file_source.go 3.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143
  1. package source
  2. import (
  3. "errors"
  4. "fmt"
  5. "github.com/emqx/kuiper/internal/conf"
  6. "github.com/emqx/kuiper/internal/pkg/filex"
  7. "github.com/emqx/kuiper/pkg/api"
  8. "github.com/emqx/kuiper/pkg/cast"
  9. "os"
  10. "path"
  11. "path/filepath"
  12. "time"
  13. )
  14. type FileType string
  15. const (
  16. JSON_TYPE FileType = "json"
  17. )
  18. var fileTypes = map[FileType]bool{
  19. JSON_TYPE: true,
  20. }
  21. type FileSourceConfig struct {
  22. FileType FileType `json:"fileType"`
  23. Path string `json:"path"`
  24. Interval int `json:"interval"`
  25. RetainSize int `json:"$retainSize"`
  26. }
  27. // The BATCH to load data from file at once
  28. type FileSource struct {
  29. file string
  30. config *FileSourceConfig
  31. }
  32. func (fs *FileSource) Close(ctx api.StreamContext) error {
  33. ctx.GetLogger().Infof("Close file source")
  34. // do nothing
  35. return nil
  36. }
  37. func (fs *FileSource) Configure(fileName string, props map[string]interface{}) error {
  38. cfg := &FileSourceConfig{}
  39. err := cast.MapToStruct(props, cfg)
  40. if err != nil {
  41. return fmt.Errorf("read properties %v fail with error: %v", props, err)
  42. }
  43. if cfg.FileType == "" {
  44. return errors.New("missing or invalid property fileType, must be 'json'")
  45. }
  46. if _, ok := fileTypes[cfg.FileType]; !ok {
  47. return fmt.Errorf("invalid property fileType: %s", cfg.FileType)
  48. }
  49. if cfg.Path == "" {
  50. return errors.New("missing property Path")
  51. }
  52. if fileName == "" {
  53. return errors.New("file name must be specified")
  54. }
  55. if !filepath.IsAbs(cfg.Path) {
  56. cfg.Path, err = conf.GetLoc(cfg.Path)
  57. if err != nil {
  58. return fmt.Errorf("invalid path %s", cfg.Path)
  59. }
  60. }
  61. fs.file = path.Join(cfg.Path, fileName)
  62. if fi, err := os.Stat(fs.file); err != nil {
  63. if os.IsNotExist(err) {
  64. return fmt.Errorf("file %s not exist", fs.file)
  65. } else if !fi.Mode().IsRegular() {
  66. return fmt.Errorf("file %s is not a regular file", fs.file)
  67. }
  68. }
  69. fs.config = cfg
  70. return nil
  71. }
  72. func (fs *FileSource) Open(ctx api.StreamContext, consumer chan<- api.SourceTuple, errCh chan<- error) {
  73. err := fs.Load(ctx, consumer)
  74. if err != nil {
  75. errCh <- err
  76. return
  77. }
  78. if fs.config.Interval > 0 {
  79. ticker := time.NewTicker(time.Millisecond * time.Duration(fs.config.Interval))
  80. logger := ctx.GetLogger()
  81. defer ticker.Stop()
  82. for {
  83. select {
  84. case <-ticker.C:
  85. logger.Debugf("Load file source again at %v", conf.GetNowInMilli())
  86. err := fs.Load(ctx, consumer)
  87. if err != nil {
  88. errCh <- err
  89. return
  90. }
  91. case <-ctx.Done():
  92. return
  93. }
  94. }
  95. }
  96. }
  97. func (fs *FileSource) Load(ctx api.StreamContext, consumer chan<- api.SourceTuple) error {
  98. switch fs.config.FileType {
  99. case JSON_TYPE:
  100. ctx.GetLogger().Debugf("Start to load from file %s", fs.file)
  101. resultMap := make([]map[string]interface{}, 0)
  102. err := filex.ReadJsonUnmarshal(fs.file, &resultMap)
  103. if err != nil {
  104. return fmt.Errorf("loaded %s, check error %s", fs.file, err)
  105. }
  106. ctx.GetLogger().Debug("Sending tuples")
  107. if fs.config.RetainSize > 0 && fs.config.RetainSize < len(resultMap) {
  108. resultMap = resultMap[(len(resultMap) - fs.config.RetainSize):]
  109. ctx.GetLogger().Debug("Sending tuples for retain size %d", fs.config.RetainSize)
  110. }
  111. for _, m := range resultMap {
  112. select {
  113. case consumer <- api.NewDefaultSourceTuple(m, nil):
  114. // do nothing
  115. case <-ctx.Done():
  116. return nil
  117. }
  118. }
  119. // Send EOF if retain size not set
  120. if fs.config.RetainSize == 0 {
  121. select {
  122. case consumer <- api.NewDefaultSourceTuple(nil, nil):
  123. // do nothing
  124. case <-ctx.Done():
  125. return nil
  126. }
  127. }
  128. ctx.GetLogger().Debug("All tuples sent")
  129. return nil
  130. }
  131. return fmt.Errorf("invalid file type %s", fs.config.FileType)
  132. }