Refactored PDF & ZIP export features for images. Add reCAPTCHA or Shib... (!102) · Merge requests · ddr / ddr-public

Refactored PDF & ZIP export features for images. Add reCAPTCHA or Shib verification to thwart bots. Use derived_image JPGs from filesystem as source instead of image server. Replace rubyzip with zip_tricks for streaming. Replace prawn with hexapdf CLI for better RAM usage. Closes DDR-2252.

This is a complete refactor of the PDF & ZIP export features for DDR image items in order to accommodate exporting 100+ page items. Summary of changes:

For either ZIP or PDF

use the new derived_image JPG derivatives stored on the filesystem as the source files instead of hitting the image server w/http requests
removes deprecated code for exports relying on the image server (e.g., absolute_urls_for_img_component_jpgs)
adds reCAPTCHA or Shib verification to prevent bots from triggering exports
adds an animated loading state to the link that indicates to a user when the export is processing
moves logic out of the CatalogController and into a new ExportFilesController
updates ddr-core to 1.6.5 in order to use SolrDocument.derived_image_file_path
adds SolrDocument.derived_image_file_paths -- an array of component JPG paths for an item

ZIP exports

replaces RubyZIP gem with ZipTricks to enable streaming the ZIP file as it builds

PDF exports

replaces Prawn with HexaPDF gem to create PDFs; uses HexaPDF's command-line utility image2pdf
this HexaPDF approach consumes only 1/3 of the system memory as Prawn did during a PDF export
now writes the PDF to a Tempfile (and removes it afterward)
no longer downscales the PDF to 1000px (this was reducing the page size but not the file size); PDF pages are now the same pixel dimensions as the source image
removes fastimage gem
note that the prawn gem has not been removed since it is presently still used for exporting AV caption files (WebVTT) as PDF

Requirements

Update ddr-admin to at least 1.13.4 to capture SolrDocument.derived_image_file_path during indexing.
Ensure image components have derived_image JPG files; run rake ddr:derived_images to retroactively create them.
Reindex image components

Add these two environment variables; see https://duke.app.box.com/notes/830416801166 for more info: