ereadifier

Single-purpose rust server that postprocesses images downloaded by suwayomi, to make them more suitable for reading on ereaders.

It can work with any reader which supports WebP decoding, though it is developed primarily for Koreader.

It listens for any POST requests, receives an input image via in the image form field, optionally resizes it to fit within the configured dimensions, and then responds with the final image encoded as a WebP.

Configuration

ereadifier is configured strictly through environment variables, as it is intended to be deployed as a container. The following configuration options are available:

• EREADIFIER_DIMENSIONS
  • "" (empty string) - Do not resize. This is the default behaviour.
  • <width>x<height> - Resize to fit within a <width> by <height> area.
  • <preset> - Resize to a named preset size. A list of device presets is available here
• EREADIFIER_ENCODE
  • lossless - Always encode as a lossless WebP. This is the default behaviour.
  • lossy - Always encode as a lossy WebP. See below to configure the target quality.
  • smallest - Encode as both lossless and lossy WebP, and select whichever has a smaller file size. This is slower due to encoding twice, but can be useful with sources which have blank pages (typically inserted to ensure an even page count for double page reading), as a fully blank page compresses significantly better as a lossless WebP.
• EREADIFIER_LOSSY_QUALITY
  • <quality> - Set the target quality for lossy encodes to <quality>. Valid range is 0 through 100, with a default of 85.
• EREADIFIER_DOUBLE_PAGE
  • ignore - Do not attempt to detect double pages. All images will target the same dimensions.
  • wider - If the provided image has a width greater than its height, assume it's a double page, and use twice the target width. This does not split the page, as suwayomi only expects one image out, but it should produce images similar in scale to if the pages had been split.
  • much_wider - As with wider, except that the threshold for being detected as a double page is a width at least 50% larger than height (i.e., 3:2 aspect or wider). This is the default behaviour.
• EREADIFIER_LISTEN
  • <address>:<port> - Address and port to listen on. Address must be a valid IPv4 or IPv6 address (not a hostname), and IPv6 addresses must be wrapped in [ and ]. Default is 0.0.0.0:80.
• EREADIFIER_DEBUG
  • yes, true, or 1 (case insensitive) - Print debug information while processing images.
  • Any other value - Do not print debug information.

Usage

Warning

This is not designed to be exposed to the public internet. You should expose it only to your local machine, or ideally only to Suwayomi, as it will process arbitrarily large inputs and could thus put you ask risk of denial of service attacks.

I recommend deploying as a container using whatever method you prefer (command, Compose, quadlet) - however, as long as you have some way of running it with any environment variables you want, it should work just the same.

Once deployed, configure Suwayomi to use it for all image processing, via Settings → Downloads → Image download processing, as shown below (change URL to match the hostname of the container where you have it deployed):

Example: running directly with Docker

docker run -p 8080:80 --rm -d \
    -e EREADIFIER_DIMENSIONS=clara \
    -e EREADIFIER_ENCODE=smallest \
    ghcr.io/rjgraffham/ereadifier:latest

Example: running with Docker Compose

services:
  ereadifier:
    image: ghcr.io/rjgraffham/ereadifier:latest
    container_name: ereadifier
    ports:
      - "8080:80"
    environment:
      EREADIFIER_DIMENSIONS: clara
      EREADIFIER_ENCODE: smallest
    restart: unless-stopped

Example: running with a Podman quadlet

[Unit]
Description=Ereadifier postprocessing server

[Install]
WantedBy=default.target

[Service]
Restart=always

[Container]
ContainerName=ereadifier
Image=ghcr.io/rjgraffham/ereadifier:latest
Environment=EREADIFIER_DIMENSIONS=clara
Environment=EREADIFIER_ENCODE=smallest
PublishPort=8080:80

Why WebP?

As this is built primarily for my own personal use, I can rely on the assumption that all readers will support WebP, and generally WebP at a similar visual quality will be smaller than JPEG or PNG, and those space savings add up with a device like an eReader which only has a small amount of onboard storage.