Motivation :-

It was the third time I missed reading my email and picking up shifts for my oncampus job. I did try creating labels and filtering them but that didn’t show any progress in improvement. It’s really stressful when I get the notification that something needs a response from me right away. I had to make sure that I was not missing this next time. It is then when I decided I need to build a system to process and remind me this.

High level view:-

• Create a rule in your work email to filter out the emails by text content/subject/sender’s email and forward it to a personal email address
• The rule should look something like

• Create a rule in the personal email address to add a label to those emails. eg. I labelled as (RSO).
• Parsing and cleaning of the emails with the assigned labels :-)
• Deploying it on a raspberry pi server

Setting up alexa_remote_control.sh (secrets.yaml)

The alexa request is based on https://github.com/walthowd/ha-alexa-tts, There is a minor change make sure to add these https://github.com/rushic24/ha-alexa-tts/commit/e9b145a01c0792b82076554579fc46e6b60a3347 if you want to try that script. I have aleady included that in this repo.

Add your login credentials to the homeassistant/secrets.yaml file, with the keys alexa_email and alexa_password. I have kept a template to make those changes.

Run the script interactively to pull your devices:

./alexa_remote_control.sh -a

And set that in ALEXA_DEVICE_NAME in reademails.py

If login fails look at /tmp/.alexa.login. Search for “password” and see if you are being prompted for the captcha. If so, you can attempt to login to Alexa manually from a browser (from the same IP) and see if that fixes the issue for you. It never did for me, so I logged in to https://alexa.amazon.com with Chrome and used the cookies.txt

Chrome cookies downloader extension
Firefox cookies downloader extension

to export my amazon cookies to /tmp/.alexa.cookie. Other individuals have reported success using the same procedure, but only when using Firefox.

Once you have ./alexa_remote_control.sh -a returning your devices you can test TTS by running:

./alexa_remote_control.sh -d "My Dot Name" -e speak:This_is_a_test!

Note: When trying out this script make sure to change the home directory (HOME="/home/dockerpi" to your home) in alexa_remote_control.sh, and restore ie back to HOME="/home/dockerpi" and also place a copy of that cookies.txt in cookies/cookies.txt for using it inside docker.

Setting up Gmail API credentials (credentials.json and token.pickle)

Go through this geekforgeeks link to set up Gmail API and downloading credentials.json. Use their template script to generate a token.pickle file. Move that to homeassistant/token.pickle for using in docker.

Testing docker

Build and run the docker container to test if everything works

docker run --rm -it $(docker build -q .)

Making the docker container as service

Create a service in /etc/systemd/system/docker.reademails.service with the template below

[Unit]
Description=read emails with RSO tag
After=docker.service
Requires=docker.service

[Service]
TimeoutStartSec=0
Restart=never
ExecStartPre=docker build -t reademails_docker /home/pi/Documents/alexa-notify-important-emails
ExecStart=docker run -t reademails_docker

[Install]
WantedBy=multi-user.target

To start this service:-

sudo systemctl start docker.reademails.service

To start this service at every boot:-

sudo systemctl enable docker.reademails.service

Wrapping things up, before reaching this stage you might need to configure :-

• secrets.yaml
• cookies.txt
• credentials.json
• token.pickle
• HOME (only during generation of secrets/tokens)
• HOME (when creating a service)
• ALEXA_DEVICE_NAME, REFRESH_RATE, MAX_TIMES_REPEAT in reademails.py
• The label variable (RSOLabelid), you can add a print statement print(labels) at the line above and select label accordingly in reademails.py

labels = service.users().labels().list(userId="me").execute()
RSOLabelid = list(filter(lambda x: x["name"] == "RSO", labels["labels"]))[0]["id"]

See it in action :)

https://user-images.githubusercontent.com/6279035/146694731-f55ce539-a4ba-40ea-a0eb-b2236114d9c6.mp4

Cool! isn’t it? What’s next?

I don’t like the voice of alexa, Lets build something like CorentinJ Real-Time-Voice-Cloning and get my emails read in Morgan Freeman’s voice :D

Github link: [https://github.com/rushic24/alexa-notify-important-emails][https://github.com/rushic24/alexa-notify-important-emails]