Changelog¶

[0.0.5] - July 30, 2024¶

register_topic() and valid_topic() methods moved into JMqtt class, where they belong, for proper object-oriented design.
Added another Shelly Plus 1 device into the myhome.py to control the heating contactor of the hot water boiler.

The cls.initialize_class() method now accepts a boolean parameter to control whether the class attempts to initialize its attributes from the configuration file. The default value is True. This update eliminates the need to create a configuration file for classes whose attributes are fully determined by other means, such as being c onfigured by another class.
The on_message() method parameter has been typed to eliminate errors with linters.
The Base class now imports MQTTMessage as MqttMsg. This is a temporary solution, but it moves the code closer to a proper abstraction that allows any MQTT implementation to be plugged in.
The LogRecord class has been registered for recording application events, such as warnings and errors, to a time series database.
Additional classes have been documented, albeit to varying degrees.
Mqtt class introduces topic registration feature, where classes that introduce new topics can register them as known valid topics. All calls to subscribe() and publish() methods check if the topic in question is registered as valid known topic. If not ValueError exception is thrown.
RVisualCrossing class reports an error if the APIKEY needed to acquire weather forecast data has not been configured in its “config/RVisualCrossing.json” file.

Bug fixes:

Setup issue: 0.0.2 required Python 3.12 for type hinting, and for _class_id : str| None = None to work. As all my attempts to upgrade my workstations to use the new python to 3.12 failed miserably, code base reverted back to Python 3.8.
Wrong signature of JApp.configure(), fixed.
JApp.register() comparison if cls.get_class_id == None fixed to if cls.get_class_id() is None
Several == None comparisons changed to ‘ is None’ to make pylint happy (need to figure out what this is about)
wrong import ordering fixed to satisfy pylint (built-in packages first, 3rd party next, juham last)
Bug in acquiring solar energy forecast from Visual Crossing fixed.

Key updates include:

Extended Shelly device support to include the ‘Shelly Motion Sensor’, which helps determine if anyone is at home and controls the hot water circulator pump accordingly.
Implemented several simulation classes for testing environments without actual sensor data:
- Shelly Motion Sensor Simulator
- Shelly Plus 1 Simulator
Replaced the Doxygen (C++) documentation tool with Python’s native Sphinx.
Fully implemented the plugin architecture; all existing classes are now genuine plugins, even though they are deployed in a single package.
Introduced a serialization system based on JSON format.
Reformatted Python code using the ‘black’ code formatter.
Automated class initialization e.g. calling of cls.load_json() and atexit.register() methods upon import.
Shortened long imports (e.g., from juham.shelly.jshellymotion import ShellyMotion) using __init__.py, resulting in cleaner imports (e.g., from juham.shelly import ShellyMotion).
Updated filename and classname conventions from C-style to Pythonic (e.g., JShellyMotion to ShellyMotion).
Added support for multiple configurations via startup arguments: –configuration-folder [path]. This allows running the same automation app on different sites or for various purposes (e.g., production, testing).
Started transition to Python type annotations. Installed mypy and wooah, I’w missed type checking so much!
Adopted f-string convention, e.g. self.log(f”Forecast for {days} days acquired”).
UML class diagrams.
[snip]

Private in-house releases, not published to the Python repository — for everyone’s sake.

Added support for more devices:
- Shelly PRO Energy Meter
- HomeWizard Water Meter
Introduced the class factory method pattern.
Added class-specific JSON configuration files, allowing automation classes to be configured instead of hardcoding settings in Python.
More refined class hierarchy, for minimizing code/functionality ratio.
First unit tests added
[snip]

Initial release featuring:

A functioning system
Some unconventional use of the Python programming language
Mosquitto MQTT via Paho
InfluxDB V3 time series recording
Python package installable via pip, based on pyproject.toml
MIT license
README, CHANGELOG, and other standard files in .rst format
A cool project name, ‘Juham™’, with a note: M currently stands for mission rather than masterpiece
Initial support for Shelly WiFi relays with temperature sensors
Some docstrings and related developer documentation generated with Doxygen