Fix unicode for socket.io 2.0

SocketIO-MacTests/SocketEngineTest.swift

@@ -60,7 +60,7 @@ class SocketEngineTest: XCTestCase {
             }
         }
 
-        engine.parseEngineMessage("{\"code\": 0, \"message\": \"Unknown transport\"}", fromPolling: false)
+        engine.parseEngineMessage("{\"code\": 0, \"message\": \"Unknown transport\"}")
         waitForExpectations(timeout: 3, handler: nil)
     }
 
@@ -71,7 +71,7 @@ class SocketEngineTest: XCTestCase {
             finalExpectation.fulfill()
         }
 
-        engine.parseEngineMessage("afafafda", fromPolling: false)
+        engine.parseEngineMessage("afafafda")
         waitForExpectations(timeout: 3, handler: nil)
     }
 
@@ -79,11 +79,13 @@ class SocketEngineTest: XCTestCase {
         let expect = expectation(description: "Engine Decodes utf8")
 
         client.on("stringTest") {data, ack in
-            XCTAssertEqual(data[0] as? String, "lïne one\nlīne \rtwo", "Failed string test")
+            XCTAssertEqual(data[0] as? String, "lïne one\nlīne \rtwo𦅙𦅛", "Failed string test")
             expect.fulfill()
         }
 
-        engine.parsePollingMessage("41:42[\"stringTest\",\"lïne one\\nlīne \\rtwo\"]")
+        let stringMessage = "42[\"stringTest\",\"lïne one\\nlīne \\rtwo𦅙𦅛\"]"
+
+        engine.parsePollingMessage("\(stringMessage.utf16.count):\(stringMessage)")
         waitForExpectations(timeout: 3, handler: nil)
     }
 
@@ -116,8 +118,8 @@ class SocketEngineTest: XCTestCase {
             expect.fulfill()
         }
 
-        engine.parseEngineMessage(packetString, fromPolling: false)
+        engine.parseEngineMessage(packetString)
-        engine.parseEngineMessage(b64String, fromPolling: false)
+        engine.parseEngineMessage(b64String)
 
         waitForExpectations(timeout: 3, handler: nil)
     }

Source/SocketEngine.swift

@@ -511,11 +511,10 @@ public final class SocketEngine : NSObject, URLSessionDelegate, SocketEnginePoll
     /// - parameter message: The message to parse.
     /// - parameter fromPolling: Whether this message is from long-polling.
     ///                          If `true` we might have to fix utf8 encoding.
-    public func parseEngineMessage(_ message: String, fromPolling: Bool) {
+    public func parseEngineMessage(_ message: String) {
         DefaultSocketLogger.Logger.log("Got message: %@", type: logType, args: message)
 
         let reader = SocketStringReader(message: message)
-        let fixedString: String
 
         if message.hasPrefix("b4") {
             return handleBase64(message: message)
@@ -527,23 +526,17 @@ public final class SocketEngine : NSObject, URLSessionDelegate, SocketEnginePoll
             return
         }
 
-        if fromPolling && type != .noop && doubleEncodeUTF8 {
-            fixedString = fixDoubleUTF8(message)
-        } else {
-            fixedString = message
-        }
-
         switch type {
         case .message:
-            handleMessage(String(fixedString.characters.dropFirst()))
+            handleMessage(String(message.characters.dropFirst()))
         case .noop:
             handleNOOP()
         case .pong:
-            handlePong(with: fixedString)
+            handlePong(with: message)
         case .open:
-            handleOpen(openData: String(fixedString.characters.dropFirst()))
+            handleOpen(openData: String(message.characters.dropFirst()))
         case .close:
-            handleClose(fixedString)
+            handleClose(message)
         default:
             DefaultSocketLogger.Logger.log("Got unknown packet type", type: logType)
         }

Source/SocketEnginePollable.swift

@@ -90,9 +90,7 @@ extension SocketEnginePollable {
         var postStr = ""
 
         for packet in postWait {
-            let len = packet.characters.count
+            postStr += "\(packet.utf16.count):\(packet)"
-
-            postStr += "\(len):\(packet)"
         }
 
         DefaultSocketLogger.Logger.log("Created POST string: %@", type: "SocketEnginePolling", args: postStr)
@@ -202,13 +200,15 @@ extension SocketEnginePollable {
     func parsePollingMessage(_ str: String) {
         guard str.characters.count != 1 else { return }
 
+        DefaultSocketLogger.Logger.log("Got poll message: %@", type: "SocketEnginePolling", args: str)
+
         var reader = SocketStringReader(message: str)
 
         while reader.hasNext {
             if let n = Int(reader.readUntilOccurence(of: ":")) {
-                parseEngineMessage(reader.read(count: n), fromPolling: true)
+                parseEngineMessage(reader.read(count: n))
             } else {
-                parseEngineMessage(str, fromPolling: true)
+                parseEngineMessage(str)
                 break
             }
         }
@@ -223,15 +223,8 @@ extension SocketEnginePollable {
     /// - parameter withData: The data associated with this message.
     public func sendPollMessage(_ message: String, withType type: SocketEnginePacketType, withData datas: [Data]) {
         DefaultSocketLogger.Logger.log("Sending poll: %@ as type: %@", type: "SocketEnginePolling", args: message, type.rawValue)
-        let fixedMessage: String
 
-        if doubleEncodeUTF8 {
+        postWait.append(String(type.rawValue) + message)
-            fixedMessage = doubleEncodeUTF8(message)
-        } else {
-            fixedMessage = message
-        }
-
-        postWait.append(String(type.rawValue) + fixedMessage)
 
         for data in datas {
             if case let .right(bin) = createBinaryDataForSend(using: data) {

Source/SocketEngineSpec.swift

@@ -125,7 +125,7 @@ import Foundation
     /// - parameter message: The message to parse.
     /// - parameter fromPolling: Whether this message is from long-polling.
     ///                          If `true` we might have to fix utf8 encoding.
-    func parseEngineMessage(_ message: String, fromPolling: Bool)
+    func parseEngineMessage(_ message: String)
 
     /// Writes a message to engine.io, independent of transport.
     ///
@@ -165,24 +165,6 @@ extension SocketEngineSpec {
         }
     }
 
-    func doubleEncodeUTF8(_ string: String) -> String {
-        if let latin1 = string.data(using: String.Encoding.utf8),
-            let utf8 = NSString(data: latin1, encoding: String.Encoding.isoLatin1.rawValue) {
-                return utf8 as String
-        } else {
-            return string
-        }
-    }
-
-    func fixDoubleUTF8(_ string: String) -> String {
-        if let utf8 = string.data(using: String.Encoding.isoLatin1),
-            let latin1 = NSString(data: utf8, encoding: String.Encoding.utf8.rawValue) {
-                return latin1 as String
-        } else {
-            return string
-        }
-    }
-
     /// Send an engine message (4)
     func send(_ msg: String, withData datas: [Data]) {
         write(msg, withType: .message, withData: datas)

Source/SocketEngineWebsocket.swift

@@ -68,7 +68,7 @@ extension SocketEngineWebsocket {
 
     /// Delegate method for when a message is received.
     public func websocketDidReceiveMessage(socket: WebSocket, text: String) {
-        parseEngineMessage(text, fromPolling: false)
+        parseEngineMessage(text)
     }
 
     /// Delegate method for when binary is received.

Source/SocketExtensions.swift

@@ -104,7 +104,7 @@ extension NSDictionary {
 
 extension String {
     func toArray() throws -> [Any] {
-        guard let stringData = data(using: .utf8, allowLossyConversion: false) else { return [] }
+        guard let stringData = data(using: .utf16, allowLossyConversion: false) else { return [] }
         guard let array = try JSONSerialization.jsonObject(with: stringData, options: .mutableContainers) as? [Any] else {
              throw JSONError.notArray
         }
@@ -113,7 +113,7 @@ extension String {
     }
 
     func toNSDictionary() throws -> NSDictionary {
-        guard let binData = data(using: .utf8, allowLossyConversion: false) else { return [:] }
+        guard let binData = data(using: .utf16, allowLossyConversion: false) else { return [:] }
         guard let json = try JSONSerialization.jsonObject(with: binData, options: .allowFragments) as? NSDictionary else {
             throw JSONError.notNSDictionary
         }

Source/SocketIOClientOption.swift

@@ -36,8 +36,8 @@ public enum SocketIOClientOption : ClientOption {
     /// An array of cookies that will be sent during the initial connection.
     case cookies([HTTPCookie])
 
-    /// The node.js socket.io currently does funky things to unicode when doing HTTP long-polling. Passing `true` in
+    /// Deprecated
-    /// this option causes the client to try and fix any bad unicode that might be sent.
+    @available(*, deprecated, message: "No longer needed in socket.io 2.0+")
     case doubleEncodeUTF8(Bool)
 
     /// Any extra HTTP headers that should be sent during the initial connection.

Source/SocketParsable.swift

@@ -107,7 +107,7 @@ extension SocketParsable where Self: SocketIOClientSpec {
             }
         }
         
-        var dataArray = message[message.characters.index(reader.currentIndex, offsetBy: 1)..<message.endIndex]
+        var dataArray = String(message.utf16[message.utf16.index(reader.currentIndex, offsetBy: 1)..<message.utf16.endIndex])!
         
         if type == .error && !dataArray.hasPrefix("[") && !dataArray.hasSuffix("]") {
             dataArray = "[" + dataArray + "]"

Source/SocketStringReader.swift

@@ -24,29 +24,29 @@
 
 struct SocketStringReader {
     let message: String
-    var currentIndex: String.Index
+    var currentIndex: String.UTF16View.Index
     var hasNext: Bool {
-        return currentIndex != message.endIndex
+        return currentIndex != message.utf16.endIndex
     }
 
     var currentCharacter: String {
-        return String(message[currentIndex])
+        return String(UnicodeScalar(message.utf16[currentIndex])!)
     }
 
     init(message: String) {
         self.message = message
-        currentIndex = message.startIndex
+        currentIndex = message.utf16.startIndex
     }
 
     @discardableResult
-    mutating func advance(by: Int) -> String.Index {
+    mutating func advance(by: Int) -> String.UTF16View.Index {
-        currentIndex = message.characters.index(currentIndex, offsetBy: by)
+        currentIndex = message.utf16.index(currentIndex, offsetBy: by)
 
         return currentIndex
     }
 
     mutating func read(count: Int) -> String {
-        let readString = message[currentIndex..<message.characters.index(currentIndex, offsetBy: count)]
+        let readString = String(message.utf16[currentIndex..<message.utf16.index(currentIndex, offsetBy: count)])!
 
         advance(by: count)
 
@@ -54,20 +54,20 @@ struct SocketStringReader {
     }
 
     mutating func readUntilOccurence(of string: String) -> String {
-        let substring = message[currentIndex..<message.endIndex]
+        let substring = message.utf16[currentIndex..<message.utf16.endIndex]
 
-        guard let foundRange = substring.range(of: string) else {
+        guard let foundIndex = substring.index(of: string.utf16.first!) else {
-            currentIndex = message.endIndex
+            currentIndex = message.utf16.endIndex
 
-            return substring
+            return String(substring)!
         }
 
-        advance(by: message.characters.distance(from: message.characters.startIndex, to: foundRange.lowerBound) + 1)
+        advance(by: substring.distance(from: substring.startIndex, to: foundIndex) + 1)
 
-        return substring.substring(to: foundRange.lowerBound)
+        return String(substring[substring.startIndex..<foundIndex])!
     }
 
     mutating func readUntilEnd() -> String {
-        return read(count: message.characters.distance(from: currentIndex, to: message.endIndex))
+        return read(count: message.utf16.distance(from: currentIndex, to: message.utf16.endIndex))
     }
 }