move implementation to internal/ directory (#828)

This commit is contained in:
Christian Schwarz
2024-10-18 19:21:17 +02:00
committed by GitHub
parent b9b9ad10cf
commit 908807bd59
360 changed files with 507 additions and 507 deletions
@@ -0,0 +1,142 @@
package retentiongrid
import (
"sort"
"time"
)
type Interval interface {
Length() time.Duration
KeepCount() int
}
const RetentionGridKeepCountAll int = -1
type Grid struct {
intervals []Interval
}
type Entry interface {
Date() time.Time
}
func NewGrid(l []Interval) *Grid {
if len(l) == 0 {
panic("must specify at least one interval")
}
// TODO Maybe check for ascending interval lengths here, although the algorithm
// itself doesn't care about that.
return &Grid{l}
}
func (g Grid) FitEntries(entries []Entry) (keep, remove []Entry) {
if len(entries) == 0 {
return
}
// determine 'now' based on youngest snapshot
// => sort youngest-to-oldest
sort.SliceStable(entries, func(i, j int) bool {
return entries[i].Date().After(entries[j].Date())
})
now := entries[0].Date()
return g.fitEntriesWithNow(now, entries)
}
type bucket struct {
keepCount int
youngerThan time.Time
olderThanOrEq time.Time
entries []Entry
}
func makeBucketFromInterval(olderThanOrEq time.Time, i Interval) bucket {
var b bucket
kc := i.KeepCount()
if kc == 0 {
panic("keep count 0 is not allowed")
}
if (kc < 0) && kc != RetentionGridKeepCountAll {
panic("negative keep counts are not allowed")
}
b.keepCount = kc
b.olderThanOrEq = olderThanOrEq
b.youngerThan = b.olderThanOrEq.Add(-i.Length())
return b
}
func (b *bucket) Contains(e Entry) bool {
d := e.Date()
olderThan := d.Before(b.olderThanOrEq)
eq := d.Equal(b.olderThanOrEq)
youngerThan := d.After(b.youngerThan)
return (olderThan || eq) && youngerThan
}
func (b *bucket) AddIfContains(e Entry) (added bool) {
added = b.Contains(e)
if added {
b.entries = append(b.entries, e)
}
return
}
func (b *bucket) RemoveYoungerSnapsExceedingKeepCount() (removed []Entry) {
if b.keepCount == RetentionGridKeepCountAll {
return nil
}
removeCount := len(b.entries) - b.keepCount
if removeCount <= 0 {
return nil
}
// sort youngest-to-oldest
sort.SliceStable(b.entries, func(i, j int) bool {
return b.entries[i].Date().After(b.entries[j].Date())
})
return b.entries[:removeCount]
}
func (g Grid) fitEntriesWithNow(now time.Time, entries []Entry) (keep, remove []Entry) {
buckets := make([]bucket, len(g.intervals))
buckets[0] = makeBucketFromInterval(now, g.intervals[0])
for i := 1; i < len(g.intervals); i++ {
buckets[i] = makeBucketFromInterval(buckets[i-1].youngerThan, g.intervals[i])
}
keep = make([]Entry, 0)
remove = make([]Entry, 0)
assignEntriesToBuckets:
for ei := 0; ei < len(entries); ei++ {
e := entries[ei]
// unconditionally keep entries that are in the future
if now.Before(e.Date()) {
keep = append(keep, e)
continue assignEntriesToBuckets
}
// add to matching bucket, if any
for bi := range buckets {
if buckets[bi].AddIfContains(e) {
continue assignEntriesToBuckets
}
}
// unconditionally remove entries older than the oldest bucket
remove = append(remove, e)
}
// now apply the `KeepCount` per bucket
for _, b := range buckets {
destroy := b.RemoveYoungerSnapsExceedingKeepCount()
remove = append(remove, destroy...)
keep = append(keep, b.entries...)
}
return
}
@@ -0,0 +1,185 @@
package retentiongrid
import (
"fmt"
"strconv"
"strings"
"testing"
"time"
"github.com/stretchr/testify/assert"
)
type testInterval struct {
length time.Duration
keepCount int
}
func (i *testInterval) Length() time.Duration { return i.length }
func (i *testInterval) KeepCount() int { return i.keepCount }
func gridFromString(gs string) (g *Grid) {
sintervals := strings.Split(gs, "|")
intervals := make([]Interval, len(sintervals))
for idx, i := range sintervals {
comps := strings.SplitN(i, ",", 2)
var durationStr, numSnapsStr string
durationStr = comps[0]
if len(comps) == 1 {
numSnapsStr = "1"
} else {
numSnapsStr = comps[1]
}
var err error
var interval testInterval
if interval.keepCount, err = strconv.Atoi(numSnapsStr); err != nil {
panic(err)
}
if interval.length, err = time.ParseDuration(durationStr); err != nil {
panic(err)
}
intervals[idx] = &interval
}
return NewGrid(intervals)
}
type testSnap struct {
Name string
ShouldKeep bool
date time.Time
}
func (ds testSnap) Date() time.Time { return ds.date }
func validateRetentionGridFitEntries(t *testing.T, now time.Time, input, keep, remove []Entry) {
snapDescr := func(d testSnap) string {
return fmt.Sprintf("%s@%s", d.Name, d.date.Sub(now))
}
t.Logf("keep list:\n")
for k := range keep {
t.Logf("\t%s\n", snapDescr(keep[k].(testSnap)))
}
t.Logf("remove list:\n")
for k := range remove {
t.Logf("\t%s\n", snapDescr(remove[k].(testSnap)))
}
t.Logf("\n\n")
for _, s := range input {
d := s.(testSnap)
descr := snapDescr(d)
t.Logf("testing %s\n", descr)
if d.ShouldKeep {
assert.Contains(t, keep, d, "expecting %s to be kept", descr)
} else {
assert.Contains(t, remove, d, "expecting %s to be removed", descr)
}
}
t.Logf("resulting list:\n")
for k := range keep {
t.Logf("\t%s\n", snapDescr(keep[k].(testSnap)))
}
}
func TestEmptyInput(t *testing.T) {
g := gridFromString("10m|10m|10m|1h")
keep, remove := g.FitEntries([]Entry{})
assert.Empty(t, keep)
assert.Empty(t, remove)
}
func TestIntervalBoundariesAndAlignment(t *testing.T) {
g := gridFromString("10m|10m|10m")
t.Logf("%#v\n", g)
now := time.Unix(0, 0)
snaps := []Entry{
testSnap{"0", true, now.Add(1 * time.Minute)}, // before now => keep unconditionally
testSnap{"1", true, now}, // 1st interval left edge => inclusive
testSnap{"2", true, now.Add(-10 * time.Minute)}, // 2nd interval left edge => inclusive
testSnap{"3", true, now.Add(-20 * time.Minute)}, // 3rd interval left edge => inclusuive
testSnap{"4", false, now.Add(-30 * time.Minute)}, // 3rd interval right edge => excludive
testSnap{"5", false, now.Add(-40 * time.Minute)}, // after last interval => remove unconditionally
}
keep, remove := g.fitEntriesWithNow(now, snaps)
validateRetentionGridFitEntries(t, now, snaps, keep, remove)
}
func TestKeepsOldestSnapsInABucket(t *testing.T) {
g := gridFromString("1m,2")
relt := func(secs int64) time.Time { return time.Unix(secs, 0) }
snaps := []Entry{
testSnap{"1", true, relt(1)},
testSnap{"2", true, relt(2)},
testSnap{"3", false, relt(3)},
testSnap{"4", false, relt(4)},
testSnap{"5", false, relt(5)},
}
now := relt(6)
keep, remove := g.FitEntries(snaps)
validateRetentionGridFitEntries(t, now, snaps, keep, remove)
}
func TestRespectsKeepCountAll(t *testing.T) {
g := gridFromString("1m,-1|1m,1")
relt := func(secs int64) time.Time { return time.Unix(secs, 0) }
snaps := []Entry{
testSnap{"a", true, relt(0)},
testSnap{"b", true, relt(-1)},
testSnap{"c", true, relt(-2)},
testSnap{"d", false, relt(-60)},
testSnap{"e", true, relt(-61)},
}
keep, remove := g.FitEntries(snaps)
validateRetentionGridFitEntries(t, relt(61), snaps, keep, remove)
}
func TestComplex(t *testing.T) {
g := gridFromString("10m,-1|10m|10m,2|1h")
t.Logf("%#v\n", g)
now := time.Unix(0, 0)
snaps := []Entry{
// pre-now must always be kept
testSnap{"1", true, now.Add(3 * time.Minute)},
// 1st interval allows unlimited entries
testSnap{"b1", true, now.Add(-6 * time.Minute)},
testSnap{"b3", true, now.Add(-8 * time.Minute)},
testSnap{"b2", true, now.Add(-9 * time.Minute)},
// 2nd interval allows 1 entry
testSnap{"a", false, now.Add(-11 * time.Minute)},
testSnap{"c", true, now.Add(-19 * time.Minute)},
// 3rd interval allows 2 entries
testSnap{"foo", true, now.Add(-25 * time.Minute)},
testSnap{"bar", true, now.Add(-26 * time.Minute)},
// this is at the left edge of the 4th interval
testSnap{"border", false, now.Add(-30 * time.Minute)},
// right in the 4th interval
testSnap{"d", true, now.Add(-1*time.Hour - 15*time.Minute)},
// on the right edge of 4th interval => not in it => delete
testSnap{"q", false, now.Add(-1*time.Hour - 30*time.Minute)},
// older then 4th interval => always delete
testSnap{"e", false, now.Add(-1*time.Hour - 31*time.Minute)},
testSnap{"f", false, now.Add(-2 * time.Hour)},
}
keep, remove := g.fitEntriesWithNow(now, snaps)
validateRetentionGridFitEntries(t, now, snaps, keep, remove)
}