So in my own time I built that functionality from scratch, initially tested with Fastmail. It's a service that can e.g. run on EC2. It uses IMAP to fetch emails, parses html/text content to get the information, connects to flights tracker APIs to get airport and flight information, then creates the calendar events using CalDAV.
An event looks like this: https://i.imgur.com/DyE78Bs.png
I have made it work with these types of emails: - Easyjet - British Airways - Kiwi - Lastminute - Gotogate - Airbnb - Eventbrite
I'm planning it to extend to more airlines, other time-based events like Meetup, cinemas, restaurants, and so on; anything you can "book a time for". Also I will test it to ensure that it works with more email and calendar providers.
I'm really happy with how it works for my purposes. I have spoken with a few people and they have shown interest in using it, and I have thought of various ways of getting them to use the service...
Which way accessing emails would be better? What would you prefer or be comfortable with? Your answers will help me decide what to work on first. Each has ups and downs, trading security with convenience. If I have enough time I will implement all:
- provide IMAP login details for your email service (this is how it works for me now but I expect not many people will be happy with that)
- create an email filter for your client to forward relevant emails to the service e.g. "from": donotreply@easyjet.com" "subject": "easyJet booking reference", then get an invite for the event (I quite like this one)
- browser extension for your email client
- get the source and run the code yourself on your machine / server after plugging in your own config
- something else?
I'm looking forward to hearing your opinions and thoughts about this.
Thanks in advance!
Suggesting users use rules to forward to your service is a good choice, it allows the user to control what you can see, and it's exactly what the user wants you to see. I personally would be nervous about letting a third party see my flight itineraries, as general opsec, but if someone wanted to do this, I think this is the best way to do so.
Browser extensions are a rough choice because if you want to support many email services, you probably will end up asking to see and modify all browsing data, which is a scary permission to give third parties, and arguably one browsers should stop handing out.
I wouldn't mind running my own apps that read my email somewhere I could monitor what they do and ensure they don't talk to any service but my email service. But you won't see widespread usage with selfhosting, and good luck monetizing it.