refactor unload and implement a destructor to clean up sessions

extensions/llamacpp-extension/src/index.ts

@@ -76,7 +76,7 @@ export default class llamacpp_extension extends AIEngine {
   private downloadManager
   private activeSessions: Map<string, sessionInfo> = new Map()
   private modelsBasePath!: string
-  private activeRequests: Map<string, AbortController> = new Map()
+  private enginesPath!: string
 
   override async onLoad(): Promise<void> {
     super.onLoad() // Calls registerEngine() from AIEngine
@@ -89,8 +89,24 @@ export default class llamacpp_extension extends AIEngine {
       await getJanDataFolderPath(),
       'models',
     ])
+
+    this.enginesPath = await joinPath([await getJanDataFolderPath(), 'llamacpp', 'engines'])
   }
 
+  override async onUnload(): Promise<void> {
+    // Terminate all active sessions
+    for (const [sessionId, _] of this.activeSessions) {
+      try {
+        await this.unload(sessionId);
+      } catch (error) {
+        console.error(`Failed to unload session ${sessionId}:`, error);
+      }
+  }
+  
+  // Clear the sessions map
+  this.activeSessions.clear();
+}
+
   // Implement the required LocalProvider interface methods
   override async list(): Promise<modelInfo[]> {
     const modelsDir = await joinPath([this.modelsBasePath, this.provider])
@@ -335,6 +351,7 @@ export default class llamacpp_extension extends AIEngine {
 
     try {
       const sInfo = await invoke<sessionInfo>('load_llama_model', {
+        server_path: this.enginesPath,
         args: args,
       })
 
@@ -348,17 +365,17 @@ export default class llamacpp_extension extends AIEngine {
     }
   }
 
-  override async unload(opts: unloadOptions): Promise<unloadResult> {
+  override async unload(sessionId: string): Promise<unloadResult> {
     try {
       // Pass the PID as the session_id
       const result = await invoke<unloadResult>('unload_llama_model', {
-        session_id: opts.sessionId, // Using PID as session ID
+        session_id: sessionId, // Using PID as session ID
       })
 
       // If successful, remove from active sessions
       if (result.success) {
-        this.activeSessions.delete(opts.sessionId)
-        console.log(`Successfully unloaded model with PID ${opts.sessionId}`)
+        this.activeSessions.delete(sessionId)
+        console.log(`Successfully unloaded model with PID ${sessionId}`)
       } else {
         console.warn(`Failed to unload model: ${result.error}`)
       }
@@ -496,8 +513,4 @@ export default class llamacpp_extension extends AIEngine {
   override getChatClient(sessionId: string): any {
     throw new Error('method not implemented yet')
   }
-
-  onUnload(): void {
-    throw new Error('Method not implemented.')
-  }
 }

src-tauri/src/core/utils/extensions/inference_llamacpp_extension/server.rs

@@ -37,33 +37,6 @@ impl serde::Serialize for ServerError {
 
 type ServerResult<T> = Result<T, ServerError>;
 
-// --- Helper function to find the server binary ---
-// -- TODO: Adjust extension engine paths
-// engine: static llama-server build (CUDA, VULKAN, SYCL, etc)
-fn get_server_path(app_handle: &AppHandle) -> ServerResult<PathBuf> {
-    let binary_name = if cfg!(windows) {
-        "llama-server.exe"
-    } else {
-        "llama-server"
-    };
-    let relative_path = PathBuf::from("engines").join(binary_name); // TODO: ADJUST THIS PATH
-
-    app_handle
-        .path()
-        .resolve(relative_path, BaseDirectory::Resource)
-        .map_err(|e| ServerError::ResourcePathError(e.to_string()))
-    // .ok_or_else(|| {
-    //     ServerError::BinaryNotFound(format!(
-    //         "Could not resolve resource path for '{}'",
-    //         if cfg!(windows) {
-    //             "engines/llama-server.exe"
-    //         } else {
-    //             "engines/llama-server"
-    //         } // TODO: ADJUST THIS PATH
-    //     ))
-    // })
-}
-
 #[derive(Debug, Serialize, Deserialize)]
 pub struct SessionInfo {
     pub session_id: String,       // opaque handle for unload/chat
@@ -82,6 +55,7 @@ pub struct UnloadResult {
 pub async fn load_llama_model(
     app_handle: AppHandle,      // Get the AppHandle
     state: State<'_, AppState>, // Access the shared state
+    server_path: String,
     args: Vec<String>,          // Arguments from the frontend
 ) -> ServerResult<SessionInfo> {
     let mut process_lock = state.llama_server_process.lock().await;
@@ -91,7 +65,6 @@ pub async fn load_llama_model(
         return Err(ServerError::AlreadyRunning);
     }
 
-    let server_path = get_server_path(&app_handle)?;
     log::info!("Attempting to launch server at path: {:?}", server_path);
     log::info!("Using arguments: {:?}", args);