From 7a9faf9512c5c06ec0e3ae9d7006b69e0c526388 Mon Sep 17 00:00:00 2001
From: jazairi <16103405+jazairi@users.noreply.github.com>
Date: Tue, 4 Aug 2026 14:26:13 -0700
Subject: [PATCH 1/6] Add JSON metadata export feature

Why these changes are being introduced:

The metadata team has requested a JSON metadata
export, as it would be easier for them to parse
than the MARC file.

Relevant ticket(s):

- [ETD-695](https://mitlibraries.atlassian.net/browse/ETD-695)

How this addresses that need:

This adds JsonExporter and JsonBatch classes, which
serialize thesis metadata and writes the metadata
to a JSON file. This workflow is invoked in the
MarcExportJob, and the output is included
alongside the MARC file in the report email.

It also creates two rake tasks: one to export
the JSON metadata of a single thesis, and one
to export all theses in a given term. This is
primarily for testing purposes.

Side effects of this change:

The export job and batch emails are not clearly
named for the time being. This feels like an
acceptable short-term risk. We plan to retire the
MARC export altogether, at which point we can
rename those files accordingly.
---
 app/jobs/marc_export_job.rb                   | 10 ++-
 app/mailers/batch_mailer.rb                   |  5 +-
 app/models/json_batch.rb                      | 27 +++++++
 app/models/json_exporter.rb                   | 60 +++++++++++++++
 .../batch_mailer/marc_batch_email.html.erb    |  2 +
 lib/tasks/metadata.rake                       | 49 +++++++++++++
 test/jobs/marc_export_job_test.rb             | 25 +++++++
 test/mailers/batch_mailer_test.rb             | 19 +++--
 test/models/json_batch_test.rb                | 73 +++++++++++++++++++
 test/models/json_exporter_test.rb             | 69 ++++++++++++++++++
 10 files changed, 327 insertions(+), 12 deletions(-)
 create mode 100644 app/models/json_batch.rb
 create mode 100644 app/models/json_exporter.rb
 create mode 100644 lib/tasks/metadata.rake
 create mode 100644 test/models/json_batch_test.rb
 create mode 100644 test/models/json_exporter_test.rb

diff --git a/app/jobs/marc_export_job.rb b/app/jobs/marc_export_job.rb
index 3d24cda7..a3af7677 100644
--- a/app/jobs/marc_export_job.rb
+++ b/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"
+    json_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
+      json_file = JsonBatch.new(theses, json_filename).build
+      BatchMailer.marc_batch_email(zip_filename, marc_zip_file, json_filename, json_file, theses).deliver_now
     ensure
-      zip_file&.close
+      marc_zip_file&.close
+      json_file&.close
     end
   end
 
diff --git a/app/mailers/batch_mailer.rb b/app/mailers/batch_mailer.rb
index b4c7d206..3d6c5e97 100644
--- a/app/mailers/batch_mailer.rb
+++ b/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)
diff --git a/app/models/json_batch.rb b/app/models/json_batch.rb
new file mode 100644
index 00000000..2592c2f1
--- /dev/null
+++ b/app/models/json_batch.rb
@@ -0,0 +1,27 @@
+class JsonBatch
+  def initialize(theses, json_filename)
+    @theses = theses
+    @json_filename = json_filename
+  end
+
+  def build
+    json_file = Tempfile.new(@json_filename)
+    write_json_file(json_file)
+    json_file
+  end
+
+  private
+
+  def write_json_file(json_file)
+    theses_data = @theses.map do |thesis|
+      JsonExporter.new(thesis).to_hash
+    end
+
+    json_output = {
+      theses: theses_data
+    }.to_json
+
+    json_file.write(JSON.pretty_generate(JSON.parse(json_output)))
+    json_file.rewind
+  end
+end
diff --git a/app/models/json_exporter.rb b/app/models/json_exporter.rb
new file mode 100644
index 00000000..a78f94ba
--- /dev/null
+++ b/app/models/json_exporter.rb
@@ -0,0 +1,60 @@
+class JsonExporter
+  def initialize(thesis)
+    @thesis = thesis
+  end
+
+  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
diff --git a/app/views/batch_mailer/marc_batch_email.html.erb b/app/views/batch_mailer/marc_batch_email.html.erb
index 0c5ae7b2..d95818d6 100644
--- a/app/views/batch_mailer/marc_batch_email.html.erb
+++ b/app/views/batch_mailer/marc_batch_email.html.erb
@@ -3,4 +3,6 @@
 <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') %>.
 
+<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>
diff --git a/lib/tasks/metadata.rake b/lib/tasks/metadata.rake
new file mode 100644
index 00000000..641895ec
--- /dev/null
+++ b/lib/tasks/metadata.rake
@@ -0,0 +1,49 @@
+namespace :metadata do
+  desc 'Generate a JSON export of a single published thesis for testing'
+  task :json_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'
+      json_exporter = JsonExporter.new(thesis)
+      json_data = json_exporter.to_hash
+
+      puts "JSON 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
+
+  desc 'Generate a JSON export batch for a specific term (e.g., "2024-June") and save to temp file'
+  task :json_export_batch, %i[term output_file] => :environment do |_t, args|
+    if args.term.blank?
+      puts 'Usage: rake metadata:json_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/json_export_#{args.term}_#{DateTime.now.utc.strftime('%H_%M')}.json").to_s
+
+    theses = Thesis.where(publication_status: 'Published').select do |t|
+      t.grad_date.year == query_date.year && t.grad_date.month == query_date.month
+    end
+
+    if theses.any?
+      json_batch = JsonBatch.new(theses, File.basename(output_file))
+      json_file = json_batch.build
+      FileUtils.cp(json_file.path, output_file)
+      json_file.close
+      puts "Exported #{theses.count} theses to: #{output_file}"
+    else
+      puts "No published theses found for #{args.term}"
+    end
+  end
+end
diff --git a/test/jobs/marc_export_job_test.rb b/test/jobs/marc_export_job_test.rb
index a9b68dba..b62ae06e 100644
--- a/test/jobs/marc_export_job_test.rb
+++ b/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
diff --git a/test/mailers/batch_mailer_test.rb b/test/mailers/batch_mailer_test.rb
index d99d8dc3..9427d796 100644
--- a/test/mailers/batch_mailer_test.rb
+++ b/test/mailers/batch_mailer_test.rb
@@ -4,19 +4,23 @@ 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
+      json_file = JsonBatch.new(theses, 'marc.json').build
+      email = BatchMailer.marc_batch_email('marc.zip', marc_zip_file, 'marc.json', json_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
@@ -24,8 +28,9 @@ class BatchMailerTest < ActionMailer::TestCase
   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
+      json_file = JsonBatch.new(theses, 'marc.json').build
+      email = BatchMailer.marc_batch_email('marc.zip', marc_zip_file, 'marc.json', json_file, theses)
       attachment = email.attachments['marc.zip']
       assert_equal 'application/zip; filename=marc.zip', attachment.content_type
     end
diff --git a/test/models/json_batch_test.rb b/test/models/json_batch_test.rb
new file mode 100644
index 00000000..2e5b12a5
--- /dev/null
+++ b/test/models/json_batch_test.rb
@@ -0,0 +1,73 @@
+require 'test_helper'
+
+class JsonBatchTest < ActiveSupport::TestCase
+  test 'builds a valid JSON file' do
+    theses = [theses(:published)]
+    batch = JsonBatch.new(theses, 'test.json')
+    json_file = batch.build
+
+    json_content = File.read(json_file.path)
+    json_data = JSON.parse(json_content)
+
+    assert_not_nil(json_data)
+    json_file.close
+  end
+
+  test 'wraps theses in a wrapper object with theses key' do
+    theses = [theses(:published)]
+    batch = JsonBatch.new(theses, 'test.json')
+    json_file = batch.build
+
+    json_content = File.read(json_file.path)
+    json_data = JSON.parse(json_content)
+
+    assert(json_data.key?('theses'))
+    assert(json_data['theses'].is_a?(Array))
+    json_file.close
+  end
+
+  test 'includes all theses in the batch' do
+    theses = [theses(:published), theses(:one)]
+    batch = JsonBatch.new(theses, 'test.json')
+    json_file = batch.build
+
+    json_content = File.read(json_file.path)
+    json_data = JSON.parse(json_content)
+
+    assert_equal(2, json_data['theses'].count)
+    json_file.close
+  end
+
+  test 'includes all required fields' do
+    theses = [theses(:published)]
+    batch = JsonBatch.new(theses, 'test.json')
+    json_file = batch.build
+
+    json_content = File.read(json_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'))
+
+    json_file.close
+  end
+
+  test 'empty theses array produces valid JSON' do
+    batch = JsonBatch.new([], 'test.json')
+    json_file = batch.build
+
+    json_content = File.read(json_file.path)
+    json_data = JSON.parse(json_content)
+
+    assert_equal(0, json_data['theses'].count)
+    json_file.close
+  end
+end
diff --git a/test/models/json_exporter_test.rb b/test/models/json_exporter_test.rb
new file mode 100644
index 00000000..38687f99
--- /dev/null
+++ b/test/models/json_exporter_test.rb
@@ -0,0 +1,69 @@
+require 'test_helper'
+
+class JsonExporterTest < ActiveSupport::TestCase
+  test 'includes correctly formatted title' do
+    thesis = theses(:published)
+    exporter = JsonExporter.new(thesis)
+    json_hash = exporter.to_hash
+    assert_equal thesis.title&.squish, json_hash[:title]
+  end
+
+  test 'includes abstract' do
+    thesis = theses(:published)
+    exporter = JsonExporter.new(thesis)
+    json_hash = exporter.to_hash
+    assert_equal thesis.abstract, json_hash[:abstract]
+  end
+
+  test 'includes grad year' do
+    thesis = theses(:published)
+    exporter = JsonExporter.new(thesis)
+    json_hash = exporter.to_hash
+    assert_equal thesis.graduation_year, json_hash[:graduation_year]
+  end
+
+  test 'includes correctly formatted dspace_url' do
+    thesis = theses(:published)
+    exporter = JsonExporter.new(thesis)
+    json_hash = exporter.to_hash
+    expected_url = "https://dspace.mit.edu/handle/#{thesis.dspace_handle}"
+    assert_equal expected_url, json_hash[:dspace_url]
+  end
+
+  test 'includes authors with nested structure' do
+    thesis = theses(:published)
+    exporter = JsonExporter.new(thesis)
+    json_hash = exporter.to_hash
+    assert json_hash[:authors].is_a?(Array)
+    assert json_hash[:authors].first.is_a?(Hash)
+    assert json_hash[:authors].first[:name]
+  end
+
+  test 'includes degrees with nested structure' do
+    thesis = theses(:published)
+    exporter = JsonExporter.new(thesis)
+    json_hash = exporter.to_hash
+    assert json_hash[:degrees].is_a?(Array)
+    assert json_hash[:degrees].first.is_a?(Hash)
+    assert json_hash[:degrees].first[:abbreviation]
+  end
+
+  test 'includes advisors with nested structure' do
+    thesis = theses(:published)
+    thesis.advisors << advisors(:first)
+    exporter = JsonExporter.new(thesis)
+    json_hash = exporter.to_hash
+    assert json_hash[:advisors].is_a?(Array)
+    assert json_hash[:advisors].first.is_a?(Hash)
+    assert json_hash[:advisors].first[:name]
+  end
+
+  test 'includes departments with nested structure' do
+    thesis = theses(:published)
+    exporter = JsonExporter.new(thesis)
+    json_hash = exporter.to_hash
+    assert json_hash[:departments].is_a?(Array)
+    assert json_hash[:departments].first.is_a?(Hash)
+    assert json_hash[:departments].first[:name]
+  end
+end

From ff7c8c3c97126058eb8db389de4ad78ea721dd2c Mon Sep 17 00:00:00 2001
From: Isra Jazairi <jazairi@mit.edu>
Date: Tue, 4 Aug 2026 19:06:40 -0400
Subject: [PATCH 2/6] Apply Copilot code review suggestions

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
---
 app/models/json_batch.rb                         | 6 ++----
 app/views/batch_mailer/marc_batch_email.html.erb | 2 +-
 lib/tasks/metadata.rake                          | 4 +---
 3 files changed, 4 insertions(+), 8 deletions(-)

diff --git a/app/models/json_batch.rb b/app/models/json_batch.rb
index 2592c2f1..15f14cd6 100644
--- a/app/models/json_batch.rb
+++ b/app/models/json_batch.rb
@@ -17,11 +17,9 @@ def write_json_file(json_file)
       JsonExporter.new(thesis).to_hash
     end
 
-    json_output = {
-      theses: theses_data
-    }.to_json
+    json_output = { theses: theses_data }
 
-    json_file.write(JSON.pretty_generate(JSON.parse(json_output)))
+    json_file.write(JSON.pretty_generate(json_output))
     json_file.rewind
   end
 end
diff --git a/app/views/batch_mailer/marc_batch_email.html.erb b/app/views/batch_mailer/marc_batch_email.html.erb
index d95818d6..336a34bb 100644
--- a/app/views/batch_mailer/marc_batch_email.html.erb
+++ b/app/views/batch_mailer/marc_batch_email.html.erb
@@ -1,7 +1,7 @@
 <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>
 
diff --git a/lib/tasks/metadata.rake b/lib/tasks/metadata.rake
index 641895ec..7d08b99c 100644
--- a/lib/tasks/metadata.rake
+++ b/lib/tasks/metadata.rake
@@ -32,9 +32,7 @@ namespace :metadata do
 
     output_file = args.output_file || Rails.root.join("tmp/json_export_#{args.term}_#{DateTime.now.utc.strftime('%H_%M')}.json").to_s
 
-    theses = Thesis.where(publication_status: 'Published').select do |t|
-      t.grad_date.year == query_date.year && t.grad_date.month == query_date.month
-    end
+    theses = Thesis.published.where(grad_date: query_date.all_month)
 
     if theses.any?
       json_batch = JsonBatch.new(theses, File.basename(output_file))

From 0fc146c2df4a3c5ab9c4d1b51c7fb2360a3d56ae Mon Sep 17 00:00:00 2001
From: jazairi <16103405+jazairi@users.noreply.github.com>
Date: Wed, 5 Aug 2026 16:46:52 -0700
Subject: [PATCH 3/6] Address code review feedback

- Rename and document new batch/export classes
- Remove unnecessary safe navigation
- Call `close!` on tempfiles to ensure immediate
cleanup
---
 app/jobs/marc_export_job.rb                   |  6 +--
 app/models/catalog_batch.rb                   | 37 +++++++++++++++++++
 .../{json_exporter.rb => catalog_exporter.rb} | 16 +++++++-
 app/models/json_batch.rb                      | 25 -------------
 lib/tasks/metadata.rake                       |  8 ++--
 test/mailers/batch_mailer_test.rb             |  4 +-
 ...on_batch_test.rb => catalog_batch_test.rb} | 12 +++---
 ...orter_test.rb => catalog_exporter_test.rb} | 18 ++++-----
 8 files changed, 75 insertions(+), 51 deletions(-)
 create mode 100644 app/models/catalog_batch.rb
 rename app/models/{json_exporter.rb => catalog_exporter.rb} (56%)
 delete mode 100644 app/models/json_batch.rb
 rename test/models/{json_batch_test.rb => catalog_batch_test.rb} (85%)
 rename test/models/{json_exporter_test.rb => catalog_exporter_test.rb} (82%)

diff --git a/app/jobs/marc_export_job.rb b/app/jobs/marc_export_job.rb
index a3af7677..cb0ef025 100644
--- a/app/jobs/marc_export_job.rb
+++ b/app/jobs/marc_export_job.rb
@@ -8,11 +8,11 @@ def perform(theses)
 
     begin
       marc_zip_file = MarcBatch.new(theses, marc_filename, zip_filename).build
-      json_file = JsonBatch.new(theses, json_filename).build
+      json_file = CatalogBatch.new(theses, json_filename).build
       BatchMailer.marc_batch_email(zip_filename, marc_zip_file, json_filename, json_file, theses).deliver_now
     ensure
-      marc_zip_file&.close
-      json_file&.close
+      marc_zip_file&.close!
+      json_file&.close!
     end
   end
 
diff --git a/app/models/catalog_batch.rb b/app/models/catalog_batch.rb
new file mode 100644
index 00000000..e219d810
--- /dev/null
+++ b/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')
+#   json_file = batch.build
+#   File.write('export.json', File.read(json_file.path))
+#   json_file.close!  # Clean up tempfile
+class CatalogBatch
+  def initialize(theses, json_filename)
+    @theses = theses
+    @json_filename = json_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
+    json_file = Tempfile.new(@json_filename)
+    write_json_file(json_file)
+    json_file
+  end
+
+  private
+
+  def write_json_file(json_file)
+    theses_data = @theses.map do |thesis|
+      CatalogExporter.new(thesis).to_hash
+    end
+
+    json_output = { theses: theses_data }
+
+    json_file.write(JSON.pretty_generate(json_output))
+    json_file.rewind
+  end
+end
diff --git a/app/models/json_exporter.rb b/app/models/catalog_exporter.rb
similarity index 56%
rename from app/models/json_exporter.rb
rename to app/models/catalog_exporter.rb
index a78f94ba..d77fd4fa 100644
--- a/app/models/json_exporter.rb
+++ b/app/models/catalog_exporter.rb
@@ -1,8 +1,20 @@
-class JsonExporter
+# 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:,
@@ -55,6 +67,6 @@ def graduation_year
   end
 
   def title
-    @thesis.title&.squish
+    @thesis.title.squish
   end
 end
diff --git a/app/models/json_batch.rb b/app/models/json_batch.rb
deleted file mode 100644
index 15f14cd6..00000000
--- a/app/models/json_batch.rb
+++ /dev/null
@@ -1,25 +0,0 @@
-class JsonBatch
-  def initialize(theses, json_filename)
-    @theses = theses
-    @json_filename = json_filename
-  end
-
-  def build
-    json_file = Tempfile.new(@json_filename)
-    write_json_file(json_file)
-    json_file
-  end
-
-  private
-
-  def write_json_file(json_file)
-    theses_data = @theses.map do |thesis|
-      JsonExporter.new(thesis).to_hash
-    end
-
-    json_output = { theses: theses_data }
-
-    json_file.write(JSON.pretty_generate(json_output))
-    json_file.rewind
-  end
-end
diff --git a/lib/tasks/metadata.rake b/lib/tasks/metadata.rake
index 7d08b99c..e352563e 100644
--- a/lib/tasks/metadata.rake
+++ b/lib/tasks/metadata.rake
@@ -1,5 +1,5 @@
 namespace :metadata do
-  desc 'Generate a JSON export of a single published thesis for testing'
+  desc 'Generate a JSON export of a single published thesis for debugging'
   task :json_export_thesis, [:thesis_id] => :environment do |_t, args|
     if args.thesis_id.blank?
       puts 'No thesis ID provided.'
@@ -9,7 +9,7 @@ namespace :metadata do
     thesis = Thesis.find(args.thesis_id)
 
     if thesis.publication_status == 'Published'
-      json_exporter = JsonExporter.new(thesis)
+      json_exporter = CatalogExporter.new(thesis)
       json_data = json_exporter.to_hash
 
       puts "JSON Export for Thesis #{args.thesis_id}:"
@@ -35,10 +35,10 @@ namespace :metadata do
     theses = Thesis.published.where(grad_date: query_date.all_month)
 
     if theses.any?
-      json_batch = JsonBatch.new(theses, File.basename(output_file))
+      json_batch = CatalogBatch.new(theses, File.basename(output_file))
       json_file = json_batch.build
       FileUtils.cp(json_file.path, output_file)
-      json_file.close
+      json_file.close!
       puts "Exported #{theses.count} theses to: #{output_file}"
     else
       puts "No published theses found for #{args.term}"
diff --git a/test/mailers/batch_mailer_test.rb b/test/mailers/batch_mailer_test.rb
index 9427d796..5f0cbf05 100644
--- a/test/mailers/batch_mailer_test.rb
+++ b/test/mailers/batch_mailer_test.rb
@@ -5,7 +5,7 @@ class BatchMailerTest < ActionMailer::TestCase
     ClimateControl.modify DISABLE_ALL_EMAIL: 'false' do
       theses = [theses(:one), theses(:two)]
       marc_zip_file = MarcBatch.new(theses, 'marc.xml', 'marc.zip').build
-      json_file = JsonBatch.new(theses, 'marc.json').build
+      json_file = CatalogBatch.new(theses, 'marc.json').build
       email = BatchMailer.marc_batch_email('marc.zip', marc_zip_file, 'marc.json', json_file, theses)
 
       # Send the email, then test that it got queued
@@ -29,7 +29,7 @@ class BatchMailerTest < ActionMailer::TestCase
     ClimateControl.modify DISABLE_ALL_EMAIL: 'false' do
       theses = [theses(:one), theses(:two)]
       marc_zip_file = MarcBatch.new(theses, 'marc.xml', 'marc.zip').build
-      json_file = JsonBatch.new(theses, 'marc.json').build
+      json_file = CatalogBatch.new(theses, 'marc.json').build
       email = BatchMailer.marc_batch_email('marc.zip', marc_zip_file, 'marc.json', json_file, theses)
       attachment = email.attachments['marc.zip']
       assert_equal 'application/zip; filename=marc.zip', attachment.content_type
diff --git a/test/models/json_batch_test.rb b/test/models/catalog_batch_test.rb
similarity index 85%
rename from test/models/json_batch_test.rb
rename to test/models/catalog_batch_test.rb
index 2e5b12a5..5ad8587c 100644
--- a/test/models/json_batch_test.rb
+++ b/test/models/catalog_batch_test.rb
@@ -1,9 +1,9 @@
 require 'test_helper'
 
-class JsonBatchTest < ActiveSupport::TestCase
+class CatalogBatchTest < ActiveSupport::TestCase
   test 'builds a valid JSON file' do
     theses = [theses(:published)]
-    batch = JsonBatch.new(theses, 'test.json')
+    batch = CatalogBatch.new(theses, 'test.json')
     json_file = batch.build
 
     json_content = File.read(json_file.path)
@@ -15,7 +15,7 @@ class JsonBatchTest < ActiveSupport::TestCase
 
   test 'wraps theses in a wrapper object with theses key' do
     theses = [theses(:published)]
-    batch = JsonBatch.new(theses, 'test.json')
+    batch = CatalogBatch.new(theses, 'test.json')
     json_file = batch.build
 
     json_content = File.read(json_file.path)
@@ -28,7 +28,7 @@ class JsonBatchTest < ActiveSupport::TestCase
 
   test 'includes all theses in the batch' do
     theses = [theses(:published), theses(:one)]
-    batch = JsonBatch.new(theses, 'test.json')
+    batch = CatalogBatch.new(theses, 'test.json')
     json_file = batch.build
 
     json_content = File.read(json_file.path)
@@ -40,7 +40,7 @@ class JsonBatchTest < ActiveSupport::TestCase
 
   test 'includes all required fields' do
     theses = [theses(:published)]
-    batch = JsonBatch.new(theses, 'test.json')
+    batch = CatalogBatch.new(theses, 'test.json')
     json_file = batch.build
 
     json_content = File.read(json_file.path)
@@ -61,7 +61,7 @@ class JsonBatchTest < ActiveSupport::TestCase
   end
 
   test 'empty theses array produces valid JSON' do
-    batch = JsonBatch.new([], 'test.json')
+    batch = CatalogBatch.new([], 'test.json')
     json_file = batch.build
 
     json_content = File.read(json_file.path)
diff --git a/test/models/json_exporter_test.rb b/test/models/catalog_exporter_test.rb
similarity index 82%
rename from test/models/json_exporter_test.rb
rename to test/models/catalog_exporter_test.rb
index 38687f99..5b631f46 100644
--- a/test/models/json_exporter_test.rb
+++ b/test/models/catalog_exporter_test.rb
@@ -1,30 +1,30 @@
 require 'test_helper'
 
-class JsonExporterTest < ActiveSupport::TestCase
+class CatalogExporterTest < ActiveSupport::TestCase
   test 'includes correctly formatted title' do
     thesis = theses(:published)
-    exporter = JsonExporter.new(thesis)
+    exporter = CatalogExporter.new(thesis)
     json_hash = exporter.to_hash
     assert_equal thesis.title&.squish, json_hash[:title]
   end
 
   test 'includes abstract' do
     thesis = theses(:published)
-    exporter = JsonExporter.new(thesis)
+    exporter = CatalogExporter.new(thesis)
     json_hash = exporter.to_hash
     assert_equal thesis.abstract, json_hash[:abstract]
   end
 
   test 'includes grad year' do
     thesis = theses(:published)
-    exporter = JsonExporter.new(thesis)
+    exporter = CatalogExporter.new(thesis)
     json_hash = exporter.to_hash
     assert_equal thesis.graduation_year, json_hash[:graduation_year]
   end
 
   test 'includes correctly formatted dspace_url' do
     thesis = theses(:published)
-    exporter = JsonExporter.new(thesis)
+    exporter = CatalogExporter.new(thesis)
     json_hash = exporter.to_hash
     expected_url = "https://dspace.mit.edu/handle/#{thesis.dspace_handle}"
     assert_equal expected_url, json_hash[:dspace_url]
@@ -32,7 +32,7 @@ class JsonExporterTest < ActiveSupport::TestCase
 
   test 'includes authors with nested structure' do
     thesis = theses(:published)
-    exporter = JsonExporter.new(thesis)
+    exporter = CatalogExporter.new(thesis)
     json_hash = exporter.to_hash
     assert json_hash[:authors].is_a?(Array)
     assert json_hash[:authors].first.is_a?(Hash)
@@ -41,7 +41,7 @@ class JsonExporterTest < ActiveSupport::TestCase
 
   test 'includes degrees with nested structure' do
     thesis = theses(:published)
-    exporter = JsonExporter.new(thesis)
+    exporter = CatalogExporter.new(thesis)
     json_hash = exporter.to_hash
     assert json_hash[:degrees].is_a?(Array)
     assert json_hash[:degrees].first.is_a?(Hash)
@@ -51,7 +51,7 @@ class JsonExporterTest < ActiveSupport::TestCase
   test 'includes advisors with nested structure' do
     thesis = theses(:published)
     thesis.advisors << advisors(:first)
-    exporter = JsonExporter.new(thesis)
+    exporter = CatalogExporter.new(thesis)
     json_hash = exporter.to_hash
     assert json_hash[:advisors].is_a?(Array)
     assert json_hash[:advisors].first.is_a?(Hash)
@@ -60,7 +60,7 @@ class JsonExporterTest < ActiveSupport::TestCase
 
   test 'includes departments with nested structure' do
     thesis = theses(:published)
-    exporter = JsonExporter.new(thesis)
+    exporter = CatalogExporter.new(thesis)
     json_hash = exporter.to_hash
     assert json_hash[:departments].is_a?(Array)
     assert json_hash[:departments].first.is_a?(Hash)

From d13b1d81e38016b351272e11a7c7c09f512ad86f Mon Sep 17 00:00:00 2001
From: jazairi <16103405+jazairi@users.noreply.github.com>
Date: Thu, 6 Aug 2026 08:11:20 -0700
Subject: [PATCH 4/6] Rename instances of json_file to catalog_file for clarity

---
 app/jobs/marc_export_job.rb       |  8 ++++----
 app/models/catalog_batch.rb       | 22 +++++++++++-----------
 lib/tasks/metadata.rake           |  8 ++++----
 test/mailers/batch_mailer_test.rb |  8 ++++----
 test/models/catalog_batch_test.rb | 30 +++++++++++++++---------------
 5 files changed, 38 insertions(+), 38 deletions(-)

diff --git a/app/jobs/marc_export_job.rb b/app/jobs/marc_export_job.rb
index cb0ef025..86e9b942 100644
--- a/app/jobs/marc_export_job.rb
+++ b/app/jobs/marc_export_job.rb
@@ -4,15 +4,15 @@ class MarcExportJob < ActiveJob::Base
   def perform(theses)
     marc_filename = "#{filename}.mrc"
     zip_filename = "#{filename}.zip"
-    json_filename = "#{filename}.json"
+    catalog_filename = "#{filename}.json"
 
     begin
       marc_zip_file = MarcBatch.new(theses, marc_filename, zip_filename).build
-      json_file = CatalogBatch.new(theses, json_filename).build
-      BatchMailer.marc_batch_email(zip_filename, marc_zip_file, json_filename, json_file, theses).deliver_now
+      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
       marc_zip_file&.close!
-      json_file&.close!
+      catalog_file&.close!
     end
   end
 
diff --git a/app/models/catalog_batch.rb b/app/models/catalog_batch.rb
index e219d810..8d74cf2e 100644
--- a/app/models/catalog_batch.rb
+++ b/app/models/catalog_batch.rb
@@ -5,33 +5,33 @@
 #
 # Example:
 #   batch = CatalogBatch.new(theses_array, 'export.json')
-#   json_file = batch.build
-#   File.write('export.json', File.read(json_file.path))
-#   json_file.close!  # Clean up tempfile
+#   catalog_file = batch.build
+#   File.write('export.json', File.read(catalog_file.path))
+#   catalog_file.close!  # Clean up tempfile
 class CatalogBatch
-  def initialize(theses, json_filename)
+  def initialize(theses, filename)
     @theses = theses
-    @json_filename = json_filename
+    @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
-    json_file = Tempfile.new(@json_filename)
-    write_json_file(json_file)
-    json_file
+    catalog_file = Tempfile.new(@filename)
+    write_catalog_file(catalog_file)
+    catalog_file
   end
 
   private
 
-  def write_json_file(json_file)
+  def write_catalog_file(catalog_file)
     theses_data = @theses.map do |thesis|
       CatalogExporter.new(thesis).to_hash
     end
 
     json_output = { theses: theses_data }
 
-    json_file.write(JSON.pretty_generate(json_output))
-    json_file.rewind
+    catalog_file.write(JSON.pretty_generate(json_output))
+    catalog_file.rewind
   end
 end
diff --git a/lib/tasks/metadata.rake b/lib/tasks/metadata.rake
index e352563e..4c35e69e 100644
--- a/lib/tasks/metadata.rake
+++ b/lib/tasks/metadata.rake
@@ -35,10 +35,10 @@ namespace :metadata do
     theses = Thesis.published.where(grad_date: query_date.all_month)
 
     if theses.any?
-      json_batch = CatalogBatch.new(theses, File.basename(output_file))
-      json_file = json_batch.build
-      FileUtils.cp(json_file.path, output_file)
-      json_file.close!
+      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}"
diff --git a/test/mailers/batch_mailer_test.rb b/test/mailers/batch_mailer_test.rb
index 5f0cbf05..6d2d019f 100644
--- a/test/mailers/batch_mailer_test.rb
+++ b/test/mailers/batch_mailer_test.rb
@@ -5,8 +5,8 @@ class BatchMailerTest < ActionMailer::TestCase
     ClimateControl.modify DISABLE_ALL_EMAIL: 'false' do
       theses = [theses(:one), theses(:two)]
       marc_zip_file = MarcBatch.new(theses, 'marc.xml', 'marc.zip').build
-      json_file = CatalogBatch.new(theses, 'marc.json').build
-      email = BatchMailer.marc_batch_email('marc.zip', marc_zip_file, 'marc.json', json_file, theses)
+      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
@@ -29,8 +29,8 @@ class BatchMailerTest < ActionMailer::TestCase
     ClimateControl.modify DISABLE_ALL_EMAIL: 'false' do
       theses = [theses(:one), theses(:two)]
       marc_zip_file = MarcBatch.new(theses, 'marc.xml', 'marc.zip').build
-      json_file = CatalogBatch.new(theses, 'marc.json').build
-      email = BatchMailer.marc_batch_email('marc.zip', marc_zip_file, 'marc.json', json_file, theses)
+      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
diff --git a/test/models/catalog_batch_test.rb b/test/models/catalog_batch_test.rb
index 5ad8587c..4c6917cb 100644
--- a/test/models/catalog_batch_test.rb
+++ b/test/models/catalog_batch_test.rb
@@ -4,46 +4,46 @@ class CatalogBatchTest < ActiveSupport::TestCase
   test 'builds a valid JSON file' do
     theses = [theses(:published)]
     batch = CatalogBatch.new(theses, 'test.json')
-    json_file = batch.build
+    catalog_file = batch.build
 
-    json_content = File.read(json_file.path)
+    json_content = File.read(catalog_file.path)
     json_data = JSON.parse(json_content)
 
     assert_not_nil(json_data)
-    json_file.close
+    catalog_file.close
   end
 
   test 'wraps theses in a wrapper object with theses key' do
     theses = [theses(:published)]
     batch = CatalogBatch.new(theses, 'test.json')
-    json_file = batch.build
+    catalog_file = batch.build
 
-    json_content = File.read(json_file.path)
+    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))
-    json_file.close
+    catalog_file.close
   end
 
   test 'includes all theses in the batch' do
     theses = [theses(:published), theses(:one)]
     batch = CatalogBatch.new(theses, 'test.json')
-    json_file = batch.build
+    catalog_file = batch.build
 
-    json_content = File.read(json_file.path)
+    json_content = File.read(catalog_file.path)
     json_data = JSON.parse(json_content)
 
     assert_equal(2, json_data['theses'].count)
-    json_file.close
+    catalog_file.close
   end
 
   test 'includes all required fields' do
     theses = [theses(:published)]
     batch = CatalogBatch.new(theses, 'test.json')
-    json_file = batch.build
+    catalog_file = batch.build
 
-    json_content = File.read(json_file.path)
+    json_content = File.read(catalog_file.path)
     json_data = JSON.parse(json_content)
 
     thesis_data = json_data['theses'].first
@@ -57,17 +57,17 @@ class CatalogBatchTest < ActiveSupport::TestCase
     assert(thesis_data.key?('graduation_year'))
     assert(thesis_data.key?('title'))
 
-    json_file.close
+    catalog_file.close
   end
 
   test 'empty theses array produces valid JSON' do
     batch = CatalogBatch.new([], 'test.json')
-    json_file = batch.build
+    catalog_file = batch.build
 
-    json_content = File.read(json_file.path)
+    json_content = File.read(catalog_file.path)
     json_data = JSON.parse(json_content)
 
     assert_equal(0, json_data['theses'].count)
-    json_file.close
+    catalog_file.close
   end
 end

From e6bf2ceae7275c5ef95d4e876f84c44ba40848e1 Mon Sep 17 00:00:00 2001
From: jazairi <16103405+jazairi@users.noreply.github.com>
Date: Thu, 6 Aug 2026 08:29:49 -0700
Subject: [PATCH 5/6] Clean up metadata rake tasks

Update task/variable names and add docstring to
recommend batch task for local development only
---
 lib/tasks/metadata.rake | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/lib/tasks/metadata.rake b/lib/tasks/metadata.rake
index 4c35e69e..64f385cd 100644
--- a/lib/tasks/metadata.rake
+++ b/lib/tasks/metadata.rake
@@ -1,6 +1,6 @@
 namespace :metadata do
-  desc 'Generate a JSON export of a single published thesis for debugging'
-  task :json_export_thesis, [:thesis_id] => :environment do |_t, args|
+  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
@@ -9,20 +9,22 @@ namespace :metadata do
     thesis = Thesis.find(args.thesis_id)
 
     if thesis.publication_status == 'Published'
-      json_exporter = CatalogExporter.new(thesis)
-      json_data = json_exporter.to_hash
+      catalog_exporter = CatalogExporter.new(thesis)
+      json_data = catalog_exporter.to_hash
 
-      puts "JSON Export for Thesis #{args.thesis_id}:"
+      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
 
-  desc 'Generate a JSON export batch for a specific term (e.g., "2024-June") and save to temp file'
-  task :json_export_batch, %i[term output_file] => :environment do |_t, args|
+  # 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:json_export_batch["2024-June","output.json"]'
+      puts 'Usage: rake metadata:catalog_export_batch["2024-June","output.json"]'
       puts 'Term format: YYYY-Month (e.g., 2024-June, 2024-September)'
       next
     end
@@ -30,7 +32,7 @@ namespace :metadata do
     year, month_name = args.term.split('-')
     query_date = Date.parse("1 #{month_name} #{year}")
 
-    output_file = args.output_file || Rails.root.join("tmp/json_export_#{args.term}_#{DateTime.now.utc.strftime('%H_%M')}.json").to_s
+    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)
 

From 43ad3a46e709bcdcef7a4ea1d4496a31da7a4993 Mon Sep 17 00:00:00 2001
From: jazairi <16103405+jazairi@users.noreply.github.com>
Date: Thu, 6 Aug 2026 08:50:05 -0700
Subject: [PATCH 6/6] Confirm correct format of titles in catalog exporter test

---
 test/models/catalog_exporter_test.rb | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/test/models/catalog_exporter_test.rb b/test/models/catalog_exporter_test.rb
index 5b631f46..2903d910 100644
--- a/test/models/catalog_exporter_test.rb
+++ b/test/models/catalog_exporter_test.rb
@@ -3,9 +3,11 @@
 class CatalogExporterTest < ActiveSupport::TestCase
   test 'includes correctly formatted title' do
     thesis = theses(:published)
+    thesis.title = "   A    weirdly \n spaced    title    "
+    thesis.save
     exporter = CatalogExporter.new(thesis)
     json_hash = exporter.to_hash
-    assert_equal thesis.title&.squish, json_hash[:title]
+    assert_equal "A weirdly spaced title", json_hash[:title]
   end
 
   test 'includes abstract' do