Only try to use languages supported by Tesseract

app/services/ocr_runner.rb

@@ -34,13 +34,13 @@ def params
     end
 
     def language
-      if parent.try(:ocr_language).blank? && parent.try(:language).blank?
-        "eng"
-      elsif parent.try(:ocr_language).blank?
-        parent.try(:language).join("+")
-      else
-        parent.try(:ocr_language).join("+")
-      end
+      return try_language(:ocr_language).join("+") unless try_language(:ocr_language).blank?
+      return try_language(:language).join("+") unless try_language(:language).blank?
+      "eng"
+    end
+
+    def try_language(field)
+      (parent.try(field) || []).select { |lang| !Tesseract.languages[lang.to_sym].nil? }
     end
 
     def parent

app/views/curation_concerns/base/_iiif_fields.html.erb

@@ -29,7 +29,7 @@
     </div>
 
     <div>
-      <%= f.input :ocr_language, collection: Tesseract.languages.map(&:reverse), input_html: { name: "#{f.object.model_name.singular}[ocr_language][]", multiple: true, class: 'resource-radio-button' } %>
+      <%= f.input :ocr_language, collection: Tesseract.languages.map(&:reverse).sort, input_html: { name: "#{f.object.model_name.singular}[ocr_language][]", multiple: true, class: 'resource-radio-button' } %>
     </div>
   <% end %>
 </div>

spec/services/ocr_runner_spec.rb

@@ -6,6 +6,7 @@
 
   before do
     allow(file_set).to receive(:generic_works).and_return([parent])
+    allow(Tesseract).to receive(:languages).and_return(eng: ["English"], ita: ["Italian"], spa: ["Spanish"])
   end
 
   describe "#language" do
@@ -29,5 +30,19 @@
         expect(subject.send(:language)).to eq('eng')
       end
     end
+
+    context "when ocr_language is an unsupported language" do
+      let(:parent) { FactoryGirl.build(:scanned_resource, ocr_language: ['xxx']) }
+      it "defaults to english" do
+        expect(subject.send(:language)).to eq('eng')
+      end
+    end
+
+    context "when ocr_language is an unsupported language, but language is supported" do
+      let(:parent) { FactoryGirl.build(:scanned_resource, language: ['spa'], ocr_language: ['xxx']) }
+      it "uses the supported language value" do
+        expect(subject.send(:language)).to eq('spa')
+      end
+    end
   end
 end