FromUnicodeScalarView deprecation message possibly references unimplemented conversion?

[kamcma]

I think the deprecation messages for FromUnicodeScalarView might reference an unimplemented conversion?

swift-parsing/Sources/Parsing/Internal/Deprecations.swift

Lines 808 to 850 in f949081

	@available(*, deprecated, message: "Use 'From(.unicodeScalars)' instead.")
	public struct FromUnicodeScalarView<Input, UnicodeScalarsParser: Parser>: Parser
	where UnicodeScalarsParser.Input == Substring.UnicodeScalarView {
	public let unicodeScalarsParser: UnicodeScalarsParser
	public let toUnicodeScalars: (Input) -> Substring.UnicodeScalarView
	public let fromUnicodeScalars: (Substring.UnicodeScalarView) -> Input
	@inlinable
	public func parse(_ input: inout Input) rethrows -> UnicodeScalarsParser.Output {
	var unicodeScalars = self.toUnicodeScalars(input)
	defer { input = self.fromUnicodeScalars(unicodeScalars) }
	return try self.unicodeScalarsParser.parse(&unicodeScalars)
	}
	}
	@available(*, deprecated, message: "Use 'From(.unicodeScalars)' instead.")
	extension FromUnicodeScalarView where Input == ArraySlice<UInt8> {
	@inlinable
	public init(@ParserBuilder<Substring.UnicodeScalarView> _ build: () -> UnicodeScalarsParser) {
	self.unicodeScalarsParser = build()
	self.toUnicodeScalars = { Substring(decoding: $0, as: UTF8.self).unicodeScalars }
	self.fromUnicodeScalars = { ArraySlice(Substring($0).utf8) }
	}
	}
	@available(*, deprecated, message: "Use 'From(.unicodeScalars)' instead.")
	extension FromUnicodeScalarView where Input == Substring {
	@inlinable
	public init(@ParserBuilder<Substring.UnicodeScalarView> _ build: () -> UnicodeScalarsParser) {
	self.unicodeScalarsParser = build()
	self.toUnicodeScalars = \.unicodeScalars
	self.fromUnicodeScalars = Substring.init
	}
	}
	@available(*, deprecated, message: "Use 'From(.unicodeScalars)' instead.")
	extension FromUnicodeScalarView where Input == Substring.UTF8View {
	@inlinable
	public init(@ParserBuilder<Substring.UnicodeScalarView> _ build: () -> UnicodeScalarsParser) {
	self.unicodeScalarsParser = build()
	self.toUnicodeScalars = { Substring($0).unicodeScalars }
	self.fromUnicodeScalars = { Substring($0).utf8 }
	}

This compiles with the deprecation warning:

let foo: some Parser<Substring.UTF8View, Void> = FromUnicodeScalarView { // ⚠️ 'FromUnicodeScalarView' is deprecated: Use 'From(.unicodeScalars)' instead.
  Peek {
    Prefix(1, while: CharacterSet.alphanumerics.contains)
  }
}

But the suggested replacement does not:

let bar: some Parser<Substring.UTF8View, Void> = From(.unicodeScalars) { // 🛑 Type 'Conversion' has no member 'unicodeScalars'
  Peek {
    Prefix(1, while: CharacterSet.alphanumerics.contains)
  }
}

Would be happy to be told I'm holding it wrong! Feel free to close/convert to discussion if so. Thanks for the incredible library.

[stephencelis]

@kamcma I think that missing conversion is just an oversight. Wanna PR one?

[kamcma]

Will do!

[kamcma]

For the record, the motivating use-case that led me to even discover the absence of these conversions mentioned in the deprecation message was wanting to use Foundation's CharacterSet.contains(_:), which operates on Unicode.Scalars, as the predicate for the Prefix(_ length:while:). Eg:

let p = Parse(input: Substring.self) {
  Peek {
    From(.unicodeScalars) {
      Prefix(1, while: CharacterSet.alphanumerics.contains)
    }
  }
  // ...
}

If there's a better way to do that, I'm all ears.

[kamcma]

Well, having played around with CharacterSet as a parser directly a little more, I may be able to accomplish this more cleanly like so:

let p = Parse(input: Substring.self) {
  Peek {
    Prefix(1).pipe {
      CharacterSet.letters
    }
  }
  CharacterSet.alphanumerics
}

Probably FromUnicodeScalarView's deprecation message still shouldn't reference an unimplemented conversion, but I may no longer need that particular conversion myself (Substring/Substring.UTF8View/ArraySlice<UInt8> to Substring.UnicodeScalars). So feel free to re-evaluate if it's a conversion you want the library to offer, and close the issue/PR if not.)