Add SourceKitObject that represents sourcekitd_object_t in Swift

[norio-nomura]

This adds SourceKitObject, SourceKitObjectConvertible and UID that salvaged from #274.
Fixes #489
This is a different solution than #488.
By using this changes, the sample code in #487 (comment) can be written as following:

func requestSymbolSuperclasses(_ kind: SymbolKind, name: String, usr: String) -> [String: Any] {
    let response: [String: Any]

    do {
        response = try Request.customRequest(request: [
            "key.request": UID(RequestKind.symbolSuperclasses),
            "key.symbol": [
                "key.indexer.arg.symbol.name": name,
                "key.indexer.arg.symbol.kind": kind.rawValue,
                "key.indexer.arg.symbol.language": "Xcode.SourceCodeLanguage.Swift",
                "key.indexer.arg.symbol.resolution": usr
            ]
        ]).send()
    } catch {
        fatalError("Symbol superclasses request failed: \(error)")
    }

    return response
}

/cc: @ileitch

[ileitch]

So much nicer! 💯

[jpsim]

This is awesome. Everything I've ever dreamed of. 😍

[jpsim]

Should update this comment.

[jpsim]

Minor: I generally avoid indenting compiler directive bodies. See https://github.com/jpsim/SourceKitten/blob/0.19.1/Source/sourcekitten/main.swift#L9-L15

[jpsim]

Thoughts on using the auto-synthesized Hashable implementation for Swift 4.1+?

extension UID: Hashable {
#if !swift(>=4.1)
    public var hashValue: Int {
        return uid.hashValue
    }

    public static func == (lhs: UID, rhs: UID) -> Bool {
        return lhs.uid == rhs.uid
    }
#endif
}

I don't feel strongly, but it seems like it might be easier to find code that can be removed once we require Swift 4.1+ to compile SourceKitten.

[jpsim]

Not doing this. Can't auto-synthesize Hashable conformance outside the source module.

[jpsim]

testDescription?

[norio-nomura]

Thanks for reviews!
Sorry, I changed it considerably other than pointed out.

• Changed SourceKitObject to class for using sourcekitd_request_release() on deinit
• Change SourceKitObjectConvertible.sourceKitObject from a property to a throwing method
• Make SourceKitObject conforms to ExpressibleByBooleanLiteral

Could you please review again?

[norio-nomura]

@jpsim Should I open another PR against this branch for changes that added after previous reviews?

[jpsim]

No need for a separate PR, I'll review this shortly. Looks mostly good, I think I just have a few pieces of small but breaking feedback

[ileitch]

Any news on this? Would be nice to have in a release :)

[jpsim]

Sorry the delay is entirely my fault. I'll review & merge this now.

[jpsim]

I renamed SourceKitObjectConvertible.sourceKitObject to sourcekitdObject to avoid confusion between 'SourceKitObject's and sourcekitd "objects". I hope that's ok with you. I'll merge this as soon as tests pass CI.

[jpsim]

Tired of waiting for CI. Will fix if I get notified of failures when tests eventually run on master.

[norio-nomura]

Some CI jobs for this PR failed since the branch has been removed before starting jobs.
I tried to re-run them, but it can't.
Since CI jobs for master branch passed, it may be ok.

[jpsim]

Yes that’s expected.