FOLIOSYNC-12: Combine downloading and syncing logic + add tests

.rubocop_todo.yml

@@ -1,12 +1,12 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2025-10-30 19:13:32 UTC using RuboCop version 1.78.0.
+# on 2026-02-18 17:54:03 UTC using RuboCop version 1.78.0.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
 # versions of RuboCop, may require this file to be generated again.
 
-# Offense count: 20
+# Offense count: 19
 # Configuration parameters: AllowedMethods, AllowedPatterns, CountRepeatedAttributes.
 Metrics/AbcSize:
   Max: 34
@@ -21,7 +21,7 @@ Metrics/ClassLength:
 Metrics/CyclomaticComplexity:
   Max: 8
 
-# Offense count: 22
+# Offense count: 24
 # Configuration parameters: CountComments, Max, CountAsOne, AllowedMethods, AllowedPatterns.
 Metrics/MethodLength:
   Exclude:
@@ -35,6 +35,7 @@ Metrics/MethodLength:
     - 'lib/folio_sync/archives_space_to_folio/job_result_processor.rb'
     - 'lib/folio_sync/archives_space_to_folio/marc_record_enhancer.rb'
     - 'lib/folio_sync/archives_space_to_folio/record_processor.rb'
+    - 'lib/folio_sync/folio_to_hyacinth/hyacinth_record_writer.rb'
     - 'lib/folio_sync/folio_to_hyacinth/marc_downloader.rb'
     - 'lib/folio_sync/folio_to_hyacinth/marc_parsing_methods/title.rb'
     - 'lib/folio_sync/rake/error_logger.rb'
@@ -66,14 +67,13 @@ Rails/FindEach:
   Exclude:
     - 'lib/tasks/test_create_or_update.rake'
 
-# Offense count: 5
+# Offense count: 3
 # This cop supports unsafe autocorrection (--autocorrect-all).
 # Configuration parameters: Include.
 # Include: **/app/**/*.rb, **/config/**/*.rb, db/**/*.rb, **/lib/**/*.rb
 Rails/Output:
   Exclude:
     - 'app/models/folio_to_hyacinth_record.rb'
-    - 'lib/folio_sync/folio_to_hyacinth/marc_parsing_methods/project.rb'
     - 'lib/folio_sync/rake/env_validator.rb'
 
 # Offense count: 1

lib/folio_sync/folio_to_hyacinth/hyacinth_record_writer.rb

@@ -0,0 +1,68 @@
+# frozen_string_literal: true
+
+module FolioSync
+  module FolioToHyacinth
+    class HyacinthRecordWriter
+      attr_reader :syncing_errors
+
+      def initialize
+        @logger = Logger.new($stdout)
+        @client = FolioSync::Hyacinth::Client.instance
+        @syncing_errors = []
+      end
+
+      # @param [String] marc_file_path
+      # @param [String] folio_hrid
+      # @param [Array] existing_records
+      def sync(marc_file_path, folio_hrid, existing_records)
+        case existing_records.length
+        when 0
+          create_new_record(marc_file_path, folio_hrid)
+        when 1
+          update_existing_record(marc_file_path, folio_hrid, existing_records.first)
+        else
+          handle_multiple_records(folio_hrid)
+        end
+      end
+
+      private
+
+      def create_new_record(marc_file_path, folio_hrid)
+        @logger.info("Creating new Hyacinth record for #{folio_hrid}")
+
+        new_record = FolioToHyacinthRecord.new(marc_file_path)
+        response = @client.create_new_record(new_record.digital_object_data, publish: true)
+
+        @logger.info("Created record for #{folio_hrid}: #{response.inspect}")
+      rescue StandardError => e
+        error_message = "Failed to create record for #{folio_hrid}: #{e.message}"
+        @logger.error(error_message)
+        @syncing_errors << error_message
+      end
+
+      def update_existing_record(marc_file_path, folio_hrid, existing_record)
+        @logger.info("Updating existing Hyacinth record for #{folio_hrid}")
+        preserved_data = { 'identifiers' => existing_record['identifiers'] }
+        updated_record = FolioToHyacinthRecord.new(marc_file_path, preserved_data)
+
+        response = @client.update_existing_record(
+          existing_record['pid'],
+          updated_record.digital_object_data,
+          publish: true
+        )
+
+        @logger.info("Updated record #{existing_record['pid']}: #{response.inspect}")
+      rescue StandardError => e
+        error_message = "Failed to update record #{existing_record['pid']} for #{folio_hrid}: #{e.message}"
+        @logger.error(error_message)
+        @syncing_errors << error_message
+      end
+
+      def handle_multiple_records(folio_hrid)
+        error_message = "Multiple Hyacinth records found for FOLIO HRID #{folio_hrid}"
+        @logger.error(error_message)
+        @syncing_errors << error_message
+      end
+    end
+  end
+end

lib/folio_sync/folio_to_hyacinth/hyacinth_synchronizer.rb

@@ -0,0 +1,59 @@
+# frozen_string_literal: true
+
+module FolioSync
+  module FolioToHyacinth
+    class HyacinthSynchronizer
+      attr_reader :downloading_errors, :syncing_errors
+
+      def initialize
+        @logger = Logger.new($stdout)
+        @downloading_errors = []
+        @syncing_errors = []
+      end
+
+      # Performs MARC downloads and syncs resources to Hyacinth
+      # @param [Integer] last_x_hours Records newer than this are synced.
+      def download_and_sync_folio_to_hyacinth_records(last_x_hours)
+        download_marc_from_folio(last_x_hours)
+        prepare_and_sync_folio_to_hyacinth_records
+      end
+
+      def clear_downloads!
+        @logger.info('Clearing downloaded MARC files...')
+        FileUtils.rm_rf(downloaded_marc_files_path)
+      end
+
+      def download_marc_from_folio(last_x_hours)
+        downloader = FolioSync::FolioToHyacinth::MarcDownloader.new
+        downloader.download_965hyacinth_marc_records(last_x_hours)
+
+        return if downloader.downloading_errors.blank?
+
+        @logger.error("Error downloading MARC records from FOLIO: #{downloader.downloading_errors}")
+        @downloading_errors = downloader.downloading_errors
+      end
+
+      def prepare_and_sync_folio_to_hyacinth_records
+        marc_files = Dir.glob(downloaded_marc_files_path)
+        @logger.info("Processing #{marc_files.count} MARC files")
+
+        marc_files.each do |marc_file_path|
+          process_marc_file(marc_file_path)
+        end
+      end
+
+      private
+
+      def downloaded_marc_files_path
+        "#{Rails.configuration.folio_to_hyacinth[:download_directory]}/*.mrc"
+      end
+
+      def process_marc_file(marc_file_path)
+        processor = FolioSync::FolioToHyacinth::MarcProcessor.new(marc_file_path)
+        processor.prepare_and_sync_folio_to_hyacinth_record!
+
+        @syncing_errors.concat(processor.syncing_errors) if processor.syncing_errors.any?
+      end
+    end
+  end
+end

lib/folio_sync/folio_to_hyacinth/marc_processor.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+class FolioSync::FolioToHyacinth::MarcProcessor
+  attr_reader :syncing_errors
+
+  def initialize(marc_file_path)
+    @marc_file_path = marc_file_path
+    @logger = Logger.new($stdout)
+    @record_syncer = FolioSync::FolioToHyacinth::HyacinthRecordWriter.new
+    @syncing_errors = []
+  end
+
+  def prepare_and_sync_folio_to_hyacinth_record!
+    folio_hrid = extract_hrid_from_filename(@marc_file_path)
+    existing_records = fetch_existing_hyacinth_records(folio_hrid)
+
+    @logger.info("Found #{existing_records.length} Hyacinth records for FOLIO HRID: #{folio_hrid}")
+
+    @record_syncer.sync(@marc_file_path, folio_hrid, existing_records)
+    @syncing_errors.concat(@record_syncer.syncing_errors) if @record_syncer.syncing_errors.any?
+  rescue StandardError => e
+    @logger.error("Failed to process #{folio_hrid}: #{e.message}")
+    @syncing_errors << "Error processing #{folio_hrid}: #{e.message}"
+  end
+
+  private
+
+  def extract_hrid_from_filename(marc_file_path)
+    File.basename(marc_file_path, '.mrc')
+  end
+
+  def fetch_existing_hyacinth_records(folio_hrid)
+    potential_clio_identifier = "clio#{folio_hrid}"
+    client = FolioSync::Hyacinth::Client.instance
+    client.find_by_identifier(
+      potential_clio_identifier,
+      { f: { digital_object_type_display_label_sim: ['Item'] } }
+    )
+  end
+end

lib/tasks/hyacinth_sync.rake

@@ -3,8 +3,42 @@
 namespace :folio_sync do
   namespace :folio_to_hyacinth do
     task run: :environment do
+      puts 'Starting Folio to Hyacinth sync task...'
+
+      modified_since = ENV['modified_since']
+
+      modified_since_sanitized =
+        if modified_since && !modified_since.strip.empty?
+          begin
+            Integer(modified_since)
+          rescue ArgumentError
+            puts 'Error: modified_since must be an integer (number of hours).'
+            exit 1
+          end
+        end
+
+      clear_downloads = ENV['clear_downloads'].nil? || ENV['clear_downloads'] == 'true'
+      puts "Will downloads be cleared? #{clear_downloads}"
+
+      synchronizer = FolioSync::FolioToHyacinth::HyacinthSynchronizer.new
+      synchronizer.clear_downloads! if clear_downloads
+      synchronizer.download_and_sync_folio_to_hyacinth_records(modified_since_sanitized)
+
+      if synchronizer.downloading_errors.any? || synchronizer.syncing_errors.any?
+        puts 'Errors encountered during Folio to Hyacinth sync:'
+        puts "Downloading Errors: #{synchronizer.downloading_errors}" if synchronizer.downloading_errors.any?
+        puts "Syncing Errors: #{synchronizer.syncing_errors}" if synchronizer.syncing_errors.any?
+
+        exit 1
+      else
+        puts 'Folio to Hyacinth sync completed successfully.'
+      end
+    end
+
+    # Downloads FOLIO MARC records, skipping the syncing step
+    task download_folio_marc_files: :environment do
       modified_since = ENV['modified_since']
-      modified_since_num =
+      modified_since_sanitized =
         if modified_since && !modified_since.strip.empty?
           begin
             Integer(modified_since)
@@ -15,7 +49,7 @@ namespace :folio_sync do
         end
 
       downloader = FolioSync::FolioToHyacinth::MarcDownloader.new
-      downloader.download_965hyacinth_marc_records(modified_since_num)
+      downloader.download_965hyacinth_marc_records(modified_since_sanitized)
 
       if downloader.downloading_errors.present?
         puts "Errors encountered during MARC download: #{downloader.downloading_errors}"
@@ -34,43 +68,19 @@ namespace :folio_sync do
       downloader.download_single_965hyacinth_marc_record(folio_hrid)
     end
 
-    # WIP: This task syncs all downloaded FOLIO MARC records to Hyacinth
+    # Syncs all previously downloaded FOLIO MARC records to Hyacinth
     task sync_to_hyacinth: :environment do
       puts 'Starting Folio to Hyacinth sync task...'
-      file_dir = Rails.root.join('tmp/working_data/development/folio_to_hyacinth/downloaded_files')
-
-      # For each MARC file in the download directory, create or update the corresponding Hyacinth record
-      Dir.glob(File.join(file_dir, '*.mrc')).each do |marc_file_path|
-        puts "Processing MARC file: #{marc_file_path}"
-
-        # Check if the record already exists in Hyacinth
-        folio_hrid = File.basename(marc_file_path, '.mrc')
-        potential_clio_identifier = "clio#{folio_hrid}"
-        client = FolioSync::Hyacinth::Client.instance
-        results = client.find_by_identifier(potential_clio_identifier,
-                                            { f: { digital_object_type_display_label_sim: ['Item'] } })
-        puts "Found #{results.length} records with identifier #{potential_clio_identifier}."
-
-        # TODO: Eventually this logic will be placed under FolioToHyacinth namespace
-        if results.empty?
-          puts 'No records found. Creating a new record in Hyacinth.'
-          new_record = FolioToHyacinthRecord.new(marc_file_path)
-          puts "New record digital object data: #{new_record.digital_object_data}"
-          response = client.create_new_record(new_record.digital_object_data, publish: true)
-          puts "Response from Hyacinth when creating record with hrid #{folio_hrid}: #{response.inspect}"
-        elsif results.length == 1
-          pid = results.first['pid']
-          puts "Found 1 record with pid: #{pid}."
-
-          # Get only the data needed for update
-          preserved_data = { 'identifiers' => results.first['identifiers'] }
-          updated_record = FolioToHyacinthRecord.new(marc_file_path, preserved_data)
-          puts "Updated record digital object data: #{updated_record.digital_object_data}"
-          response = client.update_existing_record(pid, updated_record.digital_object_data, publish: true)
-          puts "Response from Hyacinth when updating record #{pid}: #{response.inspect}"
-        else
-          puts "Error: Found multiple records with identifier #{potential_clio_identifier}."
-        end
+
+      synchronizer = FolioSync::FolioToHyacinth::HyacinthSynchronizer.new
+      synchronizer.prepare_and_sync_folio_to_hyacinth_records
+
+      if synchronizer.syncing_errors.any?
+        puts 'Errors encountered during Folio to Hyacinth sync:'
+        puts "Syncing Errors: #{synchronizer.syncing_errors}" if synchronizer.syncing_errors.any?
+        exit 1
+      else
+        puts 'Folio to Hyacinth sync completed successfully.'
       end
     end
 
@@ -91,7 +101,7 @@ namespace :folio_sync do
       end
 
       # Add 965p field with value academic_commons, ensure 965$a is set to 965hyacinth
-      marc_record.append(MARC::DataField.new('965', ' ', ' ', ['a', '965hyacinth'], ['p', 'academic_commons'], ['p', 'test']))
+      marc_record.append(MARC::DataField.new('965', ' ', ' ', ['a', '965hyacinth'], ['p', 'Test']))
       puts "Modified MARC record with new 965 field: #{marc_record.inspect}"
 
       new_filepath = Rails.root.join(Rails.configuration.folio_to_hyacinth[:download_directory], 'modified_marc.mrc')
@@ -100,7 +110,12 @@ namespace :folio_sync do
         writer.write(marc_record)
         writer.close
       end
-      puts "Final MARC record: #{marc_record.inspect}"
+      reader = MARC::Reader.new(new_filepath.to_s)
+      reader.each do |record|
+        record.fields.each_by_tag(['965']) do |field|
+          puts field
+        end
+      end
     end
 
     task create_new_hyacinth_record: :environment do