Resolve multiple sibling pseudo classes correctly

[tolgap]

This PR aims to fix issue #136 .

The issue was: when resolving multiple sibling pseudo classes like:

[ a 
    [ color (hex "#000")
    , hover
        [ color (hex "#111") ]
    , active
        [ color (hex "#222") ]
    , disabled
        [ color (hex "#333") ]
    ]
]

the code written would resolve a, and next up, it would resolve a:hover and since :hover is a complete new styleblock definition, it got appended to the list of styleblock declarations. What happens next is that :active would get resolved, but since the latest styleblock is now an a:hover, it would think it's a nested pseudo class.

I have fixed this issue, by popping the last declaration, and using that to resolve pseudo classes. And in the end, using List.tail to get rid of the a declaration. This seems to fix my issues, and the issue shows in #136 as you can see in the written tests for this.

@rtfeldman since this issue was assigned to you, did you have any other concerns about this?

[tolgap]

The tests seem to be failing because of something in test-runner's elm-package.json.

[tolgap]

I will paste the fixtures and it's output here for ease of reading:

pseudo classes:

[ (#) Page
    [ color (hex "#fff")
    , hover
        [ marginTop (px 10)
        , focus
            [ color (hex "#000") ]
        ]
    , first
        [ fontSize (em 3) ]
    , disabled
        [ marginTop (px 20) ]
    ]
]

#Page {
    color: #fff;
}

#Page:hover {
    margin-top: 10px;
}

#Page:hover:focus {
    color: #000;
}

#Page:first {
    font-size: 3em;
}

#Page:disabled {
    margin-top: 20px;
}

[rtfeldman]

@tolgap wow nice! I just upgraded the test dependencies on master, which should resolve the test failure issue - would you mind rebasing on master?

[tolgap]

@rtfeldman rebased on master

[rtfeldman]

😻 😻 😻 This looks fantastic! Really appreciate it @tolgap!

[rtfeldman]

@tolgap it looks like the second bug reported in #136 is still at large.

I reproduced it with a test case in b2a1968 (on branch reproduce-bug) if you have bandwidth to take a look. Thought I'd mention it since you did such an awesome job with this! ❤️

[tolgap]

@rtfeldman Forgot to include the explanation for that in my initial post...

I indeed could not resolve the second bug in #136 . That second issue is... quite hard to solve if I dare say 😨 . I do have some more time right now to look further into the second bug.