level3.md

This perhaps deserves a blog post, but whatever.

The challenge: What if you wanted to determine what level my Bulgarian language skills were at? Using this sample document:

{
    "name": {
        "first": "Lloyd",
        "last": "Hilaiel"
    },
    "favoriteColor": "yellow",
    "languagesSpoken": [
        {
            "language": "Bulgarian",
            "level": "advanced"
        },
        {
            "language": "English",
            "level": "native"
        },
        {
            "language": "Spanish",
            "level": "beginner"
        }
    ],
    "seatingPreference": [
        "window",
        "aisle"
    ],
    "drinkPreference": [
        "whiskey",
        "beer",
        "wine"
    ],
    "weight": 172
}

How would you craft a CSS-like selector to extract the value of the key level contained in an object where language has a value of Bulgarian? (i.e. match the string advanced)

NOW STOP. Don't read any more of this gist. Go think about it. Do you know any CSS that should apply here?

By this time, you've already thought of your own answer. A few possible (sometimes complimentary) routes are in my head.

The first solution is general and has precedent in jquery/sizzle. Specifically :has() and :contains() inspired by jquery. But maybe we'd rename :contains to :val.

.languagesSpoken :has(.language:val('Bulgarian')) .level

The other route could repurpose attribute selectors to say that they match children

.languagesSpoken object[language="Bulgarian"] .level

Note that the former is potentially much more flexible because the :has() pseudo function can take complete selectors.

Note also that the latter could be a syntactic abbreviation for the former. This is in fact how things like :first-child and :last-child are implemented in the reference implementation.

Another option could use the sibling combinator from css3:

.languagesSpoken .language:val('Bulgarian') ~ .level

Note that the former is potentially much more flexible because the :has() pseudo function can take complete selectors.

Note also that the latter could be a syntactic abbreviation for the former. This is in fact how things like :first-child and :last-child are implemented in the reference implementation.

I think you're spot-on about :has being more flexible.

:has() has been promoted from maybe to is status in the docs: http://jsonselect.org/#docs/overview

Great! I will implement it this weekend if I can find the time.

…

On Fri, May 27, 2011 at 6:07 PM, lloyd < ***@***.***>wrote: `:has()` has been promoted from maybe to is status in the docs: http://jsonselect.org/#docs/overview ## Reply to this email directly or view it on GitHub: https://gist.github.com/993928

I'll try to include tests and close this issue today: lloyd/JSONSelect#16

Good; maybe we should also add parser/deparser tests. So that we can make sure that all the parser implementations work the ~same way. Adding a deparser (transforming the ast back into a valid selector) would also be a nice feature I think. And would be especially useful during testing. Cheers

…

On Fri, May 27, 2011 at 7:03 PM, lloyd < ***@***.***>wrote: I'll try to include tests and close this issue today: lloyd/JSONSelect#16 ## Reply to this email directly or view it on GitHub: https://gist.github.com/993928

The latter could be more abbreviated even, exactly like in CSS :

.languagesSpoken [language="Bulgarian"] .level

You can also handle multiple checks like in CSS :

.languagesSpoken [language='Bulgarian'][native='yes'][present] .level

How to handle this with :has ?

.languagesSpoken :has(.language:val('Bulgarian')):has(.native:val('yes')):has('present') .level 

?

I like the potential flexibility of :has with inner selectors but only having this one to check properties would imho make JSONSelect unnecessarily more verbose, and 'diverge' from CSS, for the likely most common use cases (property presence check, property value check and other simple CSS attribute selectors).