|
| 1 | +//===----------------------------------------------------------------------===// |
| 2 | +// |
| 3 | +// This source file is part of the Swift.org open source project |
| 4 | +// |
| 5 | +// Copyright (c) 2021-2022 Apple Inc. and the Swift project authors |
| 6 | +// Licensed under Apache License v2.0 with Runtime Library Exception |
| 7 | +// |
| 8 | +// See https://swift.org/LICENSE.txt for license information |
| 9 | +// |
| 10 | +//===----------------------------------------------------------------------===// |
| 11 | + |
| 12 | +import _RegexParser |
| 13 | +@_spi(RegexBuilder) import _StringProcessing |
| 14 | + |
| 15 | +public struct CharacterClass { |
| 16 | + internal var ccc: DSLTree.CustomCharacterClass |
| 17 | + |
| 18 | + init(_ ccc: DSLTree.CustomCharacterClass) { |
| 19 | + self.ccc = ccc |
| 20 | + } |
| 21 | + |
| 22 | + init(unconverted model: _CharacterClassModel) { |
| 23 | + // FIXME: Implement in DSLTree instead of wrapping an AST atom |
| 24 | + switch model.makeAST() { |
| 25 | + case .atom(let atom): |
| 26 | + self.ccc = .init(members: [.atom(.unconverted(atom))]) |
| 27 | + default: |
| 28 | + fatalError("Unsupported _CharacterClassModel") |
| 29 | + } |
| 30 | + } |
| 31 | + |
| 32 | + init(property: AST.Atom.CharacterProperty) { |
| 33 | + // FIXME: Implement in DSLTree instead of wrapping an AST atom |
| 34 | + let astAtom = AST.Atom(.property(property), .fake) |
| 35 | + self.ccc = .init(members: [.atom(.unconverted(astAtom))]) |
| 36 | + } |
| 37 | +} |
| 38 | + |
| 39 | +extension CharacterClass: RegexComponent { |
| 40 | + public var regex: Regex<Substring> { |
| 41 | + return Regex(node: DSLTree.Node.customCharacterClass(ccc)) |
| 42 | + } |
| 43 | +} |
| 44 | + |
| 45 | +extension CharacterClass { |
| 46 | + public var inverted: CharacterClass { |
| 47 | + CharacterClass(ccc.inverted) |
| 48 | + } |
| 49 | +} |
| 50 | + |
| 51 | +extension RegexComponent where Self == CharacterClass { |
| 52 | + public static var any: CharacterClass { |
| 53 | + .init(DSLTree.CustomCharacterClass(members: [.atom(.any)])) |
| 54 | + } |
| 55 | + |
| 56 | + public static var anyGrapheme: CharacterClass { |
| 57 | + .init(unconverted: .anyGrapheme) |
| 58 | + } |
| 59 | + |
| 60 | + public static var whitespace: CharacterClass { |
| 61 | + .init(unconverted: .whitespace) |
| 62 | + } |
| 63 | + |
| 64 | + public static var digit: CharacterClass { |
| 65 | + .init(unconverted: .digit) |
| 66 | + } |
| 67 | + |
| 68 | + public static var hexDigit: CharacterClass { |
| 69 | + .init(DSLTree.CustomCharacterClass(members: [ |
| 70 | + .range(.char("A"), .char("F")), |
| 71 | + .range(.char("a"), .char("f")), |
| 72 | + .range(.char("0"), .char("9")), |
| 73 | + ])) |
| 74 | + } |
| 75 | + |
| 76 | + public static var horizontalWhitespace: CharacterClass { |
| 77 | + .init(unconverted: .horizontalWhitespace) |
| 78 | + } |
| 79 | + |
| 80 | + public static var newlineSequence: CharacterClass { |
| 81 | + .init(unconverted: .newlineSequence) |
| 82 | + } |
| 83 | + |
| 84 | + public static var verticalWhitespace: CharacterClass { |
| 85 | + .init(unconverted: .verticalWhitespace) |
| 86 | + } |
| 87 | + |
| 88 | + public static var word: CharacterClass { |
| 89 | + .init(unconverted: .word) |
| 90 | + } |
| 91 | +} |
| 92 | + |
| 93 | +extension RegexComponent where Self == CharacterClass { |
| 94 | + /// Returns a character class that matches any character in the given string |
| 95 | + /// or sequence. |
| 96 | + public static func anyOf<S: Sequence>(_ s: S) -> CharacterClass |
| 97 | + where S.Element == Character |
| 98 | + { |
| 99 | + CharacterClass(DSLTree.CustomCharacterClass( |
| 100 | + members: s.map { .atom(.char($0)) })) |
| 101 | + } |
| 102 | + |
| 103 | + /// Returns a character class that matches any unicode scalar in the given |
| 104 | + /// sequence. |
| 105 | + public static func anyOf<S: Sequence>(_ s: S) -> CharacterClass |
| 106 | + where S.Element == UnicodeScalar |
| 107 | + { |
| 108 | + CharacterClass(DSLTree.CustomCharacterClass( |
| 109 | + members: s.map { .atom(.scalar($0)) })) |
| 110 | + } |
| 111 | +} |
| 112 | + |
| 113 | +// Unicode properties |
| 114 | +extension CharacterClass { |
| 115 | + public static func generalCategory(_ category: Unicode.GeneralCategory) -> CharacterClass { |
| 116 | + guard let extendedCategory = category.extendedGeneralCategory else { |
| 117 | + fatalError("Unexpected general category") |
| 118 | + } |
| 119 | + return CharacterClass(property: |
| 120 | + .init(.generalCategory(extendedCategory), isInverted: false, isPOSIX: false)) |
| 121 | + } |
| 122 | +} |
| 123 | + |
| 124 | +/// Range syntax for characters in `CharacterClass`es. |
| 125 | +public func ...(lhs: Character, rhs: Character) -> CharacterClass { |
| 126 | + let range: DSLTree.CustomCharacterClass.Member = .range(.char(lhs), .char(rhs)) |
| 127 | + let ccc = DSLTree.CustomCharacterClass(members: [range], isInverted: false) |
| 128 | + return CharacterClass(ccc) |
| 129 | +} |
| 130 | + |
| 131 | +/// Range syntax for unicode scalars in `CharacterClass`es. |
| 132 | +@_disfavoredOverload |
| 133 | +public func ...(lhs: UnicodeScalar, rhs: UnicodeScalar) -> CharacterClass { |
| 134 | + let range: DSLTree.CustomCharacterClass.Member = .range(.scalar(lhs), .scalar(rhs)) |
| 135 | + let ccc = DSLTree.CustomCharacterClass(members: [range], isInverted: false) |
| 136 | + return CharacterClass(ccc) |
| 137 | +} |
| 138 | + |
| 139 | +extension Unicode.GeneralCategory { |
| 140 | + var extendedGeneralCategory: Unicode.ExtendedGeneralCategory? { |
| 141 | + switch self { |
| 142 | + case .uppercaseLetter: return .uppercaseLetter |
| 143 | + case .lowercaseLetter: return .lowercaseLetter |
| 144 | + case .titlecaseLetter: return .titlecaseLetter |
| 145 | + case .modifierLetter: return .modifierLetter |
| 146 | + case .otherLetter: return .otherLetter |
| 147 | + case .nonspacingMark: return .nonspacingMark |
| 148 | + case .spacingMark: return .spacingMark |
| 149 | + case .enclosingMark: return .enclosingMark |
| 150 | + case .decimalNumber: return .decimalNumber |
| 151 | + case .letterNumber: return .letterNumber |
| 152 | + case .otherNumber: return .otherNumber |
| 153 | + case .connectorPunctuation: return .connectorPunctuation |
| 154 | + case .dashPunctuation: return .dashPunctuation |
| 155 | + case .openPunctuation: return .openPunctuation |
| 156 | + case .closePunctuation: return .closePunctuation |
| 157 | + case .initialPunctuation: return .initialPunctuation |
| 158 | + case .finalPunctuation: return .finalPunctuation |
| 159 | + case .otherPunctuation: return .otherPunctuation |
| 160 | + case .mathSymbol: return .mathSymbol |
| 161 | + case .currencySymbol: return .currencySymbol |
| 162 | + case .modifierSymbol: return .modifierSymbol |
| 163 | + case .otherSymbol: return .otherSymbol |
| 164 | + case .spaceSeparator: return .spaceSeparator |
| 165 | + case .lineSeparator: return .lineSeparator |
| 166 | + case .paragraphSeparator: return .paragraphSeparator |
| 167 | + case .control: return .control |
| 168 | + case .format: return .format |
| 169 | + case .surrogate: return .surrogate |
| 170 | + case .privateUse: return .privateUse |
| 171 | + case .unassigned: return .unassigned |
| 172 | + @unknown default: return nil |
| 173 | + } |
| 174 | + } |
| 175 | +} |
| 176 | + |
| 177 | +// MARK: - Set algebra methods |
| 178 | + |
| 179 | +extension RegexComponent where Self == CharacterClass { |
| 180 | + public init(_ first: CharacterClass, _ rest: CharacterClass...) { |
| 181 | + if rest.isEmpty { |
| 182 | + self.init(first.ccc) |
| 183 | + } else { |
| 184 | + let members: [DSLTree.CustomCharacterClass.Member] = |
| 185 | + (CollectionOfOne(first) + rest).map { .custom($0.ccc) } |
| 186 | + self.init(.init(members: members)) |
| 187 | + } |
| 188 | + } |
| 189 | +} |
| 190 | + |
| 191 | +extension CharacterClass { |
| 192 | + public func union(_ other: CharacterClass) -> CharacterClass { |
| 193 | + CharacterClass(.init(members: [ |
| 194 | + .custom(self.ccc), |
| 195 | + .custom(other.ccc)])) |
| 196 | + } |
| 197 | + |
| 198 | + public func intersection(_ other: CharacterClass) -> CharacterClass { |
| 199 | + CharacterClass(.init(members: [ |
| 200 | + .intersection(self.ccc, other.ccc) |
| 201 | + ])) |
| 202 | + } |
| 203 | + |
| 204 | + public func subtracting(_ other: CharacterClass) -> CharacterClass { |
| 205 | + CharacterClass(.init(members: [ |
| 206 | + .subtraction(self.ccc, other.ccc) |
| 207 | + ])) |
| 208 | + } |
| 209 | + |
| 210 | + public func symmetricDifference(_ other: CharacterClass) -> CharacterClass { |
| 211 | + CharacterClass(.init(members: [ |
| 212 | + .symmetricDifference(self.ccc, other.ccc) |
| 213 | + ])) |
| 214 | + } |
| 215 | +} |
0 commit comments