Files
yt-dlp-webui/server/subscription/task/runner.go

173 lines
3.9 KiB
Go

package task
import (
"bytes"
"context"
"log/slog"
"os/exec"
"path/filepath"
"regexp"
"strings"
"time"
"github.com/marcopiovanello/yt-dlp-web-ui/v3/server/config"
"github.com/marcopiovanello/yt-dlp-web-ui/v3/server/internal"
"github.com/marcopiovanello/yt-dlp-web-ui/v3/server/subscription/domain"
"github.com/robfig/cron/v3"
)
type TaskRunner interface {
Submit(subcription *domain.Subscription) error
Spawner(ctx context.Context)
StopTask(id string) error
Recoverer()
}
type monitorTask struct {
Done chan struct{}
Schedule cron.Schedule
Subscription *domain.Subscription
}
type CronTaskRunner struct {
mq *internal.MessageQueue
db *internal.MemoryDB
tasks chan monitorTask
errors chan error
running map[string]*monitorTask
}
func NewCronTaskRunner(mq *internal.MessageQueue, db *internal.MemoryDB) TaskRunner {
return &CronTaskRunner{
mq: mq,
db: db,
tasks: make(chan monitorTask),
errors: make(chan error),
running: make(map[string]*monitorTask),
}
}
const (
commandTemplate = "-I1 --flat-playlist --print webpage_url $1"
getVideoIdTemplate = "--print \"%(extractor)s %(id)s\" $1"
)
var argsSplitterRe = regexp.MustCompile(`(?mi)[^\s"']+|"([^"]*)"|'([^']*)'`)
func (t *CronTaskRunner) Submit(subcription *domain.Subscription) error {
schedule, err := cron.ParseStandard(subcription.CronExpr)
if err != nil {
return err
}
job := monitorTask{
Done: make(chan struct{}),
Schedule: schedule,
Subscription: subcription,
}
t.tasks <- job
return nil
}
// Handles the entire lifecylce of a monitor job.
func (t *CronTaskRunner) Spawner(ctx context.Context) {
for req := range t.tasks {
t.running[req.Subscription.Id] = &req // keep track of the current job
go func() {
ctx, cancel := context.WithCancel(ctx) // inject into the job's context a cancellation singal
fetcherEvents := t.doFetch(ctx, &req) // retrieve the channel of events of the job
for {
select {
case <-req.Done:
slog.Info("stopping cron job and removing schedule", slog.String("url", req.Subscription.URL))
cancel()
return
case <-fetcherEvents:
slog.Info("finished monitoring channel", slog.String("url", req.Subscription.URL))
}
}
}()
}
}
// Stop a currently scheduled job
func (t *CronTaskRunner) StopTask(id string) error {
t.running[id].Done <- struct{}{}
delete(t.running, id)
return nil
}
// Notify on a channel when a fetcher has completed
func (t *CronTaskRunner) doFetch(ctx context.Context, req *monitorTask) <-chan struct{} {
events := make(chan struct{})
// generator func
go func() {
for {
sleepFor := t.fetcher(ctx, req)
events <- struct{}{}
time.Sleep(sleepFor)
}
}()
return events
}
// Perform the retrieval of the latest video of the channel.
// Returns a time.Duration containing the amount of time to the next schedule.
func (t *CronTaskRunner) fetcher(ctx context.Context, req *monitorTask) time.Duration {
slog.Info("fetching latest video for channel", slog.String("channel", req.Subscription.URL))
fetcherParams := strings.Split(strings.Replace(commandTemplate, "$1", req.Subscription.URL, 1), " ")
cmd := exec.CommandContext(
ctx,
config.Instance().DownloaderPath,
fetcherParams...,
)
stdout, err := cmd.Output()
if err != nil {
t.errors <- err
return time.Duration(0)
}
latestChannelURL := string(bytes.Trim(stdout, "\n"))
p := &internal.Process{
Url: latestChannelURL,
Params: append(
argsSplitterRe.FindAllString(req.Subscription.Params, 1),
[]string{
"--break-on-existing",
"--download-archive",
filepath.Join(config.Instance().Dir(), "archive.txt"),
}...),
AutoRemove: true,
}
t.db.Set(p)
t.mq.Publish(p)
nextSchedule := time.Until(req.Schedule.Next(time.Now()))
slog.Info(
"cron task runner next schedule",
slog.String("url", req.Subscription.URL),
slog.Any("duration", nextSchedule),
)
return nextSchedule
}
func (t *CronTaskRunner) Recoverer() {
panic("unimplemented")
}