evalai-cli

evalai-cli copied to clipboard

When an error is received while making a request to the backend, we catch it, and if it is present in the EVALAI_ERROR_CODES, we log it using the lines:

if response.status_code in EVALAI_ERROR_CODES:
    validate_token(response.json())
    echo(
        style(
            "\nError: {}\n"
            "\nUse `evalai challenges` to fetch the active challenges.\n"
            "\nUse `evalai challenge CHALLENGE phases` to fetch the "
            "active phases.\n".format(response.json()["error"]),
            fg="red",
            bold=True,
        )
    )

But response.json() does not have the key error in cases. For e.g., I made a request using an expired token and got an error. So when I logged the entire json.response(), it was something like

{'detail': 'Given token not valid for any token type', 'code': 'token_not_valid', 'messages': [{'token_class': 'RefreshToken', 'token_type': 'refresh', 'message': 'Token is invalid or expired'}]}

This does not contains a key named error

Hi @burnerlee, I run in the same error as you. I'm using a fresh install of v1.3.12.

@burnerlee nice catch, this error response is from jwt token API that we added recently. If you are interested you can pick this issue and send a PR to fix it.

@Ram81 yeah sure, I'll make a PR for this

@Ram81, should I do something like, checking if error key is present in the response, if it does, use the same value, if it doesn't then use the detail key for getting the error?

@burnerlee yes

Oh okay, it is only this jwt error log which causes this error, or there may be some other errors too? Which need to be handled 😅

For now its just the JWT API which has a different error response format. Rest of our APIs are already handled

thank you so much ❤️