Improve uts46 file parsing

- Make header a struct. - Automatically load from the owning bundle if necessary (and make UTS46 a class so Bundle(for:) works).
Wevah · Apr 18, 2020 · 6178ffc · 6178ffc
1 parent 5230a5d
commit 6178ffc
Show file tree

Hide file tree

Showing 4 changed files with 103 additions and 14 deletions.
diff --git a/Punycode.xcodeproj/project.pbxproj b/Punycode.xcodeproj/project.pbxproj
@@ -34,7 +34,7 @@
 		B27409C92443E45B00E567E7 /* Data+Extensions.swift in Sources */ = {isa = PBXBuildFile; fileRef = B27409C82443E45B00E567E7 /* Data+Extensions.swift */; };
 		B27409CA2443F66900E567E7 /* Data+Extensions.swift in Sources */ = {isa = PBXBuildFile; fileRef = B27409C82443E45B00E567E7 /* Data+Extensions.swift */; };
 		B2AF210724313283009F5EE7 /* UTS46.swift in Sources */ = {isa = PBXBuildFile; fileRef = B2AF210624313283009F5EE7 /* UTS46.swift */; };
-		B2AF21092431376E009F5EE7 /* uts46.xz in Resources */ = {isa = PBXBuildFile; fileRef = B2AF21082431376E009F5EE7 /* uts46.xz */; };
+		B2AF21092431376E009F5EE7 /* uts46 in Resources */ = {isa = PBXBuildFile; fileRef = B2AF21082431376E009F5EE7 /* uts46 */; };
 		B2B4F3D0242EB79E00CC752F /* main.swift in Sources */ = {isa = PBXBuildFile; fileRef = B2B4F3CF242EB79E00CC752F /* main.swift */; };
 		B2C3DD45241FECF200484850 /* AppDelegate.swift in Sources */ = {isa = PBXBuildFile; fileRef = B2C3DD44241FECF200484850 /* AppDelegate.swift */; };
 		B2C3DD47241FECF400484850 /* Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = B2C3DD46241FECF400484850 /* Assets.xcassets */; };
@@ -140,7 +140,7 @@
 		B26FA03C2055425C00CD6AF1 /* libicucore.tbd */ = {isa = PBXFileReference; lastKnownFileType = "sourcecode.text-based-dylib-definition"; name = libicucore.tbd; path = usr/lib/libicucore.tbd; sourceTree = SDKROOT; };
 		B27409C82443E45B00E567E7 /* Data+Extensions.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = "Data+Extensions.swift"; sourceTree = "<group>"; };
 		B2AF210624313283009F5EE7 /* UTS46.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = UTS46.swift; sourceTree = "<group>"; };
-		B2AF21082431376E009F5EE7 /* uts46.xz */ = {isa = PBXFileReference; lastKnownFileType = file; path = uts46.xz; sourceTree = "<group>"; };
+		B2AF21082431376E009F5EE7 /* uts46 */ = {isa = PBXFileReference; lastKnownFileType = file; path = uts46; sourceTree = "<group>"; };
 		B2B4F3CD242EB79E00CC752F /* icumap2code */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = icumap2code; sourceTree = BUILT_PRODUCTS_DIR; };
 		B2B4F3CF242EB79E00CC752F /* main.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = main.swift; sourceTree = "<group>"; };
 		B2C3DCC1241FE95B00484850 /* String+Punycode.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = "String+Punycode.swift"; sourceTree = "<group>"; };
@@ -347,7 +347,7 @@
 			children = (
 				B25753801F3CE8B2002606DD /* WebNSURLExtras.h */,
 				B2C59E7224278AF400183346 /* uidna-min.h */,
-				B2AF21082431376E009F5EE7 /* uts46.xz */,
+				B2AF21082431376E009F5EE7 /* uts46 */,
 				B2C3DCC0241FE94400484850 /* Swift */,
 				B2C3DCBF241FE92F00484850 /* Objective-C */,
 			);
@@ -644,7 +644,7 @@
 			files = (
 				B2C3DD47241FECF400484850 /* Assets.xcassets in Resources */,
 				B2C3DD56241FEE5A00484850 /* MainMenu.xib in Resources */,
-				B2AF21092431376E009F5EE7 /* uts46.xz in Resources */,
+				B2AF21092431376E009F5EE7 /* uts46 in Resources */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
 		};

diff --git a/Shared/Swift/String+Punycode.swift b/Shared/Swift/String+Punycode.swift
@@ -441,6 +441,8 @@ private extension String {
 	/// - Returns: The mapped string.
 	/// - Throws: `UTS46MapError`
 	func mapUTS46() throws -> String {
+		try! UTS46.loadIfNecessary()
+
 		var result = ""
 
 		for scalar in self.unicodeScalars {
@@ -478,6 +480,8 @@ private extension String {
 	///
 	/// See [RFC 5892, Appendix A.1 and A.2](https://tools.ietf.org/html/rfc5892#appendix-A).
 	var hasValidJoiners: Bool {
+		try! UTS46.loadIfNecessary()
+
 		let scalars = self.unicodeScalars
 
 		for index in scalars.indices {

diff --git a/Shared/Swift/UTS46.swift b/Shared/Swift/UTS46.swift
@@ -8,13 +8,15 @@
 import Foundation
 import Compression
 
-enum UTS46 {
+class UTS46 {
 
 	private(set) static var characterMap: [UInt32: String] = [:]
 	private(set) static var ignoredCharacters: CharacterSet = []
 	private(set) static var disallowedCharacters: CharacterSet = []
 	private(set) static var joiningTypes = [UInt32: JoiningType]()
 
+	private static var isLoaded = false
+
 	private enum Marker {
 		static let characterMap = UInt8.max
 		static let ignoredCharacters = UInt8.max - 1
@@ -38,19 +40,95 @@ enum UTS46 {
 		case badSize
 		case decompressionError
 		case badMarker
+		case unknownDataVersion
+	}
+
+	/// Identical values to `NSData.CompressionAlgorithm`.
+	enum CompressionAlgorithm: Int {
+		case lzfse = 0
+		case lz4 = 1
+		case lzma = 2
+		case zlib = 3
+	}
+
+	struct Header: RawRepresentable, CustomDebugStringConvertible {
+		var rawValue: [UInt8] {
+			return Self.signature + [UInt8(0), version, flags.rawValue]
+		}
+
+		private static let compressionMask: UInt8 = 0x07
+		private static let signature: [UInt8] = Array("UTS#46".utf8)
+
+		struct Flags: RawRepresentable {
+			var rawValue: UInt8 {
+				return (hasCRC ? hasCRCMask : 0) | UInt8(compression != nil ? compression!.rawValue + 1 : 0)
+			}
+
+			var hasCRC: Bool
+			var compression: CompressionAlgorithm?
+
+			private let hasCRCMask: UInt8 = 1 << 3
+			private let compressionMask: UInt8 = 0x7
+
+			init(rawValue: UInt8) {
+				hasCRC = rawValue & hasCRCMask != 0
+				let compressionBits = rawValue & compressionMask
+
+				if (1...4).contains(compressionBits) {
+					compression = CompressionAlgorithm(rawValue: Int(compressionBits) - 1)
+				}
+			}
+
+			init(compression: CompressionAlgorithm? = nil, hasCRC: Bool = false) {
+				self.compression = compression
+				self.hasCRC = hasCRC
+			}
+		}
+
+		let version: UInt8
+		var flags: Flags
+		var hasCRC: Bool { flags.hasCRC }
+		var compression: CompressionAlgorithm? { flags.compression }
+		var dataOffset: Int { 8 + (flags.hasCRC ? 4 : 0) }
+
+		init?<T: DataProtocol>(rawValue: T) where T.Index == Int {
+			guard rawValue.count == 8 else { return nil }
+			guard rawValue.prefix(Self.signature.count).elementsEqual(Self.signature) else { return nil }
+
+			version = rawValue[6]
+			flags = Flags(rawValue: rawValue[7])
+		}
+
+		init(compression: CompressionAlgorithm? = nil, hasCRC: Bool = false) {
+			self.version = 1
+			self.flags = Flags(compression: compression, hasCRC: hasCRC)
+		}
+
+		var debugDescription: String { "has CRC: \(hasCRC); compression: \(compression != nil ? String(describing: compression!) : "none")" }
 	}
 
-	enum CompressionAlgorithm {
-		case lzfse
-		case lz4
-		case lzma
-		case zlib
+	private static func parseHeader(from data: Data) throws -> Header? {
+		let headerData = data.prefix(8)
+
+		guard headerData.count == 8 else { throw UTS46Error.badSize }
+
+		return Header(rawValue: headerData)
 	}
 
-	static func load(from url: URL, compression: CompressionAlgorithm?) throws {
-		let compressedData = try Data(contentsOf: url)
+	static func load(from url: URL) throws {
+		let fileData = try Data(contentsOf: url)
 
-		guard let data = self.decompress(data: compressedData, algorithm: compression) else {
+		guard let header = try? parseHeader(from: fileData) else { return }
+
+		guard header.version == 1 else { throw UTS46Error.unknownDataVersion }
+
+		let offset = 8 + (header.hasCRC ? 4 : 0)
+
+		guard fileData.count > offset else { throw UTS46Error.badSize }
+
+		let compressedData = fileData[offset...]
+
+		guard let data = self.decompress(data: compressedData, algorithm: header.compression) else {
 			throw UTS46Error.decompressionError
 		}
 
@@ -75,6 +153,14 @@ enum UTS46 {
 			}
 		}
 
+		isLoaded = true
+	}
+
+	static func loadIfNecessary() throws {
+		guard !isLoaded else { return }
+		guard let url = Bundle(for: Self.self).url(forResource: "uts46", withExtension: nil) else { throw CocoaError(.fileNoSuchFile) }
+
+		try load(from: url)
 	}
 
 	private static func decompress(data: Data, algorithm: CompressionAlgorithm?) -> Data? {
@@ -92,7 +178,6 @@ enum UTS46 {
 				rawAlgorithm = COMPRESSION_ZLIB
 			default:
 				return data
-
 		}
 
 		let capacity = 100_000

diff --git a/Shared/uts46.xz → Shared/uts46 b/Shared/uts46.xz → Shared/uts46