feat: open-source 小浣熊阅读 for iOS

2026-10-24 12:51:22 +08:00 · 2026-10-14 11:37:36 +08:00
commit ffc9e9d91d
193 changed files with 41382 additions and 0 deletions
@@ -0,0 +1,189 @@
+import Foundation
+
+public enum RuleMode: String, Codable, Sendable {
+    case css
+    case xpath
+    case jsonPath
+    case regex
+    case javaScript
+    case webJavaScript
+}
+
+public enum RuleCombination: String, Codable, Sendable {
+    case single
+    case concatenate = "&&"
+    case fallback = "||"
+    case interleave = "%%"
+}
+
+public struct RuleBranch: Codable, Hashable, Sendable {
+    public var mode: RuleMode
+    public var expression: String
+    public var replacePattern: String?
+    public var replacement: String
+    public var replaceFirst: Bool
+
+    public init(
+        mode: RuleMode,
+        expression: String,
+        replacePattern: String? = nil,
+        replacement: String = "",
+        replaceFirst: Bool = false
+    ) {
+        self.mode = mode
+        self.expression = expression
+        self.replacePattern = replacePattern
+        self.replacement = replacement
+        self.replaceFirst = replaceFirst
+    }
+}
+
+public struct RulePlan: Codable, Hashable, Sendable {
+    public var source: String
+    public var combination: RuleCombination
+    public var branches: [RuleBranch]
+}
+
+public struct RuleExpressionParser: Sendable {
+    public init() {}
+
+    public func parse(_ source: String, inputLooksLikeJSON: Bool = false) throws -> RulePlan {
+        let normalized = source.trimmingCharacters(in: .newlines)
+        guard !normalized.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty else {
+            throw RuleEngineError.malformedRule(rule: source, detail: "规则不能为空")
+        }
+        let split = try splitAtTopLevel(normalized)
+        let branches = try split.parts.map { try parseBranch($0, inputLooksLikeJSON: inputLooksLikeJSON) }
+        return .init(source: source, combination: split.combination, branches: branches)
+    }
+
+    private func parseBranch(_ source: String, inputLooksLikeJSON: Bool) throws -> RuleBranch {
+        var value = String(source.drop(while: { $0.isWhitespace }))
+        var mode: RuleMode
+        if value.lowercased().hasPrefix("@css:") {
+            mode = .css
+            value.removeFirst(5)
+        } else if value.hasPrefix("@@") {
+            mode = .css
+            value.removeFirst(2)
+        } else if value.lowercased().hasPrefix("@xpath:") {
+            mode = .xpath
+            value.removeFirst(7)
+        } else if value.lowercased().hasPrefix("@json:") {
+            mode = .jsonPath
+            value.removeFirst(6)
+        } else if value.lowercased().hasPrefix("@js:") {
+            mode = .javaScript
+            value.removeFirst(4)
+        } else if value.lowercased().hasPrefix("<js>") && value.lowercased().hasSuffix("</js>") {
+            mode = .javaScript
+            value = String(value.dropFirst(4).dropLast(5))
+        } else if value.lowercased().hasPrefix("<webjs>") && value.lowercased().hasSuffix("</webjs>") {
+            mode = .webJavaScript
+            value = String(value.dropFirst(7).dropLast(8))
+        } else if value.hasPrefix(":") {
+            mode = .regex
+            value.removeFirst()
+        } else if value.hasPrefix("$.") || value.hasPrefix("$[") || inputLooksLikeJSON {
+            mode = .jsonPath
+        } else if value.hasPrefix("/") || value.hasPrefix("./") {
+            mode = .xpath
+        } else {
+            mode = .css
+        }
+
+        let replacement = value.components(separatedBy: "##")
+        let expression = replacement[0].trimmingCharacters(in: .whitespacesAndNewlines)
+        guard !expression.isEmpty || replacement.count > 1 else {
+            throw RuleEngineError.malformedRule(rule: source, detail: "规则主体不能为空")
+        }
+        return .init(
+            mode: mode,
+            expression: expression,
+            replacePattern: replacement.count > 1 ? replacement[1] : nil,
+            replacement: replacement.count > 2 ? replacement[2] : "",
+            replaceFirst: replacement.count > 3
+        )
+    }
+
+    private func splitAtTopLevel(_ source: String) throws -> (parts: [String], combination: RuleCombination) {
+        var quote: Character?
+        var escaped = false
+        var roundDepth = 0
+        var squareDepth = 0
+        var curlyDepth = 0
+        var found: RuleCombination?
+        var parts: [String] = []
+        var partStart = source.startIndex
+        var index = source.startIndex
+
+        while index < source.endIndex {
+            let character = source[index]
+            if escaped {
+                escaped = false
+                index = source.index(after: index)
+                continue
+            }
+            if character == "\\" {
+                escaped = true
+                index = source.index(after: index)
+                continue
+            }
+            if let activeQuote = quote {
+                if character == activeQuote { quote = nil }
+                index = source.index(after: index)
+                continue
+            }
+            if character == "\"" || character == "'" {
+                quote = character
+                index = source.index(after: index)
+                continue
+            }
+            switch character {
+            case "(": roundDepth += 1
+            case ")": roundDepth -= 1
+            case "[": squareDepth += 1
+            case "]": squareDepth -= 1
+            case "{": curlyDepth += 1
+            case "}": curlyDepth -= 1
+            default: break
+            }
+            if roundDepth < 0 || squareDepth < 0 || curlyDepth < 0 {
+                throw RuleEngineError.malformedRule(rule: source, detail: "规则括号未平衡")
+            }
+            if roundDepth == 0, squareDepth == 0, curlyDepth == 0 {
+                let remainder = source[index...]
+                let candidate: RuleCombination? = if remainder.hasPrefix("&&") {
+                    .concatenate
+                } else if remainder.hasPrefix("||") {
+                    .fallback
+                } else if remainder.hasPrefix("%%") {
+                    .interleave
+                } else {
+                    nil
+                }
+                if let candidate {
+                    if found == nil { found = candidate }
+                    if found == candidate {
+                        parts.append(String(source[partStart..<index]))
+                        index = source.index(index, offsetBy: 2)
+                        partStart = index
+                        continue
+                    }
+                }
+            }
+            index = source.index(after: index)
+        }
+        guard quote == nil, roundDepth == 0, squareDepth == 0, curlyDepth == 0 else {
+            throw RuleEngineError.malformedRule(rule: source, detail: "规则引号或括号未闭合")
+        }
+        if let found {
+            parts.append(String(source[partStart...]))
+            guard parts.allSatisfy({ !$0.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty }) else {
+                throw RuleEngineError.malformedRule(rule: source, detail: "组合规则存在空分支")
+            }
+            return (parts, found)
+        }
+        return ([source], .single)
+    }
+}