Check checksum against access primary before updating

app/models/child_object.rb

@@ -105,13 +105,13 @@ def checksum_matches?
       sha512_checksum == access_sha512_checksum
     # goobi
     elsif checksum.present?
-      checksum == Digest::SHA1.file(access_primary_path).to_s
+      checksum == access_sha1_checksum
     # ladybird
     elsif sha256_checksum.present?
-      sha256_checksum == Digest::SHA256.file(access_primary_path).to_s
+      sha256_checksum == access_sha256_checksum
     # ladybird
     elsif md5_checksum.present?
-      md5_checksum == Digest::MD5.file(access_primary_path).to_s
+      md5_checksum == access_md5_checksum
     else
       false
     end
@@ -130,6 +130,18 @@ def access_sha512_checksum
     Digest::SHA512.file(access_primary_path).to_s
   end
 
+  def access_sha256_checksum
+    Digest::SHA256.file(access_primary_path).to_s
+  end
+
+  def access_sha1_checksum
+    Digest::SHA1.file(access_primary_path).to_s
+  end
+
+  def access_md5_checksum
+    Digest::MD5.file(access_primary_path).to_s
+  end
+
   def access_file_size
     File.exist?(access_primary_path) ? File&.size(access_primary_path) : nil
   end

app/models/concerns/updatable.rb

@@ -53,8 +53,6 @@ def update_child_objects_caption
       attach_item(parent_object)
       add_admin_set_to_bp(sets, parent_object)
       save!
-      child_object.caption = row['caption'] unless row['caption'].nil?
-      child_object.label = row['label'] unless row['label'].nil?
       processed_fields = validate_child_field(child_object, row)
       child_object.update!(processed_fields)
       processing_event_for_child(child_object)
@@ -166,12 +164,16 @@ def remove_child_blanks(row, child_object)
 
   # CHECKS CHILD OBJECT ATTRIBUTES
   def validate_child_field(child_object, row)
-    fields = ['caption', 'label', 'sha512_checksum', 'sha256_checksum', 'checksum', 'md5_checksum']
+    fields = ['caption', 'label']
+    checksum_fields = { 'sha512' => 'sha512_checksum', 'sha256' => 'sha256_checksum', 'sha1' => 'checksum', 'md5' => 'md5_checksum' }
     row, blanks = remove_child_blanks(row, child_object)
     processed_fields = {}
     fields.each do |f|
       processed_fields[f.to_sym] = valid_regular_child_fields(row, f, child_object)
     end
+    checksum_fields.each do |k, v|
+      processed_fields[v.to_sym] = valid_checksum_child_fields(row, k, v, child_object)
+    end
     processed_fields.merge!(blanks)
     processed_fields
   end
@@ -235,22 +237,13 @@ def update_child_objects_checksum
       attach_item(parent_object)
       add_admin_set_to_bp(sets, parent_object)
       save!
-      child_object.sha512_checksum = row['sha512'] unless row['sha512'].nil?
-      child_object.sha256_checksum = row['sha256'] unless row['sha256'].nil?
-      child_object.checksum = row['sha1'] unless row['sha1'].nil?
-      child_object.md5_checksum = row['md5'] unless row['md5'].nil?
       processed_fields = validate_child_field(child_object, row)
       child_object.update!(processed_fields)
       processing_event_for_child(child_object)
       processed_child_objects_count += 1
     end
-    unique_po = po_arr.uniq(&:oid)
-    unique_po.each do |parent_object|
-      trigger_setup_metadata(parent_object)
-      processing_event_for_parent(parent_object)
-    end
     batch_processing_event("#{processed_child_objects_count} child objects updated.", "Complete")
-    batch_processing_event("Child objects that were not updated: [#{child_objects_not_updated}].", "Complete")
+    batch_processing_event("Child objects that were not updated: #{child_objects_not_updated}.", "Complete")
   end
   # rubocop:enable Metrics/AbcSize
   # rubocop:enable Metrics/CyclomaticComplexity
@@ -288,7 +281,7 @@ def remote_po_path(oid, metadata_source)
     "#{metadata_source}/#{oid}.json"
   end
 
-  # CHECKS PARENT OBJECT PERMITTED ATTRIBUTES
+  # CHECKS PARENT OBJECT REGULAR FIELDS
   def valid_regular_fields(row, field_value, parent_object)
     if row[field_value].present? && row[field_value] != parent_object.send(field_value)
       row[field_value]
@@ -297,7 +290,7 @@ def valid_regular_fields(row, field_value, parent_object)
     end
   end
 
-  # CHECKS CHILD OBJECT PERMITTED ATTRIBUTES
+  # CHECKS CHILD OBJECT REGULAR FIELDS
   def valid_regular_child_fields(row, field_value, child_object)
     if row[field_value].present? && row[field_value] != child_object.send(field_value)
       row[field_value]
@@ -306,6 +299,18 @@ def valid_regular_child_fields(row, field_value, child_object)
     end
   end
 
+  # CHECKS CHILD OBJECT CHECKSUM FIELDS
+  def valid_checksum_child_fields(row, column_name, attribute, child_object)
+    if row[column_name].present? && row[column_name] != child_object.send(attribute) && (child_object.send("access_#{attribute}") == row[column_name])
+      row[column_name]
+    elsif row[column_name].present? && row[column_name] != child_object.send(attribute) && (child_object.send("access_#{attribute}") != row[column_name])
+      process_invalid_checksum_event(column_name, child_object)
+      child_object.send("access_#{attribute}")
+    else
+      child_object.send(attribute)
+    end
+  end
+
   # CHECKS CONTROLLED VOCABULARY
   def valid_controlled_vocab_fields(row, column_name, vocab, parent_object)
     if row[column_name].present? && row[column_name] != parent_object.send(column_name) && (ParentObject.send(vocab).include? row[column_name])
@@ -334,6 +339,13 @@ def process_invalid_vocab_event(column_name, row_value, oid)
   end
   # rubocop:enable Layout/LineLength
 
+  def process_invalid_checksum_event(column_name, child_object)
+    child_object.current_batch_process = self
+    child_object.current_batch_connection = batch_connections.find_or_create_by(connectable: child_object)
+    child_object.current_batch_connection.save!
+    child_object.processing_event("Child #{child_object.oid} was updated with the #{column_name} checksum value, read from the access primary original image file.", 'update-complete')
+  end
+
   # KICKS OFF SETUP METADATA JOB AND ATTACHES BATCH PROCESS TO PARENT
   def trigger_setup_metadata(parent_object)
     parent_object.current_batch_process = self

spec/fixtures/csv/checksum_child_object_valid.csv

@@ -0,0 +1,4 @@
+oid,sha512,sha256,sha1,md5
+10736292,ddfcbb8f70ba901e979acbe0c5a716e2cb1784dae560ea396471a277caa4ce58f796adb6d64cbbdb3be3d2a2c436bc26c45b878e33a7f083c3b72272e01595b0
+67890,ddfcbb8f70ba901e979acbe0c5a716e2cb1784dae560ea396471a277caa4ce58f796adb6d64cbbdb3be3d2a2c436bc26c45b878e33a7f083c3b72272e01595b0
+12,ddfcbb8f70ba901e979acbe0c5a716e2cb1784dae560ea396471a277caa4ce58f796adb6d64cbbdb3be3d2a2c436bc26c45b878e33a7f083c3b72272e01595b0

spec/models/batch_process_update_child_objects_spec.rb

@@ -10,20 +10,25 @@
   let(:role) { FactoryBot.create(:role, name: editor) }
   let(:caption_label_csv_upload) { Rack::Test::UploadedFile.new(Rails.root.join(fixture_path, "csv", "update_child_object_caption.csv")) }
   let(:caption_label_csv_blank_value_upload) { Rack::Test::UploadedFile.new(Rails.root.join(fixture_path, "csv", "update_child_object_blank.csv")) }
-  let(:checksum_csv_upload) { Rack::Test::UploadedFile.new(Rails.root.join(fixture_path, "csv", "checksum_child_object.csv")) }
+  let(:checksum_csv_valid_upload) { Rack::Test::UploadedFile.new(Rails.root.join(fixture_path, "csv", "checksum_child_object_valid.csv")) }
+  let(:checksum_csv_invalid_upload) { Rack::Test::UploadedFile.new(Rails.root.join(fixture_path, "csv", "checksum_child_object_invalid.csv")) }
   let(:checksum_csv_blank_value_upload) { Rack::Test::UploadedFile.new(Rails.root.join(fixture_path, "csv", "checksum_child_object_blank.csv")) }
   let(:parent_object) { FactoryBot.create(:parent_object, oid: 2_002_826, admin_set_id: admin_set.id) }
   let(:initial_fixity_value) { FFaker::Number.number(digits: 20) }
+  let(:tif_sha512_fixity_value) { "ddfcbb8f70ba901e979acbe0c5a716e2cb1784dae560ea396471a277caa4ce58f796adb6d64cbbdb3be3d2a2c436bc26c45b878e33a7f083c3b72272e01595b0" }
   let(:child_object) { FactoryBot.create(:child_object, oid: 10_736_292, caption: "caption", label: "label", parent_object: parent_object, sha512_checksum: initial_fixity_value) }
   let(:child_object_2) { FactoryBot.create(:child_object, oid: 67_890, caption: "co2 caption", label: "co2 label", parent_object: parent_object, sha512_checksum: initial_fixity_value) }
   let(:child_object_3) { FactoryBot.create(:child_object, oid: 12, caption: "co3 caption", label: "co3 label", parent_object: parent_object, sha512_checksum: initial_fixity_value) }
 
   around do |example|
     perform_enqueued_jobs do
       original_path_ocr = ENV['OCR_DOWNLOAD_BUCKET']
+      access_host = ENV['ACCESS_PRIMARY_MOUNT']
       ENV['OCR_DOWNLOAD_BUCKET'] = "yul-dc-ocr-test"
+      ENV['ACCESS_PRIMARY_MOUNT'] = File.join("spec", "fixtures", "images", "access_primaries")
       example.run
       ENV['OCR_DOWNLOAD_BUCKET'] = original_path_ocr
+      ENV['ACCESS_PRIMARY_MOUNT'] = access_host
     end
   end
 
@@ -105,19 +110,61 @@
 
       with_versioning do
         it "can update child checksum based on csv import" do
+          # begin test with incorrect fixity values saved to the child object
           expect(child_object.sha512_checksum).to eq initial_fixity_value.to_s
           expect(child_object_2.sha512_checksum).to eq initial_fixity_value.to_s
           expect(child_object_3.sha512_checksum).to eq initial_fixity_value.to_s
-          checksum_batch_process.file = checksum_csv_upload
+          checksum_batch_process.file = checksum_csv_valid_upload
           checksum_batch_process.save
           updated_child_object = ChildObject.find(10_736_292)
           updated_child_object_two = ChildObject.find(67_890)
           updated_child_object_three = ChildObject.find(12)
-          expect(updated_child_object.sha512_checksum).to eq "6fe314934e4623e61084b7f590ddee5cb259db13d45901c96ac74e14a7c771164feaa3a4cdb087c0f5c1eb39d671f1040eb8c092cfd1743d07f24c081d1fcd75"
-          expect(updated_child_object_two.sha512_checksum).to eq "b04e233da2e3b76fcfe2928f73e58f61351fec489d112acba1616f4d809f83722d07fb6fe62bf21a6b1ebcc4097b64458b39ee1e35235e3604234c0b3d9840ca"
-          expect(updated_child_object_three.sha512_checksum).to eq "8ac52a77a818780d29fe390f9b69cebe5a64c06559161e2e5ba7b6f425e7cbf785b50172f5795d1fcd5c63fc99c46774eb6c470f8ca453063bf98f4470ce81b0"
+          # all tifs are the same image so matching checksums is expected
+          expect(updated_child_object.sha512_checksum).to eq tif_sha512_fixity_value
+          expect(updated_child_object_two.sha512_checksum).to eq tif_sha512_fixity_value
+          expect(updated_child_object_three.sha512_checksum).to eq tif_sha512_fixity_value
+          expect(checksum_batch_process.batch_ingest_events.count).to eq 2
+          expect(checksum_batch_process.batch_ingest_events.first.reason).to eq "3 child objects updated."
+          expect(checksum_batch_process.batch_ingest_events.last.reason).to eq "Child objects that were not updated: []."
+          expect(updated_child_object.events_for_batch_process(checksum_batch_process).count).to eq 1
+          expect(updated_child_object.events_for_batch_process(checksum_batch_process).first.reason).to eq "Child 10736292 has been updated"
+          expect(updated_child_object_two.events_for_batch_process(checksum_batch_process).count).to eq 1
+          expect(updated_child_object_two.events_for_batch_process(checksum_batch_process).first.reason).to eq "Child 67890 has been updated"
+          expect(updated_child_object_three.events_for_batch_process(checksum_batch_process).count).to eq 1
+          expect(updated_child_object_three.events_for_batch_process(checksum_batch_process).first.reason).to eq "Child 12 has been updated"
         end
 
+        # rubocop:disable Layout/LineLength
+        it "cannot update child checksum with incorrect values" do
+          # begin test with incorrect fixity values saved to the child object
+          expect(child_object.sha512_checksum).to eq initial_fixity_value.to_s
+          expect(child_object_2.sha512_checksum).to eq initial_fixity_value.to_s
+          expect(child_object_3.sha512_checksum).to eq initial_fixity_value.to_s
+          checksum_batch_process.file = checksum_csv_invalid_upload
+          checksum_batch_process.save
+          updated_child_object = ChildObject.find(10_736_292)
+          updated_child_object_two = ChildObject.find(67_890)
+          updated_child_object_three = ChildObject.find(12)
+          # ensure child object page has message to user on what succeeded or failed
+          # all tifs are the same image so matching checksums is expected
+          expect(updated_child_object.sha512_checksum).to eq tif_sha512_fixity_value
+          expect(updated_child_object_two.sha512_checksum).to eq tif_sha512_fixity_value
+          expect(updated_child_object_three.sha512_checksum).to eq tif_sha512_fixity_value
+          expect(checksum_batch_process.batch_ingest_events.count).to eq 2
+          expect(checksum_batch_process.batch_ingest_events.first.reason).to eq "3 child objects updated."
+          expect(checksum_batch_process.batch_ingest_events.last.reason).to eq "Child objects that were not updated: []."
+          expect(updated_child_object.events_for_batch_process(checksum_batch_process).count).to eq 2
+          expect(updated_child_object.events_for_batch_process(checksum_batch_process).first.reason).to eq "Child 10736292 was updated with the sha512 checksum value, read from the access primary original image file."
+          expect(updated_child_object.events_for_batch_process(checksum_batch_process).last.reason).to eq "Child 10736292 has been updated"
+          expect(updated_child_object_two.events_for_batch_process(checksum_batch_process).count).to eq 2
+          expect(updated_child_object_two.events_for_batch_process(checksum_batch_process).first.reason).to eq "Child 67890 was updated with the sha512 checksum value, read from the access primary original image file."
+          expect(updated_child_object_two.events_for_batch_process(checksum_batch_process).last.reason).to eq "Child 67890 has been updated"
+          expect(updated_child_object_three.events_for_batch_process(checksum_batch_process).count).to eq 2
+          expect(updated_child_object_three.events_for_batch_process(checksum_batch_process).first.reason).to eq "Child 12 was updated with the sha512 checksum value, read from the access primary original image file."
+          expect(updated_child_object_three.events_for_batch_process(checksum_batch_process).last.reason).to eq "Child 12 has been updated"
+        end
+        # rubocop:enable Layout/LineLength
+
         it "cannot update child checksum with _blank_ values" do
           expect(child_object_2.sha512_checksum).to eq initial_fixity_value.to_s
           checksum_batch_process.file = checksum_csv_blank_value_upload
@@ -132,7 +179,7 @@
       before do
         user.add_role(:viewer, admin_set)
         checksum_batch_process.user_id = user.id
-        checksum_batch_process.file = checksum_csv_upload
+        checksum_batch_process.file = checksum_csv_valid_upload
         checksum_batch_process.save
       end
 
@@ -141,7 +188,7 @@
           child_object = ChildObject.find(10_736_292)
           expect(child_object.sha512_checksum).to eq initial_fixity_value.to_s
           expect(checksum_batch_process.batch_ingest_events.first.reason).to eq "Skipping row [2] with child oid: 10736292, user does not have permission to update."
-          expect(checksum_batch_process.batch_ingest_events.last.reason).to eq "Child objects that were not updated: [[\"10736292\", \"67890\", \"12\"]]."
+          expect(checksum_batch_process.batch_ingest_events.last.reason).to eq "Child objects that were not updated: [\"10736292\", \"67890\", \"12\"]."
         end
       end
     end