From 0f27e3b1ed35403e9887ff08dedbe7d2f724458d Mon Sep 17 00:00:00 2001
From: Benjamin Kiah Stroud <32469930+bkiahstroud@users.noreply.github.com>
Date: Mon, 11 Nov 2024 17:01:23 -0800
Subject: [PATCH] new account setting to override Bulkrax's split pattern

Add a new Account setting that will override Bulkrax's default
multi-value split pattern. This effectively allows each tenant to have
its own default split pattern

This change relies on this Bulkrax commit to work:
- https://github.com/samvera/bulkrax/pull/984/commits/9268bbb
---
 app/models/concerns/account_settings.rb             | 13 ++++++++++++-
 app/parsers/bulkrax/application_parser_decorator.rb | 13 +++++++++++++
 config/locales/simple_form.en.yml                   |  2 +-
 3 files changed, 26 insertions(+), 2 deletions(-)
 create mode 100644 app/parsers/bulkrax/application_parser_decorator.rb

diff --git a/app/models/concerns/account_settings.rb b/app/models/concerns/account_settings.rb
index a18c066a5..3be24012a 100644
--- a/app/models/concerns/account_settings.rb
+++ b/app/models/concerns/account_settings.rb
@@ -23,6 +23,7 @@ module AccountSettings
     setting :allow_downloads, type: 'boolean', default: true
     setting :allow_signup, type: 'boolean', default: true
     setting :analytics_provider, type: 'string'
+    setting :bulkrax_split_pattern, type: 'string', default: Bulkrax.multi_value_element_split_on.source
     setting :bulkrax_validations, type: 'boolean', disabled: true
     setting :cache_api, type: 'boolean', default: false
     setting :contact_email, type: 'string', default: 'change-me-in-settings@example.com'
@@ -58,7 +59,7 @@ module AccountSettings
     validates :contact_email, :oai_admin_email,
               format: { with: URI::MailTo::EMAIL_REGEXP },
               allow_blank: true
-    validate :validate_email_format, :validate_contact_emails
+    validate :validate_email_format, :validate_contact_emails, :validate_split_pattern_regex
     validates :google_analytics_id,
               format: { with: /((UA|YT|MO)-\d+-\d+|G-[A-Z0-9]{10})/i },
               allow_blank: true
@@ -155,6 +156,16 @@ def validate_contact_emails
     end
   end
 
+  def validate_split_pattern_regex
+    return if settings['bulkrax_split_pattern'].blank?
+
+    begin
+      Regexp.new(settings['bulkrax_split_pattern'])
+    rescue RegexpError => e
+      errors.add(:bulkrax_split_pattern, e.message)
+    end
+  end
+
   def initialize_settings
     return true unless self.class.column_names.include?('settings')
     set_smtp_settings
diff --git a/app/parsers/bulkrax/application_parser_decorator.rb b/app/parsers/bulkrax/application_parser_decorator.rb
new file mode 100644
index 000000000..2d34104d3
--- /dev/null
+++ b/app/parsers/bulkrax/application_parser_decorator.rb
@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Bulkrax
+  module ApplicationParserDecorator
+    # OVERRIDE: [Bulkrax v8.3.0] Use the current Account's configured :bulkrax_split_pattern
+    # setting as the default split pattern when importing multi-valued fields
+    def multi_value_element_split_on
+      Regexp.new(Site.account&.settings&.dig('bulkrax_split_pattern')) || super
+    end
+  end
+end
+
+Bulkrax::ApplicationParser.singleton_class.send(:prepend, Bulkrax::ApplicationParserDecorator)
diff --git a/config/locales/simple_form.en.yml b/config/locales/simple_form.en.yml
index 31ee36693..2c47ad594 100644
--- a/config/locales/simple_form.en.yml
+++ b/config/locales/simple_form.en.yml
@@ -14,7 +14,6 @@ en:
         name: A single or hyphenated name used for technical aspects of the repository (e.g., "acme" or "acme-library").
         gtm_id: The ID of your Google Tag Manager account
         google_analytics_id: The ID of your Google Analytics account
-        contact_email_to: The email address that messages submitted via the contact page are sent to
         weekly_email_list: List of email addresses to email the weekly report. Leave a single space between each email
         monthly_email_list: List of email addresses to email the monthly report. Leave a single space between each email
         yearly_email_list: List of email addresses to email the yearly report. Leave a single space between each email
@@ -35,6 +34,7 @@ en:
         geonames_username: Register at http://www.geonames.org/manageaccount
         file_acl: Turn off if using a file system like samba or nfs that does not support setting access control lists
         ssl_configured: Set it true if using https
+        bulkrax_split_pattern: Default regular expression used by Bulkrax to split multiple values within the same column. <b>Omit surrounding slashes (e.g. [|;] not /[|;]/)</b>
       hyku_group:
         description: A brief summary of the role of the group
       user: