How to Create a Simple Python Chatbot on the Raspberry Pi

By dayz
on Sep 15, 2018
in Random, Raspberry Pi

359

I have recently been working on a new project that required scraping data from other websites and have been slowly learning python. I usually work with PHP but I knew python will work faster and better. I also have always wanted to create a simple chatbot that I can input text into and it would reply with something meaningful. Creating a chatbot with a huge database of words and actually understanding what is being said is not an easy task and is something that is way over my head. I was browsing Reddit and found out that you can get results in JSON format is you append ‘.json‘ to the end of the URL. This got me thinking, what if I search Reddit with a question or comment and pick a random search result and reply with one of the top comments. This would fill my urge in creating a chatbot and it would also be a good starting point if I ever wanted to go deeper into creating a database and store answers for the chatbot to get a personality of its own.

I figured the Raspberry Pi would be an easy starting point since I have a bunch of them laying around and it is already setup with a Python environment. You can easily replicate this in any Python environment but I tested out my chatbot on the Raspberry Pi.

Objective

To create a chatbot in Python on the Raspberry Pi that searches Reddit and replies using a Reddit comment

Material

You will need the following:

Instructions

Create a new file on the Raspberry Pi and call it main.py

Copy the code into main.py and save the file. To run the file open a terminal window on the Raspberry Pi and navigate to the folder the file is located in. Run the following command:

1	python main.py

You should get a prompt asking you to “Ask me anything: “. Ask the chatbot anything and see what reply you get.

Python Source Code

import requests
import json
import random

#The main function that will grab a reply
def grab_reply(question):
	#Navigate to the Search Reddit Url
	r = requests.get('https://www.reddit.com/r/AskReddit/search.json?q=' + question + '&sort=relevance&t=all', headers = {'User-agent': 'Chrome'})
	
	answers = json.loads(r.text)	#Load the JSON file
	
	Children = answers["data"]["children"]

ans_list= []

for post in Children:
		if post["data"]["num_comments"] >= 5:	#Greater then 5 or equal  comments
			ans_list.append (post["data"]["url"])
		
	#If no results are found return "I have no idea"
	if len(ans_list) == 0:
		return "I have no idea"
	
	#Pick A Random Post
	comment_url=ans_list[random.randint(0,len(ans_list)-1)] + '.json?sort=top'	#Grab Random Comment Url and Append .json to end

#Navigate to the Comments
	r = requests.get(comment_url, headers = {'User-agent': 'Chrome'})
	reply= json.loads(r.text)
	
	Children = reply[1]['data']['children']
	
	
	reply_list= []

for post in Children:
		reply_list.append(post["data"]["body"])	#Add Comments to the List
	
	if len(reply_list) == 0:
		return "I have no clue"
		
	#Return a Random Comment
	return reply_list[random.randint(0,len(reply_list)-1)]

#Main Loop, Always ask for a question
while 1:
	q=raw_input("Ask me anything: ")
	q=q.replace(" ", "+")	#Replace Spaces with + for URL encoding
	print(grab_reply(q))	#Grab and Print the Reply

import requests

import json

import random

#The main function that will grab a reply

def grab_reply(question):

#Navigate to the Search Reddit Url

r = requests.get('https://www.reddit.com/r/AskReddit/search.json?q=' + question + '&sort=relevance&t=all', headers = {'User-agent': 'Chrome'})

answers = json.loads(r.text) #Load the JSON file

Children = answers["data"]["children"]

ans_list= []

for post in Children:

if post["data"]["num_comments"] >= 5: #Greater then 5 or equal comments

ans_list.append (post["data"]["url"])

#If no results are found return "I have no idea"

if len(ans_list) == 0:

return "I have no idea"

#Pick A Random Post

comment_url=ans_list[random.randint(0,len(ans_list)-1)] + '.json?sort=top' #Grab Random Comment Url and Append .json to end

#Navigate to the Comments

r = requests.get(comment_url, headers = {'User-agent': 'Chrome'})

reply= json.loads(r.text)

Children = reply[1]['data']['children']

reply_list= []

for post in Children:

reply_list.append(post["data"]["body"]) #Add Comments to the List

if len(reply_list) == 0:

return "I have no clue"

#Return a Random Comment

return reply_list[random.randint(0,len(reply_list)-1)]

#Main Loop, Always ask for a question

while 1:

q=raw_input("Ask me anything: ")

q=q.replace(" ", "+") #Replace Spaces with + for URL encoding

print(grab_reply(q)) #Grab and Print the Reply

Input / Outputs

Below are some sample Input / Outputs that I got from my chatbot.

Input: Whats your name?

Output: My name is Lara and I never really thought it fits me, especially during my teens I always wanted a longer, more sophisticated name with a cute nickname. I have never had a short nickname, as kids the others called me Lara Croft sometimes. In addition, substitute teachers kept calling me Laura, Sarah etc. (Im German they sound kinda similar)

1 2	Input: Whats your favorite cheese? Output: Brie on a warm baguette. I'm salivating just thinking of it.

1 2	Input: Do you know you are a robot? Output: Everyone on Reddit is a bot except you.

1 2	Input: How smart are you? Output: Smart enough to know the world is full of people who know way more about every topic I know a lot about.

1 2	Input: Whats the meaning of life? Output: How the fuck would I know? I'm 5

Hopefully this Reddit Chatbot inspires you to work on something and even expand on it. You can store answers in a database for future use or you can analyze them and train the chatbot. If you use my code for anything please let me know, I would love to hear on how you expanded it.

359