Add additional Unicode API to RegexBuilder.CharacterClass #435
Conversation
This adds Unicode property APIs to CharacterClass to bring that type
more in line with what's supported via `/\p{Property=Value}/`.
natecook1000
changed the title
This adds notes about the corresponding regex syntax to all applicable CharacterClass symbols.
|
@swift-ci Please test |
|
Note: Add single API for advancing in an input's character/scalar view depending on semantic level. |
|
@swift-ci Please test |
| /// ``CharacterClass.anyNonNewline``. | ||
| /// | ||
| /// This character class is equivalent to the regex syntax "dot" | ||
| /// metacharacter in single-line mode: `(?s:.)`. |
There was a problem hiding this comment.
But that's not what this is. This is ..
| /// This character class is equivalent to the regex syntax "dot" | ||
| /// metacharacter with single-line mode disabled: `(?-s:.)`. | ||
| public static var anyNonNewline: CharacterClass { | ||
| .init(DSLTree.CustomCharacterClass(members: [.atom(.any)])) |
There was a problem hiding this comment.
.any? Aren't these two things the same?
| /// A character class that matches any single `Character`, or extended | ||
| /// grapheme cluster, regardless of the current semantic level. | ||
| /// | ||
| /// This character class is equivalent to `\X` in regex syntax. |
There was a problem hiding this comment.
Including newlines right? Is this the real "any" above?
|
|
||
| /// A character class that matches any digit. | ||
| /// | ||
| /// This character class is equivalent to `\d` in regex syntax. |
There was a problem hiding this comment.
Does it vary based on options? How is this different than any in that regard?
| @@ -72,27 +98,58 @@ extension RegexComponent where Self == CharacterClass { | |||
| ])) | |||
| } | |||
|
|
|||
| public static var horizontalWhitespace: CharacterClass { | |||
| .init(unconverted: .horizontalWhitespace) | |||
| /// A character class that matches any element that is a "word character". | |||
There was a problem hiding this comment.
Just to double check, is there any better description than "word character"? "Word character" can be mentioned as an aside but that's more of a historical note. @Azoy does Unicode have another name?
There was a problem hiding this comment.
The class of <word_character> includes all the Alphabetic values from the Unicode character database ...
https://unicode.org/reports/tr18/#RL1.4
| astCharacterProperty(.generalCategory(category.extendedGeneralCategory!)) | ||
| } | ||
|
|
||
| public static func binaryProperty( |
There was a problem hiding this comment.
Can we get this code in a different file?
| /// - Returns: The modified regular expression. | ||
| public func asciiOnlyWhitespace(_ useASCII: Bool = true) -> Regex<RegexOutput> { | ||
| wrapInOption(.asciiOnlySpace, addingIf: useASCII) | ||
| public func asciiOnlyClasses(_ kinds: RegexCharacterClassKind = .all) -> Regex<RegexOutput> { |
There was a problem hiding this comment.
What does this do to .any, properties, etc?
| @available(SwiftStdlib 5.7, *) | ||
| public struct RegexCharacterClassKind: OptionSet, Hashable { |
There was a problem hiding this comment.
Did you consider or debate whether RegexBuilder.CharacterClass should be Swift.RegexCharacterClass? Then this would be a Kind under it.
|
Closing this stale PR |
This includes revisions to the options API and additional
CharacterClasstype to bring it into alignment with the functionality that we're offering through regex literals. For example,/\p{NumericValue=1}/can be written in RegexBuilder syntax asCharacterClass.numericValue(1).