Initial QSfera import

This commit is contained in:
Курнат Андрей
2026-06-07 10:20:04 +03:00
commit 2315f25754
16485 changed files with 4826827 additions and 0 deletions
+284
View File
@@ -0,0 +1,284 @@
// Package revisions allows manipulating revisions in a storage provider.
package revisions
import (
"fmt"
"os"
"path/filepath"
"regexp"
"strings"
"sync"
"github.com/opencloud-eu/reva/v2/pkg/storage/pkg/decomposedfs/node"
"github.com/vmihailenco/msgpack/v5"
)
var (
// regex to determine if a node versioned. Examples:
// 9113a718-8285-4b32-9042-f930f1a58ac2.REV.2024-05-22T07:32:53.89969726Z
// 9113a718-8285-4b32-9042-f930f1a58ac2.REV.2024-05-22T07:32:53.89969726Z.mpk
// 9113a718-8285-4b32-9042-f930f1a58ac2.REV.2024-05-22T07:32:53.89969726Z.mlock
_versionRegex = regexp.MustCompile(`\.REV\.[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2}\.[0-9]+Z*`)
_spaceIDRegex = regexp.MustCompile(`/spaces/(.+)/nodes/`)
)
// DelBlobstore is the interface for a blobstore that can delete blobs.
type DelBlobstore interface {
Delete(node *node.Node) error
}
// Glob uses globbing to find all revision nodes in a storage provider.
func Glob(pattern string) <-chan string {
ch := make(chan string)
go func() {
defer close(ch)
nodes, err := filepath.Glob(filepath.Join(pattern))
if err != nil {
fmt.Println("error globbing", pattern, err)
return
}
if len(nodes) == 0 {
fmt.Println("no nodes found. Double check storage path")
return
}
for _, n := range nodes {
if _versionRegex.MatchString(n) {
ch <- n
}
}
}()
return ch
}
// GlobWorkers uses multiple go routine to glob all revision nodes in a storage provider.
func GlobWorkers(pattern string, depth string, remainder string) <-chan string {
wg := sync.WaitGroup{}
ch := make(chan string)
go func() {
defer close(ch)
nodes, err := filepath.Glob(pattern + depth)
if err != nil {
fmt.Println("error globbing", pattern, err)
return
}
if len(nodes) == 0 {
fmt.Println("no nodes found. Double check storage path")
return
}
for _, node := range nodes {
wg.Add(1)
go func(node string) {
defer wg.Done()
nodes, err := filepath.Glob(node + remainder)
if err != nil {
fmt.Println("error globbing", node, err)
return
}
for _, n := range nodes {
if _versionRegex.MatchString(n) {
ch <- n
}
}
}(node)
}
wg.Wait()
}()
return ch
}
// Walk walks the storage provider to find all revision nodes.
func Walk(base string) <-chan string {
ch := make(chan string)
go func() {
defer close(ch)
err := filepath.Walk(base, func(path string, info os.FileInfo, err error) error {
if err != nil {
fmt.Println("error walking", base, err)
return err
}
if !_versionRegex.MatchString(info.Name()) {
return nil
}
ch <- path
return nil
})
if err != nil {
fmt.Println("error walking", base, err)
return
}
}()
return ch
}
// List uses directory listing to find all revision nodes in a storage provider.
func List(base string, workers int) <-chan string {
ch := make(chan string)
go func() {
defer close(ch)
if err := listFolder(base, ch, make(chan struct{}, workers)); err != nil {
fmt.Println("error listing", base, err)
return
}
}()
return ch
}
// PurgeRevisions removes all revisions from a storage provider.
func PurgeRevisions(nodes <-chan string, bs DelBlobstore, dryRun, verbose bool) (int, int, int) {
countFiles := 0
countBlobs := 0
countRevisions := 0
var err error
for d := range nodes {
if !_versionRegex.MatchString(d) {
continue
}
var (
spaceID, blobID string
)
e := filepath.Ext(d)
switch e {
case ".mpk":
blobID, err = getBlobID(d)
if err != nil {
fmt.Printf("error getting blobID from %s: %v\n", d, err)
continue
}
matches := _spaceIDRegex.FindStringSubmatch(d)
if len(matches) != 2 {
fmt.Printf("error extracting spaceID from %s\n", d)
continue
}
spaceID = strings.ReplaceAll(matches[1], "/", "")
countBlobs++
case ".mlock":
// no extra action on .mlock files
default:
countRevisions++
}
if !dryRun {
if blobID != "" {
// TODO: needs spaceID for decomposeds3
if err := bs.Delete(&node.Node{BaseNode: node.BaseNode{SpaceID: spaceID}, BlobID: blobID}); err != nil {
fmt.Printf("error deleting blob %s: %v\n", blobID, err)
continue
}
}
if err := os.Remove(d); err != nil {
fmt.Printf("error removing %s: %v\n", d, err)
continue
}
}
countFiles++
if verbose {
spaceID, nodeID := getIDsFromPath(d)
if dryRun {
fmt.Println("Would delete")
fmt.Println("\tResourceID:", spaceID+"!"+nodeID)
fmt.Println("\tSpaceID:", spaceID)
fmt.Println("\tPath:", d)
if blobID != "" {
fmt.Println("\tBlob:", blobID)
}
} else {
fmt.Println("Deleted")
fmt.Println("\tResourceID:", spaceID+"!"+nodeID)
fmt.Println("\tSpaceID:", spaceID)
fmt.Println("\tPath:", d)
if blobID != "" {
fmt.Println("\tBlob:", blobID)
}
}
}
}
return countFiles, countBlobs, countRevisions
}
func listFolder(path string, ch chan<- string, workers chan struct{}) error {
workers <- struct{}{}
wg := sync.WaitGroup{}
children, err := os.ReadDir(path)
if err != nil {
<-workers
return err
}
for _, child := range children {
if child.IsDir() {
wg.Go(func() {
if err := listFolder(filepath.Join(path, child.Name()), ch, workers); err != nil {
fmt.Println("error listing", path, err)
}
})
}
if _versionRegex.MatchString(child.Name()) {
ch <- filepath.Join(path, child.Name())
}
}
<-workers
wg.Wait()
return nil
}
func getBlobID(path string) (string, error) {
b, err := os.ReadFile(path)
if err != nil {
return "", err
}
m := map[string][]byte{}
if err := msgpack.Unmarshal(b, &m); err != nil {
return "", err
}
if bid := m["user.oc.blobid"]; string(bid) != "" {
return string(bid), nil
}
return "", nil
}
func getIDsFromPath(path string) (string, string) {
rawIDs := strings.Split(path, "/nodes/")
if len(rawIDs) != 2 {
return "", ""
}
s := strings.Split(rawIDs[0], "/spaces/")
if len(s) != 2 {
return "", ""
}
n := strings.Split(rawIDs[1], ".REV.")
if len(n) != 2 {
return "", ""
}
spaceID := strings.Replace(s[1], "/", "", -1)
nodeID := strings.Replace(n[0], "/", "", -1)
return spaceID, filepath.Base(nodeID)
}
@@ -0,0 +1,170 @@
package revisions
import (
"fmt"
"io/fs"
"os"
"path/filepath"
"strconv"
"testing"
"github.com/google/uuid"
"github.com/opencloud-eu/reva/v2/pkg/storage/pkg/decomposedfs/lookup"
"github.com/test-go/testify/require"
)
var (
_basePath = "/spaces/8f/638374-6ea8-4f0d-80c4-66d9b49830a5/nodes/"
)
// func TestInit(t *testing.T) {
// initialize(10, 2)
// defer os.RemoveAll("test_temp")
// }
func TestGlob30(t *testing.T) { test(t, 10, 2, glob) }
func TestGlob80(t *testing.T) { test(t, 20, 3, glob) }
func TestGlob250(t *testing.T) { test(t, 50, 4, glob) }
func TestGlob600(t *testing.T) { test(t, 100, 5, glob) }
func TestWalk30(t *testing.T) { test(t, 10, 2, walk) }
func TestWalk80(t *testing.T) { test(t, 20, 3, walk) }
func TestWalk250(t *testing.T) { test(t, 50, 4, walk) }
func TestWalk600(t *testing.T) { test(t, 100, 5, walk) }
func TestList30(t *testing.T) { test(t, 10, 2, list2) }
func TestList80(t *testing.T) { test(t, 20, 3, list10) }
func TestList250(t *testing.T) { test(t, 50, 4, list20) }
func TestList600(t *testing.T) { test(t, 100, 5, list2) }
func TestGlobWorkers30(t *testing.T) { test(t, 10, 2, globWorkersD1) }
func TestGlobWorkers80(t *testing.T) { test(t, 20, 3, globWorkersD2) }
func TestGlobWorkers250(t *testing.T) { test(t, 50, 4, globWorkersD4) }
func TestGlobWorkers600(t *testing.T) { test(t, 100, 5, globWorkersD2) }
func BenchmarkGlob30(b *testing.B) { benchmark(b, 10, 2, glob) }
func BenchmarkWalk30(b *testing.B) { benchmark(b, 10, 2, walk) }
func BenchmarkList30(b *testing.B) { benchmark(b, 10, 2, list2) }
func BenchmarkGlobWorkers30(b *testing.B) { benchmark(b, 10, 2, globWorkersD2) }
func BenchmarkGlob80(b *testing.B) { benchmark(b, 20, 3, glob) }
func BenchmarkWalk80(b *testing.B) { benchmark(b, 20, 3, walk) }
func BenchmarkList80(b *testing.B) { benchmark(b, 20, 3, list2) }
func BenchmarkGlobWorkers80(b *testing.B) { benchmark(b, 20, 3, globWorkersD2) }
func BenchmarkGlob250(b *testing.B) { benchmark(b, 50, 4, glob) }
func BenchmarkWalk250(b *testing.B) { benchmark(b, 50, 4, walk) }
func BenchmarkList250(b *testing.B) { benchmark(b, 50, 4, list2) }
func BenchmarkGlobWorkers250(b *testing.B) { benchmark(b, 50, 4, globWorkersD2) }
func BenchmarkGlobAT600(b *testing.B) { benchmark(b, 100, 5, glob) }
func BenchmarkWalkAT600(b *testing.B) { benchmark(b, 100, 5, walk) }
func BenchmarkList2AT600(b *testing.B) { benchmark(b, 100, 5, list2) }
func BenchmarkList10AT600(b *testing.B) { benchmark(b, 100, 5, list10) }
func BenchmarkList20AT600(b *testing.B) { benchmark(b, 100, 5, list20) }
func BenchmarkGlobWorkersD1AT600(b *testing.B) { benchmark(b, 100, 5, globWorkersD1) }
func BenchmarkGlobWorkersD2AT600(b *testing.B) { benchmark(b, 100, 5, globWorkersD2) }
func BenchmarkGlobWorkersD4AT600(b *testing.B) { benchmark(b, 100, 5, globWorkersD4) }
func BenchmarkGlobAT22000(b *testing.B) { benchmark(b, 2000, 10, glob) }
func BenchmarkWalkAT22000(b *testing.B) { benchmark(b, 2000, 10, walk) }
func BenchmarkList2AT22000(b *testing.B) { benchmark(b, 2000, 10, list2) }
func BenchmarkList10AT22000(b *testing.B) { benchmark(b, 2000, 10, list10) }
func BenchmarkList20AT22000(b *testing.B) { benchmark(b, 2000, 10, list20) }
func BenchmarkGlobWorkersD1AT22000(b *testing.B) { benchmark(b, 2000, 10, globWorkersD1) }
func BenchmarkGlobWorkersD2AT22000(b *testing.B) { benchmark(b, 2000, 10, globWorkersD2) }
func BenchmarkGlobWorkersD4AT22000(b *testing.B) { benchmark(b, 2000, 10, globWorkersD4) }
func BenchmarkGlob110000(b *testing.B) { benchmark(b, 10000, 10, glob) }
func BenchmarkWalk110000(b *testing.B) { benchmark(b, 10000, 10, walk) }
func BenchmarkList110000(b *testing.B) { benchmark(b, 10000, 10, list2) }
func BenchmarkGlobWorkers110000(b *testing.B) { benchmark(b, 10000, 10, globWorkersD2) }
func benchmark(b *testing.B, numNodes int, numRevisions int, f func(string) <-chan string) {
base := initialize(numNodes, numRevisions)
defer os.RemoveAll(base)
b.ResetTimer()
for i := 0; i < b.N; i++ {
ch := f(base)
PurgeRevisions(ch, nil, false, false)
}
b.StopTimer()
}
func test(t *testing.T, numNodes int, numRevisions int, f func(string) <-chan string) {
base := initialize(numNodes, numRevisions)
defer os.RemoveAll(base)
ch := f(base)
_, _, revisions := PurgeRevisions(ch, nil, false, false)
require.Equal(t, numNodes*numRevisions, revisions, "Deleted Revisions")
}
func glob(base string) <-chan string {
return Glob(base + _basePath + "*/*/*/*/*")
}
func walk(base string) <-chan string {
return Walk(base + _basePath)
}
func list2(base string) <-chan string {
return List(base+_basePath, 2)
}
func list10(base string) <-chan string {
return List(base+_basePath, 10)
}
func list20(base string) <-chan string {
return List(base+_basePath, 20)
}
func globWorkersD1(base string) <-chan string {
return GlobWorkers(base+_basePath, "*", "/*/*/*/*")
}
func globWorkersD2(base string) <-chan string {
return GlobWorkers(base+_basePath, "*/*", "/*/*/*")
}
func globWorkersD4(base string) <-chan string {
return GlobWorkers(base+_basePath, "*/*/*/*", "/*")
}
func initialize(numNodes int, numRevisions int) string {
base := "test_temp_" + uuid.New().String()
if err := os.Mkdir(base, os.ModePerm); err != nil {
fmt.Println("Error creating test_temp directory", err)
os.RemoveAll(base)
os.Exit(1)
}
// create base path
if err := os.MkdirAll(base+_basePath, fs.ModePerm); err != nil {
fmt.Println("Error creating base path", err)
os.RemoveAll(base)
os.Exit(1)
}
for i := 0; i < numNodes; i++ {
path := lookup.Pathify(uuid.New().String(), 4, 2)
dir := filepath.Dir(path)
if err := os.MkdirAll(base+_basePath+dir, fs.ModePerm); err != nil {
fmt.Println("Error creating test_temp directory", err)
os.RemoveAll(base)
os.Exit(1)
}
if _, err := os.Create(base + _basePath + path); err != nil {
fmt.Println("Error creating file", err)
os.RemoveAll(base)
os.Exit(1)
}
for i := 0; i < numRevisions; i++ {
os.Create(base + _basePath + path + ".REV.2024-05-22T07:32:53.89969" + strconv.Itoa(i) + "Z")
}
}
return base
}