def __init__(self):
"""
Init for logger Intent engine Corpus
"""
self.engine = IntentDeterminationEngine()
self.bot = BotLogger()
self.cr = BotNextCorpus()
self.first_time = True
self.intent = []
self.train_intent()
def __init__(self):
"""
Init for logger Intent engine Corpus
"""
self.engine = IntentDeterminationEngine()
self.bot = BotLogger()
self.cr = BotNextCorpus()
self.first_time = True
self.intent = []
self.train_data = {'intent': "", 'user_text': "", 'bot_response': ""}
self.train_intent()
self.senti = SentimentIntensityAnalyzer()
from pymongo import MongoClient
from chat_logger import BotLogger
__author__ = 'Sachin Arora'
log = BotLogger()
class state_mdb:
def __init__(self):
try:
client = MongoClient(
"mongodb://*****:*****@ds027295.mlab.com:27295/chatbot")
self.db = client.chatbot
except Exception as ex:
log.log_error(ex)
def insert_user_state(self, cr):
"""
Def to insert data into db with user state
:param cr:
:return:
"""
try:
self.delete_col()
self.db.user_state.insert_one(cr)
print("Inside insert_user_state " + str(cr))
print('\nInserted data successfully\n')
class CrmnextChatBot:
def __init__(self):
"""
Init for logger Intent engine Corpus
"""
self.engine = IntentDeterminationEngine()
self.bot = BotLogger()
self.cr = BotNextCorpus()
self.first_time = True
self.intent = []
self.train_data = {'intent': "", 'user_text': "", 'bot_response': ""}
self.train_intent()
self.senti = SentimentIntensityAnalyzer()
def train_intent(self):
"""
Training intent engine to classify text intent
:return:
"""
for k in self.cr.train_intent_list.keys():
for w in self.cr.train_intent_list[k]:
self.engine.register_entity(w, k)
for w in self.cr.train_intent.keys():
self.intent.append(IntentBuilder(w) \
.require(self.cr.train_intent[w]) \
.build())
for obj in self.intent:
self.engine.register_intent_parser(obj)
def run_bot(self, conv):
"""
Thread starts from here.
:return:
"""
self.bot.log_debug("Received Json in run_bot: {}".format(conv))
msg = conv["user_text"]
isFirst = True if conv["intent_type"] == "" else False
if self.sent(msg):
if isFirst:
try:
intent = json.loads(self.intent_parser(msg))
except Exception:
intent = {
"intent_type": "",
"Card": "",
"CardLost": "",
"target": "",
"confidence": 0.0
}
conv["intent_type"] = intent["intent_type"]
return self.p_flow(self.cr.chat_data, conv)
else:
return self.p_flow(self.cr.chat_data, conv)
else:
return self.neg_res(self.cr.chat_data, conv)
def intent_parser(self, conv):
"""
Method to get intent of user text.
:param conv: user data
:return:
"""
for intent in self.engine.determine_intent(conv):
if intent.get('confidence') > 0:
return json.dumps(intent, indent=4)
else:
pass
def p_flow(self, corpus, ud):
"""
Method to implement process for user.
:param corpus: It provides data for chatbot
:param ud: User related data.
:return: response
"""
response = dict(userId="123",
user_intent="",
response_text="",
card_type="",
recommendation=[],
entities=[],
user_stage=0)
self.bot.log_debug("User Data: {}".format(ud))
p_data = corpus[ud['intent_type']][ud['user_stage']]
user_txt = self.clean_text(ud['user_text'])
if user_txt in p_data.keys():
response['response_text'] = random.choice(p_data[user_txt])
response['card_type'] = ""
elif user_txt in p_data['user_text']:
print("p_data: >>" + str(p_data))
response = self.update_res(response, p_data)
elif user_txt == 'quit':
q_data = corpus['EXIT'][0]
response['user_intent'] = q_data['intent_type']
response['user_stage'] = q_data['user_stage']
response['response_text'] = random.choice(q_data['response'])
response['card_type'] = q_data['card_type']
else:
self.build_train_cr(ud)
if 'default_speech' in p_data.keys(
) and p_data['default_speech'] is not '':
response['response_text'] = random.choice(
p_data['default_speech'])
else:
response[
'response_text'] = "Please select below option to move forward"
response['card_type'] = "#IntentOptions"
response['user_stage'] = ud['user_stage']
response['user_intent'] = ud['intent_type']
if len(p_data['recommendation']) == 0:
response['recommendation'] = self.cr.all_skills
else:
response['recommendation'] = p_data['recommendation']
print("response: " + str(response))
return response
def update_res(self, response, p_data):
"""
:param response:
:param p_data:
:return:
"""
response['response_text'] = random.choice(p_data['response'])
response['user_intent'] = p_data['intent_type']
response['card_type'] = p_data['card_type']
response['user_stage'] = p_data['user_stage']
response['entities'] = p_data['entities']
return response
def neg_res(self, cr_data, conv):
"""
Generate response if sentiment is -ve
:param cr_data:
:param conv:
:return:
"""
neg_data = cr_data["neg_sent"][0]
return dict(user_intent="",
response_text=random.choice(neg_data["response"]),
user_stage=0,
card_type='',
recommendation=[])
def sent(self, conv):
"""
Polarity is considered pos if > -0.2 (for tuning our scenario)
:param conv:
:return:
"""
t = TextBlob(conv)
vs = self.senti.polarity_scores(conv)
return False if t.polarity < -0.3 else True
def clean_text(self, statement):
"""
Remove any consecutive whitespace characters from the statement text.
"""
return re.sub(r'[^a-zA-Z0-9 ]', r'', statement).rstrip().lower()
def build_train_cr(self, ud):
intent = ud['intent_type']
if intent is not '':
self.train_data['intent'] = intent
self.train_data['user_text'] = ud['user_text']
else:
self.train_data['intent'] = ""
self.train_data['user_text'] = ud['user_text']
with open('train_data.json', 'a') as f:
json.dump(self.train_data, f, indent=True)
# bot = CrmnextChatBot()
# user_stage = 0
# intent = ''
# print(bot.clean_text("Hi I am #avinash ? "))
# while True:
#
# text = input("Enter text: ")
# reconnect = input('re-connect: ')
# print(bool(reconnect))
# data = {'userId': '123', "intent_type":intent, "user_text": text, 'user_name': 'Avinash Gaur',
# 'contactNumber': '89892398128', 'cardCount': 2, 'user_stage': user_stage, 're_connect': bool(reconnect)}
# print(data)
# print(user_stage)
# print(intent)
# d = bot.run_bot(data)
# intent = d['user_intent']
# user_stage = d['user_stage']
#
#
# print(d)
class CrmnextChatBot:
def __init__(self):
"""
Init for logger Intent engine Corpus
"""
self.engine = IntentDeterminationEngine()
self.bot = BotLogger()
self.cr = BotNextCorpus()
self.first_time = True
self.intent = []
self.train_intent()
def train_intent(self):
"""
Training intent engine to classify text intent
:return:
"""
for k in self.cr.train_intent_list.keys():
for w in self.cr.train_intent_list[k]:
self.engine.register_entity(w, k)
for w in self.cr.train_intent.keys():
self.intent.append(IntentBuilder(w) \
.require(self.cr.train_intent[w]) \
.build())
for obj in self.intent:
self.engine.register_intent_parser(obj)
def run_bot(self, conv):
"""
Thread starts from here.
:return:
"""
self.bot.log_debug("Received Json in run_bot: {}".format(conv))
msg = conv["user_text"]
isFirst = conv["re_connect"]
if self.sent(msg):
if isFirst:
try:
intent = json.loads(self.intent_parser(msg))
except Exception:
intent = {
"intent_type": "",
"Card": "",
"CardLost": "",
"target": "",
"confidence": 0.0
}
conv["intent_type"] = intent["intent_type"]
return self.p_flow(self.cr.chat_data, conv)
else:
return self.p_flow(self.cr.chat_data, conv)
else:
return self.neg_res(self.cr.chat_data, conv)
def intent_parser(self, conv):
"""
Method to get intent of user text.
:param conv: user data
:return:
"""
for intent in self.engine.determine_intent(conv):
if intent.get('confidence') > 0:
return json.dumps(intent, indent=4)
else:
pass
def p_flow(self, corpus, ud):
"""
Method to implement process for user.
:param corpus: It provides data for chatbot
:param ud: User related data.
:return: response
"""
response = dict(type="text", response_text="Sorry Not able to understand", senderId="1234", userintent="")
self.bot.log_debug("User Data: {}".format(ud))
p_data = corpus[ud['intent_type']][ud['user_stage']]
if ud['user_text'].lower() in p_data['user_text']:
response = self.update_res(response, p_data)
return response
def update_res(self, response, p_data):
"""
:param response:
:param p_data:
:return:
"""
response['response_text'] = random.choice(p_data['response'])
return response
def neg_res(self, cr_data, conv):
"""
Generate response if sentiment is -ve
:param cr_data:
:param conv:
:return:
"""
neg_data = cr_data["neg_sent"][0]
return dict(userId=conv["userId"], user_intent=neg_data["intent_type"], response_text=random.choice(neg_data["response"]),
card_type=neg_data["card_type"],user_stage=0)
def sent(self, conv):
"""
Polarity is considered pos if > -0.2 (for tuning our scenario)
:param conv:
:return:
"""
t = TextBlob(conv)
return True if t.polarity > - 0.2 else False