224 lines
8.3 KiB
Python
224 lines
8.3 KiB
Python
# Copyright 2018 Mycroft AI Inc.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
from enum import IntEnum
|
|
from abc import ABC, abstractmethod
|
|
from .mycroft_skill import MycroftSkill
|
|
|
|
from mycroft.util.file_utils import resolve_resource_file
|
|
|
|
|
|
class CQSMatchLevel(IntEnum):
|
|
EXACT = 1 # Skill could find a specific answer for the question
|
|
CATEGORY = 2 # Skill could find an answer from a category in the query
|
|
GENERAL = 3 # The query could be processed as a general quer
|
|
|
|
|
|
# Copy of CQSMatchLevel to use if the skill returns visual media
|
|
CQSVisualMatchLevel = IntEnum('CQSVisualMatchLevel',
|
|
[e.name for e in CQSMatchLevel])
|
|
|
|
|
|
def is_CQSVisualMatchLevel(match_level):
|
|
return isinstance(match_level, type(CQSVisualMatchLevel.EXACT))
|
|
|
|
|
|
"""these are for the confidence calculation"""
|
|
# how much each topic word is worth
|
|
# when found in the answer
|
|
TOPIC_MATCH_RELEVANCE = 5
|
|
|
|
# elevate relevance above all else
|
|
RELEVANCE_MULTIPLIER = 2
|
|
|
|
# we like longer articles but only so much
|
|
MAX_ANSWER_LEN_FOR_CONFIDENCE = 50
|
|
|
|
# higher number - less bias for word length
|
|
WORD_COUNT_DIVISOR = 100
|
|
|
|
|
|
class CommonQuerySkill(MycroftSkill, ABC):
|
|
"""Question answering skills should be based on this class.
|
|
|
|
The skill author needs to implement `CQS_match_query_phrase` returning an
|
|
answer and can optionally implement `CQS_action` to perform additional
|
|
actions if the skill's answer is selected.
|
|
|
|
This class works in conjunction with skill-query which collects
|
|
answers from several skills presenting the best one available.
|
|
"""
|
|
|
|
def __init__(self, name=None, bus=None):
|
|
super().__init__(name, bus)
|
|
noise_words_filepath = "text/%s/noise_words.list" % (self.lang,)
|
|
noise_words_filename = resolve_resource_file(noise_words_filepath)
|
|
self.translated_noise_words = []
|
|
try:
|
|
if noise_words_filename:
|
|
with open(noise_words_filename) as f:
|
|
read_noise_words = f.read().strip()
|
|
self.translated_noise_words = read_noise_words.split()
|
|
else:
|
|
raise FileNotFoundError
|
|
except FileNotFoundError:
|
|
self.log.warning("Missing noise_words.list file in "
|
|
f"res/text/{self.lang}")
|
|
|
|
# these should probably be configurable
|
|
self.level_confidence = {
|
|
CQSMatchLevel.EXACT: 0.9,
|
|
CQSMatchLevel.CATEGORY: 0.6,
|
|
CQSMatchLevel.GENERAL: 0.5
|
|
}
|
|
|
|
def bind(self, bus):
|
|
"""Overrides the default bind method of MycroftSkill.
|
|
|
|
This registers messagebus handlers for the skill during startup
|
|
but is nothing the skill author needs to consider.
|
|
"""
|
|
if bus:
|
|
super().bind(bus)
|
|
self.add_event('question:query', self.__handle_question_query)
|
|
self.add_event('question:action', self.__handle_query_action)
|
|
|
|
def __handle_question_query(self, message):
|
|
search_phrase = message.data["phrase"]
|
|
|
|
# First, notify the requestor that we are attempting to handle
|
|
# (this extends a timeout while this skill looks for a match)
|
|
self.bus.emit(message.response({"phrase": search_phrase,
|
|
"skill_id": self.skill_id,
|
|
"searching": True}))
|
|
|
|
# Now invoke the CQS handler to let the skill perform its search
|
|
result = self.CQS_match_query_phrase(search_phrase)
|
|
|
|
if result:
|
|
match = result[0]
|
|
level = result[1]
|
|
answer = result[2]
|
|
callback = result[3] if len(result) > 3 else None
|
|
confidence = self.__calc_confidence(
|
|
match, search_phrase, level, answer)
|
|
self.bus.emit(message.response({"phrase": search_phrase,
|
|
"skill_id": self.skill_id,
|
|
"answer": answer,
|
|
"callback_data": callback,
|
|
"conf": confidence}))
|
|
else:
|
|
# Signal we are done (can't handle it)
|
|
self.bus.emit(message.response({"phrase": search_phrase,
|
|
"skill_id": self.skill_id,
|
|
"searching": False}))
|
|
|
|
def remove_noise(self, phrase):
|
|
"""remove noise to produce essence of question"""
|
|
phrase = ' ' + phrase + ' '
|
|
for word in self.translated_noise_words:
|
|
mtch = ' ' + word + ' '
|
|
if phrase.find(mtch) > -1:
|
|
phrase = phrase.replace(mtch, " ")
|
|
phrase = ' '.join(phrase.split())
|
|
return phrase.strip()
|
|
|
|
def __calc_confidence(self, match, phrase, level, answer):
|
|
# Assume the more of the words that get consumed, the better the match
|
|
consumed_pct = len(match.split()) / len(phrase.split())
|
|
if consumed_pct > 1.0:
|
|
consumed_pct = 1.0
|
|
consumed_pct /= 10
|
|
|
|
# bonus for more sentences
|
|
num_sentences = float(float(len(answer.split("."))) / float(10))
|
|
|
|
# Add bonus if match has visuals and the device supports them.
|
|
bonus = 0.0
|
|
if is_CQSVisualMatchLevel(level) and self.gui.connected:
|
|
bonus = 0.1
|
|
|
|
# extract topic
|
|
topic = self.remove_noise(match)
|
|
|
|
# calculate relevance
|
|
answer = answer.lower()
|
|
matches = 0
|
|
for word in topic.split(' '):
|
|
if answer.find(word) > -1:
|
|
matches += TOPIC_MATCH_RELEVANCE
|
|
|
|
answer_size = len(answer.split(" "))
|
|
answer_size = min(MAX_ANSWER_LEN_FOR_CONFIDENCE, answer_size)
|
|
|
|
relevance = 0.0
|
|
if answer_size > 0:
|
|
relevance = float(float(matches) / float(answer_size))
|
|
|
|
relevance = relevance * RELEVANCE_MULTIPLIER
|
|
|
|
# extra credit for more words up to a point
|
|
wc_mod = float(float(answer_size) / float(WORD_COUNT_DIVISOR)) * 2
|
|
|
|
confidence = self.level_confidence[level] + \
|
|
consumed_pct + bonus + num_sentences + relevance + wc_mod
|
|
|
|
return confidence
|
|
|
|
def __handle_query_action(self, message):
|
|
"""Message handler for question:action.
|
|
|
|
Extracts phrase and data from message forward this to the skills
|
|
CQS_action method.
|
|
"""
|
|
if message.data["skill_id"] != self.skill_id:
|
|
# Not for this skill!
|
|
return
|
|
phrase = message.data["phrase"]
|
|
data = message.data.get("callback_data")
|
|
# Invoke derived class to provide playback data
|
|
self.CQS_action(phrase, data)
|
|
|
|
@abstractmethod
|
|
def CQS_match_query_phrase(self, phrase):
|
|
"""Analyze phrase to see if it is a play-able phrase with this skill.
|
|
|
|
Needs to be implemented by the skill.
|
|
|
|
Args:
|
|
phrase (str): User phrase, "What is an aardwark"
|
|
|
|
Returns:
|
|
(match, CQSMatchLevel[, callback_data]) or None: Tuple containing
|
|
a string with the appropriate matching phrase, the PlayMatch
|
|
type, and optionally data to return in the callback if the
|
|
match is selected.
|
|
"""
|
|
# Derived classes must implement this, e.g.
|
|
return None
|
|
|
|
def CQS_action(self, phrase, data):
|
|
"""Take additional action IF the skill is selected.
|
|
|
|
The speech is handled by the common query but if the chosen skill
|
|
wants to display media, set a context or prepare for sending
|
|
information info over e-mail this can be implemented here.
|
|
|
|
Args:
|
|
phrase (str): User phrase uttered after "Play", e.g. "some music"
|
|
data (dict): Callback data specified in match_query_phrase()
|
|
"""
|
|
# Derived classes may implement this if they use additional media
|
|
# or wish to set context after being called.
|
|
pass
|