Escaped chars in username / password are not unescaped about rust-url HOT 5 CLOSED

Which spec are you quoting? This library implements https://url.spec.whatwg.org/.

The expected value of fields in urltestdata.txt correspond to the URLUtils interface implemented by browsers, where data is percent-encoded. For example, this shows a%40b:

data:text/html,<a href="http://a@[email protected]"></a><script>document.write(document.getElementsByTagName('a')[0].username)</script>

from rust-url.

I was referring to RFC1738 which seem outdated indeed.
I've had a closer look at the whatwg url spec and I'm not sure I understand why username and password should be percent encoded when parsed (no decoding is required when serializing an URL), but that's not something to discuss here :-)

Sorry for the bother.

from rust-url.

My understanding is that this is the behavior you want on the client side, so that the fields of a parsed URL correspond to what’s gonna be sent on the network when making a request for that URL. Server-side code might make different choices.

from rust-url.

Oh, I understand it now, it does not parse a serialized url, it parses an url typed by the user. So parsing . serialization should not be identity. Thanks for the explanation.

from rust-url.

It parses for example what’s found in the href attribute of an a element, which can be anything. serialize(parse(X)) is not necessarily X for any X, but parse(serialize(parse(X))) should be the same as parse(X) I think.

from rust-url.