Handling CarrierWave/BrowseEverything controller logic

Prior to this commit, the Hyrax::WoksControllerBehavior's Valkyrie
branching logic ignored files selected via BrowseEverything.  With
this change the controller normalizes(-ish) the two methods of file
upload:

1) CarrierWave
2) BrowseEverything

An immediate follow-up step is to update
`Hyrax::UploadedFile.perform_ingest_later`.  In it's present
implementation, that method assumes a CarrierWave object.  Adding a
switching case for UploadedFiles that have a
`url_of_remote_file_for_ingest` (e.g., a file selected via
BrowseEverything) instead of the assumptive `file` attribute (e.g., a
file uploaded via CarrierWave).

Relates to #4195

Author: jeremyf

.dassie/db/migrate/20210716130305_add_url_to_hyrax_uploaded_files.hyrax.rb

@@ -0,0 +1,5 @@
+class AddUrlToHyraxUploadedFiles < ActiveRecord::Migration[5.2]
+  def change
+    add_column :uploaded_files, :url_of_remote_file_for_ingest, :text
+  end
+end

.dassie/db/schema.rb

@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema.define(version: 2020_08_21_212903) do
+ActiveRecord::Schema.define(version: 2021_07_16_130305) do
 
   # These are extensions that must be enabled in order to support this database
   enable_extension "plpgsql"
@@ -480,6 +480,7 @@
     t.string "file_set_uri"
     t.datetime "created_at", null: false
     t.datetime "updated_at", null: false
+    t.text "url_of_remote_file_for_ingest"
     t.index ["file_set_uri"], name: "index_uploaded_files_on_file_set_uri"
     t.index ["user_id"], name: "index_uploaded_files_on_user_id"
   end

.regen

@@ -1 +1 @@
-36
+37

app/controllers/concerns/hyrax/works_controller_behavior.rb

@@ -202,6 +202,7 @@ def create_work
 
         @curation_concern =
           form.validate(params[hash_key_for_curation_concern]) &&
+          normalize_uploaded_files! &&
           transactions['change_set.create_work']
           .with_step_args('work_resource.add_to_parent' => { parent_id: params[:parent_id], user: current_user },
                           'work_resource.add_file_sets' => { uploaded_files: uploaded_files, file_set_params: params[hash_key_for_curation_concern][:file_set] },
@@ -219,6 +220,7 @@ def update_work
 
         @curation_concern =
           form.validate(params[hash_key_for_curation_concern]) &&
+          normalize_uploaded_files! &&
           transactions['change_set.update_work']
           .with_step_args('work_resource.add_file_sets' => { uploaded_files: uploaded_files, file_set_params: params[hash_key_for_curation_concern][:file_set] })
           .call(form).value!
@@ -369,6 +371,44 @@ def attributes_for_actor # rubocop:disable Metrics/MethodLength
       attributes
     end
 
+    ##
+    # There are two means of "uploading" files:
+    #
+    # 1. via CarrierWave (e.g., upload in the HTML form)
+    # 2. via BrowseEverything (e.g., select files and later upload
+    #    those URLs)
+    #
+    # This method seeks to somewhat normalize the uploaded files by
+    # creating Hyrax::UploadedFile entries for the BrowseEverthing
+    # files.  Doing this minimizes branching logic up until the
+    # application stack is ready to attach the binary to the
+    # corresponding Hyrax::FileSet.
+    #
+    # @note at this point, this is a Valkyrie only path.  Perhaps a
+    # later refactor will go down the ActiveFedora path.
+    def normalize_uploaded_files!
+      # If they selected a BrowseEverything file, but then clicked the
+      # remove button, it will still show up in `selected_files`, but
+      # it will no longer be in uploaded_files. By checking the
+      # intersection, we get the files they added via BrowseEverything
+      # that they have not removed from the upload widget.
+      uploaded_files = params.fetch(:uploaded_files, [])
+      params.fetch(:selected_files, {}).each_pair do |_, value|
+        next unless uploaded_files.include?(value[:url])
+
+        # This logic mimics the behavior of CarrierWave, namely creating an ActiveRecord record for later
+        # ingest.  In this case, we're swapping out the given URL for the record ID.  Downstream, we'll
+        # act on the URL.
+        uploaded_file = UploadedFile.create!(user: current_user, url_of_remote_file_for_ingest: value[:url])
+        uploaded_files.delete(value[:url])
+        uploaded_files << uploaded_file.to_param
+      end
+      # Need to ensure that we swap out the URLs for the UploadedFile's we created above.  The uploaded_files local
+      # variable is a copy of the params[:uploaded_file] (e.g., they don't have the same object_id)
+      params[:uploaded_files] = uploaded_files unless uploaded_files.empty?
+      true
+    end
+
     def after_create_response
       respond_to do |wants|
         wants.html do

lib/generators/hyrax/templates/db/migrate/20210716002029_add_url_to_hyrax_uploaded_files.rb.erb

@@ -0,0 +1,5 @@
+class AddUrlToHyraxUploadedFiles < ActiveRecord::Migration<%= migration_version %>
+  def change
+    add_column :uploaded_files, :url_of_remote_file_for_ingest, :text
+  end
+end

spec/controllers/concerns/hyrax/works_controller_behavior_spec.rb

@@ -124,18 +124,22 @@
 
       context 'and files' do
         let(:uploads) { FactoryBot.create_list(:uploaded_file, 2, user: user) }
+        let(:url_to_file) { "https://github.com/samvera/hyrax/browse-everything-url.txt" }
+        let(:selected_files) { { "1" => { url: url_to_file } } }
 
         it 'attaches the files' do
           params = { test_simple_work: { title: 'comet in moominland' },
-                     uploaded_files: uploads.map(&:id) }
+                     # Note this is a mix of BrowseEverything files and those "uploaded" via CarrierWave.
+                     uploaded_files: uploads.map(&:id) + [url_to_file],
+                     selected_files: selected_files }
 
           get :create, params: params
 
           expect(flash[:notice]).to be_html_safe
           expect(flash[:notice]).to eq "Your files are being processed by Hyrax in the background. " \
                                        "The metadata and access controls you specified are being applied. " \
                                        "You may need to refresh this page to see these updates."
-          expect(assigns(:curation_concern)).to have_file_set_members(be_persisted, be_persisted)
+          expect(assigns(:curation_concern)).to have_file_set_members(be_persisted, be_persisted, be_persisted)
         end
 
         let(:uploads) { FactoryBot.create_list(:uploaded_file, 2, user: user) }