tinifyd

This software is not yet tested. Do NOT use it until this notice is removed. You are warned!

tinifyd is a NodeJS process that continuously monitors a filesystem path for changes to PNG and JPEG images. It compresses images on the fly via the TinyPNG API.

TinyPNG is a fantastic image compression service, but its algorithm is lossy. Repeatedly compressing the same image will result in perceptible image degradation. tinifyd remembers the SHA-1 sum of every image that is has ever optimized, along with its path so that images are optimized only once.

tinifyd is useful if you are running a CMS whose users can't be bothered to optimize images before they are uploaded. It does not replace a proper build process that optimizes images, and never will. It might save you some bandwidth (i.e., money) however.

Features

• Low CPU load; images are compressed on TinyPNG's servers
• Images are optimized only once
• Smart image type detection via mmagic
• Unoptimized image originals are backed up

Requirements

• TinyPNG API key
• Docker (if you want to run the daemon inside a container)

How to run the daemon

Via Docker

• Clone this repository

  git clone -b 0.12.0 [email protected]:itconsultis/tinifyd.git
  

• Copy src/.env.example to src/.env.

• Edit src/.env as needed.

• Run the start script from the project root.

  ./start
  

Environment variables

TINIFYD_API_KEY

Your Tinify API key

TINIFYD_SOURCE_PATH

Absolute filesystem path to the images you want to optimize.

TINIFYD_TEMP_PATH (/tmp/tinifyd)

Directory that holds temporary image copies created during an optimization. This directory must be writable by the node process.

TINIFYD_BACKUP_PATH

Image originals inside TINIFYD_SOURCE_PATH are backed up to this directory. This directory must be writable by the node process.

TINIFYD_API_DUMMY

Whether to use a dummy tinify client (which doesn't optimize images at all)

TINIFYD_CONCURRENCY

Maximum number of concurrent optimization operations tinifyd is allowed to perform at any given moment.

TINIFYD_LOCK_TIMEOUT

Locks older than this TTL (milliseconds) are deleted by the janitor plugin.

TINIFYD_OPTIMIZER_FREQUENCY

This determins how often the optimizer plugin will perform a full scan of TINIFYD_SOURCE_PATH and process unoptimized images within.

TINIFYD_DB_DATA_VOLUME

MySQL data directory (Docker host volume)

How to run tests

npm test

Workflow

Image optimization

Single image optimization procedure

• Ensure the image is a JPG or PNG
• Acquire a lock on the image path
• Compute the current hash sum of the image
• Ensure the hash sum is not in the blob table (meaning it is optimized); escape if it is
• Back up the image original
• Optimize the image via tinify
• Compute the hash sum of the optimized image
• Replace (clobber) the image original with the optimized one
• Record the optimized hash sum in the blob table
• Release the lock on image path

Batch optimization procedure

• Iterate over JPG and PNG image blobs found in the source directory
• Execute single image optimization procedure on each image that is found

Filesystem change event handling

• Listen for file change events on the source directory
• Execute single image optimization procedure on each file that changes

Locks

Locking will be pessimistic. The janitor plugin will periodically remove stale locks.

Concurrency

Each single image optimization procedure is to be queued into one ore more asynchronous task buffers. Each async buffer will have a cap on the number of concurrent optimization operations. This cap is to be configurable.

Todos

• put both node and mysql processes inside one Docker container
• tests

License

MIT