From 9268bbb262aaafee7fb098b36fd362044d6eccbc Mon Sep 17 00:00:00 2001
From: Benjamin Kiah Stroud <32469930+bkiahstroud@users.noreply.github.com>
Date: Mon, 11 Nov 2024 16:26:44 -0800
Subject: [PATCH] split pattern is accessed from the parser instead of config

Transfer responsibility of accessing the split pattern from the config
to the ApplicationParser. This enables the default split pattern to be
overwritten

Hyku will soon get an `Account` setting that allows for per-tenant
overrides for the default split pattern. This change enables that
feature

The default value is still set in the config. Existing installations
should see no difference in behavior; this change is
backwards-compatible
---
 app/matchers/bulkrax/application_matcher.rb | 2 +-
 app/models/bulkrax/csv_entry.rb             | 4 ++--
 app/parsers/bulkrax/application_parser.rb   | 4 ++++
 app/parsers/bulkrax/csv_parser.rb           | 4 ++--
 4 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/app/matchers/bulkrax/application_matcher.rb b/app/matchers/bulkrax/application_matcher.rb
index 0f7c7cca..1d4dc3c0 100644
--- a/app/matchers/bulkrax/application_matcher.rb
+++ b/app/matchers/bulkrax/application_matcher.rb
@@ -33,7 +33,7 @@ def result(_parser, content)
 
     def process_split
       if self.split.is_a?(TrueClass)
-        @result = @result.split(Bulkrax.multi_value_element_split_on)
+        @result = @result.split(ApplicationParser.multi_value_element_split_on)
       elsif self.split
         @result = @result.split(Regexp.new(self.split))
         @result = @result.map(&:strip).select(&:present?)
diff --git a/app/models/bulkrax/csv_entry.rb b/app/models/bulkrax/csv_entry.rb
index 65666890..f829d71e 100644
--- a/app/models/bulkrax/csv_entry.rb
+++ b/app/models/bulkrax/csv_entry.rb
@@ -129,7 +129,7 @@ def add_ingested_metadata
     def add_file
       self.parsed_metadata['file'] ||= []
       if record['file']&.is_a?(String)
-        self.parsed_metadata['file'] = record['file'].split(Bulkrax.multi_value_element_split_on)
+        self.parsed_metadata['file'] = record['file'].split(parser.multi_value_element_split_on)
       elsif record['file'].is_a?(Array)
         self.parsed_metadata['file'] = record['file']
       end
@@ -350,7 +350,7 @@ def collection_identifiers
       return [] unless parent_field_mapping.present? && record[parent_field_mapping].present?
 
       identifiers = []
-      split_references = record[parent_field_mapping].split(Bulkrax.multi_value_element_split_on)
+      split_references = record[parent_field_mapping].split(parser.multi_value_element_split_on)
       split_references.each do |c_reference|
         matching_collection_entries = importerexporter.entries.select do |e|
           (e.raw_metadata&.[](source_identifier) == c_reference) &&
diff --git a/app/parsers/bulkrax/application_parser.rb b/app/parsers/bulkrax/application_parser.rb
index 161a6740..e529dd30 100644
--- a/app/parsers/bulkrax/application_parser.rb
+++ b/app/parsers/bulkrax/application_parser.rb
@@ -34,6 +34,10 @@ def self.import_supported?
       true
     end
 
+    def self.multi_value_element_split_on
+      Bulkrax.multi_value_element_split_on
+    end
+
     def initialize(importerexporter)
       @importerexporter = importerexporter
       @headers = []
diff --git a/app/parsers/bulkrax/csv_parser.rb b/app/parsers/bulkrax/csv_parser.rb
index 03159ebb..d11afc35 100644
--- a/app/parsers/bulkrax/csv_parser.rb
+++ b/app/parsers/bulkrax/csv_parser.rb
@@ -338,7 +338,7 @@ def file_paths
         file_mapping = Bulkrax.field_mappings.dig(self.class.to_s, 'file', :from)&.first&.to_sym || :file
         next if r[file_mapping].blank?
 
-        r[file_mapping].split(Bulkrax.multi_value_element_split_on).map do |f|
+        r[file_mapping].split(multi_value_element_split_on).map do |f|
           file = File.join(path_to_files, f.tr(' ', '_'))
           if File.exist?(file) # rubocop:disable Style/GuardClause
             file
@@ -366,7 +366,7 @@ def unique_collection_identifier(collection_hash)
       entry_uid ||= if Bulkrax.fill_in_blank_source_identifiers.present?
                       Bulkrax.fill_in_blank_source_identifiers.call(self, records.find_index(collection_hash))
                     else
-                      collection_hash[:title].split(Bulkrax.multi_value_element_split_on).first
+                      collection_hash[:title].split(multi_value_element_split_on).first
                     end
 
       entry_uid