

The `ServiceFeature` class offers an abstract base class for this behavior. Implementing classes should define `startup(app)` and `shutdown(app)` functions, and those will be automatically called when the application starts up and shuts down.

Both `startup()` and `shutdown()` are called in an async context, making them the async counterparts of `__init__()` and `__del__()` functions.

Features must be constructed after the app is created, but before it starts running. (`service.create_app()` and `service.run(app)`)

The `add()` and `get()` functions make it easy to centrally declare a feature, and then use it in any function that has a reference to the aiohttp app.

## [events.py](./brewblox_service/events.py)

Both incoming and outgoing communication with the AMQP eventbus is handled here.

`EventListener` allows subscribing to eventbus messages. It will fire a callback when one is received. Subscriptions can be set at any time (also before the app starts running).

The listener is designed to gracefully degrade when the eventbus can't be reached. No errors will be raised, and it will periodically attempt to reconnect and restore its subscriptions.

For a practical implementation of `EventListener`, see [brewblox_history](https://github.com/BrewBlox/brewblox-history/blob/develop/brewblox_history/influx.py)

`EventPublisher` is responsible for sending new messages to the eventbus. A single publisher per application is sufficient.

In contrast with `EventListener`, the publisher will raise an exception when attempting to publish to an unreachable eventbus host.
It will attempt to reconnect for each subsequent message - no explicit connection management is required.