Request: Add an API for interacting w/ Cookies about fantoccini HOT 7 CLOSED

I might play around with it this weekend! Thanks for pointing me to that WebDriver documentation.

from fantoccini.

@jonhoo sorry I never started looking at this! The site I was scraping doubled down on their bot detection and so I pivoted the project 🤷‍♂️ I might still take a stab at it some time.

from fantoccini.

Sorry it took me a while to get back to this! I think cookie::Cookie is indeed the way to go 👍

That's awkward about Serialize. I've found the maintainers of webdriver pretty receptive to suggestions, so if you suggest it there, it'll probably be implemented pretty quickly! In the meantime though, I think it's okay to add all but one :)

from fantoccini.

Yup, I agree, that would be really handy! But unfortunately all WebDriver supports is to fetch all the cookies of the current document. Maybe that would still be useful to expose though? There are also commands for adding a cookie, and for deleting one. Together, those three may at least get you close to what you want? It should be pretty easy to add support for those to fantoccini if you want to open a PR for me to look at!

from fantoccini.

Yes, please support adding cookies :)

from fantoccini.

@jonhoo I'm working on a PR to add the 5 cookie-related functions as defined in 16. Cookies.

One challenge with defining APIs is in how to pass in/return a cookie. The two options I'm considering are:

1. Accept/return cookies as serde_json::Values

Straightforward implementation, just pass everything along to the webdriver crate and you're done. Users will need to make sure the parameters are of the correct types by checking the spec.

2. Accept/return cookies as a cookie::Cookie

This requires a bit more code within fantoccini to convert to/from JSON cookies for WebDriver. In particular, WebDriver encodes the Max-Age of a cookie as a u64 Unix timestamp, whereas cookie::Cookie requires a Duration. We'll probably need to pull in a datetime crate like chrono to be able to resolve this difference.

The two advantages of this approach are that users will be able to:

1. More easily (and safely) inspect cookies returned by WebDriver (e.g., client.get_named_cookie(&name).await?.value())
2. Construct cookies dynamically via cookie::CookieBuilder.

Thoughts?

from fantoccini.

Actually, I hit a bit of a snag while implementing the add_cookie operation... webdriver::command::AddCookieParameters does not implement Serialize, which means that fantoccini can't serialize the struct to JSON for the POST request :(

I don't actually need add_cookie, but I'm not sure if it would make sense to add all but one of the supported cookie operations.

from fantoccini.