Cal 949 catch tombstone exceptions on re ingest2 (#857)

* removed debug code

* fixed test that broke

* more fixes

* refactor to use exception and rescue when removing tombstones

* added Ldp::Gone for collections

* fixed collection tombstone detect and delete

* adjusted spacing

* expanded name, added comment for clarification of operation

Co-authored-by: darrowcoucla <[email protected]>

app/importers/actor_record_importer.rb

@@ -60,46 +60,47 @@ def update_for(existing_record:, update_record:)
   # We assume the object was created as expected if the actor stack returns true.
   def create_for(record:)
     info_stream << "event: record_import_started, row_id: #{@row_id}, ark: #{record.ark}\n"
-
     raise(ArgumentError, 'Title starts "DUPLICATE" – record will not be imported.') if record.mapper.title[0].to_s.start_with?('DUPLICATE')
-
     additional_attrs = {
       uploaded_files: create_upload_files(record),
       depositor: @depositor.user_key
     }
-
     object_type = record.mapper.metadata["Object Type"]
     created = import_type(object_type).new
     created.apply_depositor_metadata(@depositor.user_key)
     attrs = record.attributes.merge(additional_attrs)
-
     attrs = attrs.merge(member_of_collections_attributes: { '0' => { id: collection_id } }) if collection_id
 
     # Ensure nothing is passed in the files field,
     # since this is reserved for Hyrax and is where uploaded_files will be attached
     attrs.delete(:files)
     attrs.delete(:uploaded_files)
-
     based_near = attrs.delete(:based_near)
     attrs = attrs.merge(based_near_attributes: based_near_attributes(based_near)) unless based_near.nil? || based_near.empty?
-
     actor_env = Hyrax::Actors::Environment.new(created,
                                                ::Ability.new(@depositor),
                                                attrs)
     terminator = Hyrax::Actors::Terminator.new
-    middleware = Californica::IngestMiddlewareStack.build_stack.build(terminator)
-
-    if middleware.create(actor_env)
-      info_stream << "event: record_created, row_id: #{@row_id}, record_id: #{created.id}, ark: #{created.ark}\n"
-    else
-      error_messages = []
-      created.errors.each do |attr, msg|
-        error_stream << "event: validation_failed, row_id: #{@row_id}, attribute: #{attr.capitalize}, message: #{msg}, ark: #{attrs[:ark] ? attrs[:ark] : attrs}\n"
-        error_messages << msg
+    begin
+      retries ||= 0
+      middleware = Californica::IngestMiddlewareStack.build_stack.build(terminator)
+      if middleware.create(actor_env)
+        info_stream << "event: record_created, row_id: #{@row_id}, record_id: #{created.id}, ark: #{created.ark}\n"
+      else
+        error_messages = []
+        created.errors.each do |attr, msg|
+          error_stream << "event: validation_failed, row_id: #{@row_id}, attribute: #{attr.capitalize}, message: #{msg}, ark: #{attrs[:ark] ? attrs[:ark] : attrs}\n"
+          error_messages << msg
+        end
+        # Errors raised here should be rescued in the CsvRowImportJob and the
+        # message should be recorded on the CsvRow object for reporting in the UI
+        raise "Validation failed: #{error_messages.join(', ')}"
       end
-      # Errors raised here should be rescued in the CsvRowImportJob and the
-      # message should be recorded on the CsvRow object for reporting in the UI
-      raise "Validation failed: #{error_messages.join(', ')}"
+    rescue Ldp::BadRequest
+      # get the id from the ark and the uri from the id then delete the tombstone
+      tombstone_uri = "#{ActiveFedora::Base.id_to_uri(Californica::IdGenerator.id_from_ark(created.ark))}/fcr:tombstone"
+      ActiveFedora.fedora.connection.delete(tombstone_uri)
+      retry if (retries += 1) < 3
     end
   end
 end

app/importers/collection_record_importer.rb

@@ -9,8 +9,15 @@ def import_type
 
   def create_for(record:)
     info_stream << "event: collection_import_started, batch_id: #{batch_id}, record_title: #{record.respond_to?(:title) ? record.title : record}\n"
-
-    collection = Collection.find_or_create_by_ark(record.ark)
+    begin
+      retries ||= 0
+      collection = Collection.find_or_create_by_ark(record.ark)
+    rescue Ldp::Gone
+      # get the id from the ark and the uri from the id then delete the tombstone
+      tombstone_uri = "#{ActiveFedora::Base.id_to_uri(Californica::IdGenerator.id_from_ark(record.ark))}/fcr:tombstone"
+      ActiveFedora.fedora.connection.delete(tombstone_uri)
+      retry if (retries += 1) < 3
+    end
     collection.attributes = attributes_for(record: record)
     collection.recalculate_size = false
     if collection.save(update_index: false)
@@ -26,7 +33,6 @@ def create_for(record:)
 
   def update_for(existing_record:, update_record:)
     info_stream << "event: collection_update_started, batch_id: #{batch_id}, ark: #{update_record.ark}\n"
-
     collection = existing_record
     collection.attributes = attributes_for(record: update_record)
     collection.recalculate_size = false