X-Git-Url: https://git.giacomolli.uk/onechat/blobdiff_plain/eb05edcf5d575e8a3b1040debfb57423312e8320..e73e445bd78ba98271da6b5165f132299fca04b1:/turn.go?ds=inline diff --git a/turn.go b/turn.go index 051c74d..40645b1 100644 --- a/turn.go +++ b/turn.go @@ -3,14 +3,18 @@ package main import ( "context" "errors" + "fmt" + "log" "path/filepath" "sync" "time" ) -// newTurnLoop returns the loop of agent with its memory. Its calls go through -// client, and notify gets each event. The clock gives the date of a turn. -func newTurnLoop(agent *Agent, memory *Memory, client ChainClient, clock func() time.Time, notify func(Event)) (*TurnLoop, error) { +// newTurnLoop returns the loop of agent with its memory, its subagents, and the +// WhatsApp link and the schedules of its tools. Its calls go through client, +// and notify gets each event. The clock gives the date of a turn. logger gets +// the failed turns and the ends of the subagents. +func newTurnLoop(agent *Agent, memory *Memory, subagents *Subagents, whatsapp *WhatsApp, schedules *Schedules, client ChainClient, clock func() time.Time, logger *log.Logger, notify func(Event)) (*TurnLoop, error) { layout, err := loadViewLayout(filepath.Join(agent.dir, viewLayoutFileName)) if err != nil { return nil, err @@ -22,29 +26,52 @@ func newTurnLoop(agent *Agent, memory *Memory, client ChainClient, clock func() client: client, layout: layout, clock: clock, + logger: logger, notify: notify, wake: make(chan struct{}, 1), } + l.runs = &SubagentRuns{ + agent: agent, + memory: memory, + subagents: subagents, + layout: layout, + client: client, + state: l.state, + report: l.Report, + note: l.note, + fail: l.fail, + logger: logger, + notify: notify, + running: make(map[string]*SubagentRun), + } + l.tools = &Toolbox{ - log: l, - chat: memory, - workdir: agent.config.Workdir, + log: l, + chat: memory, + subagents: subagents, + runs: l.runs, + whatsapp: whatsapp, + schedules: schedules, + workdir: agent.config.Workdir, } return l, nil } // Event tells the clients what the agent does as it happens. Kind is the kind -// of a logged message (user, reply, tool, echo), or one of ready, busy, idle, -// and error. ID is the ID of a logged message. Source names the client that -// sent a user message, such as terminal or telegram, so that each client can -// show the messages of the others. +// of a logged message (user, reply, tool, echo, note, work), or one of ready, +// busy, idle, error, and subagent, which tells that the subagent Text starts a +// run. ID is the ID of a logged message. Source names the client that sent a +// user message, such as terminal or telegram, so that each client can show the +// messages of the others, or the subagent of a report. Status tells how the run +// of a report ended: finished, stopped, or failed. type Event struct { Kind string `json:"kind"` ID int `json:"id,omitempty"` Text string `json:"text,omitempty"` Source string `json:"source,omitempty"` + Status string `json:"status,omitempty"` } // TurnLoop runs the turns of one agent. A user message goes to the log when it @@ -55,20 +82,24 @@ type Event struct { // The view goes out as the messages of a ViewLayout, so the first request of a // turn reads the view of the turn before from the prompt cache. User messages // that come during a turn join the next request of its chain after the tool -// results. +// results, and so do the reports of the subagents. The subagents run until the +// loop ends. type TurnLoop struct { agent *Agent memory *Memory client ChainClient tools *Toolbox + runs *SubagentRuns layout *ViewLayout clock func() time.Time + logger *log.Logger notify func(Event) wake chan struct{} mutex sync.Mutex pending []Message cancel context.CancelFunc + failure error } // Submit logs a user message from the client source and wakes the loop. The @@ -92,6 +123,65 @@ func (l *TurnLoop) Submit(source, text string) (Message, error) { return message, nil } +// Report logs the report text of the subagent name as a work message and tells +// the clients, with the status of the run. Unless the subagent was stopped, the +// report joins the waiting messages and wakes the loop, as a user message of +// Submit does. A failed write stops the loop, because the log refuses all +// writes after it. +func (l *TurnLoop) Report(name, text, status string) { + message, err := l.Work(name, text, status) + switch { + case err != nil: + l.fail(fmt.Errorf("%w: %w", errChainLogWrite, err)) + case status != subagentStopped: + l.Wake(message) + } +} + +// Work logs text as a work message of source, such as the name of a subagent, +// and tells the clients, with status. The message starts no turn until it goes +// to Wake. +func (l *TurnLoop) Work(source, text, status string) (Message, error) { + message, err := l.memory.Append("work", text) + if err != nil { + return Message{}, err + } + + l.notify(Event{ + Kind: "work", + ID: message.ID, + Text: text, + Source: source, + Status: status, + }) + + return message, nil +} + +// Wake adds the logged messages to the waiting messages and wakes the loop +// once, as a user message of Submit does. They join the first request of a new +// turn together, or the next request of a running turn after the tool results. +func (l *TurnLoop) Wake(messages ...Message) { + if len(messages) == 0 { + return + } + + l.mutex.Lock() + l.pending = append(l.pending, messages...) + l.mutex.Unlock() + + select { + case l.wake <- struct{}{}: + default: + } +} + +// StopSubagent ends the run of the subagent name and tells if it ran. Its +// report is logged, but it starts no turn. +func (l *TurnLoop) StopSubagent(name string) (bool, error) { + return l.runs.Stop(name) +} + // Interrupt ends the running turn. A running tool is killed, and the turn sends // no further request. func (l *TurnLoop) Interrupt() bool { @@ -126,9 +216,18 @@ func (l *TurnLoop) Flush(ctx context.Context) error { return l.memory.Flush(ctx) } -// Run runs a turn each time user messages wait, until ctx ends. It returns the -// error of a failed log write, which stops the agent. +// Run runs a turn each time user messages or reports wait, until ctx ends. It +// returns the error of a failed log write, which stops the agent. At its start +// it resumes the subagents that a stop of the daemon cut, and at its end it +// ends the runs of the subagents and waits for them. func (l *TurnLoop) Run(ctx context.Context) error { + defer l.runs.Close() + + err := l.runs.Resume(ctx) + if err != nil { + return fmt.Errorf("resume subagents: %w", err) + } + for { for l.waiting() { err := l.turn(ctx) @@ -142,9 +241,21 @@ func (l *TurnLoop) Run(ctx context.Context) error { case <-ctx.Done(): return ctx.Err() } + + err := l.failed() + if err != nil { + return err + } } } +// note logs a note and tells the clients. A note starts no turn. +func (l *TurnLoop) note(text string) error { + _, err := l.record("note", "", text) + + return err +} + // record logs a message and tells the clients, with the client source of a user // message. func (l *TurnLoop) record(kind, source, text string) (Message, error) { @@ -163,7 +274,7 @@ func (l *TurnLoop) record(kind, source, text string) (Message, error) { return message, nil } -// waiting tells if user messages wait for a request. +// waiting tells if user messages or reports wait for a request. func (l *TurnLoop) waiting() bool { l.mutex.Lock() defer l.mutex.Unlock() @@ -171,10 +282,32 @@ func (l *TurnLoop) waiting() bool { return len(l.pending) > 0 } +// fail stops the loop with err. +func (l *TurnLoop) fail(err error) { + l.mutex.Lock() + l.failure = errors.Join(l.failure, err) + l.mutex.Unlock() + + select { + case l.wake <- struct{}{}: + default: + } +} + +// failed returns the error of a failed write of a report. +func (l *TurnLoop) failed() error { + l.mutex.Lock() + defer l.mutex.Unlock() + + return l.failure +} + // turn runs one turn. A failed or interrupted model call ends the turn and -// tells the clients. The messages of the turn stay in the log and the view, and -// the next turn starts only with a new message, so a lasting failure such as a -// needed login does not repeat. Only a failed log write is an error of turn. +// tells the clients. A failure also goes to the logger, so that it is not lost +// when no client is connected, but an interrupt does not. The messages of the +// turn stay in the log and the view, and the next turn starts only with a new +// message, so a lasting failure such as a needed login does not repeat. Only a +// failed log write is an error of turn. func (l *TurnLoop) turn(ctx context.Context) error { ctx, cancel := context.WithCancel(ctx) defer cancel() @@ -195,6 +328,10 @@ func (l *TurnLoop) turn(ctx context.Context) error { err := l.chain(ctx) if err != nil && !errors.Is(err, errChainLogWrite) { + if ctx.Err() == nil { + l.logger.Printf("turn failed: %v", err) + } + l.notify(Event{ Kind: "error", Text: turnError(ctx, err),