パーサを作る

.github/workflows/test.yml

@@ -8,12 +8,14 @@ on:
 
 jobs:
   test:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-22.04
     steps:
-      - uses: swift-actions/setup-swift@v2
-        with:
-          swift-version: "5.10.1"
       - uses: actions/checkout@v4
+
+      - run: sudo $GITHUB_WORKSPACE/ci/install-swift.bash
+        working-directory: /home/runner
+      - run: swift --version
+
       - run: swift package resolve
       - run: swift build
       - run: swift test

Package.swift

@@ -39,7 +39,10 @@ let package = Package(
         ),
         .testTarget(
             name: "TypeScriptASTTests",
-            dependencies: ["TypeScriptAST"]
+            dependencies: ["TypeScriptAST"],
+            swiftSettings: [
+                .enableUpcomingFeature("BareSlashRegexLiterals")
+            ]
         ),
     ]
 )

Sources/TypeScriptAST/Parser/CharacterEx.swift

@@ -0,0 +1,44 @@
+extension Character {
+    static let tab = Character("\t")
+    static let lf = Character("\n")
+    static let cr = Character("\r")
+    static let crLf = Character("\r\n")
+    static let space = Character(" ")
+    static let exclamation = Character("!")
+    static let doubleQuote = Character("\"")
+    static let dollar = Character("$")
+    static let percent = Character("%")
+    static let ampersand = Character("&")
+    static let singleQuote = Character("'")
+    static let leftParen = Character("(")
+    static let rightParen = Character(")")
+    static let asterisk = Character("*")
+    static let plus = Character("+")
+    static let comma = Character(",")
+    static let minus = Character("-")
+    static let dot = Character(".")
+    static let slash = Character("/")
+    static let _0 = Character("0")
+    static let _9 = Character("9")
+    static let colon = Character(":")
+    static let semicolon = Character(";")
+    static let leftAngleBracket = Character("<")
+    static let equal = Character("=")
+    static let rightAngleBracket = Character(">")
+    static let question = Character("?")
+    static let A = Character("A")
+    static let Z = Character("Z")
+    static let leftSquareBracket = Character("[")
+    static let backslash = Character("\\")
+    static let rightSquareBracket = Character("]")
+    static let underscore = Character("_")
+    static let backQuote = Character("`")
+    static let a = Character("a")
+    static let n = Character("n")
+    static let r = Character("r")
+    static let t = Character("t")
+    static let z = Character("z")
+    static let leftBrace = Character("{")
+    static let pipe = Character("|")
+    static let rightBrace = Character("}")
+}

Sources/TypeScriptAST/Parser/Keyword.swift

@@ -0,0 +1,8 @@
+public enum Keyword: String, Equatable & CustomStringConvertible {
+    case `import`
+    case from
+
+    public var description: String {
+        rawValue
+    }
+}

Sources/TypeScriptAST/Parser/Parser.swift

@@ -0,0 +1,73 @@
+public struct Parser {
+    public init(string: String) {
+        self.tokenizer = Tokenizer(string: string)
+    }
+
+    private var tokenizer: Tokenizer
+
+    public mutating func parse() -> TSSourceFile {
+        var es: [any ASTNode] = []
+        while let e = parseElement() {
+            es.append(e)
+        }
+        return TSSourceFile(es)
+    }
+
+    public mutating func parseElement() -> (any ASTNode)? {
+        while true {
+            switch nextToken {
+            case nil: return nil
+            case .keyword(let k):
+                switch k {
+                case .import:
+                    if let x = parseImport() {
+                        return x
+                    }
+                default:
+                    print("skip keyword: \(k)")
+                    readToken()
+                }
+            default:
+                if let x = parseExpression() {
+                    return x
+                }
+            }
+        }
+    }
+
+    private mutating func parseImport() -> TSImportDecl? {
+        let imp = readToken()
+        guard imp == .keyword(.import) else { return nil }
+        let lb = readToken()
+        guard lb == .symbol(.leftBrace) else { return nil }
+        var names: [String] = []
+        loop: while true {
+            guard let tk = readToken() else { return nil }
+            switch tk {
+            case .symbol(.rightBrace):
+                break loop
+            default:
+                names.append(tk.description)
+            }
+        }
+        let from = readToken()
+        guard from == .keyword(.from) else { return nil }
+        let pathToken = readToken()
+        guard case .stringLiteral(let path) = pathToken else { return nil }
+        return TSImportDecl(names: names, from: path)
+    }
+
+    private mutating func parseExpression() -> (any ASTNode)? {
+        readToken()
+        return nil
+    }
+
+    @discardableResult
+    private mutating func readToken() -> Token? {
+        tokenizer.read()
+    }
+
+    private var nextToken: Token? {
+        tokenizer.nextToken
+    }
+}

Sources/TypeScriptAST/Parser/Symbol.swift

@@ -0,0 +1,54 @@
+public enum Symbol: String, Hashable & CustomStringConvertible {
+    case exclamation = "!" // 21
+    case exclamationEqual = "!="
+    case exclamationEqualEqual = "!=="
+    case percent = "%" // 25
+    case percentEqual = "%="
+    case ampersand = "&" // 26
+    case ampersandAmpersand = "&&"
+    case ampersandAmpersandEqual = "&&="
+    case ampersandEqual = "&="
+    case leftParen = "(" // 28
+    case rightParen = ")" // 29
+    case asterisk = "*" // 30
+    case asteriskEqual = "*="
+    case plus = "+" // 31
+    case plusPlus = "++"
+    case plusEqual = "+="
+    case comma = "," // 32
+    case minus = "-" // 33
+    case minusMinus = "--"
+    case minusEqual = "-="
+    case dot = "." // 34
+    case slash = "/" // 35
+    case slashEqual = "/="
+    case colon = ":" // 3a
+    case semicolon = ";" // 3b
+    case leftAngleBracket = "<" // 3c
+    case leftAngleBracketLeftAngleBracket = "<<"
+    case leftAngleBracketLeftAngleBracketEqual = "<<="
+    case equal = "=" // 3d
+    case equalEqual = "=="
+    case equalEqualEqual = "==="
+    case equalRightAngleBracket = "=>"
+    case rightAngleBracket = ">" // 3e
+    case rightAngleBracketRightAngleBracket = ">>"
+    case rightAngleBracketRightAngleBracketEqual = ">>="
+    case question = "?" // 3f
+    case questionDot = "?."
+    case questionQuestion = "??"
+    case questionQuestionEqual = "??="
+    case leftSquareBracket = "[" // 5b
+    case backslash = "\\" // 5c
+    case rightSquareBracket = "]" // 5d
+    case leftBrace = "{" // 7b
+    case pipe = "|" // 7c
+    case pipeEqual = "|="
+    case pipePipe = "||"
+    case pipePipeEqual = "||="
+    case rightBrace = "}" // 7d
+
+    public var description: String {
+        rawValue
+    }
+}

Sources/TypeScriptAST/Parser/Token.swift

@@ -0,0 +1,33 @@
+public enum Token: Equatable & CustomStringConvertible & CustomDebugStringConvertible {
+    case keyword(Keyword)
+    case identifier(String)
+    case symbol(Symbol)
+    case stringLiteral(String)
+
+    public var description: String {
+        switch self {
+        case .keyword(let x): return x.description
+        case .identifier(let x): return x
+        case .symbol(let x): return x.description
+        case .stringLiteral(let x): return "\"" + Self.escapeStringLiteralContent(x) + "\""
+        }
+    }
+
+    public var debugDescription: String {
+        switch self {
+        case .keyword(let x): return "keyword(\(x))"
+        case .identifier(let x): return "identifier(\(x))"
+        case .symbol(let x): return "symbol(\(x))"
+        case .stringLiteral(let x): return "stringLiteral(\(x))"
+        }
+    }
+
+    public static func escapeStringLiteralContent(_ string: String) -> String {
+        var s = string
+        s = s.replacingOccurrences(of: "\\", with: "\\\\")
+        s = s.replacingOccurrences(of: "\"", with: "\\\"")
+        s = s.replacingOccurrences(of: "\n", with: "\\n")
+        return s
+    }
+}
+