From f6ce7ab99379f03dd0cd00257a5211dc5d0bb538 Mon Sep 17 00:00:00 2001
From: Eddie Rubeiz <erubeiz@sciencehistory.org>
Date: Wed, 18 Dec 2024 15:30:53 -0500
Subject: [PATCH] using a text field for box and folder

---
 app/controllers/catalog_controller.rb | 17 +++++++++++++++++
 app/indexers/work_indexer.rb          |  6 ++++++
 solr/config/schema.xml                |  6 ++++++
 spec/indexers/work_indexer_spec.rb    | 26 ++++++++++++++++++++++++++
 4 files changed, 55 insertions(+)

diff --git a/app/controllers/catalog_controller.rb b/app/controllers/catalog_controller.rb
index d56eb8737..799202132 100644
--- a/app/controllers/catalog_controller.rb
+++ b/app/controllers/catalog_controller.rb
@@ -361,6 +361,23 @@ def active_sort_fields
     # case for a BL "search field", which is really a dismax aggregate
     # of Solr search fields.
 
+
+    # See https://github.com/sciencehistory/scihist_digicoll/issues/2585
+    # Note that a Work can be associated with two consecutive boxes, in which case the box number is something like "34-35".
+    config.add_search_field('box') do |field|
+      field.label = 'Box'
+      field.solr_parameters = {
+        qf: 'box_tsi',
+      }
+    end
+
+    config.add_search_field('folder') do |field|
+      field.label = 'Folder'
+      field.solr_parameters = {
+        qf: 'folder_tsi',
+      }
+    end
+
     # config.add_search_field('title') do |field|
     #   # solr_parameters hash are sent to Solr as ordinary url query params.
     #   field.solr_parameters = {
diff --git a/app/indexers/work_indexer.rb b/app/indexers/work_indexer.rb
index 9d6b56ca6..bea23b3c1 100644
--- a/app/indexers/work_indexer.rb
+++ b/app/indexers/work_indexer.rb
@@ -61,6 +61,12 @@ class WorkIndexer < Kithe::Indexer
       acc.concat(DateIndexHelper.new(record).expanded_years)
     end
 
+    to_field "box_tsi",     obj_extract("physical_container"), transform( ->(v) { v.box    if v.box.present?    })
+    to_field "folder_tsi",  obj_extract("physical_container"), transform( ->(v) { v.folder if v.folder.present? })
+    to_field "box_sort",    obj_extract("physical_container"), transform( ->(v) { v.box[/\d+/]   if v.box.present?    })
+    to_field "folder_sort", obj_extract("physical_container"), transform( ->(v) { v.folder[/\d+/] if v.folder.present? })
+
+
     # For sorting by oldest first
     to_field "earliest_date" do |record, acc|
       # for Solr, we need in "xml schema" format, with 00:00:00 time, and UTC timezone
diff --git a/solr/config/schema.xml b/solr/config/schema.xml
index 0f5166959..2a8d48676 100644
--- a/solr/config/schema.xml
+++ b/solr/config/schema.xml
@@ -257,6 +257,12 @@
     <field name="latest_date" type="date_sortmissinglast" stored="true" indexed="true" multiValued="false"/>
     <field name="earliest_date" type="date_sortmissinglast" stored="true" indexed="true" multiValued="false"/>
 
+
+    <!-- Box and folder sort fields. Added by Science History Institute. -->
+    <field name="box_sort"    type="int_sortmissinglast" stored="true" indexed="true" multiValued="false"/>
+    <field name="folder_sort" type="int_sortmissinglast" stored="true" indexed="true" multiValued="false"/>
+
+
     <!-- Single-value title sort field. Used to sort collections of serials by title.
     See https://github.com/sciencehistory/scihist_digicoll/issues/2494
     If we can use the first value of (multivalued) text1_tesim instead, let's consider doing that.
diff --git a/spec/indexers/work_indexer_spec.rb b/spec/indexers/work_indexer_spec.rb
index c7d11db2f..63caec49b 100644
--- a/spec/indexers/work_indexer_spec.rb
+++ b/spec/indexers/work_indexer_spec.rb
@@ -41,6 +41,32 @@
     end
   end
 
+  # See https://github.com/sciencehistory/scihist_digicoll/issues/2585
+  describe "box and folder" do
+    let(:box_search_field) {'box_tsi'}
+    let(:box_sort_field)  {'box_sort'}
+    let(:folder_search_field) {'folder_tsi'}
+    let(:folder_sort_field)   {'folder_sort'}
+
+    let(:work_2) { create(:work, physical_container: Work::PhysicalContainer.new({"box"=>"1", "folder"=>"3"})) }
+    let(:work_3) { create(:work, physical_container: Work::PhysicalContainer.new({"box"=>"12, 34, 56", "folder"=>"56, 78, 10"})) }
+    let(:output_hash)   { WorkIndexer.new.map_record(work) }
+    let(:output_hash_2) { WorkIndexer.new.map_record(work_2) }
+    let(:output_hash_3) { WorkIndexer.new.map_record(work_3) }
+
+    it "puts the first consecutive digits, if found, into the sort fields, and everything into the search fields" do
+      expect(output_hash[box_search_field]).to eq ['Box']
+      expect(output_hash[folder_search_field]).to eq ['Folder']
+      expect(output_hash[box_sort_field]).to be_nil
+      expect(output_hash[folder_sort_field]).to be_nil
+
+      expect(output_hash_3[box_search_field]).to eq ['12, 34, 56']
+      expect(output_hash_3[folder_search_field]).to eq ['56, 78, 10']
+      expect(output_hash_3[box_sort_field]).to eq ['12']
+      expect(output_hash_3[folder_sort_field]).to eq ['56']
+    end
+  end
+
   describe "oral history" do
     let(:work) { create(:oral_history_work, format: ['text']) }