* Restore states through a JSON store
* Accept entity_id directly in restore state helper
* Keep states stored between runs for a limited time
* Remove warning
* Recreate component if discovery info is changed
* Update component instead of remove+add
* Set name and unique_id in __init__
* Update unit test
* Cleanup
* More cleanup
* Refactor according to review comments
* Change discovery_hash
* Review comments, add tests
* Fix handling of value_template
## Description:
More fixes flagged by pylint 2 that don't hurt to have before the actual pylint 2 upgrade (which I'll submit soon).
## Checklist:
- [ ] The code change is tested and works locally.
- [x] Local tests pass with `tox`. **Your PR cannot be merged unless tests pass**
* MQTT Typing
* Tiny style change
* Fixes
I should've probably really sticked to limiting myself to static typing...
* Small fix 😩
Ok, this seriously shouldn't have happened.
* Fix MQTT retained message not being re-dispatched
* Fix tests
* Use paho-mqtt for retained messages
* Improve code style
* Store list of subscribers
* Fix lint error
* Adhere to Home Assistant's logging standard
"Try to avoid brackets and additional quotes around the output to make it easier for users to parse the log."
- https://home-assistant.io/developers/development_guidelines/
* Add reconnect tests
* Fix lint error
* Introduce Subscription
Tests still need to be updated
* Use namedtuple for MQTT messages
... And fix issues
Accessing the config manually at runtime isn't ideal
* Fix MQTT __init__.py tests
* Updated usage of Mocks
* Moved tests that were testing subscriptions out of the MQTTComponent test, because of how mock.patch was used
* Adjusted the remaining tests for the MQTT clients new behavior - e.g. self.progress was removed
* Updated the async_fire_mqtt_message helper
* ✅ Update MQTT tests
* Re-introduce the MQTT subscriptions through the dispatcher for tests - quite ugly though... 🚧
* Update fixtures to use our new MQTT mock 🎨
* 📝 Update base code according to comments
* 🔨 Adjust MQTT test base
* 🔨 Update other MQTT tests
* 🍎 Fix carriage return in source files
Apparently test_mqtt_json.py and test_mqtt_template.py were written on Windows. In order to not mess up the diff, I'll just redo the carriage return.
* 🎨 Remove unused import
* 📝 Remove fire_mqtt_client_message
* 🐛 Fix using python 3.6 method
What's very interesting is that 3.4 didn't fail on travis...
* 🐛 Fix using assert directly
* Lazy loading of service descriptions
* Fix tests
* Load YAML in executor
* Return a copy of available services to allow mutations
* Remove lint
* Add zha/services.yaml
* Only cache descriptions for known services
* Remove lint
* Remove description loading during service registration
* Remove description parameter from async_register
* Test async_get_all_descriptions
* Remove lint
* Fix typos from multi-edit
* Remove unused arguments
* Remove unused import os
* Remove unused import os, part 2
* Remove unneeded coroutine decorator
* Only use executor for loading files
* Cleanups suggested in review
* Increase test coverage
* Fix races in existing tests
This applies what was the intended fix in #8336.
moves the fallback for setting client_id to the case when no mqtt config was provided at all. This should reflect the most common use case that fails.
This commit is a workaround and should be reverted when hbmqtt is fixed to allow empty client_id again.
* correct MQTT subscription filter
* wildcard handling (#) fixed
* wildcard handling (#) fixed
* added tests for topic subscription like +/something/#
* function names changed (line too long)
* using raw strings for regular expression
import order changed
