Parallel S3 Thumbnails
This tool is opinionated. I use it for an application with user uploaded pictures and frequently changing UI. Space is cheap so it makes sense to generate thumbnail versions to improve speed.
Different thumbnail versions are often named with suffixes such as _large.jpg or _small.jpg. I think it's better to explicitly state the size and lossy quality like _100x100:80.jpg.
Take an io.Reader then queue and run thumbnail jobs in parallel.
// example image id
id := "9283lkajwfeawef0000"
// somehow get an io reader
image, err := os.Open(fmt.Sprintf("tmp/%s.jpg", id))
// create worker
worker := s3jpeg.NewWorker("s3-bucket-name")
// queue parallel thumbnail jobs
worker.Queue(fmt.Sprintf("images/%s/600x600:75.jpg", id))
worker.Queue(fmt.Sprintf("images/%s/160x160:85.jpg", id))
worker.Queue(fmt.Sprintf("images/%s/160x160:45.jpg", id))
worker.Queue(fmt.Sprintf("images/%s/50x50:75.jpg", id))
// generate images and store to s3
err = worker.Run(image)
if err != nil {
// handle error
}It seems to take 1 or 2 seconds. I wrote this in a cafe tho so hopefully it's faster.
There are a number of limitations because my use case is limited. Feel free to fork.
- It only works on image files with the
width x height : qualityformat. - It only works with JPEGs.
- It only works with AWS S3 storage.
- It only works with environment variables for key & secret.
Built on top of the excellent imaging and goamz libraries.
$ go get github.com/disintegration/imaging
$ go get github.com/mitchellh/goamz
$ go get github.com/aj0strow/s3jpeg
Next moves is the CLI interface.
License: MIT