diff.go 3.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135
  1. package hercules
  2. import (
  3. "bufio"
  4. "bytes"
  5. "errors"
  6. "unicode/utf8"
  7. "github.com/sergi/go-diff/diffmatchpatch"
  8. "gopkg.in/src-d/go-git.v4"
  9. "gopkg.in/src-d/go-git.v4/plumbing"
  10. "gopkg.in/src-d/go-git.v4/plumbing/object"
  11. "gopkg.in/src-d/go-git.v4/utils/merkletrie"
  12. )
  13. // FileDiff calculates the difference of files which were modified.
  14. type FileDiff struct {
  15. }
  16. const (
  17. DependencyFileDiff = "file_diff"
  18. )
  19. type FileDiffData struct {
  20. OldLinesOfCode int
  21. NewLinesOfCode int
  22. Diffs []diffmatchpatch.Diff
  23. }
  24. func (diff *FileDiff) Name() string {
  25. return "FileDiff"
  26. }
  27. func (diff *FileDiff) Provides() []string {
  28. arr := [...]string{DependencyFileDiff}
  29. return arr[:]
  30. }
  31. func (diff *FileDiff) Requires() []string {
  32. arr := [...]string{DependencyTreeChanges, DependencyBlobCache}
  33. return arr[:]
  34. }
  35. func (diff *FileDiff) ListConfigurationOptions() []ConfigurationOption {
  36. return []ConfigurationOption{}
  37. }
  38. func (diff *FileDiff) Configure(facts map[string]interface{}) {}
  39. func (diff *FileDiff) Initialize(repository *git.Repository) {}
  40. func (diff *FileDiff) Consume(deps map[string]interface{}) (map[string]interface{}, error) {
  41. result := map[string]FileDiffData{}
  42. cache := deps[DependencyBlobCache].(map[plumbing.Hash]*object.Blob)
  43. tree_diff := deps[DependencyTreeChanges].(object.Changes)
  44. for _, change := range tree_diff {
  45. action, err := change.Action()
  46. if err != nil {
  47. return nil, err
  48. }
  49. switch action {
  50. case merkletrie.Modify:
  51. blob_from := cache[change.From.TreeEntry.Hash]
  52. blob_to := cache[change.To.TreeEntry.Hash]
  53. // we are not validating UTF-8 here because for example
  54. // git/git 4f7770c87ce3c302e1639a7737a6d2531fe4b160 fetch-pack.c is invalid UTF-8
  55. str_from, err := BlobToString(blob_from)
  56. if err != nil {
  57. return nil, err
  58. }
  59. str_to, err := BlobToString(blob_to)
  60. if err != nil {
  61. return nil, err
  62. }
  63. dmp := diffmatchpatch.New()
  64. src, dst, _ := dmp.DiffLinesToRunes(str_from, str_to)
  65. diffs := dmp.DiffMainRunes(src, dst, false)
  66. result[change.To.Name] = FileDiffData{
  67. OldLinesOfCode: len(src),
  68. NewLinesOfCode: len(dst),
  69. Diffs: diffs,
  70. }
  71. default:
  72. continue
  73. }
  74. }
  75. return map[string]interface{}{DependencyFileDiff: result}, nil
  76. }
  77. func CountLines(file *object.Blob) (int, error) {
  78. if file == nil {
  79. return -1, errors.New("Blob is nil: probably not cached.")
  80. }
  81. reader, err := file.Reader()
  82. if err != nil {
  83. return -1, err
  84. }
  85. defer checkClose(reader)
  86. var scanner *bufio.Scanner
  87. buffer := make([]byte, bufio.MaxScanTokenSize)
  88. counter := 0
  89. for scanner == nil || scanner.Err() == bufio.ErrTooLong {
  90. if scanner != nil && !utf8.Valid(scanner.Bytes()) {
  91. return -1, errors.New("binary")
  92. }
  93. scanner = bufio.NewScanner(reader)
  94. scanner.Buffer(buffer, 0)
  95. for scanner.Scan() {
  96. if !utf8.Valid(scanner.Bytes()) {
  97. return -1, errors.New("binary")
  98. }
  99. counter++
  100. }
  101. }
  102. return counter, nil
  103. }
  104. func BlobToString(file *object.Blob) (string, error) {
  105. if file == nil {
  106. return "", errors.New("Blob is nil: probably not cached.")
  107. }
  108. reader, err := file.Reader()
  109. if err != nil {
  110. return "", err
  111. }
  112. defer checkClose(reader)
  113. buf := new(bytes.Buffer)
  114. buf.ReadFrom(reader)
  115. return buf.String(), nil
  116. }
  117. func init() {
  118. Registry.Register(&FileDiff{})
  119. }