diff.go 3.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131
  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. type FileDiffData struct {
  17. OldLinesOfCode int
  18. NewLinesOfCode int
  19. Diffs []diffmatchpatch.Diff
  20. }
  21. func (diff *FileDiff) Name() string {
  22. return "FileDiff"
  23. }
  24. func (diff *FileDiff) Provides() []string {
  25. arr := [...]string{"file_diff"}
  26. return arr[:]
  27. }
  28. func (diff *FileDiff) Requires() []string {
  29. arr := [...]string{"changes", "blob_cache"}
  30. return arr[:]
  31. }
  32. func (diff *FileDiff) ListConfigurationOptions() []ConfigurationOption {
  33. return []ConfigurationOption{}
  34. }
  35. func (diff *FileDiff) Configure(facts map[string]interface{}) {}
  36. func (diff *FileDiff) Initialize(repository *git.Repository) {}
  37. func (diff *FileDiff) Consume(deps map[string]interface{}) (map[string]interface{}, error) {
  38. result := map[string]FileDiffData{}
  39. cache := deps["blob_cache"].(map[plumbing.Hash]*object.Blob)
  40. tree_diff := deps["changes"].(object.Changes)
  41. for _, change := range tree_diff {
  42. action, err := change.Action()
  43. if err != nil {
  44. return nil, err
  45. }
  46. switch action {
  47. case merkletrie.Modify:
  48. blob_from := cache[change.From.TreeEntry.Hash]
  49. blob_to := cache[change.To.TreeEntry.Hash]
  50. // we are not validating UTF-8 here because for example
  51. // git/git 4f7770c87ce3c302e1639a7737a6d2531fe4b160 fetch-pack.c is invalid UTF-8
  52. str_from, err := BlobToString(blob_from)
  53. if err != nil {
  54. return nil, err
  55. }
  56. str_to, err := BlobToString(blob_to)
  57. if err != nil {
  58. return nil, err
  59. }
  60. dmp := diffmatchpatch.New()
  61. src, dst, _ := dmp.DiffLinesToRunes(str_from, str_to)
  62. diffs := dmp.DiffMainRunes(src, dst, false)
  63. result[change.To.Name] = FileDiffData{
  64. OldLinesOfCode: len(src),
  65. NewLinesOfCode: len(dst),
  66. Diffs: diffs,
  67. }
  68. default:
  69. continue
  70. }
  71. }
  72. return map[string]interface{}{"file_diff": result}, nil
  73. }
  74. func CountLines(file *object.Blob) (int, error) {
  75. if file == nil {
  76. return -1, errors.New("Blob is nil: probably not cached.")
  77. }
  78. reader, err := file.Reader()
  79. if err != nil {
  80. return -1, err
  81. }
  82. defer checkClose(reader)
  83. var scanner *bufio.Scanner
  84. buffer := make([]byte, bufio.MaxScanTokenSize)
  85. counter := 0
  86. for scanner == nil || scanner.Err() == bufio.ErrTooLong {
  87. if scanner != nil && !utf8.Valid(scanner.Bytes()) {
  88. return -1, errors.New("binary")
  89. }
  90. scanner = bufio.NewScanner(reader)
  91. scanner.Buffer(buffer, 0)
  92. for scanner.Scan() {
  93. if !utf8.Valid(scanner.Bytes()) {
  94. return -1, errors.New("binary")
  95. }
  96. counter++
  97. }
  98. }
  99. return counter, nil
  100. }
  101. func BlobToString(file *object.Blob) (string, error) {
  102. if file == nil {
  103. return "", errors.New("Blob is nil: probably not cached.")
  104. }
  105. reader, err := file.Reader()
  106. if err != nil {
  107. return "", err
  108. }
  109. defer checkClose(reader)
  110. buf := new(bytes.Buffer)
  111. buf.ReadFrom(reader)
  112. return buf.String(), nil
  113. }
  114. func init() {
  115. Registry.Register(&FileDiff{})
  116. }