39 lines
1.4 KiB
Python
39 lines
1.4 KiB
Python
from UserActType import UserActType
|
|
from nlu import nlu
|
|
from dst import DST
|
|
from dp import DP
|
|
from nlg import NLG
|
|
|
|
if __name__ == '__main__':
|
|
|
|
dst = DST()
|
|
dp = DP()
|
|
nlg = NLG()
|
|
|
|
print("Witamy w restauracji πzza. W czym mogę pomóc?")
|
|
|
|
while True:
|
|
user_input = input("$")
|
|
# get user act frame from user input with Natural Language Understanding
|
|
user_act_frame = nlu(user_input)
|
|
# print('NLU', user_act_frame)
|
|
# update Dialogue State Tracker with new user act frame
|
|
dst.update(user_act_frame)
|
|
state, last_user_act, last_system_act, slots = dst.get_dialogue_state_tracker_state()
|
|
# print('state', state)
|
|
# print('last_user_act', last_user_act)
|
|
# print('last_system_act', last_system_act)
|
|
# print('slots', slots)
|
|
|
|
# get system act frame which decides what's next from Dialogue Policy
|
|
system_act_frame = dp.update_system_action(state, last_user_act, last_system_act, slots)
|
|
dst.update_last_system_act(system_act_frame)
|
|
# print('system_act_frame', system_act_frame)
|
|
|
|
# generate response based on system act frame
|
|
system_response = nlg.generate_response(state, last_user_act, last_system_act, slots, system_act_frame)
|
|
print('BOT:', system_response)
|
|
|
|
if user_act_frame['act'] == UserActType['bye']:
|
|
break
|