Refined UI. Fixed connection errors as well as reduced latency to rem…

…ote model.

Sidekick.xcodeproj/project.pbxproj

@@ -581,6 +581,13 @@
 			path = "GPU Info";
 			sourceTree = "<group>";
 		};
+		2048941B2CE76ABD00473107 /* View Modifiers */ = {
+			isa = PBXGroup;
+			children = (
+			);
+			path = "View Modifiers";
+			sourceTree = "<group>";
+		};
 		204C79542CB765C100EED58E /* Profile */ = {
 			isa = PBXGroup;
 			children = (
@@ -762,6 +769,7 @@
 				20571F752CBD5FF4004F3344 /* Settings */,
 				20B9E6132CB1458500BA0AD2 /* Misc */,
 				20E48C272CB560A30085E7D6 /* Styles */,
+				2048941B2CE76ABD00473107 /* View Modifiers */,
 			);
 			path = Views;
 			sourceTree = "<group>";

Sidekick/Localizable.xcstrings

@@ -1191,12 +1191,12 @@
         }
       }
     },
-    "Remote Model on Server" : {
+    "Remote Model: " : {
       "localizations" : {
         "zh-Hans" : {
           "stringUnit" : {
             "state" : "translated",
-            "value" : "远程模型"
+            "value" : "远端模型： "
           }
         }
       }
@@ -1301,6 +1301,16 @@
         }
       }
     },
+    "Save as Image" : {
+      "localizations" : {
+        "zh-Hans" : {
+          "stringUnit" : {
+            "state" : "translated",
+            "value" : "导出图片"
+          }
+        }
+      }
+    },
     "Say hi" : {
       "localizations" : {
         "zh-Hans" : {
@@ -1641,12 +1651,12 @@
         }
       }
     },
-    "The endpoint on the server used for inference. This endpoint must be accessible from this device, and provide an OpenAI compatible API. (e.g. http://localhost:11434, where http://localhost:11434/v1/chat/completions is accessible)" : {
+    "The endpoint on the server used for inference. This endpoint must be accessible from this device, and provide an OpenAI compatible API. (e.g. http://localhost:11434, where http://localhost:11434/v1/chat/completions is accessible)\n\nTo ensure privacy and security of your data, host your own server." : {
       "localizations" : {
         "zh-Hans" : {
           "stringUnit" : {
             "state" : "translated",
-            "value" : "用于推理的服务器上的端点。此端点必须可从此设备访问，并提供与 OpenAI 兼容的 API。 （例如 http://localhost:11434，其中 http://localhost:11434/v1/chat/completions 可被访问）"
+            "value" : "用于推理的服务器上的端点。此端点必须可从此设备访问，并提供与 OpenAI 兼容的 API。 （例如 http://localhost:11434，其中 http://localhost:11434/v1/chat/completions 可被访问）\n\n为保障隐私和数据安全，我们建议你托管您自己的服务器。"
           }
         }
       }

Sidekick/Logic/Types/Conversation/Message.swift

@@ -14,7 +14,8 @@ public struct Message: Identifiable, Codable, Hashable {
 	init(
 		text: String,
 		sender: Sender,
-		usedServer: Bool? = nil
+		model: String? = nil,
+		usedServer: Bool = false
 	) {
 		self.id = UUID()
 		self.text = text.replacingOccurrences(
@@ -29,13 +30,13 @@ public struct Message: Identifiable, Codable, Hashable {
 		self.startTime = .now
 		self.lastUpdated = .now
 		self.outputEnded = false
+		var modelName: String = model ?? String(
+			localized: "Unknown"
+		)
 		if usedServer == true {
-			self.model = String(localized: "Remote Model on Server")
-		} else {
-			self.model = Settings.modelUrl?.lastPathComponent ?? String(
-				localized: "Unknown"
-			)
+			modelName = String(localized: "Remote Model: ") + modelName
 		}
+		self.model = modelName
 	}
 
 	/// Stored property for `Identifiable` conformance

Sidekick/Logic/llama.cpp/LlamaServer.swift

@@ -45,16 +45,20 @@ public actor LlamaServer {
 	private var process: Process = Process()
 
 	/// Function to get path to llama-server
-	private func url(_ path: String) async -> URL {
+	private func url(_ path: String) async -> (
+		url: URL,
+		usingRemoteServer: Bool
+	) {
 		// Check endpoint
 		let endpoint: String = InferenceSettings.endpoint
 		let urlString: String
-		if await !Self.remoteServerIsReachable() || !InferenceSettings.useServer {
+		let notUsingServer: Bool = await !Self.remoteServerIsReachable() || !InferenceSettings.useServer
+		if notUsingServer {
 			 urlString = "\(scheme)://\(host):\(port)\(path)"
 		} else {
 			urlString = "\(endpoint)\(path)"
 		}
-		return URL(string: urlString)!
+		return (URL(string: urlString)!, !notUsingServer)
 	}
 
 	/// Function to check if the remote server is reachable
@@ -65,12 +69,19 @@ public actor LlamaServer {
 		let endpointUrl: URL = URL(
 			string: "\(InferenceSettings.endpoint)/health"
 		)!
-		let serverHealth = ServerHealth()
-		await serverHealth.updateURL(endpointUrl)
-		await serverHealth.check()
-		let score: Double = await serverHealth.score
-		let serverIsHealthy: Bool = score > 0.25
-		return serverIsHealthy
+		do {
+			let response: (data: Data, URLResponse) = try await URLSession.shared.data(
+				from: endpointUrl
+			)
+			let decoder: JSONDecoder = JSONDecoder()
+			let healthStatus: HealthResponse = try decoder.decode(
+				HealthResponse.self,
+				from: response.data
+			)
+			return healthStatus.isHealthy
+		} catch {
+			return false
+		}
 	}
 
 	/// Function to start a monitor process that will terminate the server when our app dies
@@ -99,10 +110,6 @@ public actor LlamaServer {
 
 	/// Function to start the `llama-server` process
 	private func startServer() async throws {
-		// If a server is used, exit
-		if await Self.remoteServerIsReachable() && InferenceSettings.useServer {
-			return
-		}
 		// If server is running, exit
 		guard !process.isRunning, let modelPath = self.modelPath else { return }
 		await stopServer()
@@ -179,22 +186,29 @@ public actor LlamaServer {
 		progressHandler: (@Sendable (String) -> Void)? = nil
 	) async throws -> CompleteResponse {
 
+		let rawUrl = await self.url("/v1/chat/completions")
+
 		let start: CFAbsoluteTime = CFAbsoluteTimeGetCurrent()
-		try await startServer()
+		if !rawUrl.usingRemoteServer {
+			try await startServer()
+		}
 
 		// Hit localhost for completion
 		async let params = ChatParameters(
 			messages: messages,
 			systemPrompt: systemPrompt,
 			similarityIndex: similarityIndex
 		)
-		var request = await URLRequest(
-			url: url("/v1/chat/completions")
+		var request = URLRequest(
+			url: rawUrl.url
 		)
 		request.httpMethod = "POST"
 		request.setValue("application/json", forHTTPHeaderField: "Content-Type")
 		request.setValue("text/event-stream", forHTTPHeaderField: "Accept")
 		request.setValue("keep-alive", forHTTPHeaderField: "Connection")
+		if rawUrl.usingRemoteServer {
+			request.setValue("nil", forHTTPHeaderField: "ngrok-skip-browser-warning")
+		}
 		request.httpBody = await params.toJSON().data(using: .utf8)
 		// Use EventSource to receive server sent events
 		eventSource = EventSource(request: request)
@@ -265,7 +279,8 @@ public actor LlamaServer {
 			responseStartSeconds: responseDiff,
 			predictedPerSecond: Double(tokens) / generationTime,
 			modelName: modelName,
-			nPredicted: tokens
+			nPredicted: tokens,
+			usedServer: rawUrl.usingRemoteServer
 		)
 	}
 
@@ -280,7 +295,7 @@ public actor LlamaServer {
 		serverErrorMessage = ""
 
 		let serverHealth = ServerHealth()
-		await serverHealth.updateURL(url("/health"))
+		await serverHealth.updateURL(url("/health").url)
 		await serverHealth.check()
 
 		var timeout = 30
@@ -302,6 +317,14 @@ public actor LlamaServer {
 		}
 	}
 
+	struct HealthResponse: Codable {
+
+		var status: String
+
+		var isHealthy: Bool { self.status == "ok" }
+
+	}
+
 	struct StreamMessage: Codable {
 		let content: String?
 	}
@@ -332,6 +355,7 @@ public actor LlamaServer {
 		var predictedPerSecond: Double?
 		var modelName: String?
 		var nPredicted: Int?
+		var usedServer: Bool
 	}
 
 }

Sidekick/Views/Conversation/Controls/Input Field/PromptInputField.swift

@@ -145,8 +145,6 @@ struct PromptInputField: View {
 	private func getResponse(
 		tempResources: [TemporaryResource]
 	) async {
-		// Check if remote server was used
-		let usedServer: Bool = await LlamaServer.remoteServerIsReachable() && InferenceSettings.useServer
 		// If processing, use recursion to update
 		if (model.status == .processing || model.status == .coldProcessing) {
 			Task {
@@ -201,7 +199,8 @@ struct PromptInputField: View {
 			var responseMessage: Message = Message(
 				text: "",
 				sender: .assistant,
-				usedServer: usedServer
+				model: response.modelName,
+				usedServer: response.usedServer
 			)
 			responseMessage.update(
 				newText: response.text,

Sidekick/Views/Conversation/Messages/Message/MessageBackgroundView.swift

@@ -9,34 +9,36 @@ import SwiftUI
 
 struct MessageBackgroundView: View {
 
-	@Environment(\.colorScheme) private var colorScheme
+	private let cornerRadius: CGFloat = 13
+	private let borderWidth: CGFloat = 0.5
 
-	var shadowColor: SwiftUI.Color {
-		return colorScheme == .dark ? .white : .black
+	var body: some View {
+		unevenRoundedRectangle(cornerRadius)
+			.fill(
+				Color(nsColor: .textBackgroundColor)
+			)
+			.padding(borderWidth)
+			.background {
+				unevenRoundedRectangle(cornerRadius + borderWidth)
+					.fill(Color.secondary)
+					.opacity(0.5)
+			}
 	}
 
-	var shadowRadius: CGFloat {
-		return colorScheme == .dark ? 2.5 : 0
-	}
-
-    var body: some View {
+	private func unevenRoundedRectangle(
+		_ cornerRadius: CGFloat
+	) -> some Shape {
 		UnevenRoundedRectangle(
 			cornerRadii: .init(
 				topLeading: 0,
-				bottomLeading: 12,
-				bottomTrailing: 12,
-				topTrailing: 12
+				bottomLeading: cornerRadius,
+				bottomTrailing: cornerRadius,
+				topTrailing: cornerRadius
 			),
 			style: .circular
 		)
-		.fill(
-			Color(nsColor: .textBackgroundColor)
-		)
-		.shadow(
-			color: shadowColor,
-			radius: shadowRadius
-		)
-    }
+	}
+
 }
 
 #Preview {

Sidekick/Views/Conversation/Messages/MessageShareMenu.swift

@@ -64,7 +64,7 @@ struct MessageShareMenu<MessagesView: View>: View {
 	}
 
 	var pngButton: some View {
-		Button("Save as Image") {
+		Button {
 			VStack(
 				alignment: .leading,
 				spacing: 15
@@ -75,6 +75,8 @@ struct MessageShareMenu<MessagesView: View>: View {
 			.background(Color.gray)
 			.frame(width: 1000)
 			.generatePng()
+		} label: {
+			Text("Save as Image")
 		}
 	}
 

Sidekick/Views/Settings/InferenceSettingsView.swift

@@ -219,7 +219,7 @@ struct InferenceSettingsView: View {
 				Text("Endpoint")
 					.font(.title3)
 					.bold()
-				Text("The endpoint on the server used for inference. This endpoint must be accessible from this device, and provide an OpenAI compatible API. (e.g. http://localhost:11434, where http://localhost:11434/v1/chat/completions is accessible)")
+				Text("The endpoint on the server used for inference. This endpoint must be accessible from this device, and provide an OpenAI compatible API. (e.g. http://localhost:11434, where http://localhost:11434/v1/chat/completions is accessible)\n\nTo ensure privacy and security of your data, host your own server.")
 					.font(.caption)
 			}
 			Spacer()