为了测试 Python 聊天机器人的性能,我们使用了现成的问题和答案 预配置的电报机器人服务器. 该机器人可以回答预先准备的问题——answers.txt 文件中包含的预先准备的答案,并保留日志。
让我们继续安装:
source my-tel-bot/bin/activate
pip install fuzzywuzzy
pip install python-Levenshtein
mkdir /root/my-tel-bot/data
chmod 775 /root/my-tel-bot/data
让我们准备一个包含问题和答案的文件:
nano /root/my-tel-bot/data/answers.txt
u: hello
Welcome to the chatbot
u: what is your name
Bro Bot!
u: how old are you
relatively few
u: what can you do
You can add to my database in the answers.txt file
Заменим наш демо бот на существующий:
/root/my-tel-bot/bot.py
import telebot
import os
from fuzzywuzzy import fuzz
# Create a bot, write your own token
bot = telebot.TeleBot('')
# Loading a list of phrases and answers into an array
mas=[]
if os.path.exists('data/answers.txt'):
f=open('data/answers.txt', 'r', encoding='UTF-8')
for x in f:
if(len(x.strip()) > 2):
mas.append(x.strip().lower())
f.close()
# Using fuzzywuzzy, we calculate the most similar phrase and give the next element of the list as an answer
def answer(text):
try:
text=text.lower().strip()
if os.path.exists('data/answers.txt'):
a = 0
n = 0
nn = 0
for q in mas:
if('u: ' in q):
# Using fuzzywuzzy we get how similar two strings are
aa=(fuzz.token_sort_ratio(q.replace('u: ',''), text))
if(aa > a and aa!= a):
a = aa
nn = n
n = n + 1
s = mas[nn + 1]
return s
else:
return 'Error'
except:
return 'Error'
# /Start Command
@bot.message_handler(commands=["start"])
def start(m, res=False):
bot.send_message(m.chat.id, 'I am in touch. Write me hello )')
# Receiving messages from a user
@bot.message_handler(content_types=["text"])
def handle_text(message):
# Logging
f=open('data/' + str(message.chat.id) + '_log.txt', 'a', encoding='UTF-8')
s=answer(message.text)
f.write('u: ' + message.text + '\n' + s +'\n')
f.close()
# Sending a response
bot.send_message(message.chat.id, s)
# Launching the bot
bot.polling(none_stop=True, interval=0)
完成后,不要忘记在脚本中指定您的密钥令牌并重新启动服务:
service my-tel-bot restart
暂时没有评论