job-search/internal/fetcher/fetcher.go

package fetcher

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

	"git.avsdev.uk/Go/job-search/internal/config"
	"git.avsdev.uk/Go/job-search/internal/util"
)

// ErrNoNewReleases is returned by a source's task when there was nothing new
// to fetch; it wraps util.ErrWorkerNoWork.
var ErrNoNewReleases = fmt.Errorf("%w: no new releases", util.ErrWorkerNoWork)

// Option configures a Fetcher.
type Option func(*Fetcher)

// WithLogger makes Run, its workers and their clients log to logger instead
// of slog.Default().
func WithLogger(logger *slog.Logger) Option {
	return func(f *Fetcher) {
		f.log = logger
	}
}

// logger returns the logger Run logs to: the given one, or slog.Default().
// It isn't tagged here, as it is also given to the clients the workers
// create, which add their own component; see tagged.
func (f *Fetcher) logger() *slog.Logger {
	if f.log == nil {
		return slog.Default()
	}
	return f.log
}

// tagged returns logger tagged with the fetcher's component, for the lines
// Run and its workers log themselves.
func tagged(logger *slog.Logger) *slog.Logger {
	return logger.With("component", "fetcher")
}

// Fetcher runs a worker for each enabled source. Create it with New and
// start it with Run.
type Fetcher struct {
	conf    *config.JSCSources
	log     *slog.Logger
	started time.Time

	// backgroundCtx is the context the workers run under. backgroundCancel
	// stops them: Run calls it when its ctx is cancelled, when a worker
	// fails, and as it returns.
	backgroundCtx    context.Context
	backgroundCancel context.CancelFunc
}

// New returns a Fetcher for the sources conf configures. It logs to
// slog.Default() unless given WithLogger.
func New(conf *config.JSCSources, opts ...Option) *Fetcher {
	fetcher := &Fetcher{
		conf:    conf,
		started: time.Now(),
	}
	for _, option := range opts {
		option(fetcher)
	}

	fetcher.backgroundCtx, fetcher.backgroundCancel = context.WithCancel(
		context.Background(),
	)

	return fetcher
}

// Run starts a worker for each enabled source and blocks until every worker
// has stopped. Workers stop when ctx is cancelled, or when any worker stops
// with an error, which starts a shutdown of all of them. The errors are
// logged as they happen and returned joined; nil means a clean shutdown. Run
// cancels the Fetcher's background context as it returns, so it can only be
// called once.
func (f *Fetcher) Run(ctx context.Context, since time.Time) error {
	defer f.backgroundCancel()

	// Cancelling ctx stops the workers, through the background context.
	stopAfter := context.AfterFunc(ctx, f.backgroundCancel)
	defer stopAfter()

	logger := f.logger()

	var (
		wg   sync.WaitGroup
		mu   sync.Mutex
		errs []error
	)

	// start runs one source's worker until f.backgroundCtx is cancelled. Each
	// worker is given logger untagged: it tags its own lines with tagged, and
	// passes logger as it is to any client it creates.
	start := func(name string, run func(ctx context.Context) error) {
		wg.Add(1)
		go func() {
			defer wg.Done()
			if err := run(f.backgroundCtx); err != nil {
				tagged(logger).Error(
					"source stopped, shutting down",
					"source", name,
					"error", err,
				)
				mu.Lock()
				errs = append(errs, fmt.Errorf("%s: %w", name, err))
				mu.Unlock()
				f.backgroundCancel()
			}
		}()
	}

	if f.conf.FindATender.Enabled {
		start("FindATender", func(ctx context.Context) error {
			return runFindATender(ctx, logger, &f.conf.FindATender, since)
		})
	}
	if f.conf.FTSMock.Enabled {
		start("FTSMock", func(ctx context.Context) error {
			return runFTSMock(ctx, logger, &f.conf.FTSMock, since)
		})
	}
	if f.conf.Sleeper.Enabled {
		start("Sleeper", func(ctx context.Context) error {
			return runSleeper(ctx, logger, &f.conf.Sleeper, since)
		})
	}

	// Every worker has stopped once Wait returns, so errs is no longer
	// written.
	wg.Wait()

	return errors.Join(errs...)
}