feat: pid file writing and clearing and improved error handling for load state

Author: 35C4n0r

lib/httpapi/server.go

@@ -350,16 +350,11 @@ func (s *Server) StartSnapshotLoop(ctx context.Context) {
 		for {
 			currentStatus := s.conversation.Status()
 
-			// Send initial prompt when agent becomes stable for the first time
+			// Send initial prompt & load state when agent becomes stable for the first time
 			if convertStatus(currentStatus) == AgentStatusStable {
 
 				if !s.stateLoadComplete && s.statePersistenceCfg.LoadState {
-					_, err := s.conversation.LoadState(s.statePersistenceCfg.StateFile)
-					if err != nil {
-						s.logger.Warn("Failed to load state file", "path", s.statePersistenceCfg.StateFile, "err", err)
-					} else {
-						s.logger.Info("Successfully loaded state", "path", s.statePersistenceCfg.StateFile)
-					}
+					_, _ = s.conversation.LoadState(s.statePersistenceCfg.StateFile)
 					s.stateLoadComplete = true
 				}
 				if !s.conversation.InitialPromptSent {
@@ -612,6 +607,11 @@ func (s *Server) Start() error {
 		Handler: s.router,
 	}
 
+	// Write PID file if configured
+	if err := s.writePIDFile(); err != nil {
+		return xerrors.Errorf("failed to write PID file: %w", err)
+	}
+
 	return s.srv.ListenAndServe()
 }
 
@@ -626,6 +626,9 @@ func (s *Server) Stop(ctx context.Context) error {
 		}
 	}
 
+	// Clean up PID file
+	s.cleanupPIDFile()
+
 	// Clean up temporary directory
 	s.cleanupTempDir()
 
@@ -644,6 +647,43 @@ func (s *Server) cleanupTempDir() {
 	}
 }
 
+// writePIDFile writes the current process ID to the configured PID file
+func (s *Server) writePIDFile() error {
+	if s.statePersistenceCfg.PidFile == "" {
+		return nil
+	}
+
+	pid := os.Getpid()
+	pidContent := fmt.Sprintf("%d\n", pid)
+
+	// Create directory if it doesn't exist
+	dir := filepath.Dir(s.statePersistenceCfg.PidFile)
+	if err := os.MkdirAll(dir, 0o755); err != nil {
+		return xerrors.Errorf("failed to create PID file directory: %w", err)
+	}
+
+	// Write PID file
+	if err := os.WriteFile(s.statePersistenceCfg.PidFile, []byte(pidContent), 0o644); err != nil {
+		return xerrors.Errorf("failed to write PID file: %w", err)
+	}
+
+	s.logger.Info("Wrote PID file", "pidFile", s.statePersistenceCfg.PidFile, "pid", pid)
+	return nil
+}
+
+// cleanupPIDFile removes the PID file if it exists
+func (s *Server) cleanupPIDFile() {
+	if s.statePersistenceCfg.PidFile == "" {
+		return
+	}
+
+	if err := os.Remove(s.statePersistenceCfg.PidFile); err != nil && !os.IsNotExist(err) {
+		s.logger.Error("Failed to remove PID file", "pidFile", s.statePersistenceCfg.PidFile, "error", err)
+	} else if err == nil {
+		s.logger.Info("Removed PID file", "pidFile", s.statePersistenceCfg.PidFile)
+	}
+}
+
 // HandleSignals sets up signal handlers for:
 // - SIGTERM, SIGINT, SIGHUP: save conversation state, then close the process
 // - SIGUSR1: save conversation state without exiting
@@ -664,6 +704,9 @@ func (s *Server) HandleSignals(ctx context.Context, process *termexec.Process) {
 			}
 		}
 
+		// Clean up PID file
+		s.cleanupPIDFile()
+
 		// Now close the process
 		if err := process.Close(s.logger, 5*time.Second); err != nil {
 			s.logger.Error("Error closing process", "signal", sig, "error", err)

lib/screentracker/pty_conversation.go

@@ -442,22 +442,26 @@ func (c *PTYConversation) LoadState(stateFile string) ([]ConversationMessage, er
 
 	// Check if file exists
 	if _, err := os.Stat(stateFile); os.IsNotExist(err) {
+		c.cfg.Logger.Info("No previous state to load (file does not exist)", "path", stateFile)
 		return nil, nil
 	}
 
 	// Read state file
 	data, err := os.ReadFile(stateFile)
 	if err != nil {
+		c.cfg.Logger.Warn("Failed to load state file", "path", stateFile, "err", err)
 		return nil, xerrors.Errorf("failed to read state file: %w", err)
 	}
 
 	if len(data) == 0 {
-		return nil, xerrors.Errorf("failed to read state file: empty state file")
+		c.cfg.Logger.Info("No previous state to load (file is empty)", "path", stateFile)
+		return nil, nil
 	}
 
 	var agentState AgentState
 	if err := json.Unmarshal(data, &agentState); err != nil {
-		return nil, xerrors.Errorf("failed to unmarshal state: %w", err)
+		c.cfg.Logger.Warn("Failed to load state file (corrupted or invalid JSON)", "path", stateFile, "err", err)
+		return nil, xerrors.Errorf("failed to unmarshal state (corrupted or invalid JSON): %w", err)
 	}
 
 	c.InitialPromptSent = agentState.InitialPromptSent
@@ -470,6 +474,7 @@ func (c *PTYConversation) LoadState(stateFile string) ([]ConversationMessage, er
 		c.firstStableSnapshot = c.cfg.FormatMessage(strings.TrimSpace(snapshots[len(snapshots)-1].screen), "")
 	}
 
+	c.cfg.Logger.Info("Successfully loaded state", "path", stateFile, "messages", len(c.messages))
 	return c.messages, nil
 }