**Update Lingua Franca to v0.4.1**
The update from Lingua Franca v0.2.x to v0.4.x includes few
breaking changes.
- Some API methods have been updated.
- Mycroft-core tests have been updated to reflect improvements in
Lingua Franca's formatting and parsing.
- add LF default lang setting method to config.locale
including warning that this method will change in the future
- Add TODO's for 21.08 - moving more methods to LF
- simplify loading and setting default of languages in Skills service
- Remove unneeded wrappers for Lingua Franca functions
- Fix documentation of format and parse utils
- Fix test warnings
- Remove unused ContextManager import from mycroft.skills.intent_service
- Remove Old 'intent_failure' -> fallback handler
- Remove handling of fallback messages without the fallback_range
parameter
Add empty skill directories to a set and remove them whenever they are
no longer empty. Empty skill directories only appear once in the log.
==== Fixed Issues ====
log polution - #2750
The api methods are now much easier to use, almost transparent. The
current caveat is that only "standarad" python types are acceptable
(int, float, str, list, bool, None) due to the json serialization.
- api methods are now created with the skill_api_method decorator
- both arguments and keyword arguments are sent to the api method
instead of the message object
- api methods now uses a normal return statement instead of having to
handle creating response messages on the bus.
For example if the datetime skill wants to make the datetime string
fetchable simply add the skill_api_method decorator to the
get_display_date method.
@skill_api_method
def get_display_date(self, day=None, location=None):
"""Returns the date and time as a string."""
[...]
The methods return value will be sent back to the caller and can be used
from a skill through
datetime = SkillApi.get('mycroft-date-time.mycroftai')
self.log.info(datetime.get_display_date())
The skill api allows skills to define a public api which can easily be
accessed by other skills over the message bus just as easy as working
with a normal object.
The skill api object is generated from the skill's public_api property. It's a dict where each key is turned into a method on the api object. The method is defined as
"api_method": {
"type": message type string
"func": handler method
"help": help string for cli
}
Example skill:
class Test2(MycroftSkill):
def __init__(self):
MycroftSkill.__init__(self)
self.public_api = {
'speak': {
'type': 't2.speak',
'func': self.handle_speak,
'help': 'speak the test sentence\nand another line\n\nlast'
},
'speak2': {
'type': 't2.speak2',
'func': self.handle_speak2,
'help': 'speak the other sentence'
}
}
def handle_speak(self, message):
self.speak('This is a test')
self.bus.emit(message.response(data=None))
def handle_speak2(self, message):
self.speak('This is another test')
self.bus.emit(message.response(data=None))
Previously Mycroft reported mycroft.ready when Padatious had
finished it's first training run. This check has been shifted
to the enclosure service, and starts checking once the Padatious
training is complete.
Currently checks on readiness of audio, speech and skills services.
ProcessStatus tracks the process status and allows callbacks on changes
and status queries over the messagebus.
StatusCallbackMap is used to setup the callbacks
ProcessState is an enum tracking the different states.
Mycroft always mangeled any local settings changes at startup.
This caches the last settings from home on disk and only updates skills
_if_ there is a change.
This means that as if a member is changed locally (manually edited or
changed by the skill itself) it will be kept until another change is made
to the skill on Home.
Previously Mycroft reported mycroft.ready when Padatious had
finished it's first training run. This check has been shifted
to the enclosure service, and starts checking once the Padatious
training is complete.
Currently checks on readiness of audio, speech and skills services.
ProcessStatus tracks the process status and allows callbacks on changes
and status queries over the messagebus.
StatusCallbackMap is used to setup the callbacks
ProcessState is an enum tracking the different states.
self._loaded_status was accessed from a message handler, if a message is
received at the precisely wrong time an exception may occur. This fixes
the init order.
Also moves the skill_updater creation to be fore the handlers are
registered.
Simplify the handle_utterance into a list of intent matching functions
run in order until a match is found.
The resolution order is
- Converse
- Padatious High Confidence
- Adapt
- Fallback High priority
- Padatious Medium Confidence
- Fallback Medium priority
- Padatious Last ditch effore
- Fallback Low priority
This collects the many parts of where intent is handled into a single
location handling the entire flow.
The idea is that, in the end, all the skill calling should be done from
this method. The main intent parsers does this but the converse and
fallback still calls directly.
Kris Gesling
Ken Smith
dzekem christa
Kris Gesling
Åke Forslund
Chris Veilleux
neonandrii
jarbasal
Diego Asterio
JarbasAI
Tim Gates
katridi