Prevent pseudo selector args from prefixing

[toddself]

This change prevents the argument to a pseudo selector from being prefixed
(since it shouldn't be!).

Also added provisions to run the tests via npm test

I'm not sure if this is the best way to solve this, but it does fix #4.

We might not care that the first parent selector is a selector and only need to determine if the parent's parent is a pseudo selector.

I still feel like this might be a bug in the parser, however, it's late and I don't have time to delve into that :)

[ahdinosaur]

hey @toddself

this pull request is awesome, thanks for working on this. 😸

code changes look good, but to honest i'm not sure i completely understand what's happening here. css is confusing, it's a mystery how my previous self even wrote the current code here. so just wondering, are there any other test cases you can think of that might fall into this same bug?

anyways, great work!

[toddself]

not a problem!

Any of the CSS psuedo selectors which take an argument:

• :dir
• :lang
• :nth-* (last-child, child, of-type, last-of-type)

The difficulty here, I realize, is that :not should get it's arguments prefixed:

.myel :not(p) {}

Should likely transform to:

#hello-world .myel :not(#hello-world p) {}

Let me update the PR to handle this case and add cases for other elements as well just to make sure our t's are crossed and i's dotted

[toddself]

OK this has been updated to test all the pseudo selectors which take arguments and handle :not correctly with a prefix.

The logic feels a little gnarly, but I'm not sure there's a better way since we're looking at the argument to the selector when we need to make the determination?

Anyway, the tests pass :)

[yoshuawuyts]

Looks quite reasonable to me; everything seems tested now - merging! ✨

[yoshuawuyts]

Published as v1.0.4 ✨