Skip to content

Instantly share code, notes, and snippets.

@Fustrate
Created November 11, 2020 03:41
Show Gist options
  • Save Fustrate/cf1b3ce8b227e385287963c23edb8c72 to your computer and use it in GitHub Desktop.
Save Fustrate/cf1b3ce8b227e385287963c23edb8c72 to your computer and use it in GitHub Desktop.
Explicit caching for Prawn PDF images
# frozen_string_literal: true
module Prawn
# `cached_image` needs to know if it's in a cloned document or the original so that it can store
# the image object data properly.
class ClonedDocument < Document
attr_reader :root_document
def initialize(root_document, **options, &block)
@root_document = root_document
super(**options, &block)
end
protected
# Make sure the image cache used is the parent document's cache, not a temporary
# one that'll be lost after the group ends.
def cached_images
@root_document.cached_images
end
end
# This is mostly just a rewrite of `prawn-grouping` to fit my needs. The only necessary change is
# that it creates a ClonedDocument instead of a Document.
module Grouping
def group
# create a temporary document with current context and offset
cloned_pdf = create_box_clone
update_cloned_pdf cloned_pdf
yield cloned_pdf
start_new_page if cloned_pdf.page_count > 1
yield self
end
private
def create_box_clone
Prawn::ClonedDocument.new(
self,
page_size: state.page.size,
page_layout: state.page.layout
)
end
def update_cloned_pdf(cloned)
cloned.margin_box = @bounding_box.dup
cloned.text_formatter = @text_formatter.dup
# cloned.page.margins = page.margins.dup
cloned.font_families.merge! font_families
cloned.font font.family
cloned.font_size font_size
cloned.default_leading = default_leading
cloned.y = y
end
end
end
Prawn::Document.extensions << Prawn::Grouping
# frozen_string_literal: true
module Prawn
module Images
def cached_image(cache_key, **options)
Prawn.verify_options(%i[at position vposition height width scale fit], options)
unless cached_images.key?(cache_key)
raise ArgumentError, "Cached image #{cache_key} not found" unless block_given?
yield(->(file) { cached_images[cache_key] = cacheable_image_data(file) })
end
pdf_obj, info = cached_images[cache_key]
embed_image(pdf_obj, info, options)
info
end
# This is just `image` with all of the caching logic bypassed.
def uncached_image(file, **options)
Prawn.verify_options(%i[at position vposition height width scale fit], options)
pdf_obj, info = cacheable_image_data(file)
embed_image(pdf_obj, info, options)
info
end
protected
def cached_images
@cached_images ||= {}
end
# This doesn't *perform* caching, it's just data that *can be* cached. It's basically just the
# contents of `build_image_object`'s `else` branch.
def cacheable_image_data(file)
image_content = verify_and_read_image(file)
# Build the image object
info = Prawn.image_handler.find(image_content).new(image_content)
# Bump PDF version if the image requires it
renderer.min_version(info.min_pdf_version) if info.respond_to?(:min_pdf_version)
# Add the image to the PDF and register it in case we see it again. This must be given
# the original PDF, not a ClonedDocument, or the actual image data will be lost before
# it's actually used.
image_obj = info.build_pdf_object(respond_to?(:root_document) ? root_document : self)
[image_obj, info]
end
end
end
# frozen_string_literal: true
# Inside the prawn template:
# An IO/File/string can be used, just like with the original `image` method. This method will only
# read the file from disk once.
pdf.cached_image('vendor_logo', scale: 0.25) do |block|
block.call 'path/to/vendor_logo.png'
end
# Active Storage attachments need to be opened first and passed the yielded block
pdf.cached_image("user_avatar_#{user.id}", height: 50) do |block|
user.avatar.open(&block)
end
# This will never be cached, and will be read from disk every time.
pdf.uncached_image('some/args/as/image.png', width: 180)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment