How to fix 'NoneType' object has no attribute 'json' in recursive function

Question

def getJoke(idList):
    joke=requests.get("https://official-joke-api.appspot.com/jokes/programming/random")
    if(joke.json()[0]['id'] in idList):
        getJoke(idList)
    else:
        return joke

idList=[]
for i in range(10):
    response=getJoke(idList)
    idList.append(response.json()[0]['id'])
    print('Setup: '+response.json()[0]['setup']+'\nPunchline: '+response.json()[0]['punchline']+'\n')

On the line idList.append(response.json()[0]['id']) I am getting the error 'NoneType' object has no attribute 'json' and I do not know why.

If it is a NoneType, chances are your request may have either failed or returned nothing. — Stick-Figure, Apr 11 '21 at 13:35
`getJoke(idList)` => `return getJoke(idList)`. This is a rather odd use case for recursion... If you can share your problem/expected output, we can probably offer a better way to do it. — ggorlen, Apr 11 '21 at 13:35

ggorlen · Answer 1 · 2021-04-11T15:30:22.070

It appears you want 10 unique random jokes. Instead of recursion, I'd use a loop (to iterate until you have 10 jokes) and a set (to keep track of duplicate ids):

import requests

def get_joke(
    url="https://official-joke-api.appspot.com/jokes/programming/random"
):
    response = requests.get(url)
    response.raise_for_status()
    return response.json()[0]
    
def get_unique_random_jokes(n):
    ids_used = set()
    jokes = []

    while len(jokes) < n:
        joke = get_joke()

        if joke["id"] not in ids_used:
            ids_used.add(joke["id"])
            jokes.append(joke)

    return jokes       

if __name__ == "__main__":
    jokes = get_unique_random_jokes(n=10)

    for joke in jokes:
        print(f'Setup: {joke["setup"]}\nPunchline: {joke["punchline"]}\n')

If you do use recursion, you'll need to return the response of the recursive call, e.g. return getJoke(idList)

A couple other remarks:

I'd cache the result of response.json()[0] in a variable to reduce parsing and make the code prettier.
idList is mutated by the getJoke function, so state management is shared between client and the function. This is hard to reason about. Functions should be self-contained and stateless if possible.
Recursion can blow the stack, is confusing (the bug here is testament to this) and a poor fit for "try again" semantics. Stick to recursion for stuff like tree processing and divide & conquer algorithms.
Although it's a rather premature optimization, keep in mind that in is linear on lists and constant time on sets. Sets are generally the best way to check for membership.

For requests, response.raise_for_status() raises a better error to the caller than .json() would. For a not found request, you'd get

requests.exceptions.HTTPError: 404 Client Error: Not Found for url: 
https://official-joke-api.appspot.com/jokes/programming/asdasd

instead of the irrelevant

json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)

Follow PEP-8's recommendations of snake_case instead of camelCase and spacing around operators, foo=bar => foo = bar.
When making multiple requests to the same domain, use requests.Session to speed things up (exercise for the reader).

How to fix 'NoneType' object has no attribute 'json' in recursive function

1 Answers1