* Refactor translations to reduce dict lookups
All of our cache lookups used:
`cache[language][O(component)][category]`
The cache was designed as
`cache[language][component][category][flatted_key]`
The lookups are now
`cache[language][category][O(component)]`
The cache is now stored as
`cache[language][category][component][flatted_key]`
This allows the catch fetch to avoid looking up
the category each loop
* already a set, and we do not mutate
* add placeholder support to entity name translation
* add negativ tests
* make property also available via description
* fix doc string in translation_placeholders()
* fix detection of placeholder
* validate placeholders for localized strings
* add test
* Cache translation_placeholders property
* Make translation_placeholders uncondotionally return dict
* Fall back to unsubstituted name in case of mismatch
* Only replace failing translations with English
* Update snapshots
* Blow up on non stable releases
* Fix test
* Update entity.py
---------
Co-authored-by: Erik <erik@montnemery.com>
Co-authored-by: Joost Lekkerkerker <joostlek@outlook.com>
* Add Valve integration.
This adds the valve integration discussed in https://github.com/home-assistant/architecture/discussions/975
Most of the code is taken from the cover integration but simplified since valves
can't tilt.
There are a couple outstanding errors I'm not sure how to solve and prevents
me from even making this commit without `--no-verify`.
* Apply PR feedback
* Apply more feedback: Intruduce the bare minimum
* Remove file commited by mistake
* Hopefully this fixes tests
* Match cover's typing and mypy settings
* Change some configuration files
* Fix test
* Increase code coverage a little
* Code coverate inproved to 91%
* 95% code coverage
* Coverate up to 97%
* Coverage 98%
* Apply PR feedback
* Even more feedback
* Add line I shouldn't have removed
* Derive closed/open state from current position
* Hopefully last feedback
* Update homeassistant/components/valve/__init__.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* Update homeassistant/components/valve/__init__.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* Remove unnecesary translation
* Remove unused method arguments
* Complete code coverage
* Update homeassistant/components/valve/__init__.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* Simplify tests
* Update homeassistant/components/valve/__init__.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* Apply last feedback
* Update tests/components/valve/test_init.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* Update tests/components/valve/test_init.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* Update tests/testing_config/custom_components/test/valve.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* More feedback
* Apply suggestion
* And more feedback
* Apply feedback
* Remove commented code
* Reverse logic to unindent
* Update homeassistant/components/valve/__init__.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* Update homeassistant/components/valve/__init__.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* Implement stop valve for Mock valve
* Fix tests now that I've implemented stop_valve
* Assert it's neither opening nor closing
* Use current position instead
* Avoid scheduling executor when opening or closing
* Fix incorrect bitwise operation
* Simplify toggle
* Remove uneeded partial functions
* Make is_last_toggle_direction_open private
* Remove valve from test custom integration
* Improve test coverage
* Address review comments
* Address review comments
* Address review comments
* Update homeassistant/components/valve/__init__.py
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
* Update tests
---------
Co-authored-by: Martin Hjelmare <marhje52@gmail.com>
Co-authored-by: Erik <erik@montnemery.com>
Reduce overhead connect dispatcher
- We tend to have 1000s (or 10000s) of connected dispatchers which
makes these prime targets to reduce overhead/memory
- Instead of creating new functions to wrap log exceptions each time
use partials which reuses the function body and only create new
arguments
Previous optimizations #103307#93602
* Keep capabilities up to date in the entity registry
* Warn if entities update their capabilities very often
* Fix updating of device class
* Stop tracking capability updates once flooding is logged
* Only sync registry if state changed
* Improve test
* Revert "Only sync registry if state changed"
This reverts commit 1c52571596c06444df234d4b088242b494b630f2.
* Avoid calculating device class twice
* Address review comments
* Revert using dataclass
* Fix unintended revert
* Add helper method
* Drop use of regex in helpers.extract_domain_configs
* Update test
* Revert test update
* Add domain_from_config_key helper
* Add validator
* Address review comment
* Update snapshots
* Inline domain_from_config_key in validator
* Improve OAuth error handling in configuration flows
* Update strings for all integrations that use oauth2 config flow
* Remove invalid_auth strings
* Revert change to release
* Revert close change in aiohttp mock
* Differentiate between warnings and errors in check_config helper
* Update tests
* Treat configuration errors in frontend and its dependencies as errors
* Improve test coverage
* Address review comments
* Improve test coverage
* Improve test coverage
* Address review comments
* Add comment
The contextmanager decorator creates a new context manager every
time its run, but since we only have a single context var, we can
use the same one every time. Creating the contextmanager was roughly
20% of the time time of the template render
I was a bit suprised to find it creates a new context manager
object every time https://stackoverflow.com/questions/34872535/why-contextmanager-is-slow
* Prevent accidentally reusing an entity object
* Fix group reload service
* Revert "Fix group reload service"
* Improve test
* Add tests aserting entity can't be reused
* add supports_response to platform entity services
* support multiple entities in entity_service_call
* support legacy response format for service calls
* revert changes to script/shell_command
* add back test for multiple responses for legacy service
* remove SupportsResponse.ONLY_LEGACY
* Apply suggestion
Co-authored-by: Allen Porter <allen.porter@gmail.com>
* test for entity_id remove None
* revert Apply suggestion
* return EntityServiceResponse from _handle_entity_call
* Use asyncio.gather
* EntityServiceResponse not Optional
* styling
---------
Co-authored-by: Allen Porter <allen.porter@gmail.com>
* Avoid looking up the callable type for HassJob when we already know it
When we connect the frontend we call async_listen with run_immediately MANY
times when we already know the job type (it will always be a callback). This
reduces the latency to get the frontend going
* missing coverage
This was being built every time state was written. When
a robo vac is operating it writes state often which mean building
the icon string over and over again when it rarely changes.
async_has_matching_flow is more likely to be True than hass.is_stopping
This does not make much difference but it was adding noise to a profile
that I am digging into to look for another issue
* Refactor entity service calls to reduce complexity
gets rid of the noqa C901
* Refactor entity service calls to reduce complexity
gets rid of the noqa C901
* short
* Use loop time to set context
loop time is faster than utcnow, and since its only used internally it can
be switched without a breaking change
* fix mocking
* Allow specifying a custom log function for template render
* Bypass template cache when reporting errors + fix tests
* Send errors as events
* Fix logic for creating new TemplateEnvironment
* Add strict mode back
* Only send error events if report_errors is True
* Force test of websocket_api only
* Debug test
* Run pytest with higher verbosity
* Timeout after 1 minute, enable syslog output
* Adjust timeout
* Add debug logs
* Fix unsafe call to WebSocketHandler._send_message
* Remove debug code
* Improve test coverage
* Revert accidental change
* Include severity in error events
* Remove redundant information from error events
Lars R
Erik Montnemery
J. Nick Koston
Michael Hansen
Marc Mueller
Jan-Philipp Benecke
Petru Paler
Michael
Martijn van der Pol
Robert Resch
Bram Kragten
Ville Skyttä
Miguel Camba
Jan Bouwhuis
Paul Bottein
schelv
Aarni Koskela
Ruslan Sayfutdinov
Allen Porter
Kevin Stillhammer
Raman Gupta
G Johansson
Nicolas van de Walle
Joost Lekkerkerker
Dave T
Joakim Plate