leaks.go 3.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133
  1. package main
  2. import (
  3. "bytes"
  4. "encoding/json"
  5. "fmt"
  6. "io/ioutil"
  7. "log"
  8. "os"
  9. "os/exec"
  10. "os/signal"
  11. "strings"
  12. "sync"
  13. "syscall"
  14. )
  15. type LeakElem struct {
  16. Line string `json:"line"`
  17. Commit string `json:"commit"`
  18. }
  19. func start(_ *Options, repoURL string) {
  20. c := make(chan os.Signal, 2)
  21. signal.Notify(c, os.Interrupt, syscall.SIGTERM)
  22. err := exec.Command("git", "clone", repoURL).Run()
  23. if err != nil {
  24. log.Fatalf("failed to clone repo %v", err)
  25. }
  26. repoName := getLocalRepoName(repoURL)
  27. if err = os.Chdir(repoName); err != nil {
  28. log.Fatal(err)
  29. }
  30. go func() {
  31. <-c
  32. cleanup(repoName)
  33. os.Exit(1)
  34. }()
  35. report := getLeaks(repoName)
  36. cleanup(repoName)
  37. reportJSON, _ := json.MarshalIndent(report, "", "\t")
  38. err = ioutil.WriteFile(fmt.Sprintf("%s_leaks.json", repoName), reportJSON, 0644)
  39. if err != nil {
  40. log.Fatalf("Can't write to file: %s", err)
  41. }
  42. }
  43. // getLocalRepoName generates the name of the local clone folder based on the given URL
  44. func getLocalRepoName(url string) string {
  45. splitSlashes := strings.Split(url, "/")
  46. name := splitSlashes[len(splitSlashes)-1]
  47. name = strings.TrimSuffix(name, ".git")
  48. splitColons := strings.Split(name, ":")
  49. name = splitColons[len(splitColons)-1]
  50. return name
  51. }
  52. func cleanup(repoName string) {
  53. if err := os.Chdir(appRoot); err != nil {
  54. log.Fatalf("failed cleaning up repo. Does the repo exist? %v", err)
  55. }
  56. err := exec.Command("rm", "-rf", repoName).Run()
  57. if err != nil {
  58. log.Fatal(err)
  59. }
  60. }
  61. func getLeaks(repoName string) []LeakElem {
  62. var (
  63. out []byte
  64. err error
  65. commitWG sync.WaitGroup
  66. gitLeakReceiverWG sync.WaitGroup
  67. concurrent = 100
  68. semaphoreChan = make(chan struct{}, concurrent)
  69. gitLeaks = make(chan LeakElem)
  70. report []LeakElem
  71. )
  72. go func(commitWG *sync.WaitGroup, gitLeakReceiverWG *sync.WaitGroup) {
  73. for gitLeak := range gitLeaks {
  74. fmt.Println(gitLeak)
  75. report = append(report, gitLeak)
  76. gitLeakReceiverWG.Done()
  77. }
  78. }(&commitWG, &gitLeakReceiverWG)
  79. out, err = exec.Command("git", "rev-list", "--all", "--remotes", "--topo-order").Output()
  80. if err != nil {
  81. log.Fatalf("error retrieving commits%v\n", err)
  82. }
  83. commits := bytes.Split(out, []byte("\n"))
  84. commitWG.Add(len(commits))
  85. for _, currCommitB := range commits {
  86. currCommit := string(currCommitB)
  87. go func(currCommit string, repoName string, commitWG *sync.WaitGroup, gitLeakReceiverWG *sync.WaitGroup) {
  88. defer commitWG.Done()
  89. var leakPrs bool
  90. if err := os.Chdir(fmt.Sprintf("%s/%s", appRoot, repoName)); err != nil {
  91. log.Fatal(err)
  92. }
  93. commitCmp := fmt.Sprintf("%s^!", currCommit)
  94. semaphoreChan <- struct{}{}
  95. out, err := exec.Command("git", "diff", commitCmp).Output()
  96. <-semaphoreChan
  97. if err != nil {
  98. return
  99. }
  100. lines := checkRegex(string(out))
  101. if len(lines) == 0 {
  102. return
  103. }
  104. for _, line := range lines {
  105. leakPrs = checkEntropy(line)
  106. if leakPrs {
  107. gitLeakReceiverWG.Add(1)
  108. gitLeaks <- LeakElem{line, currCommit}
  109. }
  110. }
  111. }(currCommit, repoName, &commitWG, &gitLeakReceiverWG)
  112. }
  113. commitWG.Wait()
  114. gitLeakReceiverWG.Wait()
  115. return report
  116. }