galaxy-game/mail/internal/worker/attempt_worker.go

package worker

import (
	"context"
	"errors"
	"fmt"
	"log/slog"
	"sync"

	"galaxy/mail/internal/service/executeattempt"
)

// AttemptExecutionService executes one claimed in-progress attempt.
type AttemptExecutionService interface {
	// Execute runs one claimed attempt through provider execution and durable
	// state mutation.
	Execute(context.Context, executeattempt.WorkItem) error
}

// AttemptWorkerPoolConfig stores the dependencies used by AttemptWorkerPool.
type AttemptWorkerPoolConfig struct {
	// Concurrency stores how many workers run concurrently.
	Concurrency int

	// WorkQueue stores the claimed attempt handoff channel produced by the
	// scheduler.
	WorkQueue <-chan executeattempt.WorkItem

	// Service executes one claimed attempt.
	Service AttemptExecutionService
}

// AttemptWorkerPool executes claimed attempts concurrently.
type AttemptWorkerPool struct {
	concurrency int
	workQueue   <-chan executeattempt.WorkItem
	service     AttemptExecutionService
	logger      *slog.Logger
}

// NewAttemptWorkerPool constructs one attempt worker pool.
func NewAttemptWorkerPool(cfg AttemptWorkerPoolConfig, logger *slog.Logger) (*AttemptWorkerPool, error) {
	switch {
	case cfg.Concurrency <= 0:
		return nil, errors.New("new attempt worker pool: concurrency must be positive")
	case cfg.WorkQueue == nil:
		return nil, errors.New("new attempt worker pool: nil work queue")
	case cfg.Service == nil:
		return nil, errors.New("new attempt worker pool: nil attempt execution service")
	}
	if logger == nil {
		logger = slog.Default()
	}

	return &AttemptWorkerPool{
		concurrency: cfg.Concurrency,
		workQueue:   cfg.WorkQueue,
		service:     cfg.Service,
		logger:      logger.With("component", "attempt_worker_pool", "concurrency", cfg.Concurrency),
	}, nil
}

// Run starts the attempt worker pool and blocks until ctx is canceled or one
// worker returns an execution error.
func (pool *AttemptWorkerPool) Run(ctx context.Context) error {
	if ctx == nil {
		return errors.New("run attempt worker pool: nil context")
	}
	if err := ctx.Err(); err != nil {
		return err
	}
	if pool == nil {
		return errors.New("run attempt worker pool: nil pool")
	}

	pool.logger.Info("attempt worker pool started")
	defer pool.logger.Info("attempt worker pool stopped")

	runCtx, cancel := context.WithCancel(ctx)
	defer cancel()

	errs := make(chan error, pool.concurrency)
	var waitGroup sync.WaitGroup

	for index := 0; index < pool.concurrency; index++ {
		waitGroup.Add(1)
		go func(workerIndex int) {
			defer waitGroup.Done()
			if err := pool.runWorker(runCtx, workerIndex); err != nil {
				errs <- err
			}
		}(index)
	}

	done := make(chan struct{})
	go func() {
		waitGroup.Wait()
		close(done)
	}()

	select {
	case <-ctx.Done():
		cancel()
		<-done
		return ctx.Err()
	case err := <-errs:
		cancel()
		<-done
		return err
	case <-done:
		if ctx.Err() != nil {
			return ctx.Err()
		}
		return errors.New("run attempt worker pool: workers exited without shutdown")
	}
}

func (pool *AttemptWorkerPool) runWorker(ctx context.Context, workerIndex int) error {
	pool.logger.Debug("attempt worker started", "worker_index", workerIndex)
	defer pool.logger.Debug("attempt worker stopped", "worker_index", workerIndex)

	for {
		select {
		case <-ctx.Done():
			return ctx.Err()
		case item, ok := <-pool.workQueue:
			if !ok {
				return nil
			}
			if err := pool.service.Execute(ctx, item); err != nil {
				return fmt.Errorf("attempt worker %d: %w", workerIndex, err)
			}
		}
	}
}

// Shutdown stops the attempt worker pool within ctx. The pool does not own
// additional resources beyond its run loop.
func (pool *AttemptWorkerPool) Shutdown(ctx context.Context) error {
	if ctx == nil {
		return errors.New("shutdown attempt worker pool: nil context")
	}
	if pool == nil {
		return nil
	}

	return nil
}