Authentication - Best Practices

Introduction #

The OAuth 2.0 authentication/authorization flow differs from the traditional username/password flow still heavily used in healthcare interoperability. To help users new to writing applications using OAuth, here are some best practices recommended by Canvas.

The Access Token: Don’t lose it, reuse it #

Instead of this #

For some connections requiring a username and password, it may be necessary to use those credentials every time a message is sent to that system. In that type of flow, you might need to write something like:

import requests

token = get_new_auth_token()
patient = get_patient(token, patient_id)

updated_patient_body = update_patient(patient, name="Frank")

token = get_new_auth_token()
update_patient(token, updated_patient_body)

token = get_new_auth_token()
schedule_appointment(token, patient=patient, type="telehealth")


Repeated, unnecessary calls to a token request function double the number of requests this script generates.

Do this #

Regardless of the authentication method used, a successful request includes "expires_in": in the response body. The corresponding value will be an integer for the number of seconds until the token will be expired in Canvas. By reusing the token and eliminating redundant token requests, there is a performance gain from reducing the overall number of requests. The system is also more secure - having fewer tokens reduces the overall surface area for illegitamate access

General Guidelines:

  • store the access token as securely as the client id and secret
  • depending on your use case, it may be best to store a token just for that session or to reuse it until it nears expiration
  • store the expiration datetime as well and check it before requesting a new token
  • use a refresh token if you provided in the authentication response for the authentication flow being used

There’s any number of ways to do this in your language of choice. In Python, your code may look more like:

import os
import requests

from datetime import datetime, timedelta
from urllib.parse import urlencode

# we should get these from environment variables or a secure location at runtime, not in the code

def get_new_auth_token():
    # Auth tokens are requested from the EMR instance, not the FHIR API.
    url = FUMAGE_BASE_URL.replace("fumage-", "")

    payload = urlencode(
            "grant_type": "client_credentials",
            "client_id": f"{CLIENT_ID}",
            "client_secret": f"{CLIENT_SECRET}"
    headers = { "Content-Type": "application/x-www-form-urlencoded", }

    response ="{url}/auth/token/", headers=headers, data=payload)

    if response.status_code == 200:
        access_token = response.json()["access_token"]
        expiration_date = + timedelta(seconds=response.json()["expires_in"])

        os.environ["CANVAS_API_ACCESS_TOKEN"] = access_token
        os.environ["CANVAS_ACESS_TOKEN_EXPIRATION_DATE"] = expiration_date.strftime("%m/%d/%y %H:%M:%S")

        return response.json()["access_token"]
        raise Exception(f"Could not acquire new auth token: {response.text}")

if __name__ == '__main__':
    access_token = os.getenv("CANVAS_API_ACCESS_TOKEN")
    if expiration_date := os.getenv("CANVAS_ACCESS_TOKEN_EXPIRATION_DATE"):
        expiration_date = datetime.strptime(expiration_date, "%m/%d/%y %H:%M:%S")
        expiration_date =

     # only request a new token when we do not have a token or the one we have has expired
    if not access_token or expiration_date <=
        access_token = get_new_auth_token()

    headers = { "Authorization": f"Bearer {access_token}" }

    response = requests.get(f"{FUMAGE_BASE_URL}/Patient?name=Briddle", headers=headers)

    if response.status_code == 401:
        # attempt to acquire a new token 1 time if we get a 401 - maybe it was manually expired or
        # we have the wrong expiration date
        access_token = get_new_auth_token()
        headers = { "Authorization": f"Bearer {access_token}" }
        response = requests.get(f"{FUMAGE_BASE_URL}/Patient?name=Briddle", headers=headers)

        if response.status_code == 401:
            # limit retries but throw a specific exception for authentication related issues
            raise Exception(f"Cannot authenticate to Canvas after 1 retry: {response.text}")
        elif response.status_code != 200:
            # capture other issues as well
            raise Exception(f"{response.text}")

    patients = response.json()["entry"]
    for patient in patients:
        patient_id = patient["resource"]["id"]
        response = requests.get(f"{FUMAGE_BASE_URL}/Appointment?patient=Patient/{patient_id}")

        # do something with the appointments - save locally, modify and update, etc.

Takeaway #