job.go

//go:generate mockgen -destination=mocks/job.go -package=gocronmocks . Job
package gocron

import (
	"context"
	"errors"
	"fmt"
	"math/rand"
	"strings"
	"time"

	"github.com/google/uuid"
	"github.com/jonboulle/clockwork"
	"github.com/robfig/cron/v3"
	"golang.org/x/exp/slices"
)

// internalJob stores the information needed by the scheduler
// to manage scheduling, starting and stopping the job
type internalJob struct {
	ctx    context.Context
	cancel context.CancelFunc
	id     uuid.UUID
	name   string
	tags   []string
	jobSchedule

	// as some jobs may queue up, it's possible to
	// have multiple nextScheduled times
	nextScheduled []time.Time

	lastRun            time.Time
	function           any
	parameters         []any
	timer              clockwork.Timer
	singletonMode      bool
	singletonLimitMode LimitMode
	limitRunsTo        *limitRunsTo
	startTime          time.Time
	startImmediately   bool
	// event listeners
	afterJobRuns          func(jobID uuid.UUID, jobName string)
	beforeJobRuns         func(jobID uuid.UUID, jobName string)
	afterJobRunsWithError func(jobID uuid.UUID, jobName string, err error)
	afterJobRunsWithPanic func(jobID uuid.UUID, jobName string, recoverData any)
	afterLockError        func(jobID uuid.UUID, jobName string, err error)

	locker Locker
}

// stop is used to stop the job's timer and cancel the context
// stopping the timer is critical for cleaning up jobs that are
// sleeping in a time.AfterFunc timer when the job is being stopped.
// cancelling the context keeps the executor from continuing to try
// and run the job.
func (j *internalJob) stop() {
	if j.timer != nil {
		j.timer.Stop()
	}
	j.cancel()
}

// task stores the function and parameters
// that are actually run when the job is executed.
type task struct {
	function   any
	parameters []any
}

// Task defines a function that returns the task
// function and parameters.
type Task func() task

// NewTask provides the job's task function and parameters.
func NewTask(function any, parameters ...any) Task {
	return func() task {
		return task{
			function:   function,
			parameters: parameters,
		}
	}
}

// limitRunsTo is used for managing the number of runs
// when the user only wants the job to run a certain
// number of times and then be removed from the scheduler.
type limitRunsTo struct {
	limit    uint
	runCount uint
}

// -----------------------------------------------
// -----------------------------------------------
// --------------- Job Variants ------------------
// -----------------------------------------------
// -----------------------------------------------

// JobDefinition defines the interface that must be
// implemented to create a job from the definition.
type JobDefinition interface {
	setup(j *internalJob, l *time.Location, now time.Time) error
}

var _ JobDefinition = (*cronJobDefinition)(nil)

type cronJobDefinition struct {
	crontab     string
	withSeconds bool
}

func (c cronJobDefinition) setup(j *internalJob, location *time.Location, _ time.Time) error {
	var withLocation string
	if strings.HasPrefix(c.crontab, "TZ=") || strings.HasPrefix(c.crontab, "CRON_TZ=") {
		withLocation = c.crontab
	} else {
		// since the user didn't provide a timezone default to the location
		// passed in by the scheduler. Default: time.Local
		withLocation = fmt.Sprintf("CRON_TZ=%s %s", location.String(), c.crontab)
	}

	var (
		cronSchedule cron.Schedule
		err          error
	)

	if c.withSeconds {
		p := cron.NewParser(cron.SecondOptional | cron.Minute | cron.Hour | cron.Dom | cron.Month | cron.Dow | cron.Descriptor)
		cronSchedule, err = p.Parse(withLocation)
	} else {
		cronSchedule, err = cron.ParseStandard(withLocation)
	}
	if err != nil {
		return errors.Join(ErrCronJobParse, err)
	}

	j.jobSchedule = &cronJob{cronSchedule: cronSchedule}
	return nil
}

// CronJob defines a new job using the crontab syntax: `* * * * *`.
// An optional 6th field can be used at the beginning if withSeconds
// is set to true: `* * * * * *`.
// The timezone can be set on the Scheduler using WithLocation, or in the
// crontab in the form `TZ=America/Chicago * * * * *` or
// `CRON_TZ=America/Chicago * * * * *`
func CronJob(crontab string, withSeconds bool) JobDefinition {
	return cronJobDefinition{
		crontab:     crontab,
		withSeconds: withSeconds,
	}
}

var _ JobDefinition = (*durationJobDefinition)(nil)

type durationJobDefinition struct {
	duration time.Duration
}

func (d durationJobDefinition) setup(j *internalJob, _ *time.Location, _ time.Time) error {
	if d.duration == 0 {
		return ErrDurationJobIntervalZero
	}
	j.jobSchedule = &durationJob{duration: d.duration}
	return nil
}

// DurationJob defines a new job using time.Duration
// for the interval.
func DurationJob(duration time.Duration) JobDefinition {
	return durationJobDefinition{
		duration: duration,
	}
}

var _ JobDefinition = (*durationRandomJobDefinition)(nil)

type durationRandomJobDefinition struct {
	min, max time.Duration
}

func (d durationRandomJobDefinition) setup(j *internalJob, _ *time.Location, _ time.Time) error {
	if d.min >= d.max {
		return ErrDurationRandomJobMinMax
	}

	j.jobSchedule = &durationRandomJob{
		min:  d.min,
		max:  d.max,
		rand: rand.New(rand.NewSource(time.Now().UnixNano())), // nolint:gosec
	}
	return nil
}

// DurationRandomJob defines a new job that runs on a random interval
// between the min and max duration values provided.
//
// To achieve a similar behavior as tools that use a splay/jitter technique
// consider the median value as the baseline and the difference between the
// max-median or median-min as the splay/jitter.
//
// For example, if you want a job to run every 5 minutes, but want to add
// up to 1 min of jitter to the interval, you could use
// DurationRandomJob(4*time.Minute, 6*time.Minute)
func DurationRandomJob(minDuration, maxDuration time.Duration) JobDefinition {
	return durationRandomJobDefinition{
		min: minDuration,
		max: maxDuration,
	}
}

// DailyJob runs the job on the interval of days, and at the set times.
// By default, the job will start the next available day, considering the last run to be now,
// and the time and day based on the interval and times you input. This means, if you
// select an interval greater than 1, your job by default will run X (interval) days from now
// if there are no atTimes left in the current day. You can use WithStartAt to tell the
// scheduler to start the job sooner.
func DailyJob(interval uint, atTimes AtTimes) JobDefinition {
	return dailyJobDefinition{
		interval: interval,
		atTimes:  atTimes,
	}
}

var _ JobDefinition = (*dailyJobDefinition)(nil)

type dailyJobDefinition struct {
	interval uint
	atTimes  AtTimes
}

func (d dailyJobDefinition) setup(j *internalJob, location *time.Location, _ time.Time) error {
	atTimesDate, err := convertAtTimesToDateTime(d.atTimes, location)
	switch {
	case errors.Is(err, errAtTimesNil):
		return ErrDailyJobAtTimesNil
	case errors.Is(err, errAtTimeNil):
		return ErrDailyJobAtTimeNil
	case errors.Is(err, errAtTimeHours):
		return ErrDailyJobHours
	case errors.Is(err, errAtTimeMinSec):
		return ErrDailyJobMinutesSeconds
	}

	ds := dailyJob{
		interval: d.interval,
		atTimes:  atTimesDate,
	}
	j.jobSchedule = ds
	return nil
}

var _ JobDefinition = (*weeklyJobDefinition)(nil)

type weeklyJobDefinition struct {
	interval      uint
	daysOfTheWeek Weekdays
	atTimes       AtTimes
}

func (w weeklyJobDefinition) setup(j *internalJob, location *time.Location, _ time.Time) error {
	var ws weeklyJob
	ws.interval = w.interval

	if w.daysOfTheWeek == nil {
		return ErrWeeklyJobDaysOfTheWeekNil
	}

	daysOfTheWeek := w.daysOfTheWeek()

	slices.Sort(daysOfTheWeek)
	ws.daysOfWeek = daysOfTheWeek

	atTimesDate, err := convertAtTimesToDateTime(w.atTimes, location)
	switch {
	case errors.Is(err, errAtTimesNil):
		return ErrWeeklyJobAtTimesNil
	case errors.Is(err, errAtTimeNil):
		return ErrWeeklyJobAtTimeNil
	case errors.Is(err, errAtTimeHours):
		return ErrWeeklyJobHours
	case errors.Is(err, errAtTimeMinSec):
		return ErrWeeklyJobMinutesSeconds
	}
	ws.atTimes = atTimesDate

	j.jobSchedule = ws
	return nil
}

// Weekdays defines a function that returns a list of week days.
type Weekdays func() []time.Weekday

// NewWeekdays provide the days of the week the job should run.
func NewWeekdays(weekday time.Weekday, weekdays ...time.Weekday) Weekdays {
	return func() []time.Weekday {
		weekdays = append(weekdays, weekday)
		return weekdays
	}
}

// WeeklyJob runs the job on the interval of weeks, on the specific days of the week
// specified, and at the set times.
//
// By default, the job will start the next available day, considering the last run to be now,
// and the time and day based on the interval, days and times you input. This means, if you
// select an interval greater than 1, your job by default will run X (interval) weeks from now
// if there are no daysOfTheWeek left in the current week. You can use WithStartAt to tell the
// scheduler to start the job sooner.
func WeeklyJob(interval uint, daysOfTheWeek Weekdays, atTimes AtTimes) JobDefinition {
	return weeklyJobDefinition{
		interval:      interval,
		daysOfTheWeek: daysOfTheWeek,
		atTimes:       atTimes,
	}
}

var _ JobDefinition = (*monthlyJobDefinition)(nil)

type monthlyJobDefinition struct {
	interval       uint
	daysOfTheMonth DaysOfTheMonth
	atTimes        AtTimes
}

func (m monthlyJobDefinition) setup(j *internalJob, location *time.Location, _ time.Time) error {
	var ms monthlyJob
	ms.interval = m.interval

	if m.daysOfTheMonth == nil {
		return ErrMonthlyJobDaysNil
	}

	var daysStart, daysEnd []int
	for _, day := range m.daysOfTheMonth() {
		if day > 31 || day == 0 || day < -31 {
			return ErrMonthlyJobDays
		}
		if day > 0 {
			daysStart = append(daysStart, day)
		} else {
			daysEnd = append(daysEnd, day)
		}
	}
	daysStart = removeSliceDuplicatesInt(daysStart)
	slices.Sort(daysStart)
	ms.days = daysStart

	daysEnd = removeSliceDuplicatesInt(daysEnd)
	slices.Sort(daysEnd)
	ms.daysFromEnd = daysEnd

	atTimesDate, err := convertAtTimesToDateTime(m.atTimes, location)
	switch {
	case errors.Is(err, errAtTimesNil):
		return ErrMonthlyJobAtTimesNil
	case errors.Is(err, errAtTimeNil):
		return ErrMonthlyJobAtTimeNil
	case errors.Is(err, errAtTimeHours):
		return ErrMonthlyJobHours
	case errors.Is(err, errAtTimeMinSec):
		return ErrMonthlyJobMinutesSeconds
	}
	ms.atTimes = atTimesDate

	j.jobSchedule = ms
	return nil
}

type days []int

// DaysOfTheMonth defines a function that returns a list of days.
type DaysOfTheMonth func() days

// NewDaysOfTheMonth provide the days of the month the job should
// run. The days can be positive 1 to 31 and/or negative -31 to -1.
// Negative values count backwards from the end of the month.
// For example: -1 == the last day of the month.
//
//	-5 == 5 days before the end of the month.
func NewDaysOfTheMonth(day int, moreDays ...int) DaysOfTheMonth {
	return func() days {
		moreDays = append(moreDays, day)
		return moreDays
	}
}

type atTime struct {
	hours, minutes, seconds uint
}

func (a atTime) time(location *time.Location) time.Time {
	return time.Date(0, 0, 0, int(a.hours), int(a.minutes), int(a.seconds), 0, location)
}

// AtTime defines a function that returns the internal atTime
type AtTime func() atTime

// NewAtTime provide the hours, minutes and seconds at which
// the job should be run
func NewAtTime(hours, minutes, seconds uint) AtTime {
	return func() atTime {
		return atTime{hours: hours, minutes: minutes, seconds: seconds}
	}
}

// AtTimes define a list of AtTime
type AtTimes func() []AtTime

// NewAtTimes provide the hours, minutes and seconds at which
// the job should be run
func NewAtTimes(atTime AtTime, atTimes ...AtTime) AtTimes {
	return func() []AtTime {
		atTimes = append(atTimes, atTime)
		return atTimes
	}
}

// MonthlyJob runs the job on the interval of months, on the specific days of the month
// specified, and at the set times. Days of the month can be 1 to 31 or negative (-1 to -31), which
// count backwards from the end of the month. E.g. -1 is the last day of the month.
//
// If a day of the month is selected that does not exist in all months (e.g. 31st)
// any month that does not have that day will be skipped.
//
// By default, the job will start the next available day, considering the last run to be now,
// and the time and month based on the interval, days and times you input.
// This means, if you select an interval greater than 1, your job by default will run
// X (interval) months from now if there are no daysOfTheMonth left in the current month.
// You can use WithStartAt to tell the scheduler to start the job sooner.
//
// Carefully consider your configuration!
//   - For example: an interval of 2 months on the 31st of each month, starting 12/31
//     would skip Feb, April, June, and next run would be in August.
func MonthlyJob(interval uint, daysOfTheMonth DaysOfTheMonth, atTimes AtTimes) JobDefinition {
	return monthlyJobDefinition{
		interval:       interval,
		daysOfTheMonth: daysOfTheMonth,
		atTimes:        atTimes,
	}
}

var _ JobDefinition = (*oneTimeJobDefinition)(nil)

type oneTimeJobDefinition struct {
	startAt OneTimeJobStartAtOption
}

func (o oneTimeJobDefinition) setup(j *internalJob, _ *time.Location, now time.Time) error {
	sortedTimes := o.startAt(j)
	slices.SortStableFunc(sortedTimes, ascendingTime)
	// keep only schedules that are in the future
	idx, found := slices.BinarySearchFunc(sortedTimes, now, ascendingTime)
	if found {
		idx++
	}
	sortedTimes = sortedTimes[idx:]
	if !j.startImmediately && len(sortedTimes) == 0 {
		return ErrOneTimeJobStartDateTimePast
	}
	j.jobSchedule = oneTimeJob{sortedTimes: sortedTimes}
	return nil
}

// OneTimeJobStartAtOption defines when the one time job is run
type OneTimeJobStartAtOption func(*internalJob) []time.Time

// OneTimeJobStartImmediately tells the scheduler to run the one time job immediately.
func OneTimeJobStartImmediately() OneTimeJobStartAtOption {
	return func(j *internalJob) []time.Time {
		j.startImmediately = true
		return []time.Time{}
	}
}

// OneTimeJobStartDateTime sets the date & time at which the job should run.
// This datetime must be in the future (according to the scheduler clock).
func OneTimeJobStartDateTime(start time.Time) OneTimeJobStartAtOption {
	return func(j *internalJob) []time.Time {
		return []time.Time{start}
	}
}

// OneTimeJobStartDateTimes sets the date & times at which the job should run.
// At least one of the date/times must be in the future (according to the scheduler clock).
func OneTimeJobStartDateTimes(times ...time.Time) OneTimeJobStartAtOption {
	return func(j *internalJob) []time.Time {
		return times
	}
}

// OneTimeJob is to run a job once at a specified time and not on
// any regular schedule.
func OneTimeJob(startAt OneTimeJobStartAtOption) JobDefinition {
	return oneTimeJobDefinition{
		startAt: startAt,
	}
}

// -----------------------------------------------
// -----------------------------------------------
// ----------------- Job Options -----------------
// -----------------------------------------------
// -----------------------------------------------

// JobOption defines the constructor for job options.
type JobOption func(*internalJob) error

// WithDistributedJobLocker sets the locker to be used by multiple
// Scheduler instances to ensure that only one instance of each
// job is run.
func WithDistributedJobLocker(locker Locker) JobOption {
	return func(j *internalJob) error {
		if locker == nil {
			return ErrWithDistributedJobLockerNil
		}
		j.locker = locker
		return nil
	}
}

// WithEventListeners sets the event listeners that should be
// run for the job.
func WithEventListeners(eventListeners ...EventListener) JobOption {
	return func(j *internalJob) error {
		for _, eventListener := range eventListeners {
			if err := eventListener(j); err != nil {
				return err
			}
		}
		return nil
	}
}

// WithLimitedRuns limits the number of executions of this job to n.
// Upon reaching the limit, the job is removed from the scheduler.
func WithLimitedRuns(limit uint) JobOption {
	return func(j *internalJob) error {
		j.limitRunsTo = &limitRunsTo{
			limit:    limit,
			runCount: 0,
		}
		return nil
	}
}

// WithName sets the name of the job. Name provides
// a human-readable identifier for the job.
func WithName(name string) JobOption {
	// TODO use the name for metrics and future logging option
	return func(j *internalJob) error {
		if name == "" {
			return ErrWithNameEmpty
		}
		j.name = name
		return nil
	}
}

// WithSingletonMode keeps the job from running again if it is already running.
// This is useful for jobs that should not overlap, and that occasionally
// (but not consistently) run longer than the interval between job runs.
func WithSingletonMode(mode LimitMode) JobOption {
	return func(j *internalJob) error {
		j.singletonMode = true
		j.singletonLimitMode = mode
		return nil
	}
}

// WithStartAt sets the option for starting the job at
// a specific datetime.
func WithStartAt(option StartAtOption) JobOption {
	return func(j *internalJob) error {
		return option(j)
	}
}

// StartAtOption defines options for starting the job
type StartAtOption func(*internalJob) error

// WithStartImmediately tells the scheduler to run the job immediately
// regardless of the type or schedule of job. After this immediate run
// the job is scheduled from this time based on the job definition.
func WithStartImmediately() StartAtOption {
	return func(j *internalJob) error {
		j.startImmediately = true
		return nil
	}
}

// WithStartDateTime sets the first date & time at which the job should run.
// This datetime must be in the future.
func WithStartDateTime(start time.Time) StartAtOption {
	return func(j *internalJob) error {
		if start.IsZero() || start.Before(time.Now()) {
			return ErrWithStartDateTimePast
		}
		j.startTime = start
		return nil
	}
}

// WithTags sets the tags for the job. Tags provide
// a way to identify jobs by a set of tags and remove
// multiple jobs by tag.
func WithTags(tags ...string) JobOption {
	return func(j *internalJob) error {
		j.tags = tags
		return nil
	}
}

// -----------------------------------------------
// -----------------------------------------------
// ------------- Job Event Listeners -------------
// -----------------------------------------------
// -----------------------------------------------

// EventListener defines the constructor for event
// listeners that can be used to listen for job events.
type EventListener func(*internalJob) error

// BeforeJobRuns is used to listen for when a job is about to run and
// then run the provided function.
func BeforeJobRuns(eventListenerFunc func(jobID uuid.UUID, jobName string)) EventListener {
	return func(j *internalJob) error {
		if eventListenerFunc == nil {
			return ErrEventListenerFuncNil
		}
		j.beforeJobRuns = eventListenerFunc
		return nil
	}
}

// AfterJobRuns is used to listen for when a job has run
// without an error, and then run the provided function.
func AfterJobRuns(eventListenerFunc func(jobID uuid.UUID, jobName string)) EventListener {
	return func(j *internalJob) error {
		if eventListenerFunc == nil {
			return ErrEventListenerFuncNil
		}
		j.afterJobRuns = eventListenerFunc
		return nil
	}
}

// AfterJobRunsWithError is used to listen for when a job has run and
// returned an error, and then run the provided function.
func AfterJobRunsWithError(eventListenerFunc func(jobID uuid.UUID, jobName string, err error)) EventListener {
	return func(j *internalJob) error {
		if eventListenerFunc == nil {
			return ErrEventListenerFuncNil
		}
		j.afterJobRunsWithError = eventListenerFunc
		return nil
	}
}

// AfterJobRunsWithPanic is used to listen for when a job has run and
// returned panicked recover data, and then run the provided function.
func AfterJobRunsWithPanic(eventListenerFunc func(jobID uuid.UUID, jobName string, recoverData any)) EventListener {
	return func(j *internalJob) error {
		if eventListenerFunc == nil {
			return ErrEventListenerFuncNil
		}
		j.afterJobRunsWithPanic = eventListenerFunc
		return nil
	}
}

// AfterLockError is used to when the distributed locker returns an error and
// then run the provided function.
func AfterLockError(eventListenerFunc func(jobID uuid.UUID, jobName string, err error)) EventListener {
	return func(j *internalJob) error {
		if eventListenerFunc == nil {
			return ErrEventListenerFuncNil
		}
		j.afterLockError = eventListenerFunc
		return nil
	}
}

// -----------------------------------------------
// -----------------------------------------------
// ---------------- Job Schedules ----------------
// -----------------------------------------------
// -----------------------------------------------

type jobSchedule interface {
	next(lastRun time.Time) time.Time
}

var _ jobSchedule = (*cronJob)(nil)

type cronJob struct {
	cronSchedule cron.Schedule
}

func (j *cronJob) next(lastRun time.Time) time.Time {
	return j.cronSchedule.Next(lastRun)
}

var _ jobSchedule = (*durationJob)(nil)

type durationJob struct {
	duration time.Duration
}

func (j *durationJob) next(lastRun time.Time) time.Time {
	return lastRun.Add(j.duration)
}

var _ jobSchedule = (*durationRandomJob)(nil)

type durationRandomJob struct {
	min, max time.Duration
	rand     *rand.Rand
}

func (j *durationRandomJob) next(lastRun time.Time) time.Time {
	r := j.rand.Int63n(int64(j.max - j.min))
	return lastRun.Add(j.min + time.Duration(r))
}

var _ jobSchedule = (*dailyJob)(nil)

type dailyJob struct {
	interval uint
	atTimes  []time.Time
}

func (d dailyJob) next(lastRun time.Time) time.Time {
	firstPass := true
	next := d.nextDay(lastRun, firstPass)
	if !next.IsZero() {
		return next
	}
	firstPass = false

	startNextDay := time.Date(lastRun.Year(), lastRun.Month(), lastRun.Day()+int(d.interval), 0, 0, 0, lastRun.Nanosecond(), lastRun.Location())
	return d.nextDay(startNextDay, firstPass)
}

func (d dailyJob) nextDay(lastRun time.Time, firstPass bool) time.Time {
	for _, at := range d.atTimes {
		// sub the at time hour/min/sec onto the lastScheduledRun's values
		// to use in checks to see if we've got our next run time
		atDate := time.Date(lastRun.Year(), lastRun.Month(), lastRun.Day(), at.Hour(), at.Minute(), at.Second(), lastRun.Nanosecond(), lastRun.Location())

		if firstPass && atDate.After(lastRun) {
			// checking to see if it is after i.e. greater than,
			// and not greater or equal as our lastScheduledRun day/time
			// will be in the loop, and we don't want to select it again
			return atDate
		} else if !firstPass && !atDate.Before(lastRun) {
			// now that we're looking at the next day, it's ok to consider
			// the same at time that was last run (as lastScheduledRun has been incremented)
			return atDate
		}
	}
	return time.Time{}
}

var _ jobSchedule = (*weeklyJob)(nil)

type weeklyJob struct {
	interval   uint
	daysOfWeek []time.Weekday
	atTimes    []time.Time
}

func (w weeklyJob) next(lastRun time.Time) time.Time {
	firstPass := true
	next := w.nextWeekDayAtTime(lastRun, firstPass)
	if !next.IsZero() {
		return next
	}
	firstPass = false

	startOfTheNextIntervalWeek := (lastRun.Day() - int(lastRun.Weekday())) + int(w.interval*7)
	from := time.Date(lastRun.Year(), lastRun.Month(), startOfTheNextIntervalWeek, 0, 0, 0, 0, lastRun.Location())
	return w.nextWeekDayAtTime(from, firstPass)
}

func (w weeklyJob) nextWeekDayAtTime(lastRun time.Time, firstPass bool) time.Time {
	for _, wd := range w.daysOfWeek {
		// checking if we're on the same day or later in the same week
		if wd >= lastRun.Weekday() {
			// weekDayDiff is used to add the correct amount to the atDate day below
			weekDayDiff := wd - lastRun.Weekday()
			for _, at := range w.atTimes {
				// sub the at time hour/min/sec onto the lastScheduledRun's values
				// to use in checks to see if we've got our next run time
				atDate := time.Date(lastRun.Year(), lastRun.Month(), lastRun.Day()+int(weekDayDiff), at.Hour(), at.Minute(), at.Second(), lastRun.Nanosecond(), lastRun.Location())

				if firstPass && atDate.After(lastRun) {
					// checking to see if it is after i.e. greater than,
					// and not greater or equal as our lastScheduledRun day/time
					// will be in the loop, and we don't want to select it again
					return atDate
				} else if !firstPass && !atDate.Before(lastRun) {
					// now that we're looking at the next week, it's ok to consider
					// the same at time that was last run (as lastScheduledRun has been incremented)
					return atDate
				}
			}
		}
	}
	return time.Time{}
}

var _ jobSchedule = (*monthlyJob)(nil)

type monthlyJob struct {
	interval    uint
	days        []int
	daysFromEnd []int
	atTimes     []time.Time
}

func (m monthlyJob) next(lastRun time.Time) time.Time {
	daysList := make([]int, len(m.days))
	copy(daysList, m.days)

	daysFromEnd := m.handleNegativeDays(lastRun, daysList, m.daysFromEnd)
	next := m.nextMonthDayAtTime(lastRun, daysFromEnd, true)
	if !next.IsZero() {
		return next
	}

	from := time.Date(lastRun.Year(), lastRun.Month()+time.Month(m.interval), 1, 0, 0, 0, 0, lastRun.Location())
	for next.IsZero() {
		daysFromEnd = m.handleNegativeDays(from, daysList, m.daysFromEnd)
		next = m.nextMonthDayAtTime(from, daysFromEnd, false)
		from = from.AddDate(0, int(m.interval), 0)
	}

	return next
}

func (m monthlyJob) handleNegativeDays(from time.Time, days, negativeDays []int) []int {
	var out []int
	// getting a list of the days from the end of the following month
	// -1 == the last day of the month
	firstDayNextMonth := time.Date(from.Year(), from.Month()+1, 1, 0, 0, 0, 0, from.Location())
	for _, daySub := range negativeDays {
		day := firstDayNextMonth.AddDate(0, 0, daySub).Day()
		out = append(out, day)
	}
	out = append(out, days...)
	slices.Sort(out)
	return out
}

func (m monthlyJob) nextMonthDayAtTime(lastRun time.Time, days []int, firstPass bool) time.Time {
	// find the next day in the month that should run and then check for an at time
	for _, day := range days {
		if day >= lastRun.Day() {
			for _, at := range m.atTimes {
				// sub the day, and the at time hour/min/sec onto the lastScheduledRun's values
				// to use in checks to see if we've got our next run time
				atDate := time.Date(lastRun.Year(), lastRun.Month(), day, at.Hour(), at.Minute(), at.Second(), lastRun.Nanosecond(), lastRun.Location())

				if atDate.Month() != lastRun.Month() {
					// this check handles if we're setting a day not in the current month
					// e.g. setting day 31 in Feb results in March 2nd
					continue
				}

				if firstPass && atDate.After(lastRun) {
					// checking to see if it is after i.e. greater than,
					// and not greater or equal as our lastScheduledRun day/time
					// will be in the loop, and we don't want to select it again
					return atDate
				} else if !firstPass && !atDate.Before(lastRun) {
					// now that we're looking at the next month, it's ok to consider
					// the same at time that was  lastScheduledRun (as lastScheduledRun has been incremented)
					return atDate
				}
			}
			continue
		}
	}
	return time.Time{}
}

var _ jobSchedule = (*oneTimeJob)(nil)

type oneTimeJob struct {
	sortedTimes []time.Time
}

// next finds the next item in a sorted list of times using binary-search.
//
// example: sortedTimes: [2, 4, 6, 8]
//
// lastRun: 1 => [idx=0,found=false] => next is 2 - sorted[idx] idx=0
// lastRun: 2 => [idx=0,found=true] => next is 4 - sorted[idx+1] idx=1
// lastRun: 3 => [idx=1,found=false] => next is 4 - sorted[idx] idx=1
// lastRun: 4 => [idx=1,found=true] => next is 6 - sorted[idx+1] idx=2
// lastRun: 7 => [idx=3,found=false] => next is 8 - sorted[idx] idx=3
// lastRun: 8 => [idx=3,found=found] => next is none
// lastRun: 9 => [idx=3,found=found] => next is none
func (o oneTimeJob) next(lastRun time.Time) time.Time {
	idx, found := slices.BinarySearchFunc(o.sortedTimes, lastRun, ascendingTime)
	// if found, the next run is the following index
	if found {
		idx++
	}
	// exhausted runs
	if idx >= len(o.sortedTimes) {
		return time.Time{}
	}

	return o.sortedTimes[idx]
}

// -----------------------------------------------
// -----------------------------------------------
// ---------------- Job Interface ----------------
// -----------------------------------------------
// -----------------------------------------------

// Job provides the available methods on the job
// available to the caller.
type Job interface {
	// ID returns the job's unique identifier.
	ID() uuid.UUID
	// LastRun returns the time of the job's last run
	LastRun() (time.Time, error)
	// Name returns the name defined on the job.
	Name() string
	// NextRun returns the time of the job's next scheduled run.
	NextRun() (time.Time, error)
	// NextRuns returns the requested number of calculated next run values.
	NextRuns(int) ([]time.Time, error)
	// RunNow runs the job once, now. This does not alter
	// the existing run schedule, and will respect all job
	// and scheduler limits. This means that running a job now may
	// cause the job's regular interval to be rescheduled due to
	// the instance being run by RunNow blocking your run limit.
	RunNow() error
	// Tags returns the job's string tags.
	Tags() []string
}

var _ Job = (*job)(nil)

// job is the internal struct that implements
// the public interface. This is used to avoid
// leaking information the caller never needs
// to have or tinker with.
type job struct {
	id            uuid.UUID
	name          string
	tags          []string
	jobOutRequest chan jobOutRequest
	runJobRequest chan runJobRequest
}

func (j job) ID() uuid.UUID {
	return j.id
}

func (j job) LastRun() (time.Time, error) {
	ij := requestJob(j.id, j.jobOutRequest)
	if ij == nil || ij.id == uuid.Nil {
		return time.Time{}, ErrJobNotFound
	}
	return ij.lastRun, nil
}

func (j job) Name() string {
	return j.name
}

func (j job) NextRun() (time.Time, error) {
	ij := requestJob(j.id, j.jobOutRequest)
	if ij == nil || ij.id == uuid.Nil {
		return time.Time{}, ErrJobNotFound
	}
	if len(ij.nextScheduled) == 0 {
		return time.Time{}, nil
	}
	// the first element is the next scheduled run with subsequent
	// runs following after in the slice
	return ij.nextScheduled[0], nil
}

func (j job) NextRuns(count int) ([]time.Time, error) {
	ij := requestJob(j.id, j.jobOutRequest)
	if ij == nil || ij.id == uuid.Nil {
		return nil, ErrJobNotFound
	}

	lengthNextScheduled := len(ij.nextScheduled)
	if lengthNextScheduled == 0 {
		return nil, nil
	} else if count <= lengthNextScheduled {
		return ij.nextScheduled[:count], nil
	}

	out := make([]time.Time, count)
	for i := 0; i < count; i++ {
		if i < lengthNextScheduled {
			out[i] = ij.nextScheduled[i]
			continue
		}

		from := out[i-1]
		out[i] = ij.next(from)
	}

	return out, nil
}

func (j job) Tags() []string {
	return j.tags
}

func (j job) RunNow() error {
	ctx, cancel := context.WithTimeout(context.Background(), time.Second)
	defer cancel()
	resp := make(chan error, 1)

	select {
	case j.runJobRequest <- runJobRequest{
		id:      j.id,
		outChan: resp,
	}:
	case <-time.After(100 * time.Millisecond):
		return ErrJobRunNowFailed
	}
	var err error
	select {
	case <-ctx.Done():
		return ErrJobRunNowFailed
	case errReceived := <-resp:
		err = errReceived
	}
	return err
}