when archiving a submission delete old files and zip master file if big

lib/ontologies_linked_data/models/ontology_submission.rb

@@ -15,7 +15,6 @@ class OntologySubmission < LinkedData::Models::Base
       include LinkedData::Concerns::SubmissionProcessable
       include LinkedData::Concerns::OntologySubmission::MetadataExtractor
 
-      FILES_TO_DELETE = ['labels.ttl', 'mappings.ttl', 'obsolete.ttl', 'owlapi.xrdf', 'errors.log']
       FLAT_ROOTS_LIMIT = 1000
 
       model :ontology_submission, name_with: lambda { |s| submission_id_generator(s) }

lib/ontologies_linked_data/services/submission_process/operations/submission_archiver.rb

@@ -2,29 +2,49 @@ module LinkedData
   module Services
     class OntologySubmissionArchiver < OntologySubmissionProcess
 
-      FILES_TO_DELETE = %w[labels.ttl mappings.ttl obsolete.ttl owlapi.xrdf errors.log]
-
+      FILES_TO_DELETE = ['labels.ttl', 'mappings.ttl', 'obsolete.ttl', 'owlapi.xrdf', 'errors.log']
+      FOLDERS_TO_DELETE = ['unzipped']
+      FILE_SIZE_ZIPPING_THRESHOLD = 100 * 1024 * 1024 # 100MB
 
       def process
-        submission_archive
+        archive_submission
       end
 
       private
-      def submission_archive
+
+      def archive_submission
+        @submission.ontology.bring(:submissions)
+        submissions = @submission.ontology.submissions
+        return if submissions.nil?
+
+        submissions.each { |s| s.bring(:submissionId) }
+        submission = submissions.sort { |a, b| b.submissionId <=> a.submissionId }.first
+
+        return unless @submission.submissionId < submission.submissionId
+
         @submission.submissionStatus = nil
         status = LinkedData::Models::SubmissionStatus.find("ARCHIVED").first
         @submission.add_submission_status(status)
 
+        @submission.unindex
 
         # Delete everything except for original ontology file.
-        @submission.ontology.bring(:submissions)
-        submissions = @submission.ontology.submissions
-        unless submissions.nil?
-          submissions.each { |s| s.bring(:submissionId) }
-          submission = submissions.sort { |a, b| b.submissionId <=> a.submissionId }.first
-          # Don't perform deletion if this is the most recent submission.
-          delete_old_submission_files if @submission.submissionId < submission.submissionId
-        end
+        delete_old_submission_files
+        @submission.uploadFilePath = zip_submission_uploaded_file
+      end
+
+      def zip_submission_uploaded_file
+        @submission.bring(:uploadFilePath) if @submission.bring?(:uploadFilePath)
+        return @submission.uploadFilePath if @submission.zipped?
+
+        return @submission.uploadFilePath if @submission.uploadFilePath.nil? || @submission.uploadFilePath.empty?
+
+        return @submission.uploadFilePath if File.size(@submission.uploadFilePath) < FILE_SIZE_ZIPPING_THRESHOLD
+
+        old_path = @submission.uploadFilePath
+        zip_file = Utils::FileHelpers.zip_file(old_path)
+        FileUtils.rm(old_path, force: true)
+        zip_file
       end
 
       def delete_old_submission_files
@@ -33,10 +53,11 @@ def delete_old_submission_files
         submission_files.push(@submission.csv_path)
         submission_files.push(@submission.parsing_log_path) unless @submission.parsing_log_path.nil?
         FileUtils.rm(submission_files, force: true)
+        submission_folders = FOLDERS_TO_DELETE.map { |f| File.join(path_to_repo, f) }
+        submission_folders.each { |d| FileUtils.remove_dir(d) if File.directory?(d) }
       end
 
     end
 
-
   end
 end

lib/ontologies_linked_data/services/submission_process/submission_process.rb

@@ -1,4 +1,4 @@
-module LinkedData
+  module LinkedData
   module Services
     class OntologySubmissionProcess
 
@@ -7,6 +7,10 @@ def initialize(submission)
       end
 
       def process(logger, options = {})
+        call
+      end
+
+      def call
         raise NotImplementedError
       end
     end

lib/ontologies_linked_data/utils/file.rb

@@ -7,6 +7,14 @@
 module LinkedData
   module Utils
     module FileHelpers
+
+      class GzipFile
+        attr_accessor :name
+        def initialize(gz)
+          self.name = gz.orig_name
+        end
+      end
+
 
       def self.zip?(file_path)
         file_path = file_path.to_s
@@ -78,6 +86,21 @@ def self.unzip(file_path, dst_folder)
         extracted_files
       end
 
+      def self.zip_file(file_path)
+        return file_path if self.zip?(file_path)
+
+        zip_file_path = "#{file_path}.zip"
+        Zip::File.open(zip_file_path, Zip::File::CREATE) do |zipfile|
+          # Add the file to the zip
+          begin
+            zipfile.add(File.basename(file_path), file_path)
+          rescue Zip::EntryExistsError
+          end
+
+        end
+        zip_file_path
+      end
+
       def self.automaster?(path, format)
         self.automaster(path, format) != nil
       end