globgrinder

This library will watch (poll every 10 seconds) for files matching a given glob pattern, give you a chance to do something with it, then move the file to a "processed" directory.

Once a file is found, it's path will be thrown on a supplied channel for you to synchronously process however you'd like. When you are done doing your thing, put a bool on the done channel to signal globgrinder to give you the next file.

The file processing, though, is asynchronous; meaning, you can run multiple globgrinders at once, even if their globs overlap. Only one globgrinder will process a file. This is accomplished by atomically moving the file to a processing directory created within your output directory.

The processing directories will be created as needed.

pattern := "../some_dir/*.txt"
outDir := "./processed"

gg, err := globgrinder.New(pattern, outDir)
if err != nil {
  log.Panic(err)
}

process := make(chan string)
done := make(chan bool)

go gg.Run(process, done)

for path := range process {

  // Do something with the file

  // The files are being processed, so they'll have the suffix *.grinder
  // If you just want the filename, you can use the .Path(path string) convenience function.

  log.Printf("Processing file: %v\n", gg.Path(path))

  done <- true
}

Notes

• Be careful to not have files in your chosen processed directory match your glob pattern. It will cause a recursion you really don't want.
• If you specify your processed directory as an empty string "", your files will just be deleted after processing.