Merge pull request pulibrary#661 from pulibrary/repo-ocr

Adding OCR to the repository

app/services/ocr_runner.rb

@@ -11,13 +11,22 @@ def from_file(filename)
 
   def from_datastream
     Hydra::Derivatives::TempfileService.create(resource.original_file) do |f|
-      from_file(f.path)
+      ocr_output = from_file(f.path)
+      attach_ocr(ocr_filename(ocr_output))
     end
-    resource.update_index
+    resource.save
   end
 
   private
 
+    def attach_ocr(filename)
+      Hydra::Works::AddFileToFileSet.call(resource, File.open(filename), :extracted_text)
+    end
+
+    def ocr_filename(ocr_output)
+      ocr_output.first[:url].sub(/^file:/, '')
+    end
+
     def creator_factory
       OCRCreator
     end

spec/models/file_set_spec.rb

@@ -44,7 +44,7 @@
 
       expect(path).to exist
     end
-    it "creates full text and indexes it" do
+    it "creates full text, attaches it to the object, and indexes it" do
       allow_any_instance_of(described_class).to receive(:warn) # suppress virus check warnings
       allow(Hydra::Derivatives::Jpeg2kImageDerivatives).to receive(:create).and_return(true)
       file = File.open(Rails.root.join("spec", "fixtures", "files", "page18.tif"))
@@ -55,6 +55,11 @@
 
       expect(ocr_path).to exist
       expect(subject.to_solr["full_text_tesim"]).to eq "yo"
+
+      # verify that ocr has been added to the FileSet
+      subject.reload
+      expect(subject.files.size).to eq(2)
+      expect(subject.files.last.content).to include "<div class='ocr_page'"
     end
     after do
       FileUtils.rm_rf(path.parent) if path.exist?