Merge remote-tracking branch 'origin/dev' into dev

This commit is contained in:
Vanessa 2022-06-05 00:48:45 +08:00
commit bb09e1e102
3 changed files with 135 additions and 72 deletions

View file

@ -278,7 +278,7 @@ func CreateLocalBackup() (err error) {
return
}
err = genCloudIndex(newBackupDir, map[string]bool{})
err = genFullCloudIndex(newBackupDir, map[string]bool{})
if nil != err {
return
}
@ -364,7 +364,7 @@ func UploadBackup() (err error) {
util.PushEndlessProgress(Conf.Language(61))
util.LogInfof("uploading backup...")
start := time.Now()
wroteFiles, transferSize, err := ossUpload(localDirPath, "backup", "", false)
wroteFiles, transferSize, err := ossUpload(localDirPath, "backup", "not exist", false, map[string]bool{}, map[string]bool{})
if nil == err {
elapsed := time.Now().Sub(start).Seconds()
util.LogInfof("uploaded backup [wroteFiles=%d, transferSize=%s] in [%.2fs]", wroteFiles, humanize.Bytes(transferSize), elapsed)

View file

@ -318,17 +318,36 @@ func ossDownload0(localDirPath, cloudDirPath, fetch string, fetchedFiles *int, t
return
}
func ossUpload(localDirPath, cloudDirPath, cloudDevice string, boot bool) (wroteFiles int, transferSize uint64, err error) {
func ossUpload(localDirPath, cloudDirPath, cloudDevice string, boot bool, removedSyncList, upsertedSyncList map[string]bool) (wroteFiles int, transferSize uint64, err error) {
if !gulu.File.IsExist(localDirPath) {
return
}
var cloudFileList map[string]*CloudIndex
localDevice := Conf.System.ID
syncIgnoreList := getSyncIgnoreList()
excludes := map[string]bool{}
ignores := syncIgnoreList.Values()
for _, p := range ignores {
relPath := p.(string)
relPath = pathSha256Short(relPath, "/")
relPath = filepath.Join(localDirPath, relPath)
excludes[relPath] = true
}
localFileList, getLocalFileListErr := getLocalFileListOSS(cloudDirPath, excludes)
if "" != localDevice && localDevice == cloudDevice {
//util.LogInfof("cloud device is the same as local device, get index from local")
cloudFileList, err = getLocalFileListOSS(cloudDirPath)
if nil != err {
if nil == getLocalFileListErr {
cloudFileList = map[string]*CloudIndex{}
// 深拷贝一次,避免后面和 localFileList 对比时引用值相同
for p, idx := range localFileList {
cloudFileList[p] = &CloudIndex{
Hash: idx.Hash,
Size: idx.Size,
}
}
} else {
util.LogInfof("get local index failed [%s], get index from cloud", err)
cloudFileList, err = getCloudFileListOSS(cloudDirPath)
}
@ -339,7 +358,12 @@ func ossUpload(localDirPath, cloudDirPath, cloudDevice string, boot bool) (wrote
return
}
localUpserts, cloudRemoves, err := cloudUpsertRemoveListOSS(localDirPath, cloudFileList)
localUpserts, cloudRemoves, err := cloudUpsertRemoveListOSS(localDirPath, cloudFileList, localFileList, removedSyncList, upsertedSyncList, excludes)
if nil != err {
return
}
err = ossRemove0(cloudDirPath, cloudRemoves)
if nil != err {
return
}
@ -400,8 +424,6 @@ func ossUpload(localDirPath, cloudDirPath, cloudDevice string, boot bool) (wrote
if needPushProgress {
util.PushMsg(Conf.Language(105), 3000)
}
err = ossRemove0(cloudDirPath, cloudRemoves)
return
}
@ -584,20 +606,27 @@ func getCloudSync(cloudDir string) (assetSize, backupSize int64, device string,
return
}
func getLocalFileListOSS(dirPath string) (ret map[string]*CloudIndex, err error) {
func getLocalFileListOSS(dirPath string, excludes map[string]bool) (ret map[string]*CloudIndex, err error) {
dir := "sync"
if !strings.HasPrefix(dirPath, "sync") {
dir = "backup"
}
data, err := os.ReadFile(filepath.Join(util.WorkspaceDir, dir, "index.json"))
localDirPath := filepath.Join(util.WorkspaceDir, dir)
indexPath := filepath.Join(localDirPath, "index.json")
if !gulu.File.IsExist(indexPath) {
err = genFullCloudIndex(localDirPath, excludes)
if nil != err {
return
}
}
data, err := os.ReadFile(indexPath)
if nil != err {
return
}
if err = gulu.JSON.UnmarshalJSON(data, &ret); nil != err {
return
}
err = gulu.JSON.UnmarshalJSON(data, &ret)
return
}
@ -696,44 +725,26 @@ func localUpsertRemoveListOSS(localDirPath string, cloudFileList map[string]*Clo
return
}
func cloudUpsertRemoveListOSS(localDirPath string, cloudFileList map[string]*CloudIndex) (localUpserts, cloudRemoves []string, err error) {
func cloudUpsertRemoveListOSS(localDirPath string, cloudFileList, localFileList map[string]*CloudIndex, removedSyncList, upsertedSyncList, excludes map[string]bool) (localUpserts, cloudRemoves []string, err error) {
localUpserts, cloudRemoves = []string{}, []string{}
unchanged := map[string]bool{}
for cloudFile, cloudIdx := range cloudFileList {
localCheckPath := filepath.Join(localDirPath, cloudFile)
if !gulu.File.IsExist(localCheckPath) {
cloudRemoves = append(cloudRemoves, cloudFile)
continue
}
localHash, hashErr := util.GetEtag(localCheckPath)
if nil != hashErr {
util.LogErrorf("get local file [%s] hash failed: %s", localCheckPath, hashErr)
err = hashErr
return
}
if localHash == cloudIdx.Hash {
unchanged[localCheckPath] = true
}
}
syncIgnoreList := getSyncIgnoreList()
excludes := map[string]bool{}
ignores := syncIgnoreList.Values()
for _, p := range ignores {
relPath := p.(string)
relPath = pathSha246(relPath, "/")
relPath = filepath.Join(localDirPath, relPath)
excludes[relPath] = true
}
delete(unchanged, filepath.Join(localDirPath, "index.json")) // 同步偶尔报错 `The system cannot find the path specified.` https://github.com/siyuan-note/siyuan/issues/4942
err = genCloudIndex(localDirPath, excludes)
if nil != err {
if err = genIncCloudIndex(localDirPath, &localFileList, removedSyncList, upsertedSyncList, excludes); nil != err {
return
}
unchanged := map[string]bool{}
for cloudFile, cloudIdx := range cloudFileList {
localIdx := localFileList[cloudFile]
if nil == localIdx {
cloudRemoves = append(cloudRemoves, cloudFile)
continue
}
if localIdx.Hash == cloudIdx.Hash {
unchanged[filepath.Join(localDirPath, cloudFile)] = true
}
}
delete(unchanged, filepath.Join(localDirPath, "index.json")) // 同步偶尔报错 `The system cannot find the path specified.` https://github.com/siyuan-note/siyuan/issues/4942
filepath.Walk(localDirPath, func(path string, info fs.FileInfo, err error) error {
if localDirPath == path || info.IsDir() {
return nil

View file

@ -135,8 +135,9 @@ func SyncData(boot, exit, byHand bool) {
WaitForWritingFiles()
writingDataLock.Lock()
var err error
var removedSyncList, upsertedSyncList map[string]bool
// 将 data 变更同步到 sync
if err = workspaceData2SyncDir(); nil != err {
if removedSyncList, upsertedSyncList, err = workspaceData2SyncDir(); nil != err {
msg := fmt.Sprintf(Conf.Language(80), formatErrorMsg(err))
Conf.Sync.Stat = msg
util.PushErrMsg(msg, 7000)
@ -245,7 +246,7 @@ func SyncData(boot, exit, byHand bool) {
return
}
wroteFiles, transferSize, err := ossUpload(localSyncDirPath, "sync/"+Conf.Sync.CloudName, device, boot)
wroteFiles, transferSize, err := ossUpload(localSyncDirPath, "sync/"+Conf.Sync.CloudName, device, boot, removedSyncList, upsertedSyncList)
if nil != err {
util.PushClearMsg()
IncWorkspaceDataVer() // 上传失败的话提升本地版本,以备下次上传
@ -560,7 +561,7 @@ func syncDirUpsertWorkspaceData(downloadedFiles []string) (err error) {
// 2. 将 sync 中新增/修改的文件解密后拷贝到 data 中
func syncDir2WorkspaceData(boot bool) (upsertFiles, removeFiles []string, err error) {
start := time.Now()
unchanged, removeFiles, err := unchangedSyncList()
unchanged, removeFiles, err := calcUnchangedSyncList()
if nil != err {
return
}
@ -590,18 +591,17 @@ func syncDir2WorkspaceData(boot bool) (upsertFiles, removeFiles []string, err er
// workspaceData2SyncDir 将 data 的数据更新到 sync 中。
// 1. 删除 sync 中多余的文件
// 2. 将 data 中新增/修改的文件加密后拷贝到 sync 中
func workspaceData2SyncDir() (err error) {
func workspaceData2SyncDir() (removedSyncList, upsertedSyncList map[string]bool, err error) {
start := time.Now()
filesys.ReleaseAllFileLocks()
passwd := Conf.E2EEPasswd
unchanged, removed, err := unchangedDataList(passwd)
unchangedDataList, removedSyncList, err := calcUnchangedDataList(passwd)
if nil != err {
return
}
_ = removed // TODO: 支持多设备操作不同文档后云端同步合并 https://github.com/siyuan-note/siyuan/issues/5092
encryptedDataDir, err := prepareSyncData(passwd, unchanged)
encryptedDataDir, upsertedSyncList, err := prepareSyncData(passwd, unchangedDataList)
if nil != err {
util.LogErrorf("encrypt data dir failed: %s", err)
return
@ -623,9 +623,9 @@ type CloudIndex struct {
Size int64 `json:"size"`
}
func genCloudIndex(localDirPath string, excludes map[string]bool) (err error) {
// genCloudIndex 全量生成云端索引文件。
func genFullCloudIndex(localDirPath string, excludes map[string]bool) (err error) {
cloudIndex := map[string]*CloudIndex{}
// TODO: 优化云端同步资源占用 https://github.com/siyuan-note/siyuan/issues/5093
err = filepath.Walk(localDirPath, func(path string, info fs.FileInfo, err error) error {
if nil != err {
return err
@ -665,6 +665,46 @@ func genCloudIndex(localDirPath string, excludes map[string]bool) (err error) {
return
}
// genCloudIndex 增量生成云端索引文件。
func genIncCloudIndex(localDirPath string, localFileList *map[string]*CloudIndex, removes, upserts, excludes map[string]bool) (err error) {
for remove, _ := range removes {
delete(*localFileList, remove)
}
for exclude, _ := range excludes {
delete(*localFileList, filepath.ToSlash(strings.TrimPrefix(exclude, localDirPath)))
}
for upsert, _ := range upserts {
path := filepath.Join(localDirPath, upsert)
if excludes[path] {
continue
}
info, statErr := os.Stat(path)
if nil != statErr {
util.LogErrorf("stat file [%s] failed: %s", path, statErr)
return statErr
}
hash, hashErr := util.GetEtag(path)
if nil != hashErr {
util.LogErrorf("get file [%s] hash failed: %s", path, hashErr)
return hashErr
}
(*localFileList)[upsert] = &CloudIndex{Hash: hash, Size: info.Size()}
}
data, err := gulu.JSON.MarshalJSON(localFileList)
if nil != err {
util.LogErrorf("marshal sync cloud index failed: %s", err)
return
}
if err = os.WriteFile(filepath.Join(localDirPath, "index.json"), data, 0644); nil != err {
util.LogErrorf("write sync cloud index failed: %s", err)
return
}
return
}
func recoverSyncData(modified map[string]bool) (decryptedDataDir string, upsertFiles []string, err error) {
passwd := Conf.E2EEPasswd
decryptedDataDir = filepath.Join(util.WorkspaceDir, "incremental", "sync-decrypt")
@ -764,7 +804,7 @@ func recoverSyncData(modified map[string]bool) (decryptedDataDir string, upsertF
return
}
func prepareSyncData(passwd string, unchangedList map[string]bool) (encryptedDataDir string, err error) {
func prepareSyncData(passwd string, unchangedDataList map[string]bool) (encryptedDataDir string, upsertedSyncList map[string]bool, err error) {
encryptedDataDir = filepath.Join(util.WorkspaceDir, "incremental", "sync-encrypt")
if err = os.RemoveAll(encryptedDataDir); nil != err {
return
@ -792,12 +832,12 @@ func prepareSyncData(passwd string, unchangedList map[string]bool) (encryptedDat
p := plainP
if !strings.HasPrefix(plainP, ".siyuan") { // 配置目录下都用明文,其他文件需要映射文件名
p = pathSha246(p, string(os.PathSeparator))
p = pathSha256Short(p, string(os.PathSeparator))
}
metaJSON[filepath.ToSlash(p)] = filepath.ToSlash(plainP)
// 如果不是新增或者修改则跳过
if unchangedList[path] {
if unchangedDataList[path] {
return nil
}
@ -857,6 +897,7 @@ func prepareSyncData(passwd string, unchangedList map[string]bool) (encryptedDat
}
}
upsertedSyncList = map[string]bool{}
// 检查文件是否全部已经编入索引
err = filepath.Walk(encryptedDataDir, func(path string, info fs.FileInfo, _ error) error {
if encryptedDataDir == path {
@ -869,6 +910,10 @@ func prepareSyncData(passwd string, unchangedList map[string]bool) (encryptedDat
util.LogErrorf("not found sync path in meta [%s]", path)
return errors.New(Conf.Language(27))
}
if !info.IsDir() {
upsertedSyncList["/"+path] = true
}
return nil
})
if nil != err {
@ -881,7 +926,8 @@ func prepareSyncData(passwd string, unchangedList map[string]bool) (encryptedDat
}
data, err = encryption.AESGCMEncryptBinBytes(data, passwd)
if nil != err {
return "", errors.New("encrypt file failed")
util.LogErrorf("encrypt file failed: %s", err)
return
}
meta := filepath.Join(encryptedDataDir, pathJSON)
if err = os.WriteFile(meta, data, 0644); nil != err {
@ -907,8 +953,8 @@ func modifiedSyncList(unchangedList map[string]bool) (ret map[string]bool) {
return
}
// unchangedSyncList 获取 data 和 sync 一致(没有修改过)的文件列表,并删除 data 中不存在于 sync 中的多余文件。
func unchangedSyncList() (ret map[string]bool, removes []string, err error) {
// calcUnchangedSyncList 获取 data 和 sync 一致(没有修改过)的文件列表,并删除 data 中不存在于 sync 中的多余文件。
func calcUnchangedSyncList() (ret map[string]bool, removes []string, err error) {
syncDir := Conf.Sync.GetSaveDir()
meta := filepath.Join(syncDir, pathJSON)
if !gulu.File.IsExist(meta) {
@ -935,7 +981,7 @@ func unchangedSyncList() (ret map[string]bool, removes []string, err error) {
ignores := syncIgnoreList.Values()
for _, p := range ignores {
relPath := p.(string)
relPath = pathSha246(relPath, "/")
relPath = pathSha256Short(relPath, "/")
relPath = filepath.Join(syncDir, relPath)
excludes[relPath] = true
}
@ -956,7 +1002,7 @@ func unchangedSyncList() (ret map[string]bool, removes []string, err error) {
plainP := strings.TrimPrefix(path, util.DataDir+sep)
dataP := plainP
dataP = pathSha246(dataP, sep)
dataP = pathSha256Short(dataP, sep)
syncP := filepath.Join(syncDir, dataP)
if excludes[syncP] {
@ -993,8 +1039,8 @@ func unchangedSyncList() (ret map[string]bool, removes []string, err error) {
return
}
// unchangedDataList 获取 sync 和 data 一致(没有修改过)的文件列表,并删除 sync 中不存在于 data 中的多余文件。
func unchangedDataList(passwd string) (unchangedList map[string]bool, removeList map[string]bool, err error) {
// calcUnchangedDataList 计算 sync 和 data 一致(没有修改过)的文件列表 unchangedDataList,并删除 sync 中不存在于 data 中的多余文件 removedSyncList
func calcUnchangedDataList(passwd string) (unchangedDataList map[string]bool, removedSyncList map[string]bool, err error) {
syncDir := Conf.Sync.GetSaveDir()
meta := filepath.Join(syncDir, pathJSON)
if !gulu.File.IsExist(meta) {
@ -1015,8 +1061,8 @@ func unchangedDataList(passwd string) (unchangedList map[string]bool, removeList
return
}
unchangedList = map[string]bool{}
removeList = map[string]bool{}
unchangedDataList = map[string]bool{}
removedSyncList = map[string]bool{}
filepath.Walk(syncDir, func(path string, info fs.FileInfo, _ error) error {
if syncDir == path || pathJSON == info.Name() {
return nil
@ -1026,7 +1072,7 @@ func unchangedDataList(passwd string) (unchangedList map[string]bool, removeList
encryptedP = filepath.ToSlash(encryptedP)
decryptedP := metaJSON[encryptedP]
if "" == decryptedP { // 理论上不会发生
removeList[path] = true
removedSyncList[path] = true
if gulu.File.IsDir(path) {
return filepath.SkipDir
}
@ -1035,7 +1081,7 @@ func unchangedDataList(passwd string) (unchangedList map[string]bool, removeList
dataP := filepath.Join(util.DataDir, decryptedP)
dataP = filepath.FromSlash(dataP)
if !gulu.File.IsExist(dataP) { // data 已经删除的文件
removeList[path] = true
removedSyncList[path] = true
if gulu.File.IsDir(path) {
return filepath.SkipDir
}
@ -1045,22 +1091,28 @@ func unchangedDataList(passwd string) (unchangedList map[string]bool, removeList
stat, _ := os.Stat(dataP)
dataModTime := stat.ModTime()
if info.ModTime() == dataModTime {
unchangedList[dataP] = true
unchangedDataList[dataP] = true
return nil
}
return nil
})
tmp := map[string]bool{}
// 在 sync 中删除 data 中已经删除的文件
for remove, _ := range removeList {
for remove, _ := range removedSyncList {
if strings.HasSuffix(remove, "index.json") {
continue
}
p := strings.TrimPrefix(remove, syncDir)
p = filepath.ToSlash(p)
tmp[p] = true
if err = os.RemoveAll(remove); nil != err {
util.LogErrorf("remove [%s] failed: %s", remove, err)
}
}
removedSyncList = tmp
return
}
@ -1291,7 +1343,7 @@ func getSyncIgnoreList() (ret *hashset.Set) {
return
}
func pathSha246(p, sep string) string {
func pathSha256Short(p, sep string) string {
buf := bytes.Buffer{}
parts := strings.Split(p, sep)
for i, part := range parts {