|
| 1 | +struct DSLTree { |
| 2 | +} |
| 3 | + |
| 4 | +extension DSLTree { |
| 5 | + indirect enum Node: _TreeNode { |
| 6 | + /// ... | ... | ... |
| 7 | + case alternation([Node]) |
| 8 | + |
| 9 | + /// ... ... |
| 10 | + case concatenation([Node]) |
| 11 | + |
| 12 | + /// (...) |
| 13 | + case group(AST.Group.Kind, Node) |
| 14 | + |
| 15 | + /// (?(cond) true-branch | false-branch) |
| 16 | + case conditional( |
| 17 | + AST.Conditional.Condition.Kind, Node, Node) |
| 18 | + |
| 19 | + case quantification( |
| 20 | + AST.Quantification.Amount, |
| 21 | + AST.Quantification.Kind, |
| 22 | + Node) |
| 23 | + |
| 24 | + case customCharacterClass(CustomCharacterClass) |
| 25 | + |
| 26 | + case atom(Atom) |
| 27 | + |
| 28 | + /// Comments, non-semantic whitespace, etc |
| 29 | + // TODO: Do we want this? Could be interesting |
| 30 | + case trivia(String) |
| 31 | + |
| 32 | + // TODO: Probably some atoms, built-ins, etc. |
| 33 | + |
| 34 | + case empty |
| 35 | + |
| 36 | + case stringLiteral(String) |
| 37 | + |
| 38 | + /// An embedded literal |
| 39 | + case regexLiteral(AST) |
| 40 | + |
| 41 | + // MARK: - Tree conversions |
| 42 | + |
| 43 | + /// The target of AST conversion. |
| 44 | + /// |
| 45 | + /// Keeps original AST around for rich syntatic and source information |
| 46 | + case convertedRegexLiteral(Node, AST) |
| 47 | + |
| 48 | + // Fall-back for when conversion fails |
| 49 | + case unconvertedRegexLiteral(AST) |
| 50 | + |
| 51 | + // MARK: - Extensibility points |
| 52 | + |
| 53 | + /// A capturing group (TODO: is it?) with a transformation function |
| 54 | + case groupTransform( |
| 55 | + AST.Group.Kind, |
| 56 | + Node, |
| 57 | + CaptureTransform) |
| 58 | + |
| 59 | + case consumer(_ConsumerInterface) |
| 60 | + |
| 61 | + case consumerValidator(_ConsumerValidatorInterface) |
| 62 | + |
| 63 | + // TODO: Would this just boil down to a consumer? |
| 64 | + case characterPredicate(_CharacterPredicateInterface) |
| 65 | + } |
| 66 | +} |
| 67 | + |
| 68 | +extension DSLTree { |
| 69 | + struct CustomCharacterClass { |
| 70 | + var members: [Member] |
| 71 | + var isInverted: Bool |
| 72 | + |
| 73 | + enum Member { |
| 74 | + case atom(Atom) |
| 75 | + case range(Atom, Atom) |
| 76 | + case custom(CustomCharacterClass) |
| 77 | + |
| 78 | + indirect case intersection(CustomCharacterClass, CustomCharacterClass) |
| 79 | + indirect case subtraction(CustomCharacterClass, CustomCharacterClass) |
| 80 | + indirect case symmetricDifference(CustomCharacterClass, CustomCharacterClass) |
| 81 | + } |
| 82 | + } |
| 83 | + |
| 84 | + enum Atom { |
| 85 | + case char(Character) |
| 86 | + case scalar(Unicode.Scalar) |
| 87 | + |
| 88 | + case unconverted(AST.Atom) |
| 89 | + } |
| 90 | +} |
| 91 | + |
| 92 | +// CollectionConsumer |
| 93 | +typealias _ConsumerInterface = ( |
| 94 | + String, Range<String.Index> |
| 95 | +) -> String.Index? |
| 96 | + |
| 97 | +// Type producing consume |
| 98 | +typealias _ConsumerValidatorInterface = ( |
| 99 | + String, Range<String.Index> |
| 100 | +) -> (Any, Any.Type, String.Index)? |
| 101 | + |
| 102 | +// Character-set (post grapheme segmentation) |
| 103 | +typealias _CharacterPredicateInterface = ( |
| 104 | + (Character) -> Bool |
| 105 | +) |
| 106 | + |
| 107 | +/* |
| 108 | + |
| 109 | + TODO: Use of syntactic types, like group kinds, is a |
| 110 | + little suspect. We may want to figure out a model here. |
| 111 | + |
| 112 | + TODO: Do capturing groups need explicit numbers? |
| 113 | + |
| 114 | + TODO: Are storing closures better/worse than existentials? |
| 115 | + |
| 116 | + */ |
| 117 | + |
| 118 | +extension DSLTree.Node { |
| 119 | + var children: [DSLTree.Node]? { |
| 120 | + switch self { |
| 121 | + |
| 122 | + case let .alternation(v): return v |
| 123 | + case let .concatenation(v): return v |
| 124 | + |
| 125 | + case let .convertedRegexLiteral(n, _): |
| 126 | + // Treat this transparently |
| 127 | + return n.children |
| 128 | + |
| 129 | + case let .group(_, n): return [n] |
| 130 | + case let .groupTransform(_, n, _): return [n] |
| 131 | + case let .quantification(_, _, n): return [n] |
| 132 | + |
| 133 | + case let .conditional(_, t, f): return [t,f] |
| 134 | + |
| 135 | + case .trivia, .empty, .stringLiteral, .regexLiteral, |
| 136 | + .consumer, .consumerValidator, .characterPredicate, |
| 137 | + .customCharacterClass, .unconvertedRegexLiteral, |
| 138 | + .atom: |
| 139 | + return [] |
| 140 | + } |
| 141 | + } |
| 142 | +} |
| 143 | + |
| 144 | +extension DSLTree.Node { |
| 145 | + var ast: AST? { |
| 146 | + switch self { |
| 147 | + case let .regexLiteral(ast): return ast |
| 148 | + case let .unconvertedRegexLiteral(ast): return ast |
| 149 | + case let .convertedRegexLiteral(_, ast): return ast |
| 150 | + default: return nil |
| 151 | + } |
| 152 | + } |
| 153 | +} |
| 154 | + |
| 155 | +extension AST { |
| 156 | + /// Converts an AST node to a `convertedRegexLiteral` node. |
| 157 | + var dslTreeNode: DSLTree.Node { |
| 158 | + func wrap(_ node: DSLTree.Node) -> DSLTree.Node { |
| 159 | + switch node { |
| 160 | + case .convertedRegexLiteral: |
| 161 | + assertionFailure("Double wrapping?") |
| 162 | + default: |
| 163 | + break |
| 164 | + } |
| 165 | + // TODO: Should we do this for the |
| 166 | + // single-concatenation child too, or should? |
| 167 | + // we wrap _that_? |
| 168 | + return .convertedRegexLiteral(node, self) |
| 169 | + } |
| 170 | + |
| 171 | + // Convert the top-level node without wrapping |
| 172 | + func convert() -> DSLTree.Node { |
| 173 | + switch self { |
| 174 | + case let .alternation(v): |
| 175 | + let children = v.children.map(\.dslTreeNode) |
| 176 | + return .alternation(children) |
| 177 | + |
| 178 | + case let .concatenation(v): |
| 179 | + // Coalesce adjacent children who can produce a |
| 180 | + // string literal representation |
| 181 | + let astChildren = v.children |
| 182 | + func coalesce( |
| 183 | + _ idx: inout Array<AST>.Index |
| 184 | + ) -> String? { |
| 185 | + var result = "" |
| 186 | + while idx < astChildren.endIndex { |
| 187 | + let atom: AST.Atom? = astChildren[idx].as() |
| 188 | + guard let str = atom?.literalStringValue else { |
| 189 | + break |
| 190 | + } |
| 191 | + result += str |
| 192 | + astChildren.formIndex(after: &idx) |
| 193 | + } |
| 194 | + return result.isEmpty ? nil : result |
| 195 | + } |
| 196 | + |
| 197 | + // No need to nest single children concatenations |
| 198 | + if astChildren.count == 1 { |
| 199 | + return astChildren.first!.dslTreeNode |
| 200 | + } |
| 201 | + |
| 202 | + // Check for a single child post-coalescing |
| 203 | + var idx = astChildren.startIndex |
| 204 | + if let str = coalesce(&idx), |
| 205 | + idx == astChildren.endIndex |
| 206 | + { |
| 207 | + return .stringLiteral(str) |
| 208 | + } |
| 209 | + |
| 210 | + // Coalesce adjacent string children |
| 211 | + var curIdx = astChildren.startIndex |
| 212 | + var children = Array<DSLTree.Node>() |
| 213 | + while curIdx < astChildren.endIndex { |
| 214 | + if let str = coalesce(&curIdx) { |
| 215 | + // TODO: Track source info... |
| 216 | + children.append(.stringLiteral(str)) |
| 217 | + } else { |
| 218 | + children.append(astChildren[curIdx].dslTreeNode) |
| 219 | + children.formIndex(after: &curIdx) |
| 220 | + } |
| 221 | + } |
| 222 | + return .concatenation(children) |
| 223 | + |
| 224 | + case let .group(v): |
| 225 | + let child = v.child.dslTreeNode |
| 226 | + return .group(v.kind.value, child) |
| 227 | + |
| 228 | + case let .conditional(v): |
| 229 | + let trueBranch = v.trueBranch.dslTreeNode |
| 230 | + let falseBranch = v.falseBranch.dslTreeNode |
| 231 | + return .conditional( |
| 232 | + v.condition.kind, trueBranch, falseBranch) |
| 233 | + |
| 234 | + case let .quantification(v): |
| 235 | + let child = v.child.dslTreeNode |
| 236 | + return .quantification( |
| 237 | + v.amount.value, v.kind.value, child) |
| 238 | + |
| 239 | + case let .quote(v): |
| 240 | + return .stringLiteral(v.literal) |
| 241 | + |
| 242 | + case let .trivia(v): |
| 243 | + return .trivia(v.contents) |
| 244 | + |
| 245 | + case let .atom(v): |
| 246 | + if let str = v.literalStringValue { |
| 247 | + return .stringLiteral(str) |
| 248 | + } |
| 249 | + return .atom(v.dslTreeAtom) |
| 250 | + |
| 251 | + case let .customCharacterClass(ccc): |
| 252 | + return .customCharacterClass(ccc.dslTreeClass) |
| 253 | + |
| 254 | + case .empty(_): |
| 255 | + return .empty |
| 256 | + |
| 257 | + case let .groupTransform(v, transform): |
| 258 | + let child = v.child.dslTreeNode |
| 259 | + return .groupTransform( |
| 260 | + v.kind.value, child, transform) |
| 261 | + } |
| 262 | + } |
| 263 | + |
| 264 | + return wrap(convert()) |
| 265 | + } |
| 266 | +} |
| 267 | + |
| 268 | +extension AST.CustomCharacterClass { |
| 269 | + var dslTreeClass: DSLTree.CustomCharacterClass { |
| 270 | + // TODO: Not quite 1-1 |
| 271 | + func convert( |
| 272 | + _ member: Member |
| 273 | + ) -> DSLTree.CustomCharacterClass.Member { |
| 274 | + switch member { |
| 275 | + case let .custom(ccc): |
| 276 | + return .custom(ccc.dslTreeClass) |
| 277 | + |
| 278 | + case let .range(r): |
| 279 | + return .range( |
| 280 | + r.lhs.dslTreeAtom, r.rhs.dslTreeAtom) |
| 281 | + |
| 282 | + case let .atom(a): |
| 283 | + return .atom(a.dslTreeAtom) |
| 284 | + |
| 285 | + case let .quote(q): |
| 286 | + // TODO: Probably should flatten instead of nest |
| 287 | + return .custom(.init( |
| 288 | + members: q.literal.map { .atom(.char($0)) }, |
| 289 | + isInverted: false)) |
| 290 | + |
| 291 | + case let .setOperation(lhs, op, rhs): |
| 292 | + let lhs = DSLTree.CustomCharacterClass( |
| 293 | + members: lhs.map(convert), |
| 294 | + isInverted: false) |
| 295 | + let rhs = DSLTree.CustomCharacterClass( |
| 296 | + members: rhs.map(convert), |
| 297 | + isInverted: false) |
| 298 | + |
| 299 | + switch op.value { |
| 300 | + case .subtraction: |
| 301 | + return .subtraction(lhs, rhs) |
| 302 | + case .intersection: |
| 303 | + return .intersection(lhs, rhs) |
| 304 | + case .symmetricDifference: |
| 305 | + return .symmetricDifference(lhs, rhs) |
| 306 | + } |
| 307 | + } |
| 308 | + } |
| 309 | + |
| 310 | + return .init( |
| 311 | + members: members.map(convert), |
| 312 | + isInverted: self.isInverted) |
| 313 | + } |
| 314 | +} |
| 315 | + |
| 316 | +extension AST.Atom { |
| 317 | + var dslTreeAtom: DSLTree.Atom { |
| 318 | + if case let .scalar(s) = kind { |
| 319 | + return .scalar(s) |
| 320 | + } |
| 321 | + if case let .char(c) = kind { |
| 322 | + return .char(c) |
| 323 | + } |
| 324 | + |
| 325 | + // TODO: More conversions |
| 326 | + return .unconverted(self) |
| 327 | + } |
| 328 | +} |
0 commit comments