feat: update simple model call

Author: AtharvaVaidya

Package.resolved

@@ -62,4 +62,61 @@ public final actor CompilerClient {
     public func makeStreamingSession() -> StreamConfiguration {
         configuration.streamingChat
     }
+
+    /// Generate text from a prompt using the specified model
+    /// - Parameters:
+    ///   - prompt: The input prompt
+    ///   - model: The model configuration to use
+    ///   - systemPrompt: Optional system prompt to set context
+    /// - Returns: The complete model response including tokens used, finish reason, etc.
+    public func generateText(
+        prompt: String,
+        using model: StreamConfiguration,
+        systemPrompt: String? = nil
+    ) async throws -> CompletionResponse {
+        try await makeModelCallWithResponse(
+            using: model.metadata,
+            systemPrompt: systemPrompt,
+            userPrompt: prompt
+        )
+    }
+
+    /// Stream text generation from a prompt
+    /// - Parameters:
+    ///   - prompt: The input prompt
+    ///   - model: The model configuration to use
+    ///   - systemPrompt: Optional system prompt to set context
+    /// - Returns: An async stream of response chunks with metadata
+    public func streamText(
+        prompt: String,
+        using model: StreamConfiguration,
+        systemPrompt: String? = nil
+    ) async -> AsyncThrowingStream<String, Error> {
+        let message = Message(role: .user, content: prompt)
+        let messages = systemPrompt.map { [Message(role: .system, content: $0), message] } ?? [message]
+        return makeStreamingModelCall(using: model.metadata, messages: messages)
+    }
+
+    /// Process a natural language command into structured function calls
+    /// - Parameters:
+    ///   - command: The natural language command to process
+    /// - Returns: Array of functions with their parameters
+    /// - Note: You must specify the Parameters type when calling this function, either through type annotation or explicit generic parameter:
+    ///   ```swift
+    ///   // Option 1: Type annotation
+    ///   let functions: [Function<MyParameters>] = try await client.processFunctionCall("Add todo")
+    ///
+    ///   // Option 2: Explicit generic
+    ///   let functions = try await client.processFunctionCall<MyParameters>("Add todo")
+    ///   ```
+    public func processFunctionCall<Parameters: Decodable & Sendable>(
+        _ command: String
+    ) async throws -> [Function<Parameters>] {
+        // We use an empty state since this is the simplified version
+        try await processFunction(command, for: EmptyState(), using: "")
+    }
+}
+
+private struct EmptyState: Encodable, Sendable {
+    // Empty state for simplified function calls
 }

Sources/CompilerSwiftAI/CompilerClient.swift

@@ -71,12 +71,37 @@ struct StreamRequest: ModelCallRequestBase {
     }
 }
 
-/// Response format for completion calls
-struct CompletionResponse: Codable, Sendable {
-    let content: String
+public struct CompletionResponse: Codable, Sendable {
+    private struct Choice: Codable {
+        struct Message: Codable {
+            let content: String
+        }
+
+        let message: Message
+    }
+
+    private let choices: [Choice]
+
+    /// The generated text content
+    public var content: String {
+        choices.first?.message.content ?? ""
+    }
 }
 
-/// Response format for streaming calls - each chunk
-struct StreamChunk: Codable, Sendable {
-    let content: String
+/// Response format for streaming chunks
+public struct StreamChunk: Codable, Sendable {
+    private struct Choice: Codable {
+        struct Message: Codable {
+            let content: String
+        }
+
+        let message: Message
+    }
+
+    private let choices: [Choice]
+
+    /// The generated text content
+    public var content: String {
+        choices.first?.message.content ?? ""
+    }
 }

Sources/CompilerSwiftAI/Model Calling/ModelCall.swift

@@ -14,6 +14,7 @@ You are a helpful AI Assistant. Be direct, concise, and friendly. Always format
 class ChatViewModel: Transcribable {
     public var isRecording = false
     public var transcribedText = ""
+    public var rmsLevel: Float = 0.0
     public var authStatus: SFSpeechRecognizerAuthorizationStatus = .notDetermined
     public var error: Error?
 
@@ -105,11 +106,11 @@ class ChatViewModel: Transcribable {
                     let stream = try await transcriber.startStream()
                     isRecording = true
 
-                    for try await partialResult in stream {
-                        switch partialResult {
-                        case let .rms(float):
-                            print("rms: \(float)")
-                        case let .transcription(string):
+                    for try await signal in stream {
+                        switch signal {
+                        case .rms(let float):
+                            self.rmsLevel = float
+                        case .transcription(let string):
                             self._userInput = string
                         }
                     }

Sources/CompilerSwiftAI/UI/Chat/ChatView/ChatViewModel.swift

@@ -9,6 +9,7 @@ import Transcriber
 class FunctionChatViewModel<AppState: Encodable & Sendable, Parameters: Decodable & Sendable>: Transcribable {
     public var isRecording = false
     public var transcribedText = ""
+    public var rmsLevel: Float = 0
     public var authStatus: SFSpeechRecognizerAuthorizationStatus = .notDetermined
     public var error: Error?
 
@@ -41,12 +42,12 @@ class FunctionChatViewModel<AppState: Encodable & Sendable, Parameters: Decodabl
                 do {
                     isRecording = true
                     let stream = try await transcriber.startStream()
-
+                    
                     for try await signal in stream {
                         switch signal {
-                        case let .rms(float):
-                            print("float: \(float)")
-                        case let .transcription(string):
+                        case .rms(let float):
+                            rmsLevel = float
+                        case .transcription(let string):
                             inputText = string
                         }
                     }