Add an option for K-anonymity of URLs about ackee

[electerious]

We would need some kind of extension that would allow users to write custom rules as I don't think we can find defaults that work for everyone. It might be better to block those URLs instead of just hiding them.

My first idea: An env variable that lets you specify Regex patterns to block. Ackee will check those before writing them to the database. This however won't allow you to just remove parts of a URL.

(Alternatively, Ackee could remove the query string from the URL, and that would solve 90% of the above problem, but eg. search terms or marketing campaigns would not be trackable.)

Ackee already applies small adjustments to URLs before writing them to the database. I can't do too much by default as queries are relevant for some users. It's also possible to build the routing of a webapp via queries for example.

Any other ideas? What do you think about Regex patterns?

[emschwartz]

I also have a use case where I don't want to track the full URLs. I opted to modify the URL before it's sent to the server by ackeeTracker:

const url = new URL(window.location.href)
url.hash = ''
tracker.record({
    siteLocation: url.toString()
})

@gcampax would something like that work for your use case?

[gcampax]

Yeah I think the problem is that, as @electerious notes, queries are sometimes relevant and sometimes not. Often, the URL path itself can be sensitive, if it contains say the ID or hash of a document (think Google Doc like link). So none of these solutions would be complete.
I think removing all URLs that have not been seen >= K times (for a configurable K) is a much better approach, and it has the advantage of being an easy-to-defend standard in protecting privacy.

[electerious]

I don't think that there's a good way to build an option that works for all users. I would suggest to modify the URL before tracking the visit with ackee-tracker. It would be great if someone could add a page to the docs/ that shows how to do so (with examples).

Here's an example: #113 (comment)

Would be helpful for #114, #113 and this issue.

[gcampax]

Ok, I think it's clear you're not interested in adding this feature. That is fine, and I think the workarounds you suggested are acceptable. Thank you for the all the help, and for this project in this first place. I'm going to close this issue.

[electerious]

I've created a new issue. Would be cool if someone could write a documentation for this workaround: #163