Merge pull request #90 from prey/chore/remove-tmp-hocr

Temporary hocr file is deleted after the file is processed.

Author: dannnylo

CHANGELOG.md

@@ -1,6 +1,10 @@
 # Changes
-## v3.1.3
 
+## v3.1.4
+# Changed
+* Temporary hocr file is deleted after the file is processed.
+
+## v3.1.3
 * Fixed a configuration error that wouldn't allow you to do different kinds of calls on the same object, for example calling .to_box and then .to_s would result in unexpected behavior.
 
 ## v3.1.2

lib/rtesseract/base.rb

@@ -9,5 +9,9 @@ module Base
     def temp_file_path
       Pathname.new(Dir.tmpdir).join("rtesseract_#{SecureRandom.uuid}").to_s
     end
+
+    def remove_tmp_file(absolute_file_path)
+      File.delete(absolute_file_path) if File.file?(absolute_file_path)
+    end
   end
 end

lib/rtesseract/box.rb

@@ -9,7 +9,10 @@ def run(source, errors, options)
         options = options.merge({ tessedit_create_hocr: 1 })
 
         RTesseract::Command.new(source, temp_file_path, errors, options).run do |output_path|
-          parse(File.read("#{output_path}.hocr"))
+          filename = "#{output_path}.hocr"
+          content = File.read(filename)
+          remove_tmp_file(filename)
+          parse(content)
         end
       end
 

spec/rtesseract/box_spec.rb

@@ -13,4 +13,10 @@
   it 'bounding box' do
     expect(instance.to_box).to include(word: 'you', x_start: 69, y_start: 17, x_end: 100, y_end: 31, confidence: 96)
   end
+
+  it 'removes the temp hocr file' do
+    initial_count = Dir["#{Dir.tmpdir}/*"].length
+    instance.to_box
+    expect(initial_count).to eql(Dir["#{Dir.tmpdir}/*"].length)
+  end
 end