Testing Flows

Programmatically test your flows

You may already be using your app's /status API endpoint to monitor it's uptime and latency, but did you know you can also use it to test your flows? Here's how.

Triggering a flow

Let's review what the /status endpoint looks like:

https://grid.meya.ai/gateway/v2/status/<APP_ID>/__status__?user_id=monitor&text=__status__

🚧

Caching

You must set Cache-Control: no-cache in the header when testing any endpoint to ensure you don't get a cached result.

Replace APP_ID with your app's actual ID, which can be found on your app's Settings page.

You can leave the user_id to monitor. This ensures that your testing will only use a single Meya user, instead of generating a new one every time.

You can fire a flow by changing the value of the text query parameter. For example, you would use text=_custom to trigger this custom.yaml flow:

triggers:
  - keyword: _custom

steps:
  - say: first step
  - flow_set:
      x: (@ 2 * 2 * 2 )
  - say: second step x=(@ flow.x )

📘

Using keywords

To guarantee you trigger the right flow, consider adding a keyword trigger to each of your flows. Make the keyword something a user is unlikely to say, like _test_welcome_flow.

This is the Orb simulator output for the above flow:

888

And here's the response you'd get from your API call:

{
  "ok": true,
  "text": "first step second step x=8",
  "time": "2020-09-09 19:42:21.875174"
}

Automated testing

With the /status endpoint, you can write a script to run a series of tests of your flows and compare the response text with the expected text. Run the script as a cron job and now you have ongoing automated testing of your flows!

Here's a sample Python script that stores the test phrases and their expected responses in a dictionary:

import requests

URL = "https://grid.meya.ai/gateway/v2/status/APP_ID/__status__?user_id=monitor&text="
HEADERS = {
    "Cache-Control": "no-cache",
}
tests = {
    "_custom": "first step second step x=8",
    "hi": "Hi! What would you like to do?",
}

def main():
    for test in tests:
        res = requests.get(f"{URL}{test}", headers=HEADERS)
        if res.status_code != 200:
            print(f"Test <{test}> failed! Status code: {res.status_code}")
            continue

        data = res.json()
        actual = data.get("text")
        expected = tests[test]
        if actual != expected:
            print(f"Test <{test}> failed! Expected: {expected}; Actual: {actual}")
        else:
            print(f"Test <{test}> passed.")

if __name__ == "__main__":
    main()

Automated browser testing

Another option for testing your app is to use a tool like Selenium, which allows you to automate browser interactions. With Selenium, you can record your interactions with the browser and replay them later.

The image below shows a basic example. When the play button is clicked, it opens a new browser window, opens the Orb chat, and sends the word hi to the app.

838

A simple test in Selenium


What’s Next