Add JSON metadata export feature

app/jobs/marc_export_job.rb

@@ -4,11 +4,15 @@ class MarcExportJob < ActiveJob::Base
   def perform(theses)
     marc_filename = "#{filename}.mrc"
     zip_filename = "#{filename}.zip"
+    catalog_filename = "#{filename}.json"
+
     begin
-      zip_file = MarcBatch.new(theses, marc_filename, zip_filename).build
-      BatchMailer.marc_batch_email(zip_filename, zip_file, theses).deliver_now
+      marc_zip_file = MarcBatch.new(theses, marc_filename, zip_filename).build
+      catalog_file = CatalogBatch.new(theses, catalog_filename).build
+      BatchMailer.marc_batch_email(zip_filename, marc_zip_file, catalog_filename, catalog_file, theses).deliver_now
     ensure
-      zip_file&.close
+      marc_zip_file&.close!
+      catalog_file&.close!
     end
   end
 

app/mailers/batch_mailer.rb

@@ -1,13 +1,14 @@
 class BatchMailer < ApplicationMailer
-  def marc_batch_email(marc_zip_filename, marc_zip_file, theses)
+  def marc_batch_email(marc_zip_filename, marc_zip_file, json_filename, json_file, theses)
     return unless ENV.fetch('DISABLE_ALL_EMAIL', 'true') == 'false' # allows PR builds to disable emails
 
     @theses = theses
     attachments[marc_zip_filename.to_s] = File.binread(marc_zip_file)
+    attachments[json_filename.to_s] = File.read(json_file)
     mail(from: "MIT Libraries <#{ENV['ETD_APP_EMAIL']}>",
          to: ENV['METADATA_ADMIN_EMAIL'],
          cc: ENV['MAINTAINER_EMAIL'],
-         subject: 'ETD MARC batch export')
+         subject: 'ETD metadata batch export')
   end
 
   def proquest_export_email(json_blob, csv_blob, thesis_count, budget_report_count)

app/models/catalog_batch.rb

@@ -0,0 +1,37 @@
+# Generates a JSON metadata file from a collection of theses to add to the Libraries catalog.
+#
+# Produces a tempfile containing a JSON object with a 'theses' array, where each thesis is
+# exported via CatalogExporter.
+#
+# Example:
+#   batch = CatalogBatch.new(theses_array, 'export.json')
+#   catalog_file = batch.build
+#   File.write('export.json', File.read(catalog_file.path))
+#   catalog_file.close!  # Clean up tempfile
+class CatalogBatch
+  def initialize(theses, filename)
+    @theses = theses
+    @filename = filename
+  end
+
+  # Builds and returns a Tempfile containing the JSON metadata export. The file is ready to read
+  # (file pointer rewound after writing). Caller is responsible for closing the file.
+  def build
+    catalog_file = Tempfile.new(@filename)
+    write_catalog_file(catalog_file)
+    catalog_file
+  end
+
+  private
+
+  def write_catalog_file(catalog_file)
+    theses_data = @theses.map do |thesis|
+      CatalogExporter.new(thesis).to_hash
+    end
+
+    json_output = { theses: theses_data }
+
+    catalog_file.write(JSON.pretty_generate(json_output))
+    catalog_file.rewind
+  end
+end

app/models/catalog_exporter.rb

@@ -0,0 +1,72 @@
+# Exports a single thesis as a hash for JSON serialization.
+#
+# Transforms a thesis record into a flat-ish structure with nested arrays for repeating fields
+# (authors, advisors, degrees, departments).
+#
+# Example:
+#   exporter = CatalogExporter.new(thesis)
+#   hash = exporter.to_hash
+#   # => { title: "...", abstract: "...", authors: [{name: "..."}, ...], ... }
+class CatalogExporter
+  def initialize(thesis)
+    @thesis = thesis
+  end
+
+  # Returns a hash representation of the thesis with all fields required by the metadata team.
+  # Includes: title, abstract, graduation_year, dspace_url, advisors, authors, degrees, and
+  # departments. Array fields are normalized to hashes with relevant metadata.
+  def to_hash
+    {
+      abstract:,
+      advisors:,
+      authors:,
+      degrees:,
+      departments:,
+      dspace_url:,
+      graduation_year:,
+      title:
+    }
+  end
+
+  private
+
+  def abstract
+    @thesis.abstract
+  end
+
+  def advisors
+    @thesis.advisors.map do |advisor|
+      { name: advisor.name }
+    end
+  end
+
+  def authors
+    @thesis.authors.map do |author|
+      { name: author.user.preferred_name }
+    end
+  end
+
+  def degrees
+    @thesis.degrees.map do |degree|
+      { abbreviation: degree.abbreviation }
+    end
+  end
+
+  def departments
+    @thesis.departments.map do |department|
+      { name: department.name_dspace }
+    end
+  end
+
+  def dspace_url
+    "https://dspace.mit.edu/handle/#{@thesis.dspace_handle}"
+  end
+
+  def graduation_year
+    @thesis.graduation_year
+  end
+
+  def title
+    @thesis.title.squish
+  end
+end

app/views/batch_mailer/marc_batch_email.html.erb

@@ -1,6 +1,8 @@
 <p>Hello,</p>
 
 <p>Attached is a metadata export of <%= @theses.count %> theses generated on
-<%= Date.current.strftime('%A, %B %d, %Y') %> at <%= Time.now.strftime('%r %Z') %>.
+<%= Date.current.strftime('%A, %B %d, %Y') %> at <%= Time.now.strftime('%r %Z') %>.</p>
+
+<p>This export includes both MARC (in zip) and JSON.</p>
 
 <p>Please contact the ETD team at <%= ENV['THESIS_ADMIN_EMAIL'] %> with any questions.</p>

lib/tasks/metadata.rake

@@ -0,0 +1,49 @@
+namespace :metadata do
+  desc 'Generate a catalog export of a single published thesis for debugging'
+  task :catalog_export_thesis, [:thesis_id] => :environment do |_t, args|
+    if args.thesis_id.blank?
+      puts 'No thesis ID provided.'
+      next
+    end
+
+    thesis = Thesis.find(args.thesis_id)
+
+    if thesis.publication_status == 'Published'
+      catalog_exporter = CatalogExporter.new(thesis)
+      json_data = catalog_exporter.to_hash
+
+      puts "Catalog Export for Thesis #{args.thesis_id}:"
+      puts JSON.pretty_generate(json_data)
+    else
+      puts "Thesis status of #{thesis.publication_status} cannot be exported. Only published theses can be exported."
+    end
+  end
+
+  # This task is recommended for local development only. On Heroku (or other ephemeral filesystems),
+  # files saved to disk will be deleted when the dyno restarts, making them inaccessible.
+  desc 'Generate a catalog export batch for a specific term (e.g., "2024-June") and save to temp file'
+  task :catalog_export_batch, %i[term output_file] => :environment do |_t, args|
+    if args.term.blank?
+      puts 'Usage: rake metadata:catalog_export_batch["2024-June","output.json"]'
+      puts 'Term format: YYYY-Month (e.g., 2024-June, 2024-September)'
+      next
+    end
+
+    year, month_name = args.term.split('-')
+    query_date = Date.parse("1 #{month_name} #{year}")
+
+    output_file = args.output_file || Rails.root.join("tmp/catalog_export_#{args.term}_#{DateTime.now.utc.strftime('%H_%M')}.json").to_s
+
+    theses = Thesis.published.where(grad_date: query_date.all_month)
+
+    if theses.any?
+      catalog_batch = CatalogBatch.new(theses, File.basename(output_file))
+      catalog_file = catalog_batch.build
+      FileUtils.cp(catalog_file.path, output_file)
+      catalog_file.close!
+      puts "Exported #{theses.count} theses to: #{output_file}"
+    else
+      puts "No published theses found for #{args.term}"
+    end
+  end
+end

test/jobs/marc_export_job_test.rb

@@ -21,4 +21,29 @@ class MarcExportJobTest < ActiveJob::TestCase
       end
     end
   end
+
+  test 'sent email includes both MARC and JSON attachments' do
+    ClimateControl.modify DISABLE_ALL_EMAIL: 'false' do
+      theses = [theses(:one)]
+      Timecop.freeze(Time.utc(2022, 2, 14, 17, 10, 0)) do
+        email = MarcExportJob.perform_now(theses)
+        assert_equal 2, email.attachments.count
+        filenames = email.attachments.map(&:filename)
+        assert(filenames.include?('marc_220214_17_10.zip'))
+        assert(filenames.include?('marc_220214_17_10.json'))
+      end
+    end
+  end
+
+  test 'JSON attachment is valid JSON' do
+    ClimateControl.modify DISABLE_ALL_EMAIL: 'false' do
+      theses = [theses(:one)]
+      email = MarcExportJob.perform_now(theses)
+      json_attachment = email.attachments.find { |a| a.filename.ends_with?('.json') }
+      assert_not_nil(json_attachment)
+
+      json_data = JSON.parse(json_attachment.body.to_s)
+      assert(json_data.key?('theses'))
+    end
+  end
 end

test/mailers/batch_mailer_test.rb

@@ -4,28 +4,33 @@ class BatchMailerTest < ActionMailer::TestCase
   test 'sends emails for MARC batch exports' do
     ClimateControl.modify DISABLE_ALL_EMAIL: 'false' do
       theses = [theses(:one), theses(:two)]
-      zip_file =  MarcBatch.new(theses, 'marc.xml', 'marc.zip').build
-      email = BatchMailer.marc_batch_email('marc.zip', zip_file, theses)
+      marc_zip_file = MarcBatch.new(theses, 'marc.xml', 'marc.zip').build
+      catalog_file = CatalogBatch.new(theses, 'marc.json').build
+      email = BatchMailer.marc_batch_email('marc.zip', marc_zip_file, 'marc.json', catalog_file, theses)
 
       # Send the email, then test that it got queued
       assert_emails 1 do
         email.deliver_now
       end
 
-      # Make sure it was sent to the right person with the expected attachment.
+      # Make sure it was sent to the right person with the expected attachments.
       assert_equal ['app@example.com'], email.from
       assert_equal ['test-metadata@example.com'], email.to
-      assert_equal 'ETD MARC batch export', email.subject
-      assert_equal 'marc.zip', email.attachments.first.filename
+      assert_equal 'ETD metadata batch export', email.subject
+      assert_equal 2, email.attachments.count
+      filenames = email.attachments.map(&:filename)
+      assert_includes filenames, 'marc.zip'
+      assert_includes filenames, 'marc.json'
       assert_includes '2 theses', email.body.to_s
     end
   end
 
   test 'zip file is attached with correct mimetype' do
     ClimateControl.modify DISABLE_ALL_EMAIL: 'false' do
       theses = [theses(:one), theses(:two)]
-      zip_file =  MarcBatch.new(theses, 'marc.xml', 'marc.zip').build
-      email = BatchMailer.marc_batch_email('marc.zip', zip_file, theses)
+      marc_zip_file = MarcBatch.new(theses, 'marc.xml', 'marc.zip').build
+      catalog_file = CatalogBatch.new(theses, 'marc.json').build
+      email = BatchMailer.marc_batch_email('marc.zip', marc_zip_file, 'marc.json', catalog_file, theses)
       attachment = email.attachments['marc.zip']
       assert_equal 'application/zip; filename=marc.zip', attachment.content_type
     end

test/models/catalog_batch_test.rb

@@ -0,0 +1,73 @@
+require 'test_helper'
+
+class CatalogBatchTest < ActiveSupport::TestCase
+  test 'builds a valid JSON file' do
+    theses = [theses(:published)]
+    batch = CatalogBatch.new(theses, 'test.json')
+    catalog_file = batch.build
+
+    json_content = File.read(catalog_file.path)
+    json_data = JSON.parse(json_content)
+
+    assert_not_nil(json_data)
+    catalog_file.close
+  end
+
+  test 'wraps theses in a wrapper object with theses key' do
+    theses = [theses(:published)]
+    batch = CatalogBatch.new(theses, 'test.json')
+    catalog_file = batch.build
+
+    json_content = File.read(catalog_file.path)
+    json_data = JSON.parse(json_content)
+
+    assert(json_data.key?('theses'))
+    assert(json_data['theses'].is_a?(Array))
+    catalog_file.close
+  end
+
+  test 'includes all theses in the batch' do
+    theses = [theses(:published), theses(:one)]
+    batch = CatalogBatch.new(theses, 'test.json')
+    catalog_file = batch.build
+
+    json_content = File.read(catalog_file.path)
+    json_data = JSON.parse(json_content)
+
+    assert_equal(2, json_data['theses'].count)
+    catalog_file.close
+  end
+
+  test 'includes all required fields' do
+    theses = [theses(:published)]
+    batch = CatalogBatch.new(theses, 'test.json')
+    catalog_file = batch.build
+
+    json_content = File.read(catalog_file.path)
+    json_data = JSON.parse(json_content)
+
+    thesis_data = json_data['theses'].first
+
+    assert(thesis_data.key?('abstract'))
+    assert(thesis_data.key?('advisors'))
+    assert(thesis_data.key?('authors'))
+    assert(thesis_data.key?('degrees'))
+    assert(thesis_data.key?('departments'))
+    assert(thesis_data.key?('dspace_url'))
+    assert(thesis_data.key?('graduation_year'))
+    assert(thesis_data.key?('title'))
+
+    catalog_file.close
+  end
+
+  test 'empty theses array produces valid JSON' do
+    batch = CatalogBatch.new([], 'test.json')
+    catalog_file = batch.build
+
+    json_content = File.read(catalog_file.path)
+    json_data = JSON.parse(json_content)
+
+    assert_equal(0, json_data['theses'].count)
+    catalog_file.close
+  end
+end