add spec, clarify code matching in comments

lib/dor/text_extraction/cocina_updater.rb

@@ -150,9 +150,12 @@ def find_resource(path)
         end
       end
 
-      # Find a Cocina resource for the given path by looking for a resource that has a matching filename "stem"
+      # Find a Cocina resource for the given file path of new files in the workspace by
+      # looking for a resource that has a matching filename "stem"
       # OR a matching filename with an extension added to the end of the stem,
-      # e.g. "file1_tiff.xml" would be added to the same resource that contains "file1.tiff"
+      # e.g. "file1.xml" would be added to the same resource as "file1.tiff"
+      # OR "file1_tiff.xml" would be added to the same resource that contains "file1.tiff" if a resource with "file1.xml" does not exist
+      # Nothing is done if neither the stem nor the stem with extension is found in the Cocina
       # see https://github.com/sul-dlss/common-accessioning/issues/1443
       # @param path {String} - the path to look for
       # @return {FileSet, nil} - the resource or nil if it is not found
@@ -250,9 +253,15 @@ def stem(path)
       end
 
       # get the full filename of a file path by extracting the extension from the last part of the filename (e.g. '/user/path/file1_mp4.vtt' => 'file1.mp4')
+      # note that this assumes that the extension is the last part of the filename after the last underscore even if it is not a valid extension
+      # while this could theoretically cause us to match the file to the wrong resource, it is unlikely to happen in practice since we are
+      # (1) first looking for exact filename matches and (2) only iterating and adding files in the workspace that were generated by OCR or speech to text
       def extracted_filename_with_extension(path)
         basename_with_extension_stem = stem(path)
-        extension_from_filename = basename_with_extension_stem.include?('_') ? basename_with_extension_stem.split('_').last : ''
+
+        return nil unless basename_with_extension_stem.include?('_')
+
+        extension_from_filename = basename_with_extension_stem.split('_').last
         basename_without_extension_stem = basename_with_extension_stem.gsub("_#{extension_from_filename}", '')
         "#{basename_without_extension_stem}.#{extension_from_filename}"
       end

spec/lib/dor/text_extraction/cocina_updater_spec.rb

@@ -0,0 +1,82 @@
+# frozen_string_literal: true
+
+require 'spec_helper'
+
+RSpec.describe Dor::TextExtraction::CocinaUpdater do
+  subject(:updater) { described_class.new(dro:, logger:) }
+
+  let(:druid) { 'druid:bc123df4567' }
+  let(:object_type) { Cocina::Models::ObjectType.image }
+  let(:dro) { instance_double(Cocina::Models::DRO, externalIdentifier: druid, dro?: true, version: 1, type: object_type, structural:) }
+  let(:logger) { instance_double(Logger) }
+  let(:path) { "/some/server/path/#{filename}" }
+  let(:tiff_file) { instance_double(Cocina::Models::File, filename: 'file1.tiff') }
+  let(:other_tiff_file) { instance_double(Cocina::Models::File, filename: 'file2.tiff') }
+  let(:tiff_fileset) { instance_double(Cocina::Models::FileSet, structural: tiff_fileset_structural) }
+  let(:other_tiff_fileset) { instance_double(Cocina::Models::FileSet, structural: other_tiff_fileset_structural) }
+  let(:tiff_fileset_structural) { instance_double(Cocina::Models::FileSetStructural, contains: [tiff_file]) }
+  let(:other_tiff_fileset_structural) { instance_double(Cocina::Models::FileSetStructural, contains: [other_tiff_file]) }
+  let(:structural) { instance_double(Cocina::Models::DROStructural, contains: [tiff_fileset, other_tiff_fileset]) }
+
+  describe '#find_resource_with_stem' do
+    context 'when file has matching stem' do
+      let(:filename) { 'file1.xml' }
+
+      it 'finds resource with matching stem' do
+        expect(updater.send(:find_resource_with_stem, path)).to eq tiff_fileset
+      end
+    end
+
+    context 'when file has extension in stem' do
+      let(:filename) { 'file1_tiff.xml' }
+
+      it 'finds resource with matching filename' do
+        expect(updater.send(:find_resource_with_stem, path)).to eq tiff_fileset
+      end
+    end
+
+    context 'when no matching stem exists' do
+      let(:filename) { 'file3.xml' }
+
+      it 'returns nil' do
+        expect(updater.send(:find_resource_with_stem, path)).to be_nil
+      end
+    end
+  end
+
+  describe '#extracted_filename_with_extension' do
+    context 'when no underscores exist' do
+      let(:filename) { 'file1.vtt' }
+
+      it 'returns nil' do
+        expect(updater.send(:extracted_filename_with_extension, path)).to be_nil
+      end
+    end
+
+    context 'when the file extension is included in the filename' do
+      let(:filename) { 'file1_tiff.vtt' }
+
+      it 'extracts filename with original extension' do
+        expect(updater.send(:extracted_filename_with_extension, path)).to eq 'file1.tiff'
+      end
+    end
+
+    context 'when the file extension is included in the filename and there are other underscors in the filename too' do
+      let(:filename) { 'file1_cool_file_tiff.vtt' }
+
+      it 'extracts filename with original extension' do
+        expect(updater.send(:extracted_filename_with_extension, path)).to eq 'file1_cool_file.tiff'
+      end
+    end
+
+    context 'when there are underscores in the filename but none are the extension' do
+      let(:filename) { 'file1_cool_file.vtt' }
+
+      # NOTE: this will not throw an exception, and in theory could end up inadventantely matching a file in a resource
+      # that it should not, but this is a rare edge case ... it'll probably not match anything and be skipped
+      it 'extracts a filename with something it thinks is an extension, but it actually is not' do
+        expect(updater.send(:extracted_filename_with_extension, path)).to eq 'file1_cool.file'
+      end
+    end
+  end
+end