document.rb in ocr-file-0.0.4

- old
+ new

@@ -3,10 +3,11 @@
     # TODO: Skewness / text orientation detection
     # TODO: Better handwriting analysis
 
     ACCEPTED_IMAGE_TYPES = ['png', 'jpeg', 'jpg', 'tiff', 'bmp']
     PAGE_BREAK = "\n\r\n" # TODO: Make configurable
+    EFFECTS_TO_REMOVE = ['', 'norm', 'remove_shadow', 'bw']
     DEFAULT_CONFIG = {
       # Images from PDF
       filetype: 'png',
       quality: 100,
       dpi: 300,
@@ -21,24 +22,28 @@
       type_of_ocr: OcrFile::OcrEngines::CloudVision::DOCUMENT_TEXT_DETECTION,
       ocr_engine: 'tesseract', # 'cloud-vision'
       # Image Pre-Processing
       image_preprocess: true,
       effects: ['despeckle', 'deskew', 'enhance', 'sharpen', 'remove_shadow', 'bw'],
+      automatic_reprocess: true,
       # PDF to Image Processing
       optimise_pdf: true,
       extract_pdf_images: true, # if false will screenshot each PDF page
       temp_filename_prefix: 'image',
       # Console Output
       verbose: true,
+      timing: true,
     }
 
     attr_reader :original_file_path,
       :filename,
       :save_file_path,
       :final_save_file,
       :config,
-      :ocr_engine
+      :ocr_engine,
+      :start_time,
+      :end_time
 
     # save_file_path will also generate a tmp path for tmp files. Expected folder path
     # TODO: Add in more input validation
     def initialize(original_file_path:, save_file_path:, config: DEFAULT_CONFIG)
       @original_file_path = original_file_path
@@ -67,80 +72,54 @@
       !pdf? && !image?
     end
 
     # Trigger OCR pipeline
     def to_pdf
-      if pdf?
-        create_temp_folder
-        image_paths = extract_image_paths_from_pdf(@original_file_path)
+      @start_time = Time.now
+      find_best_image_processing if config[:automatic_reprocess] && !text?
 
-        pdfs_to_merge = []
-
-        image_paths.each do |image_path|
-          pdfs_to_merge << @ocr_engine.ocr_to_pdf(process_image(image_path), options: @config)
-        end
-
-        merged_pdf = OcrFile::ImageEngines::PdfEngine.merge(pdfs_to_merge)
-
-        OcrFile::ImageEngines::PdfEngine
-          .save_pdf(merged_pdf, "#{@final_save_file}.pdf", optimise: @config[:optimise_pdf])
+      if pdf?
+        ocr_pdf_to_searchable_pdf
       elsif text?
-        text = ::OcrFile::FileHelpers.open_text_file(@original_file_path)
-        pdf_file = OcrFile::ImageEngines::PdfEngine.pdf_from_text(text, @config)
-
-        OcrFile::ImageEngines::PdfEngine
-          .save_pdf(pdf_file, "#{@final_save_file}.pdf", optimise: @config[:optimise_pdf])
+        text_to_pdf
       else # is an image
         ocr_image_to_pdf
       end
 
       close
+
+      @end_time = Time.now
+      print_time
     end
 
     def to_text
-      if pdf?
-        create_temp_folder
-        image_paths = extract_image_paths_from_pdf(@original_file_path)
+      @start_time = Time.now
+      return ::OcrFile::FileHelpers.open_text_file(@original_file_path) if text?
 
-        image_paths.each do |image_path|
-          text = @ocr_engine.ocr_to_text(process_image(image_path), options: @config)
-          ::OcrFile::FileHelpers.append_file("#{@final_save_file}.txt", "#{text}#{PAGE_BREAK}")
-        end
-      elsif text?
-        ::OcrFile::FileHelpers.open_text_file(@original_file_path)
-      else # is an image
-        ocr_image_to_text(save: true)
-      end
-
+      find_best_image_processing(save: true)
       close
+
+      @end_time = Time.now
+      print_time
     end
 
     def to_s
-      if pdf?
-        create_temp_folder
-        image_paths = extract_image_paths_from_pdf(@original_file_path)
+      @start_time = Time.now
+      return ::OcrFile::FileHelpers.open_text_file(@original_file_path) if text?
 
-        text = ''
+      text = find_best_image_processing(save: false)
 
-        image_paths.each do |image_path|
-          text = "#{text}#{PAGE_BREAK}#{@ocr_engine.ocr_to_text(process_image(image_path), options: @config)}"
-        end
+      close
 
-        close
-        text
-      elsif text?
-        ::OcrFile::FileHelpers.open_text_file(@original_file_path)
-      else # is an image
-        text = ocr_image_to_text(save: false)
+      @end_time = Time.now
+      print_time
 
-        close
-        text
-      end
+      text
     end
 
     def close
-      ::OcrFile::FileHelpers.clear_folder(@temp_folder_path)
+      # ::OcrFile::FileHelpers.clear_folder(@temp_folder_path)
     end
 
     private
 
     def extract_image_paths_from_pdf(file_path)
@@ -183,23 +162,99 @@
       )
 
       image_processor.convert!
     end
 
+    def ocr_pdf_to_searchable_pdf
+      create_temp_folder
+      image_paths = extract_image_paths_from_pdf(@original_file_path)
+
+      pdfs_to_merge = []
+
+      image_paths.each do |image_path|
+        pdfs_to_merge << @ocr_engine.ocr_to_pdf(process_image(image_path), options: @config)
+      end
+
+      merged_pdf = OcrFile::ImageEngines::PdfEngine.merge(pdfs_to_merge)
+
+      OcrFile::ImageEngines::PdfEngine
+        .save_pdf(merged_pdf, "#{@final_save_file}.pdf", optimise: @config[:optimise_pdf])
+    end
+
+    def text_to_pdf
+      text = ::OcrFile::FileHelpers.open_text_file(@original_file_path)
+      pdf_file = OcrFile::ImageEngines::PdfEngine.pdf_from_text(text, @config)
+
+      OcrFile::ImageEngines::PdfEngine
+        .save_pdf(pdf_file, "#{@final_save_file}.pdf", optimise: @config[:optimise_pdf])
+    end
+
     def ocr_image_to_pdf
+      find_best_image_processing if config[:automatic_reprocess]
+
       pdf_document = @ocr_engine.ocr_to_pdf(process_image(@original_file_path), options: @config)
       OcrFile::ImageEngines::PdfEngine
         .save_pdf(pdf_document, "#{@final_save_file}.pdf", optimise: @config[:optimise_pdf])
     end
 
-    def ocr_image_to_text(save: true)
+    def ocr_pdf_to_text(save:)
+      create_temp_folder
+      image_paths = extract_image_paths_from_pdf(@original_file_path)
+
+      text = ''
+
+      image_paths.each do |image_path|
+        text = "#{text}#{PAGE_BREAK}#{@ocr_engine.ocr_to_text(process_image(image_path), options: @config)}"
+      end
+
+      if save
+        ::OcrFile::FileHelpers.append_file("#{@final_save_file}.txt", "#{text}#{PAGE_BREAK}")
+      else
+        text
+      end
+    end
+
+    def ocr_image_to_text(save:)
+      create_temp_folder
       text = @ocr_engine.ocr_to_text(process_image(@original_file_path), options: @config)
 
       if save
         ::OcrFile::FileHelpers.append_file("#{@final_save_file}.txt", text)
       else
         text
       end
+    end
+
+    def ocr_file_to_text(save:)
+      if pdf? &&
+        ocr_pdf_to_text(save: save)
+      else # is an image
+        ocr_image_to_text(save: save)
+      end
+    end
+
+    def find_best_image_processing(save:)
+      ocr_file_to_text(save: save) if !config[:automatic_reprocess]
+
+      text = ''
+      effects_to_test = [''] + (EFFECTS_TO_REMOVE - (EFFECTS_TO_REMOVE - config[:effects]))
+      effects_to_test.each do |effect|
+        config[:effects] = config[:effects] - [effect]
+
+        text = ocr_file_to_text(save: false)
+        break if OcrFile::TextEngines::ResultProcessor.new(text).valid_words?
+      end
+
+      # Adds in extra operations which is unfortunately inefficient
+      if save
+        ocr_file_to_text(save: save)
+      else
+        text
+      end
+    end
+
+    def print_time
+      puts "Total Time: #{end_time-start_time} secs.\n\n" if config[:timing]
     end
 
     def find_ocr_engine(engine_id)
       ocr_engine_constants
         .map { |c| ocr_module(c) }