IIIF image server in ruby
Clone or download
Latest commit 9dcd95f Mar 5, 2018



Gem Version Coverage Status Dependency Status

A Ruby IIIF image server as a rails engine


RIIIF depends on Imagemagick so you must install that first. On a mac using Homebrew you can follow these instructions:

ImageMagick (7.0.4) may be installed with a few options:

  • --with-ghostscript Compile with Ghostscript for Postscript/PDF support
  • --with-tiff Compile with libtiff support for TIFF files
  • --with-openjpeg Compile with openjpeg2 support for jpeg2000
brew install imagemagick --with-ghostscript --with-tiff --with-openjpeg

Install the gem

Add this line to your application's Gemfile:

gem 'riiif'

And then execute:

$ bundle

Or install it yourself as:

$ gem install riiif


Images on the servers file system.

By default Riiif is set to load images from the filesystem using the Riiif::FileSystemFileResolver. You can configure how this resolver finds the files by setting this property:

    Riiif::Image.file_resolver.base_path = '/opt/repository/images/'

When the Id passed in is "foo_image", then it will look for an image file using this glob:


Images retrieved over HTTP

It's preferable to use files on the filesystem, because this avoids the overhead of downloading the file. If this is unavoidable, Riiif can be configured to fetch files from the network. To enable this behavior, configure Riiif to use an alternative resolver:

      Riiif::Image.file_resolver = Riiif::HTTPFileResolver.new

Then we configure the resolver with a mechanism for mapping the provided id to a url:

      Riiif::Image.file_resolver.id_to_uri = lambda do |id|

If you need to use HTTP basic authentication you can enable it like this:

      Riiif::Image.file_resolver.basic_auth_credentials = ['username', 's0s3kr3t']

This file resolver caches the network files, so you will want to clear out the old files or the cache will expand until you run out of disk space. Using a script like this would be a good idea: https://github.com/pulibrary/loris/blob/607567b921404a15a2111fbd7123604f4fdec087/bin/loris-cache_clean.sh By default the cache is located in tmp/network_files. You can set the cache path like this: Riiif::Image.file_resolver.cache_path = '/var/cache'

Kakadu (for faster jp2 decoding)

To configure Riiif to use Kakadu set:

Riiif::Engine.config.kakadu_enabled = true

See benchmark for details


To use GraphicsMagick instead of ImageMagick

Riiif::ImagemagickCommandFactory.external_command = "gm convert"
Riiif::ImageMagickInfoExtractor.external_command  = "gm identify"

You will of course need to install GraphicsMagick on your system.


Add the routes to your application by inserting the following line into config/routes.rb

  mount Riiif::Engine => '/image-service', as: 'riiif'

Then you can make requests like this:

Route helpers

It is prefereable that you use the provided route helpers to build these URIs. Here's an example:

  image_tag(Riiif::Engine.routes.url_helpers.image_path(file_id, size: ',600'))

Using default images for missing and unauthorized requests

If there is a request for an id that doesn't exist, a 404 will be returned. You can optionally return an image with this 404 by setting this in your initializer:

Riiif.not_found_image = 'path/to/image.png'

If the request is unauthorized, a 401 will be returned, and a custom error image can also be configured.

Riiif.unauthorized_image = 'path/to/unauthorized_image.png'

You can do this to create a default Riiif::Image to use (useful for passing "missing" images to openseadragon_collection_viewer):

Riiif::Image.new('no_image', Riiif::File.new(Riiif.not_found_image))


The controller will call an authorization service with the controller context. This service must have a method can?(action, image) which returns a boolean. The default service is the RIIIF::NilAuthrorizationService which permits all requests.

In this example we've dissallowed all requests:

class NoService
  def initalize(controller)

  def can?(action, image)

Riiif::Image.authorization_service = NoService

Integration with Hydra/Fedora

Create an initializer like this in config/initializers/riiif_initializer.rb

# Tell RIIIF to get files via HTTP (not from the local disk)
Riiif::Image.file_resolver = Riiif::HTTPFileResolver.new

# This tells RIIIF how to resolve the identifier to a URI in Fedora
Riiif::Image.file_resolver.id_to_uri = lambda do |id|
  ActiveFedora::Base.id_to_uri(CGI.unescape(id)).tap do |url|
    logger.info "Riiif resolved #{id} to #{url}"

# In order to return the info.json endpoint, we have to have the full height and width of
# each image. If you are using hydra-file_characterization, you have the height & width
# cached in Solr. The following block directs the info_service to return those values:
Riiif::Image.info_service = lambda do |id, file|
  # id will look like a path to a pcdm:file
  # (e.g. rv042t299%2Ffiles%2F6d71677a-4f80-42f1-ae58-ed1063fd79c7)
  # but we just want the id for the FileSet it's attached to.

  # Capture everything before the first slash
  fs_id = id.sub(/\A([^\/]*)\/.*/, '\1')
  resp = ActiveFedora::SolrService.get("id:#{fs_id}")
  doc = resp['response']['docs'].first
  raise "Unable to find solr document with id:#{fs_id}" unless doc

  # You’ll want default values if you make thumbnails of PDFs or other
  # file types that `identify` won’t return dimensions for
    height: doc["height_is"] || 100,
    width: doc["width_is"] || 100,
    format: doc["mime_type_ssi"],

def logger

# Note that this is translated to an `expires` argument to the
# ActiveSupport::Cache::Store in use, by default the host application's
# Rails.cache. Some cache stores may not automatically purge expired content,
# such as the default FileStore.
# http://guides.rubyonrails.org/caching_with_rails.html#cache-stores
Riiif::Engine.config.cache_duration = 30.days

Special note for Passenger and Apache users

If you are running riiif in Passenger under Apache, you must set the following in your virtual host definition:

AllowEncodedSlashes NoDecode

You may also need to set the following in your virtual host definition, either at the top level, or within a Location block for a specific path. See the Passenger configuration reference for more info.

PassengerAllowEncodedSlashes on

An alternative approach to PassengerAllowEncodedSlashes on is to explicitly decode the url, like this:

require "uri"
fs_id = URI.decode(id).sub(/\A([^\/]*)\/.*/, '\1')

Running the tests

First, build the engine

rake engine_cart:generate

Run the tests

rake spec

For more information

see the IIIF spec: