Attributeerror: str Object has no attribute 'Confidence'

Question

I’m developing a chatbot and I wish it would only respond if it had determined level of confidence in the answer.

# -*- codding: utf-8 -*-
import os
import telebot
from chatterbot import ChatBot
from chatterbot.trainers import ListTrainer

bot = telebot.TeleBot("TEM UM CÓDIGO DO TELEGRAM AQUI")

#Futaba original
def bot_convencional(message):
    chatbot = ChatBot("Futaba")
    trainer = ListTrainer(chatbot)
    for arquivos in os.listdir('arquivos'):
        chats = open('arquivos/' + arquivos, 'r').readlines()
        trainer.train(chats)

    resposta = chatbot.get_response(message) #ESSE GET DENTRO DO IF DA ÚLTIMA FUNÇÃO
    resposta = str(resposta) 

    mensagem = open("arquivos/teste", "w")
    mensagem.write(resposta)
    mensagem.close()

#Recebe e envia uma resposta inicial
@bot.message_handler(commands = ["help", "start"])
def enviar_mensagem(message):
    bot.reply_to(message, "Hey")

#Recebe qualquer outra mensagem
@bot.message_handler(func=lambda message:True)
def mensagem(message):
    bot_convencional(message.text)
    resposta = open("arquivos/teste", "r")
    resposta = resposta.read()

    if float(resposta.confidence) > 0.5:
        bot.reply_to(message, resposta)
    else:
        bot.reply_to(message, "Como você está se sentindo?")
bot.polling()

The problem is that when I try to use the Confidence gives the error

if float(response.Confidence) > 0.5: Attributeerror: str Object has no attribute 'Confidence'

I’ve used Confidence on another chatbot and I don’t understand why you’re giving this problem this time. Can someone help me?

I wanted to put the bot_conventional function get.Response inside the message if.

Answer 1

The problem is that you store a Python object in the form of string in the file, later read it also as string and use it as if it were the same initial object. At the moment you save the object as string in the file you no longer have it as an object of the original type; it will become an instance object of string, with the fields of string. There is no field confidence in string, maybe only in the initial type of your object.

You save an object as string in the archive:

#Futaba original
def bot_convencional(message):
    ...
    resposta = chatbot.get_response(message) #ESSE GET DENTRO DO IF DA ÚLTIMA FUNÇÃO
    resposta = str(resposta) 

    mensagem = open("arquivos/teste", "w")
    mensagem.write(resposta)
    mensagem.close()

You read the file and access the field confidence of a string, generating the error cited in the question:

#Recebe qualquer outra mensagem
@bot.message_handler(func=lambda message:True)
def mensagem(message):
    ...
    resposta = open("arquivos/teste", "r")
    resposta = resposta.read()

    if float(resposta.confidence) > 0.5:
        bot.reply_to(message, resposta)
    else:
        bot.reply_to(message, "Como você está se sentindo?")

If you need to persist an object in a file to use it later, you will need to use some serialization technique. In Python there is a native library pickle that does this.

Save a serialized object to file:

obj = {'foo': 'bar'}

with open('arquivo.pickle', 'wb') as stream:
    pickle.dump(obj, stream)

Read a serialized object from the file:

with open('arquivo.pickle', 'rb') as stream:
    obj = pickle.load(stream)

print(obj['foo'])  # 'bar'

^{See working on Repl.it}

Answer 2

To the confidence I needed to access the chatbot.get_response(message). For this, I created a variable called corpus to store this corpus = chatbot.get_response(message), I have made it global and if of confidence.

FUNCTION STORING THE CHATBOT

def bot_convencional(message):
    chatbot = ChatBot("Futaba")
    trainer = ListTrainer(chatbot)
    for arquivos in os.listdir('arquivos'):
        chats = open('arquivos/' + arquivos, 'r').readlines()
        trainer.train(chats)

    resposta = chatbot.get_response(message)
    resposta = str(resposta)
    global corpus
    corpus = chatbot.get_response(message)

    mensagem = open("arquivos/teste", "w")
    mensagem.write(resposta)
    mensagem.close()

FUNCTION RESPONDING TO MESSAGES

#Recebe qualquer outra mensagem
@bot.message_handler(func=lambda message:True)
def mensagem(message):
    bot_convencional(message.text)
    resposta = open("arquivos/teste", "r")
    resposta = resposta.read()

    if float(corpus.confidence) > 0.5:
        bot.reply_to(message, resposta)
    else:
        bot.reply_to(message, "Como você está se sentindo?")
bot.polling()

I don’t know if this is right, but it worked.