babybutton

babybutton is a project to create a physical button that can be put on your wall and send customizable messages when it's pressed. It was created when we were in the process of adopting our child, when we could get a phone call at any moment telling us we needed to be anywhere in the US within 24 hours. We wanted a single button to push to notify our jobs, our parents, the people who would house-and-pet-sit for us, and anyone else we wanted to know. There was a solid chance we would have no time to spare updating everyone by hand, so we automated it.

Supported Communication Methods

We needed to contact people using the following methods, so that's what babybutton can speak. More methods are pretty trivial to add, so if it's missing something you want, feel free to fork and add it.

• Text messages (through Twilio)
• Slack (sending as a user, not a bot)
• Discord (sending as a bot user)
• Matrix (sending as a user)

Configuration

babybutton is configured with an HCL configuration file. The configuration is expected to be passed as the only argument when running babybutton:

$ babybutton /path/to/config.hcl

Sensitive values, like configuration parameters for connecting to the messaging services, are stored in Vault. A vault configuration block is required, and only one may be specified:

vault {
  # address is optional and defaults to http://127.0.0.1:8200, or whatever the
  # environment variable VAULT_ADDR is set to
  address = "https://my.vault.cluster.local:8200/"

  # mount_path is required, and should be set to the mount path of your v2 KV
  # secret store that secrets are written to
  mount_path = "babybutton"
}

The message services you want to use, who you want to receive the messages, and what messages you want to send are all set in this configuration file.

An optional defaults block lets you set the default message that will be sent unless it is overridden:

defaults {
  # messages are parsed for emoji using the same format as
  # https://github.com/github/gemoji
  message = ":rotating_light: The button was pressed! :rotating_light:"
}

Text Messages

To have someone receive a text message when you push the button, you'll need to configure them as a recipient in your config file:

# "Label" is how they'll be referred to in logs. Also, it helps make the config
# file readable instead of a collection of inscrutable numbers. Consider using
# something like the person's name.
sms "Label" {
  # any number Twilio accepts
  # I don't risk it and just use the country code and full 7 digit number with
  no punctuation
  number = "+12345678"

  # the message you want this person to get when the button is pushed
  # if this isn't set, defaults.message must be
  message = ":rotating_light: The button was pressed so you get a text!"
}

You can have any number of sms blocks in your config, they just need to have a unique label.

Slack Messages

To have someone receive a Slack message when you push the button, you'll need to configure them as a recipient in your config file:

# "Label" is how they'll be referred to in logs. Also, it helps make the config
# file readable instead of a collection of inscrutable numbers. Consider using
# something like the person's name.
slack_message "Label" {
  # the channel ID (how to find IDs: https://stackoverflow.com/a/57246565)
  channel_id = "E1HVSW6DS"

  # the message you want this person to get when the button is pushed
  # if this isn't set, defaults.message must be
  message = ":rotating_light: The button was pressed so you get a Slack message!"
}

You can have any number of slack_message blocks in your config, they just need to have a unique label.

Discord Messages

To have someone receive a Discord message when you push the button, you'll need to configure them as a recipient in your config file:

# "Label" is how they'll be referred to in logs. Also, it helps make the config
# file readable instead of a collection of inscrutable numbers. Consider using
# something like the person's name.
discord_message "Label" {
  # the channel ID (enable developer mode in settings, right click the channel
  to copy IDs)
  channel_id = "454654654893132168494"

  # the message you want this person to get when the button is pushed
  # if this isn't set, defaults.message must be
  message = ":rotating_light: The button was pressed so you get a Discord message!"
}

You can have any number of discord_message blocks in your config, they just need to have a unique label.

Matrix Messages

To have someone receive a Matrix message when you push the button, you'll need to configure them as a recipient in your config file:

# "Label" is how they'll be referred to in logs. Also, it helps make the config
# file readable instead of a collection of matrix IDs. Consider using something
# like the person's name.
matrix_message "Label" {
  # The room ID or alias
  # for unaliased rooms, do the "Share room" thing
  # in the matrix.to link, the room ID will be the bit after the "#/" and
  # before the "?"
  channel_id = "!ysdkjhsASfalkfjasfiohF:myserver.com"

  # the message you want this person to get when the button is pushed
  # if this isn't set, defaults.message must be
  message = ":rotating_light: The button was pressed so you get a Matrix message!"
}

You can have any number of matrix_message blocks in your config, they just need to have a unique label.

Buying the Hardware

Before we can do get this all set up, you're going to need some hardware. At the very least, a button to push, and a computer to run the software that handles the button being pushed.

You can use whatever you want, but for our build, we used this random button we found on Amazon. It had no way to mount it to the wall easily, so we just put a screw right through the plastic. We used a Raspberry Pi 4 model B as the computer. We also used a Raspberry Pi PoE Hat to power the Raspberry Pi over Ethernet, which conveniently let us plug in a single cable and be done with it. Given we had Ethernet in the garage where we wanted to wire this up, this worked out quite nicely, but is not strictly speaking necessary. We put the Raspberry Pi in a SecurePi case which mounted neatly to the wall.

Setting Up The Button

Creating a Vault Secret Engine

The Vault secret engine in use is expected to be a key/value secret engine, version 2. Set one up using the following command:

$ vault secrets enable --path babybutton -version=2 kv

Replace babybutton with the mount path you want to use for your secrets.

Creating a Vault Policy

Set up a policy to give the button access to the data you're storing for it:

path "babybutton/data/*" {
  capabilities = ["read"]
}

Again, replace babybutton with the mount path you want to use for your secrets.

This will grant access to any token using this policy to read the data, but not write it.

Creating a Vault Token

To provision a token for the button to use, run the following command:

$ vault token create -policy=babybutton -display-name "Baby Button Raspberry Pi"

-policy should be set to the ID of the policy you created in the last step.

-display-name is a name that will remind you what this token is for and what is using it.

This will return a token. Set this token as the value of VAULT_TOKEN when starting the babybutton binary. If you're using systemd, you'll want to replace INSERT_TOKEN_HERE in the babybutton.service file in this repo with the token returned here.

Setting Up a Twilio Account

If you want to send text messages, you'll need a Twilio account. Sign up (it's free!) and follow their onboarding flow. You'll probably want to upgrade your account and purchase a number for text messages to be coming from. The default $20 funding minimum is probably plenty to work with.

On the dashboard for your project, you'll see an ACCOUNT SID and AUTH TOKEN. Run the following command to put them into Vault:

$ vault kv put babybutton/twilio account_sid="your account sid here" auth_token="your auth token here" number="your twilio number here"

Remember to replace babybutton with the mount path of your Vault secrets engine.

When it comes to phone numbers, you're probably safe with a variety of formats, but I like to play it safe and use the +12345678910 format, including the leading area code, with no spaces or punctuation. It seems to work for me.

Setting Up a Slack App

To be able to send Slack messages, you're going to need to create a Slack app. This app will send messages as you (so it can send DMs on your behalf, which is what I have it configured to do).

Head to the apps dashboard and click the big green "Create New App" button. Give it a name and assign it to a workspace.

On the Basic Information screen, use the "Add features and functionality" prompt to choose "Permissions". This will take you to an "OAuth & Permissions" page. In "User Token Scopes", click "Add an OAuth Scope" and pick "chat:write".

You'll need to install your app to your workspace now. Click the button for that. Depending on your workspace, you may need to beg your administrator to allow you to install the app.

You'll now see the "OAuth Access Token" field filled out under "OAuth Tokens for Your Team" on the "OAuth & Permissions" page. That's the access token you'll want to be copying. Put it in Vault using the following command:

$ vault kv put babybutton/slack auth_token="your auth token here"

Remember to replace babybutton with whatever your Vault mount point is.

Setting Up a Discord App

To send messages using Discord, you'll need an app and a bot user. Head to the applications dashboard. Click the shiny "New Application" button. Give your application a name.

In the app settings, click on the Bot link in the side menu. Click the "Add Bot" button. Go ahead and uncheck "Public Bot"--you'll be the only one needing to add this bot to a server.

Go ahead and click the OAuth2 link in the side menu. Under "Scopes", choose "bot". "Bot Permissions" will appear. Select "Send Messages" and "Mention Everyone". (If you want to be able to use @everyone in the message you're sending. Leave it off if not.) Scroll back up and copy the link under "Scopes". Go ahead and go to that URL, and authorize the application and pick the server to add it to. Hey, your bot will join the server! Neat.

Before you can use the bot, and I know this is super bizarre, you need to connect it to the websocket gateway at least once. You should only have to do this the one time. No, I don't know why, but you do, or it throws a very confusing "access denied" error in your face. This is how I figured out how to do this.

Back in the app dashboard, click on the "Bot" link in the menu, and copy the token.

Finally, put the bot credentials in Vault:

$ vault kv put babybutton/discord token="your bot token here"

Don't forget to replace babybutton with your Vault mount point.

Setting Up a Matrix User

You're going to want a separate Matrix user to send these messages. Go ahead and create the user, however you normally do that.

You're going to need to get an access token for that user. You do that by logging in. I use the httpie CLI tool for that, but you can use whatever you want to send the HTTP request:

$ http -v POST https://yourhomeserverurl.com/_matrix/client/r0/login type="m.login.password" user="yourmatrixuser" password="yourmatrixpassword"

Replace yourhomeserverurl.com with your homeserver's URL. Replace yourmatrixuser with your Matrix user's username. Replace yourmatrixpassword with the password for your Matrix user. The response will contain an access token. Put that in Vault:

$ vault kv put babybutton/matrix homeserver_url="https://myhomeserverurl.com" user_id="mymatrixuser" access_token="myaccesstoken"

Replace babybutton with your Vault mount point. Replace myhomeserverurl.com with your homeserver's URL. Replace mymatrixuser with your Matrix user's username. Replace myaccesstoken with the access token you just retrieved.

Setting Up a Raspberry Pi

Our Raspberry Pi uses Raspberry Pi OS Lite to run headless. The only things to remember here: you should enable SSH access on the Pi so you can log into it. This can be done by creating an ssh file in the /boot directory of the SD card. On boot, you're going to want to change the default password (using passwd or the configuration utility) and optionally change the hostname to something like babybutton, which will make connecting easier if you're on a network with mDNS support. This should also be achievable through the configuration utility.

Setting Up inputexec

We're going to be using inputexec to launch the babybutton binary when the button is pushed. That takes a little setup.

Find the Device

First things first, we need to identify what device represents your button. The easiest way to do this is to run lsusb without your button plugged in, plug in your button, and run lsusb again. Whichever new device appeared the second time is your button.

Install inputexec

We're going to need inputexec, but unfortunately, it's somewhat out of date. We're going to need to build it from source.

First, we'll need git:

$ sudo apt-get install git-core

Next, we'll need to clone inputexec:

$ git clone https://github.com/rbarrois/inputexec

Finally, we're going to need to path the out of date Python to work with Python 3. Apply this patch:

diff --git a/inputexec/cli.py b/inputexec/cli.py
index 34d9df8..8ac8ad0 100644
--- a/inputexec/cli.py
+++ b/inputexec/cli.py
@@ -119,7 +119,7 @@ class Setup(object):
   else:
       return line_readers.LineReader(src,
	   pattern=args.format_pattern,
-                end_line=args.format_endline.decode('string_escape'),
+                end_line=args.format_endline.encode('utf8').decode('unicode_escape'),
       )

def make_executor(self, args):
@@ -137,7 +137,7 @@ class Setup(object):
       return executors.BlockingExcutor(command_map=commands)
   else:
       return executors.PrintingExecutor('-',
-                end_line=args.format_endline.decode('string_escape'),
+                end_line=args.format_endline.encode('utf8').decode('unicode_escape'),
       )

def make_runner(self, args):

Finally, we should be all set. Install the now-patched inputexec:

$ python3 setup.py install

Find Keypress Code

With inputexec now installed, we need to find what keypress code your button is configured to send. We can do this by running inputexec with certain arguments:

$ inputexec --action-mode=print --source-file=/dev/input/by-id/YOURDEVICE

Replace YOURDEVICE with your device, obtained using lsusb above. Once this is running, go ahead and push your button. inputexec will log which keypress code it receives.

Update actions.ini

Modify the actions.ini file to change the keypress code being listened for to the one your button sends. Replace keypress.KEY_F12 with whatever was output in the step above.

Setting Up systemd

Now that we're all set, it's time to set up inputexec to start listening for button presses every time the Raspberry Pi boots up.

Modify the included babybutton.service file. Change the usb-SIGMACHIP_USB_Keyboard-event-kbd to your device as found through lsusb and used earlier with inputexec.

Move everything into place:

• babybutton.service goes in /etc/systemd/system as babybutton.service
• actions.ini goes in /usr/local/etc as inputexec-actions.ini
• The babybutton binary goes in /usr/local/bin as babybutton
• Your config file goes in /usr/local/etc as babybutton.hcl

Reload systemd to pick up on your new service:

$ sudo systemctl daemon-reload

Enable your new service:

$ sudo systemctl enable babybutton

Start your new service:

$ sudo systemctl start babybutton

Push the Button!

You should be all set up! Push the button, and your messages should get sent.

Debug

Logs are available via journalctl -u babybutton.service -a.