Weather monitor

I built myself a simple weather alert that sends rain warnings to my smartphone, by using the OpenWeatherMap API and telegram-send.

There’s really not much to it. Once I had the idea that I wanted rain warnings and that they should be messages on my phone, everything was quick and simple. There are two main problems I needed to solve to get the warnings: (1) find the data and (2) get them on my phone.

The first, getting data, I solved by using the OpenWeatherMap API, which is free for my purposes. I had looked at a few different APIs, none of which included what I needed or were too cumbersome to use for this use casewhich doesn’t mean I wouldn’t use them for projects with different requirements..
What brought me to OpenWeatherMap initially was their minute-by-minute forecast for the next sixty minutes, which is exactly what I wanted. Alas, it is only available in the US, so my alerts are less fine-grained since I use the hourly forecasts.

Once I had registered with OpenWeatherMap, the call to their OneCall-API is very nice and simple:

URL = "https://api.openweathermap.org/data/2.5/onecall?lat=52.5201631&lon=13.3691966&exclude=daily&appid=yourapikeyhere"
requests.get(URL)

Simply pluck your coordinates from Google Maps and see weather information for the place. Perfect, exactly what I need.

The second problem, messages to my phone, I had already solved with telegram-send some time before. It’s a command-line tool for sending messages to a telegram bot, that you can subscribe to on your telegram account or accounts. I use that quite often, it’s an excellent tool for simple monitoring tasks. I call that via a subprocess call:

subprocess.run(["/home/shezi/.bin/tele", message])

Yes, I have a shortcut for telegram-send since nine characters (including a dash, of all things) is really too much. There is also a python-internal API, but who has time for that?

Plugging both parts together gives me a simple weather warning that sends me a message when there’s rain in the next hour, and when the rain stops.

Here’s the complete code:

from datetime import datetime
import time
import subprocess

import requests


def send(message):
	"""Send the message via telegram-send."""

	subprocess.run(["/home/shezi/.bin/tele", message])


URL = "https://api.openweathermap.org/data/2.5/onecall?lat=52.5201631&lon=13.3691966&exclude=daily&appid=yourapikeyhere"

def run():
	send("starte Regen-Service")

	was_rain = False

	while True:
		try:
			r = requests.get(URL)
			data = r.json()
			rain_branch = data['hourly'][0].get('rain')
			print(datetime.now(), rain_branch)

			will_rain = rain_branch is not None

			if was_rain and will_rain:
				pass
			elif was_rain and not will_rain:
				send(f"Regen hört auf innerhalb der nächsten Stunde")
				was_rain = False
			elif not was_rain and will_rain:
				rain_amount = rain_branch.get('1h')
				send(f"Regen innerhalb der nächsten Stunde: {rain_amount:.2f}")
				was_rain = True
			elif not was_rain and not will_rain:
				pass

		except Exception as e:
			send(f"Could not parse results: {e}")

		time.sleep(1800)


if __name__ == '__main__':
	try:
		run()
	finally:
		send("beende Regen-Service")

That’s really all there is to it. It works brilliantly, apart from the occasional Could not parse results, which I should probably deactivate at some pointOh, and for operations, I use the best tool there is: screen..

But on a meta-level, this simple program goes much deeper: For some time, I have been thinking about how “things” should be represented in my digital life. And for many of these things, they should be events that appear somewhere. Things like “you have an appointment” or “you should go to the dentist in the next four weeks” or “there will be rain” or “something happened on the production servers”. Currently, there is really no way to have these events collected somewhere and visible, so I rely on many different systems to keep track of them: a calendar, my brain, specific apps that I need to check, email. I don’t want to keep track of all of this, so I’m going to try and consolidate these events into a stream that is more easily manageable and consumable.

Weather events are just the beginning.