You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

execute_diff.go 5.6 kB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240
  1. package spacesyncer
  2. import (
  3. "context"
  4. "io"
  5. "time"
  6. "gitlink.org.cn/cloudream/common/pkgs/ioswitch/exec"
  7. "gitlink.org.cn/cloudream/common/pkgs/logger"
  8. "gitlink.org.cn/cloudream/common/pkgs/trie"
  9. "gitlink.org.cn/cloudream/common/utils/math2"
  10. clitypes "gitlink.org.cn/cloudream/jcs-pub/client/types"
  11. "gitlink.org.cn/cloudream/jcs-pub/common/pkgs/ioswitch2"
  12. "gitlink.org.cn/cloudream/jcs-pub/common/pkgs/ioswitch2/parser"
  13. stgtypes "gitlink.org.cn/cloudream/jcs-pub/common/pkgs/storage/types"
  14. )
  15. func executeDiff(syncer *SpaceSyncer, task *task, mode *clitypes.SpaceSyncModeDiff) {
  16. log := logger.WithField("Mod", logMod).WithField("TaskID", task.Task.TaskID)
  17. startTime := time.Now()
  18. log.Infof("begin full sync task")
  19. defer func() {
  20. log.Infof("full sync task finished, time: %v", time.Since(startTime))
  21. }()
  22. srcSpace := syncer.spaceMeta.Get(task.Task.SrcUserSpaceID)
  23. if srcSpace == nil {
  24. log.Warnf("src space %v not found", task.Task.SrcUserSpaceID)
  25. return
  26. }
  27. if len(task.Task.Dests) > 1 {
  28. log.Warnf("diff mode only support one dest now")
  29. }
  30. dstSpace := syncer.spaceMeta.Get(task.Task.Dests[0].DestUserSpaceID)
  31. if dstSpace == nil {
  32. log.Warnf("dest space %v not found", task.Task.Dests[0].DestUserSpaceID)
  33. return
  34. }
  35. srcBase, err := syncer.stgPool.GetBaseStore(srcSpace)
  36. if err != nil {
  37. log.Warnf("get src base store error: %v", err)
  38. return
  39. }
  40. dstBase, err := syncer.stgPool.GetBaseStore(dstSpace)
  41. if err != nil {
  42. log.Warnf("get dst base store error: %v", err)
  43. return
  44. }
  45. filter := buildFilter(task)
  46. srcReader := srcBase.ReadDir(task.Task.SrcPath)
  47. dstReader := dstBase.ReadDir(task.Task.Dests[0].DestPath)
  48. dirTree := trie.NewTrie[srcDstDirEntry]()
  49. for {
  50. e, err := srcReader.Next()
  51. if err == io.EOF {
  52. break
  53. }
  54. if err != nil {
  55. log.Warnf("read src dir: %v", err)
  56. return
  57. }
  58. if !filter(e) {
  59. continue
  60. }
  61. rela := e.Path.Clone()
  62. rela.DropFrontN(task.Task.SrcPath.Len())
  63. ne := e
  64. ne.Path = rela.Clone()
  65. if !filter(ne) {
  66. continue
  67. }
  68. diffCreateSrcNode(dirTree, rela, &e)
  69. }
  70. for {
  71. e, err := dstReader.Next()
  72. if err == io.EOF {
  73. break
  74. }
  75. if err != nil {
  76. log.Warnf("read dst dir: %v", err)
  77. return
  78. }
  79. if !filter(e) {
  80. continue
  81. }
  82. rela := e.Path.Clone()
  83. rela.DropFrontN(task.Task.Dests[0].DestPath.Len())
  84. ne := e
  85. ne.Path = rela.Clone()
  86. if !filter(ne) {
  87. continue
  88. }
  89. diffCreateDstNode(dirTree, rela, &e)
  90. }
  91. var willSync []stgtypes.DirEntry
  92. var willMkdirs []clitypes.JPath
  93. dirTree.Iterate(func(path []string, node *trie.Node[srcDstDirEntry], isWordNode bool) trie.VisitCtrl {
  94. if node.Value.src == nil {
  95. // 目前不支持删除多余文件
  96. return trie.VisitContinue
  97. }
  98. if node.Value.src.IsDir {
  99. if node.Value.dst == nil {
  100. if node.IsEmpty() {
  101. willMkdirs = append(willMkdirs, clitypes.PathFromComps(path...))
  102. }
  103. }
  104. } else {
  105. if node.Value.dst == nil {
  106. // 目标路径不存在(不是文件也不是目录),需要同步
  107. if node.IsEmpty() {
  108. willSync = append(willSync, *node.Value.src)
  109. }
  110. } else if !node.Value.dst.IsDir {
  111. // 目标路径是个文件,但文件指纹不同,需要同步
  112. if !cmpFile(mode, node.Value.src, node.Value.dst) {
  113. willSync = append(willSync, *node.Value.src)
  114. }
  115. }
  116. // 目标路径是个目录,则不进行同步
  117. }
  118. return trie.VisitContinue
  119. })
  120. willSyncCnt := len(willSync)
  121. for len(willSync) > 0 {
  122. syncs := willSync[:math2.Min(len(willSync), 50)]
  123. willSync = willSync[len(syncs):]
  124. ft := ioswitch2.NewFromTo()
  125. for _, s := range syncs {
  126. ft.AddFrom(ioswitch2.NewFromBaseStore(*srcSpace, s.Path))
  127. rela := s.Path.Clone()
  128. rela.DropFrontN(task.Task.SrcPath.Len())
  129. dstPath := task.Task.Dests[0].DestPath.ConcatNew(rela)
  130. to := ioswitch2.NewToBaseStore(*dstSpace, dstPath)
  131. to.Option.ModTime = s.ModTime
  132. ft.AddTo(to)
  133. }
  134. planBld := exec.NewPlanBuilder()
  135. err := parser.Parse(ft, planBld)
  136. if err != nil {
  137. log.Warnf("parse fromto: %v", err)
  138. return
  139. }
  140. execCtx := exec.NewWithContext(task.Context)
  141. exec.SetValueByType(execCtx, syncer.stgPool)
  142. _, err = planBld.Execute(execCtx).Wait(context.Background())
  143. if err != nil {
  144. log.Warnf("execute plan: %v", err)
  145. return
  146. }
  147. }
  148. log.Infof("%v files synced", willSyncCnt)
  149. if !task.Task.Options.NoEmptyDirectories && len(willMkdirs) > 0 {
  150. for _, p := range willMkdirs {
  151. rela := p.Clone()
  152. rela.DropFrontN(task.Task.SrcPath.Len())
  153. dstPath := task.Task.Dests[0].DestPath.ConcatNew(rela)
  154. err := dstBase.Mkdir(dstPath)
  155. if err != nil {
  156. log.Warnf("mkdir: %v", err)
  157. continue
  158. }
  159. }
  160. }
  161. }
  162. func diffCreateSrcNode(tree *trie.Trie[srcDstDirEntry], path clitypes.JPath, e *stgtypes.DirEntry) {
  163. var ptr = &tree.Root
  164. for _, c := range path.Comps() {
  165. if ptr.Value.src != nil && ptr.Value.src.IsDir {
  166. ptr.Value.src = nil
  167. }
  168. ptr = ptr.Create(c)
  169. }
  170. ptr.Value.src = e
  171. }
  172. func diffCreateDstNode(tree *trie.Trie[srcDstDirEntry], path clitypes.JPath, e *stgtypes.DirEntry) {
  173. var ptr = &tree.Root
  174. for _, c := range path.Comps() {
  175. if ptr.Value.src != nil && ptr.Value.src.IsDir {
  176. ptr.Value.src = nil
  177. }
  178. if ptr.Value.dst != nil && ptr.Value.dst.IsDir {
  179. ptr.Value.dst = nil
  180. }
  181. ptr = ptr.Create(c)
  182. }
  183. ptr.Value.dst = e
  184. }
  185. type srcDstDirEntry struct {
  186. src *stgtypes.DirEntry
  187. dst *stgtypes.DirEntry
  188. }
  189. func cmpFile(diff *clitypes.SpaceSyncModeDiff, src, dst *stgtypes.DirEntry) bool {
  190. if diff.IncludeSize && src.Size != dst.Size {
  191. return false
  192. }
  193. if diff.IncludeModTime && src.ModTime != dst.ModTime {
  194. return false
  195. }
  196. return true
  197. }

本项目旨在将云际存储公共基础设施化,使个人及企业可低门槛使用高效的云际存储服务(安装开箱即用云际存储客户端即可,无需关注其他组件的部署),同时支持用户灵活便捷定制云际存储的功能细节。