gitea源码

git.go 5.9KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201
  1. // Copyright 2019 The Gitea Authors. All rights reserved.
  2. // SPDX-License-Identifier: MIT
  3. package code
  4. import (
  5. "context"
  6. "strconv"
  7. "strings"
  8. repo_model "code.gitea.io/gitea/models/repo"
  9. "code.gitea.io/gitea/modules/git"
  10. "code.gitea.io/gitea/modules/git/gitcmd"
  11. "code.gitea.io/gitea/modules/indexer/code/internal"
  12. "code.gitea.io/gitea/modules/log"
  13. "code.gitea.io/gitea/modules/setting"
  14. )
  15. func getDefaultBranchSha(ctx context.Context, repo *repo_model.Repository) (string, error) {
  16. stdout, _, err := gitcmd.NewCommand("show-ref", "-s").AddDynamicArguments(git.BranchPrefix+repo.DefaultBranch).RunStdString(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
  17. if err != nil {
  18. return "", err
  19. }
  20. return strings.TrimSpace(stdout), nil
  21. }
  22. // getRepoChanges returns changes to repo since last indexer update
  23. func getRepoChanges(ctx context.Context, repo *repo_model.Repository, revision string) (*internal.RepoChanges, error) {
  24. status, err := repo_model.GetIndexerStatus(ctx, repo, repo_model.RepoIndexerTypeCode)
  25. if err != nil {
  26. return nil, err
  27. }
  28. needGenesis := len(status.CommitSha) == 0
  29. if !needGenesis {
  30. hasAncestorCmd := gitcmd.NewCommand("merge-base").AddDynamicArguments(status.CommitSha, revision)
  31. stdout, _, _ := hasAncestorCmd.RunStdString(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
  32. needGenesis = len(stdout) == 0
  33. }
  34. if needGenesis {
  35. return genesisChanges(ctx, repo, revision)
  36. }
  37. return nonGenesisChanges(ctx, repo, revision)
  38. }
  39. func isIndexable(entry *git.TreeEntry) bool {
  40. if !entry.IsRegular() && !entry.IsExecutable() {
  41. return false
  42. }
  43. name := strings.ToLower(entry.Name())
  44. for _, g := range setting.Indexer.ExcludePatterns {
  45. if g.Match(name) {
  46. return false
  47. }
  48. }
  49. for _, g := range setting.Indexer.IncludePatterns {
  50. if g.Match(name) {
  51. return true
  52. }
  53. }
  54. return len(setting.Indexer.IncludePatterns) == 0
  55. }
  56. // parseGitLsTreeOutput parses the output of a `git ls-tree -r --full-name` command
  57. func parseGitLsTreeOutput(stdout []byte) ([]internal.FileUpdate, error) {
  58. entries, err := git.ParseTreeEntries(stdout)
  59. if err != nil {
  60. return nil, err
  61. }
  62. idxCount := 0
  63. updates := make([]internal.FileUpdate, len(entries))
  64. for _, entry := range entries {
  65. if isIndexable(entry) {
  66. updates[idxCount] = internal.FileUpdate{
  67. Filename: entry.Name(),
  68. BlobSha: entry.ID.String(),
  69. Size: entry.Size(),
  70. Sized: true,
  71. }
  72. idxCount++
  73. }
  74. }
  75. return updates[:idxCount], nil
  76. }
  77. // genesisChanges get changes to add repo to the indexer for the first time
  78. func genesisChanges(ctx context.Context, repo *repo_model.Repository, revision string) (*internal.RepoChanges, error) {
  79. var changes internal.RepoChanges
  80. stdout, _, runErr := gitcmd.NewCommand("ls-tree", "--full-tree", "-l", "-r").AddDynamicArguments(revision).RunStdBytes(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
  81. if runErr != nil {
  82. return nil, runErr
  83. }
  84. var err error
  85. changes.Updates, err = parseGitLsTreeOutput(stdout)
  86. return &changes, err
  87. }
  88. // nonGenesisChanges get changes since the previous indexer update
  89. func nonGenesisChanges(ctx context.Context, repo *repo_model.Repository, revision string) (*internal.RepoChanges, error) {
  90. diffCmd := gitcmd.NewCommand("diff", "--name-status").AddDynamicArguments(repo.CodeIndexerStatus.CommitSha, revision)
  91. stdout, _, runErr := diffCmd.RunStdString(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
  92. if runErr != nil {
  93. // previous commit sha may have been removed by a force push, so
  94. // try rebuilding from scratch
  95. log.Warn("git diff: %v", runErr)
  96. if err := (*globalIndexer.Load()).Delete(ctx, repo.ID); err != nil {
  97. return nil, err
  98. }
  99. return genesisChanges(ctx, repo, revision)
  100. }
  101. var changes internal.RepoChanges
  102. var err error
  103. updatedFilenames := make([]string, 0, 10)
  104. updateChanges := func() error {
  105. cmd := gitcmd.NewCommand("ls-tree", "--full-tree", "-l").AddDynamicArguments(revision).
  106. AddDashesAndList(updatedFilenames...)
  107. lsTreeStdout, _, err := cmd.RunStdBytes(ctx, &gitcmd.RunOpts{Dir: repo.RepoPath()})
  108. if err != nil {
  109. return err
  110. }
  111. updates, err1 := parseGitLsTreeOutput(lsTreeStdout)
  112. if err1 != nil {
  113. return err1
  114. }
  115. changes.Updates = append(changes.Updates, updates...)
  116. return nil
  117. }
  118. lines := strings.SplitSeq(stdout, "\n")
  119. for line := range lines {
  120. line = strings.TrimSpace(line)
  121. if len(line) == 0 {
  122. continue
  123. }
  124. fields := strings.Split(line, "\t")
  125. if len(fields) < 2 {
  126. log.Warn("Unparseable output for diff --name-status: `%s`)", line)
  127. continue
  128. }
  129. filename := fields[1]
  130. if len(filename) == 0 {
  131. continue
  132. } else if filename[0] == '"' {
  133. filename, err = strconv.Unquote(filename)
  134. if err != nil {
  135. return nil, err
  136. }
  137. }
  138. switch status := fields[0][0]; status {
  139. case 'M', 'A':
  140. updatedFilenames = append(updatedFilenames, filename)
  141. case 'D':
  142. changes.RemovedFilenames = append(changes.RemovedFilenames, filename)
  143. case 'R', 'C':
  144. if len(fields) < 3 {
  145. log.Warn("Unparseable output for diff --name-status: `%s`)", line)
  146. continue
  147. }
  148. dest := fields[2]
  149. if len(dest) == 0 {
  150. log.Warn("Unparseable output for diff --name-status: `%s`)", line)
  151. continue
  152. }
  153. if dest[0] == '"' {
  154. dest, err = strconv.Unquote(dest)
  155. if err != nil {
  156. return nil, err
  157. }
  158. }
  159. if status == 'R' {
  160. changes.RemovedFilenames = append(changes.RemovedFilenames, filename)
  161. }
  162. updatedFilenames = append(updatedFilenames, dest)
  163. default:
  164. log.Warn("Unrecognized status: %c (line=%s)", status, line)
  165. }
  166. // According to https://learn.microsoft.com/en-us/troubleshoot/windows-client/shell-experience/command-line-string-limitation#more-information
  167. // the command line length should less than 8191 characters, assume filepath is 256, then 8191/256 = 31, so we use 30
  168. if len(updatedFilenames) >= 30 {
  169. if err := updateChanges(); err != nil {
  170. return nil, err
  171. }
  172. updatedFilenames = updatedFilenames[0:0]
  173. }
  174. }
  175. if len(updatedFilenames) > 0 {
  176. if err := updateChanges(); err != nil {
  177. return nil, err
  178. }
  179. }
  180. return &changes, err
  181. }