libis-format 1.3.0 → 2.0.0

data/lib/libis/format/converter/jp2_converter.rb

@@ -1,5 +1,4 @@
-# frozen_string_literal: true
-
+# encoding: utf-8
 require 'libis-tools'
 require 'fileutils'
 require 'libis/format/config'
@@ -8,31 +7,32 @@ require_relative 'base'
 module Libis
   module Format
     module Converter
+
       class Jp2Converter < Libis::Format::Converter::Base
+
         def self.input_types
-          %i[TIFF JPG PNG BMP GIF PDF]
+          [:TIFF, :JPG, :PNG, :BMP, :GIF, :PDF]
         end
 
         def self.output_types(format = nil)
-          return [] unless input_types.include?(format)
-
+          return [] unless input_types.include?(format) if format
           [:JP2]
         end
 
         def initialize
           super
           @options = {
-            color_xform: false,
-            error_resilience: :ALL,
-            lossless: true,
-            progression_order: 'RLCP',
-            tile_size: [1024, 1024],
-            codeblock_size: [6, 6]
+              color_xform: false,
+              error_resilience: :ALL,
+              lossless: true,
+              progression_order: 'RLCP',
+              tile_size: [1024, 1024],
+              codeblock_size: [6, 6],
           }
         end
 
         def j2kdriver(_)
-          # force usage of this converter
+          #force usage of this converter
         end
 
         def color_xform(flag = true)
@@ -68,36 +68,38 @@ module Libis
 
           @options.each do |key, value|
             case key
-            when :color_xform
-              options << '--set-output-j2k-color-xform' << (value ? 'YES' : 'NO')
-            when :error_resilience
-              options << '--set-output-j2k-error-resilience' << value.to_s
-            when :lossless
-              if value
-                options << '--set-output-j2k-xform' << 'R53' << '5' << '--set-output-j2k-ratio' << '0'
+              when :color_xform
+                options << '--set-output-j2k-color-xform' << (value ? 'YES' : 'NO')
+              when :error_resilience
+                options << '--set-output-j2k-error-resilience' << value.to_s
+              when :lossless
+                if value
+                  options << '--set-output-j2k-xform' << 'R53' << '5' << '--set-output-j2k-ratio' << '0'
+                else
+                  options << '--set-output-j2k-xform' << 'I97' << '--set-output-j2k-psnr' << '46'
+                end
+              when :progression_order
+                options << '--set-output-j2k-progression-order' << value.to_s
+              when :tile_size
+                options << '--set-output-j2k-tile-size' << value[0].to_s << value[1].to_s
+              when :codeblock_size
+                options << '--set-output-j2k-codeblock-size' << value[0].to_s << value[1].to_s
               else
-                options << '--set-output-j2k-xform' << 'I97' << '--set-output-j2k-psnr' << '46'
-              end
-            when :progression_order
-              options << '--set-output-j2k-progression-order' << value.to_s
-            when :tile_size
-              options << '--set-output-j2k-tile-size' << value[0].to_s << value[1].to_s
-            when :codeblock_size
-              options << '--set-output-j2k-codeblock-size' << value[0].to_s << value[1].to_s
-            else # rubocop:disable Style/EmptyElse
-              # do nothing
+                #do nothing
             end
           end
 
-          result = Libis::Tools::Command.run(
-            Libis::Format::Config[:j2k_cmd],
-            '--input-file-name', source,
-            '--set-output-type', 'JP2',
-            *options,
-            '--output-file-name', target
+
+          Libis::Tools::Command.run(
+              Libis::Format::Config[:j2k_cmd],
+              '--input-file-name', source,
+              '--set-output-type', 'JP2',
+              *options,
+              '--output-file-name', target,
+
           )
 
-          result.merge(files: [target], converter: self.class.name)
+          target
         end
       end
     end

data/lib/libis/format/converter/office_converter.rb

@@ -1,50 +1,56 @@
-# frozen_string_literal: true
+# encoding: utf-8
 
 require_relative 'base'
 
 require 'libis/format/tool/office_to_pdf'
-require 'libis/format/type_database'
+require 'libis/format/library'
 
 module Libis
   module Format
     module Converter
+
       class OfficeConverter < Libis::Format::Converter::Base
+
         def self.input_types
-          %i[
-            TXT
-            RTF
-            HTML
-            MSDOC
-            MSDOT
-            MSXLS
-            MSPPT
-            MSDOCX
-            MSDOTX
-            MSXLSX
-            MSPPTX
-            WORDPERFECT
-            OO_WRITER
-            OO_IMPRESS
-            OO_CALC
+          [
+              :TXT,
+              :RTF,
+              :HTML,
+              :MSDOC,
+              :MSDOT,
+              :MSXLS,
+              :MSPPT,
+              :MSDOCX,
+              :MSDOTX,
+              :MSXLSX,
+              :MSPPTX,
+              :WORDPERFECT,
+              :OO_WRITER,
+              :OO_IMPRESS,
+              :OO_CALC
           ]
         end
 
         def self.output_types(format = nil)
-          return [] unless input_types.include?(format)
-
+          return [] unless input_types.include?(format) if format
           [:PDF]
         end
 
         def office_convert(_)
-          # force usage of this converter
+          #force usage of this converter
         end
 
         def convert(source, target, format, opts = {})
           super
 
-          Format::Tool::OfficeToPdf.run(source, target)
+          return nil unless Format::Tool::OfficeToPdf.run(source, target)
+
+          target
+
         end
+
       end
+
     end
   end
 end

data/lib/libis/format/converter/pdf_assembler.rb

@@ -0,0 +1,66 @@
+# encoding: utf-8
+
+require_relative 'base'
+
+require 'libis/format/tool/pdf_merge'
+
+module Libis
+  module Format
+    module Converter
+
+      # noinspection DuplicatedCode
+      class PdfAssembler < Libis::Format::Converter::Base
+
+        def self.input_types
+          [:PDF]
+        end
+
+        def self.output_types(format = nil)
+          return [] unless input_types.include?(format) if format
+          [:PDF]
+        end
+
+        def self.category
+          :assembler
+        end
+
+        def pdf_assemnble(_)
+          #force usage of this converter
+        end
+
+        def convert(source, target, format, opts = {})
+          super
+
+          result = if source.is_a? Array
+                     assemble(source, target)
+                   elsif File.directory?(source)
+                     source_list = Dir[File.join(source, '**', '*')].reject {|p| File.directory? p}
+                     assemble(source_list, target)
+                   else
+                     assemble([source], target)
+                   end
+          return nil unless result
+
+          result
+        end
+
+        private
+
+        def assemble(source, target)
+
+          result = Libis::Format::Tool::PdfMerge.run(source, target)
+
+          unless result[:err].empty?
+            error("PdfMerge encountered errors:\n%s", result[:err].join(join("\n")))
+            return nil
+          end
+
+          target
+        end
+
+      end
+
+    end
+
+  end
+end

data/lib/libis/format/converter/pdf_converter.rb

@@ -1,103 +1,52 @@
-# frozen_string_literal: true
+# encoding: utf-8
 
 require_relative 'base'
 
 require 'libis/tools/extend/hash'
 require 'libis/format/tool/pdf_copy'
 require 'libis/format/tool/pdf_to_pdfa'
-require 'libis/format/tool/pdfa_validator'
 require 'libis/format/tool/pdf_optimizer'
 
 module Libis
   module Format
     module Converter
+
+      # noinspection DuplicatedCode
       class PdfConverter < Libis::Format::Converter::Base
+
         def self.input_types
           [:PDF]
         end
 
         def self.output_types(format = nil)
-          return [] unless input_types.include?(format)
+          return [] unless input_types.include?(format) if format
+          [:PDF, :PDFA]
+        end
 
-          %i[PDF PDFA]
+        def title(v)
+          @options[:title] = v.blank? ? nil : v
         end
 
-        def pdf_convert(_)
-          # force usage of this converter
+        def author(v)
+          @options[:author] = v.blank? ? nil : v
         end
 
-        # Set metadata for Pdf file
-        #
-        # valid metadata keys are):
-        #       - title
-        #       - author
-        #       - creator
-        #       - keywords
-        #       - subject
-        #
-        # @param [Hash] values list of metadata values to set
-        def metadata(values = {})
-          values.key_strings_to_symbols!
-          values.each do |k, v|
-            next unless %i[title author creator keywords subject].include?(k)
-
-            @options["md_#{k}"] = v
-          end
+        def creator(v)
+          @options[:creator] = v.blank? ? nil : v
         end
 
-        # Select a partial list of pages
-        # @param [String] selection as described in com.itextpdf.text.pdf.SequenceList: [!][o][odd][e][even]start-end
-        def range(selection)
-          @options[:ranges] = selection
+        def keywords(v)
+          @options[:keywords] = v.blank? ? nil : v
         end
 
-        # Create or use a watermark image.
-        #
-        # The watermark options are (use symbols):
-        #     - text: text to create a watermark from
-        #     - file: watermark image to use
-        #     - rotation: rotation of the watermark text (in degrees; integer number)
-        #     - size: font size of the watermark text
-        #     - opacity: opacity of the watermark (fraction 0.0 - 1.0)
-        #     - gap: size of the gap between watermark instances. Integer value is absolute size in points (1/72 inch).
-        #            Fractions are percentage of widht/height.
-        # If both options are given, the file will be used as-is if it exists and is a valid image file. Otherwise the
-        # file will be created or overwritten with a newly created watermark image.
-        #
-        # The created watermark file will be a PNG image with transparent background containing the supplied text
-        # slanted by 30 degrees counter-clockwise.
-        #
-        # @param [Hash] options Hash of options for watermark creation.
-        def watermark(options = {})
-          options.key_strings_to_symbols!
-          if options[:file] && File.exist?(options[:file])
-            @options['wm_image'] = options[:file]
-          else
-            @options['wm_text'] = (options[:text] || '© LIBIS').split('\n')
-            @options['wm_text_rotation'] = options[:rotation] if options[:rotation]
-            @options['wm_font_size'] = options[:size] if options[:size]
-          end
-          @options['wm_opacity'] = options[:opacity] || '0.3'
-          @options['wm_gap_ratio'] = options[:gap] if options[:gap].to_s =~ /^\s*(0+\.\d+|1\.0+)\s*$/
-          @options['wm_gap_size'] = options[:gap] if options[:gap].to_s =~ /^\s*\d+\s*$/
+        def subject(v)
+          @options[:subject] = v.blank? ? nil : v
         end
 
-        # Optimize the PDF
-        #
-        # This reduces the graphics quality to a level in order to limit file size. This option relies on the
-        # presence of ghostscript and takes one argument: the quality level. It should be one of:
-        #
-        # - 0 : lowest quality (Acrobat Distiller 'Screen Optimized' equivalent)
-        # - 1 : medium quality (Acrobat Distiller 'eBook' equivalent)
-        # - 2 : good quality
-        # - 3 : high quality (Acrobat Distiller 'Print Optimized' equivalent)
-        # - 4 : highest quality (Acrobat Distiller 'Prepress Optimized' equivalent)
-        #
-        # Note that the optimization is intended to be used with PDF's containing high-resolution images.
-        #
-        # @param [Integer] setting quality setting. [0-4]
-        def optimize(setting = 1)
-          @options['optimize'] = %w[screen ebook default printer prepress][setting] if (0..4).include?(setting)
+        # Select a partial list of pages
+        # @param [String] selection as described in com.itextpdf.text.pdf.SequenceList: [!][o][odd][e][even]start-end
+        def range(selection)
+          @options[:ranges] = selection.blank? ? nil : selection
         end
 
         def convert(source, target, format, opts = {})
@@ -105,50 +54,40 @@ module Libis
 
           result = nil
 
-          if (quality = @options.delete('optimize'))
-            result = optimize_pdf(source, target, quality)
-            return nil unless result
-
-            source = result
-          end
-
           unless @options.empty?
             result = convert_pdf(source, target)
             return nil unless result
-
             source = result
           end
 
-          result = pdf_to_pdfa(source, target) if (format == :PDFA) && source
+          if format == :PDFA and source
+            result = pdf_to_pdfa(source, target)
+          end
 
-          {
-            files: [result],
-            converter: self.class.name
-          }
-        end
+          result
 
-        def optimize_pdf(source, target, quality)
-          using_temp(target) do |tmpname|
-            result = Libis::Format::Tool::PdfOptimizer.run(source, tmpname, quality)
-            unless (result[:status]).zero?
-              error("Pdf optimization encountered errors:\n%s", (result[:err] + result[:out]).join("\n"))
-              next nil
-            end
-            tmpname
-          end
         end
 
+        OPTIONS_TABLE = {
+            title: 'md_title',
+            author: 'md_author',
+            creator: 'md_creator',
+            keywords: 'md_keywords',
+            subject: 'md_subject'
+        }
+
         def convert_pdf(source, target)
+
           using_temp(target) do |tmpname|
             result = Libis::Format::Tool::PdfCopy.run(
-              source, tmpname,
-              @options.map do |k, v|
-                if v.nil?
-                  nil
-                else
-                  ["--#{k}", (v.is_a?(Array) ? v : v.to_s)]
-                end
-              end.flatten
+                source, tmpname,
+                @options.map {|k, v|
+                  if v.nil?
+                    nil
+                  else
+                    k = OPTIONS_TABLE[k] || k
+                    ["--#{k}", (v.is_a?(Array) ? v : v.to_s)]
+                  end}.compact.flatten
             )
             unless result[:err].empty?
               error("Pdf conversion encountered errors:\n%s", result[:err].join(join("\n")))
@@ -156,26 +95,26 @@ module Libis
             end
             tmpname
           end
+
         end
 
         def pdf_to_pdfa(source, target)
+
           using_temp(target) do |tmpname|
             result = Libis::Format::Tool::PdfToPdfa.run source, tmpname
-
-            if result[:command][:status] != 0
-              error("Pdf/A conversion encountered errors:\n%s", (result[:command][:out] + result[:command][:err]).join("\n"))
+            if result[:status] != 0
+              error("Pdf/A conversion encountered errors:\n%s", result[:err].join("\n"))
               next nil
             else
-              r = Libis::Format::Tool::PdfaValidator.run tmpname
-              if r[:status] != 0
-                error "Pdf/A file failed to validate with following errors:\n%s", (r[:err] || r[:out] || []).join("\n")
-                next nil
-              end
+              warn("Pdf/A conversion warnings:\n%s", result[:err].join("\n")) unless result[:err].empty?
             end
             tmpname
           end
+
         end
+
       end
+
     end
   end
 end

data/lib/libis/format/converter/pdf_optimizer.rb

@@ -0,0 +1,70 @@
+# encoding: utf-8
+
+require_relative 'base'
+
+require 'libis/tools/extend/hash'
+require 'libis/format/tool/pdf_copy'
+require 'libis/format/tool/pdf_to_pdfa'
+require 'libis/format/tool/pdf_optimizer'
+
+module Libis
+  module Format
+    module Converter
+
+      class PdfOptimizer < Libis::Format::Converter::Base
+
+        def self.input_types
+          [:PDF]
+        end
+
+        def self.output_types(format = nil)
+          return [] unless input_types.include?(format) if format
+          [:PDF]
+        end
+
+        def pdf_optimize(_)
+          #force usage of this converter
+        end
+
+        # Optimize the PDF
+        #
+        # This reduces the graphics quality to a level in order to limit file size. This option relies on the
+        # presence of ghostscript and takes one argument: the quality level. It should be one of:
+        #
+        # - 0 : lowest quality (Acrobat Distiller 'Screen Optimized' equivalent)
+        # - 1 : medium quality (Acrobat Distiller 'eBook' equivalent)
+        # - 2 : good quality
+        # - 3 : high quality (Acrobat Distiller 'Print Optimized' equivalent)
+        # - 4 : highest quality (Acrobat Distiller 'Prepress Optimized' equivalent)
+        #
+        # Note that the optimization is intended to be used with PDF's containing high-resolution images.
+        #
+        # @param [Integer] setting quality setting. [0-4]
+        def quality(setting = 1)
+          @quality = %w(screen ebook default printer prepress)[setting] if (0..4) === setting
+        end
+
+        def convert(source, target, format, opts = {})
+          super
+
+           optimize_pdf(source, target, @quality || 'ebook')
+
+        end
+
+        def optimize_pdf(source, target, quality)
+
+          using_temp(target) do |tmpname|
+            result = Libis::Format::Tool::PdfOptimizer.run(source, tmpname, quality)
+            unless result[:status] == 0
+              error("Pdf optimization encountered errors:\n%s", (result[:err] + result[:out]).join("\n"))
+              next nil
+            end
+            tmpname
+          end
+        end
+
+      end
+
+    end
+  end
+end

data/lib/libis/format/converter/pdf_splitter.rb

@@ -0,0 +1,65 @@
+# encoding: utf-8
+
+require_relative 'base'
+
+require 'libis/format/tool/pdf_split'
+
+module Libis
+  module Format
+    module Converter
+
+      # noinspection DuplicatedCode
+      class PdfSplitter < Libis::Format::Converter::Base
+
+        def self.input_types
+          [:PDF]
+        end
+
+        def self.output_types(format = nil)
+          return [] unless input_types.include?(format) if format
+          [:PDF]
+        end
+
+        def pdf_split(_)
+          #force usage of this converter
+        end
+
+        def self.category
+          :splitter
+        end
+
+        # Split at given page. If omitted or nil, the source PDF will be split at every page
+        def page(v)
+          @page = v unless v.blank
+        end
+
+        def convert(source, target, format, opts = {})
+          super
+
+          result = split(source, target)
+          return nil unless result
+
+          result
+        end
+
+        private
+
+        def split(source, target)
+
+          options = @page ? ['--page', @page] : ['--every_page']
+          using_temp(target) do |tmpname|
+            result = Libis::Format::Tool::PdfSplit.run(source, tmpname, *options)
+            unless result[:err].empty?
+              error("Pdf split encountered errors:\n%s", result[:err].join(join("\n")))
+              next nil
+            end
+            tmpname
+          end
+
+        end
+
+      end
+
+    end
+  end
+end