This small change adds a check that the intent has been registered
before removing it.
If an unregistered intent was removed padatious would silently throw an
exception due to a list operation error. But when run synchronously from
the skill tester this silent exception was actually loud causing skills
to fail to load.
Padatious is registered as a FallbackSkill and was pushing settings to
the backend.
This change allows Padatious to flag to the MycroftSkill class
that no settings should be created. This is done via the new optional
MycroftSkill parameter use_settings.
Previously Padatious intent matches were performed on non-normalized text, meaning that things like "what's the weather" wouldn't match a Padatious intent but
"what is the weather" would.
The "utterance" in Adapt intent data will still be non-normalized even if the intent match occurred on a normalized utterance. Retaining the existing behavior.
The "intent_failure" data. In there, "utterance" is always the raw version, "norm_utt" is the normalized one.
Also added better debugging info for intent matching to the log.
Also addresses a rare issue with the old code where the Adapt context could
have been updated even if the Adapt intent wasn't actually invoked due to
a higher Padatious intent match.
The loose (conf > 0.5) Padatious match was previously occurring as Fallback
priority 99. The AIML fallback at priority 90 would consume lots of
utterances, interferring with many skills. Now Padatious runs at priority
89.
Additionally, added documentation of the intent and fallback system, including
guidelines for priorities.
Padatious was accepting fairly low confidence matches (0.5). This
would match a phrase such as "Where is the Empire State Building?" to the
intent "Where were you born?" (Which is a 0.54 match)
Now there are two passes mad on the Padatious fallback. The first is
looking for high confidence matches (> 0.8). Then other fallbacks -- such
as the CommonQuery fallbacks -- have an opportunity to consume the
utterance. If they don't consume it, Padatious is invoked again with
looser confidence before finally invoking the Unknown fallback.
Now the eyes go to a khaki color and "< < < LOADING < < <" displays
during the skill startup. When Padatious finishes the first training
we now emit a "mycroft.ready" on the messagebus.
A change to the skill_mark_1 to look for "mycroft.ready" instead of
"mycroft.skills.initialized" provides a good visual change to show that
it is ready for use.
* Deprecate self.config in skills
Skills should contain their own settings, the self.config concept is being
deprecated.
Also removed the defaults set for several old MycroftAI skills. The 19.02
version of these Skills initializes the default values using:
```python
self.settings["key"] = default
```
* Update padatious config to work with the config property.
The registered intents are now stored in a list. When a detach_skill message is received the list is checked for matching intents and the intents are removed
* Add protection for naive skill authors
It is fairly common for new skill authors to attempt actions in the __init__()
method which are not legal yet, as the Skill has not been fully connected to
the Mycroft system. This adds an @property protection layer for the two most common
issues:
* Accessing MycroftSkill.bus
* Accessing MycroftSkill.enclosure
Now those are properties instead of variables and provide appropriate warnings
when used before they exist.
Also enhancing the handling of error logs in the CLI to highlight problems such
as this:
* Color "- ERROR -" log messages in red
* Retaining leading characters from log messages, improving readability in formatted messages
Allow a Padatious intent to override Adapt when it is VERY
certain that the utterance is directed at it. (95% confidence
or greater.) Right now that only occurs if the intent match
for the given phrase is perfect.
This solves this kind of issue:
* Adapt: Matching on "Set" and "Alarm"
* Padatious: Handling "is an alarm set"
* Fix logic error for when no Padatious intent
At startup, Padatious was blocking in the fallback while the
training occurred. As a result, any attempts to use Mycroft
during that period would queue up rather than giving the
user feedback. Now it immediately returns False, allowing
user notification to occur elsewhere.
This commit officially switches the mycroft-core repository from
GPLv3.0 licensing to Apache 2.0. All dependencies on GPL'ed code
have been removed and we have contacted all previous contributors
with still-existing code in the repository to agree to this change.
Going forward, all contributors will sign a Contributor License
Agreement (CLA) by visiting https://mycroft.ai/cla, then they will
be included in the Mycroft Project's overall Contributor list,
found at: https://github.com/MycroftAI/contributors. This cleanly
protects the project, the contributor and all who use the technology
to build upon.
Futher discussion can be found at this blog post:
https://mycroft.ai/blog/right-license/
This commit also removes all __author__="" from the code. These
lines are painful to maintain and the etiquette surrounding their
maintainence is unclear. Do you remove a name from the list if the
last line of code the wrote gets replaced? Etc. Now all
contributors are publicly acknowledged in the aforementioned repo,
and actual authorship is maintained by Github in a much more
effective and elegant way!
Finally, a few references to "Mycroft AI" were changed to the correct
legal entity name "Mycroft AI Inc."
==== Fixed Issues ====
#403 Update License.md and file headers to Apache 2.0
#400 Update LICENSE.md
==== Documentation Notes ====
Deprecated the ScheduledSkill and ScheduledCRUDSkill classes.
These capabilities have been superceded by the more flexible MycroftSkill
class methods schedule_event(), schedule_repeating_event(), update_event(),
and cancel_event().
Åke Forslund
jarbasal
andlo
Steve Penrod
Steve Penrod
Matthew D. Scholefield
Matthew D. Scholefield