def parse_all(user_input: str, culture: str) -> List[ModelResult]:
return [
# Number recognizer - This function will find any number from the input
# E.g "I have two apples" will return "2".
NumberRecognizer.recognize_number(user_input, culture),
# Ordinal number recognizer - This function will find any ordinal number
# E.g "eleventh" will return "11".
NumberRecognizer.recognize_ordinal(user_input, culture),
# Percentage recognizer - This function will find any number presented as percentage
# E.g "one hundred percents" will return "100%"
NumberRecognizer.recognize_percentage(user_input, culture),
# Age recognizer - This function will find any age number presented
# E.g "After ninety five years of age, perspectives change" will return "95 Year"
NumberWithUnitRecognizer.recognize_age(user_input, culture),
# Currency recognizer - This function will find any currency presented
# E.g "Interest expense in the 1988 third quarter was $ 75.3 million" will return "75300000 Dollar"
NumberWithUnitRecognizer.recognize_currency(user_input, culture),
# Dimension recognizer - This function will find any dimension presented
# E.g "The six-mile trip to my airport hotel that had taken 20 minutes earlier in the day took more than three hours." will return "6 Mile"
NumberWithUnitRecognizer.recognize_dimension(user_input, culture),
# Temperature recognizer - This function will find any temperature presented
# E.g "Set the temperature to 30 degrees celsius" will return "30 C"
NumberWithUnitRecognizer.recognize_temperature(user_input, culture),
# DateTime recognizer - This function will find any Date even if its write in colloquial language -
# E.g "I'll go back 8pm today" will return "2017-10-04 20:00:00"
DateTimeRecognizer.recognize_datetime(user_input, culture)
]
async def on_recognize(
self,
turn_context: TurnContext,
state: Dict[str, object],
options: PromptOptions,
) -> PromptRecognizerResult:
if not turn_context:
raise TypeError("turn_context can’t be none")
if turn_context.activity.type == ActivityTypes.message:
utterance = turn_context.activity.text
turn_context.activity.locale = self._default_locale
recognizer_result = PromptRecognizerResult()
recognizer = NumberRecognizer(turn_context.activity.locale)
if (self._prompt_type == NumberWithTypePromptType.Ordinal):
model = recognizer.get_ordinal_model()
elif (self._prompt_type == NumberWithTypePromptType.Percentage):
model = recognizer.get_percentage_model()
model_result = model.parse(utterance)
if len(model_result) > 0 and len(model_result[0].resolution) > 0:
recognizer_result.succeeded = True
recognizer_result.value = model_result[0].resolution["value"]
return recognizer_result
def perform_ner(self, filepath: str, entity_type: str = 'number', update: bool = False) -> str:
newfilepath = os.path.join(
self.storage_path, entity_type + '_' + os.path.basename(filepath))
if not update and os.path.exists(newfilepath):
return newfilepath
wfile = open(newfilepath, 'w')
filepath = os.path.abspath(filepath)
rfile = open(filepath, 'r')
recognizer = NumberRecognizer(Culture.English)
model = recognizer.get_number_model()
for line in rfile:
answer = next(rfile)
answer_json = json.loads(answer)
text = str(answer_json['body'])
try:
result = model.parse(text)
if result:
for x in result:
if x.type_name == entity_type:
wfile.write(line)
wfile.write(answer)
break
except Exception as e:
print(e)
return newfilepath
def _recognize_ordinal(utterance: str, culture: str) -> List[ModelResult]:
model: OrdinalModel = NumberRecognizer(culture).get_ordinal_model(
culture)
return list(
map(ChoiceRecognizers._found_choice_constructor,
model.parse(utterance)))