libis-format 1.3.4 → 2.0.0

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

@@ -1,143 +1,52 @@
-# frozen_string_literal: true
+# encoding: utf-8
 
 require_relative 'base'
 
 require 'libis/tools/extend/hash'
-require 'libis/format/tool/pdf_tool'
+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)
-
-          %i[PDF PDFA]
-        end
-
-        def pdf_convert(_)
-          # force usage of this converter
-        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[:metadata] ||= {})[k] = v
-          end
+          return [] unless input_types.include?(format) if format
+          [:PDF, :PDFA]
         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[:select] = {range: [selection].flatten.compact.join(',')}
+        def title(v)
+          @options[:title] = 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
-        #     - image: same as above
-        #     - 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] || options[:image]
-            watermark_image(options)
-          elsif options[:text]
-            watermark_text(options)
-          elsif options[:banner]
-            watermark_banner(options)
-          end
+        def author(v)
+          @options[:author] = v.blank? ? nil : v
         end
 
-        def watermark_image(options = {})
-          options.key_strings_to_symbols!
-          @options[:watermark] = {command: 'image'}
-          @options[:watermark][:data] = options[:file] || options[:image]
-          @options[:watermark][:opacity] = options[:opacity] || '0.3'
-
+        def creator(v)
+          @options[:creator] = v.blank? ? nil : v
         end
 
-        def watermark_text(options = {})
-          options.key_strings_to_symbols!
-          @options[:watermark] = {command: 'text'}
-          @options[:watermark][:data] = (options[:text] || '© LIBIS').split('\n')
-          @options[:watermark][:rotation] = options[:rotation] if options[:rotation]
-          @options[:watermark][:size] = options[:size] if options[:size]
-          @options[:watermark][:gap] = options[:gap] if options[:gap].to_s =~ /^\s*\d+\s*$/
-          @options[:watermark][:padding] = options[:gap] if options[:gap].to_s =~ /^\s*(0+\.\d+|1\.0+)\s*$/
-          @options[:watermark][:padding] = options[:padding] if options[:padding]
-          @options[:watermark][:opacity] = options[:opacity] || '0.3'
+        def keywords(v)
+          @options[:keywords] = v.blank? ? nil : v
         end
 
-        # Create a vertical banner to the right side of each page
-        #
-        # The banner options are:
-        # - banner: text to put in the banner
-        # - add_filename: append filename to the text (use any value to enable)
-        # - fontsize: size of the font (in points)
-        # - width: width of the banner
-        # - (background|text)_color_(red|green|blue): color components of background and text
-        def watermark_banner(options = {})
-          options.key_strings_to_symbols!
-          @options[:watermark] = {command: 'banner'}
-          @options[:watermark][:data] = (options[:banner] || '© LIBIS')
-          @options[:watermark][:add_filename] = !!options[:add_filename]
-          @options[:watermark][:size] = options[:fontsize] if options[:fontsize]
-          @options[:watermark][:width] = options[:width] if options[:width]
-          @options[:watermark][:background_red] = options[:background_color_red] if options[:background_color_red]
-          @options[:watermark][:background_green] = options[:background_color_green] if options[:background_color_green]
-          @options[:watermark][:background_blue] = options[:background_color_blue] if options[:background_color_blue]
-          @options[:watermark][:text_red] = options[:text_color_red] if options[:text_color_red]
-          @options[:watermark][:text_green] = options[:text_color_green] if options[:text_color_green]
-          @options[:watermark][:text_blue] = options[:text_color_blue] if options[:text_color_blue]
+        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 (Acrobat Distiller 'Default' equivalent)
-        # - 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 = {})
@@ -147,122 +56,65 @@ module Libis
 
           unless @options.empty?
             result = convert_pdf(source, target)
+            return nil unless result
             source = result
           end
 
-          if source && (quality = @options.delete(:optimize))
-            result = optimize_pdf(source, target, quality)
-            source = result
-          end
-
-          if source && (format == :PDFA)
+          if format == :PDFA and source
             result = pdf_to_pdfa(source, target)
           end
 
-          {
-            files: [result],
-            converter: self.class.name
-          }
-        end
-
-        protected
+          result
 
-        def optimize_pdf(source, target, quality)
-          using_temp(target) do |tmpname|
-            result = Libis::Format::Tool::PdfOptimizer.run(source, tmpname, quality)
-            unless result[:err].empty?
-              error("Pdf optimization encountered errors:\n%s", (result[:err] + result[:out]).join("\n"))
-              return nil
-            end
-            tmpname
-          end
         end
 
-        def convert_pdf(source, target)
-          result = source
-          result = add_watermark(result, target, @options[:watermark]) if @options[:watermark]
-          result = add_metadata(result, target, @options[:metadata]) if @options[:metadata]
-          result = select_range(result, target, @options[:select]) if @options[:select]
-          return result
-        end
-
-        def options_to_args(options)
-          options.map do |k, v|
-            key = "--#{k.to_s.tr_s('_', '-')}"
-            value = v
-            case value
-            when TrueClass
-              value = nil
-            when FalseClass
-              value = key = nil
-            when Array
-              value = value.map(&:to_s)
-            else
-              value = v.to_s
-            end
-            [key, value]
-          end.flatten.compact
-        end
-
-        def add_watermark(source, target, options)
-          command = options.delete(:command)
-          data = [options.delete(:data)].flatten.compact
-          args = data + options_to_args(options)
-
-          using_temp(target) do |tmpname|
-            result = Libis::Format::Tool::PdfTool.run(['watermark', command], source, tmpname, *args)
-            unless result[:err].empty?
-              error("Pdf watermarking encountered errors:\n%s", result[:err].join(join("\n")))
-              return nil
-            end
-            tmpname
-          end
-        end
+        OPTIONS_TABLE = {
+            title: 'md_title',
+            author: 'md_author',
+            creator: 'md_creator',
+            keywords: 'md_keywords',
+            subject: 'md_subject'
+        }
 
-        def add_metadata(source, target, options)
-          args = options_to_args(options)
+        def convert_pdf(source, target)
 
           using_temp(target) do |tmpname|
-            result = Libis::Format::Tool::PdfTool.run('metadata', source, tmpname, *args)
+            result = Libis::Format::Tool::PdfCopy.run(
+                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 metadata encountered errors:\n%s", result[:err].join(join("\n")))
-              return nil
+              error("Pdf conversion encountered errors:\n%s", result[:err].join(join("\n")))
+              next nil
             end
             tmpname
           end
-        end
 
-        def select_range(source, target, options)
-          args = options_to_args(options)
-
-          using_temp(target) do |tmpname|
-            result = Libis::Format::Tool::PdfTool.run('select', source, tmpname, *args)
-            unless result[:err].empty?
-              error("Pdf select encountered errors:\n%s", result[:err].join(join("\n")))
-              return nil
-            end
-            tmpname
-          end
         end
 
         def pdf_to_pdfa(source, target)
+
           using_temp(target) do |tmpname|
             result = Libis::Format::Tool::PdfToPdfa.run source, tmpname
-
-            unless result[:status].zero?
-              error("Pdf/A conversion encountered errors:\n%s", (result[:out] + result[:err]).join("\n"))
-              return nil
+            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")
-                return 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

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

@@ -0,0 +1,110 @@
+# 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 PdfWatermarker < 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_watermark(_)
+          #force usage of this converter
+        end
+
+        def initialize
+          super
+          @options[:text] = '© LIBIS'
+          @options[:opacity] = '0.3'
+        end
+
+        def file(v)
+          @options[:file] = v.blank? ? nil : v
+        end
+
+        def text(v)
+          @options[:text] = v
+        end
+
+        def rotation(v)
+          @options[:rotation] = v unless v.blank?
+        end
+
+        def size(v)
+          @options[:size] = v unless v.blank?
+        end
+
+        def opacity(v)
+          @options[:opacity] = v unless v.blank?
+        end
+
+        def gap_size(v)
+          @options[:gap_size] = v
+        end
+
+        def gap_ratio(v)
+          @options[:gap_ratio] = v
+        end
+
+        def convert(source, target, format, opts = {})
+          super
+
+          result = convert_pdf(source, target)
+          return nil unless result
+
+          result
+
+        end
+
+        OPTIONS_TABLE = {
+            file: 'wm_image',
+            text: 'wm_text',
+            rotation: 'wm_text_rotation',
+            size: 'wm_font_size',
+            opacity: 'wm_opacity',
+            gap_size: 'wm_gap_size',
+            gap_ratio: 'wm_gap_ratio'
+        }
+        # noinspection DuplicatedCode
+        def convert_pdf(source, target)
+
+          using_temp(target) do |tmpname|
+            result = Libis::Format::Tool::PdfCopy.run(
+                source, tmpname,
+                @options.map {|k, v|
+                  if v.nil?
+                    nil
+                  else
+                    v = v.split('\n') unless v.blank? if k == :text
+                    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")))
+              next nil
+            end
+            tmpname
+          end
+
+        end
+
+      end
+
+    end
+  end
+end

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

@@ -1,5 +1,6 @@
-# frozen_string_literal: true
+# coding: utf-8
 
+require 'set'
 require 'singleton'
 
 require 'libis/tools/logger'
@@ -10,6 +11,7 @@ require_relative 'chain'
 module Libis
   module Format
     module Converter
+
       class Repository
         include Singleton
         include ::Libis::Tools::Logger
@@ -22,15 +24,15 @@ module Libis
           @converters_glob = File.join(File.dirname(__FILE__), '*_converter.rb')
         end
 
-        def self.register(converter_class)
+        def Repository.register(converter_class)
           instance.converters.add? converter_class
         end
 
-        def self.get_converters # rubocop:disable Naming/AccessorMethodName
+        def Repository.get_converters
           instance.get_converters
         end
 
-        def get_converters # rubocop:disable Naming/AccessorMethodName
+        def get_converters
           if converters.empty?
             Dir.glob(converters_glob).each do |filename|
               # noinspection RubyResolve
@@ -40,12 +42,12 @@ module Libis
           converters
         end
 
-        def self.get_converter_chain(src_type, tgt_type, operations = {})
+        def Repository.get_converter_chain(src_type, tgt_type, operations = {})
           instance.get_converter_chain src_type, tgt_type, operations
         end
 
         def get_converter_chain(src_type, tgt_type, operations = {})
-          msg = "conversion from #{src_type} to #{tgt_type}"
+          msg = "conversion from #{src_type.to_s} to #{tgt_type.to_s}"
           chain_list = find_chains src_type, tgt_type, operations
           # if chain_list.length > 1
           #   warn "Found more than one conversion chain for #{msg}. Picking the first one."
@@ -68,6 +70,7 @@ module Libis
         end
 
         def build_chains(chain)
+
           found = []
           chains = [chain]
 
@@ -78,15 +81,18 @@ module Libis
               new_chains += chains.map { |c| c.append(converter) }.flatten
             end
 
-            found = new_chains.select(&:valid?)
+            found = new_chains.select { |c| c.valid?}
             return found unless found.empty?
 
             chains = new_chains
           end
 
           found
+
         end
+
       end
+
     end
   end
 end

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

@@ -1,38 +1,44 @@
-# frozen_string_literal: true
+# encoding: utf-8
 
 require_relative 'base'
 
 require 'libis/format/tool/spreadsheet_to_ods'
-require 'libis/format/type_database'
+require 'libis/format/library'
 
 module Libis
   module Format
     module Converter
+
       class SpreadsheetConverter < Libis::Format::Converter::Base
+
         def self.input_types
-          %i[
-            MSXLS
-            MSXLSX
-            OO_CALC
+          [
+              :MSXLS,
+              :MSXLSX,
+              :OO_CALC
           ]
         end
 
         def self.output_types(format = nil)
-          return [] unless input_types.include?(format)
-
+          return [] unless input_types.include?(format) if format
           [:OO_CALC]
         end
 
         def spreadsheet_convert(_)
-          # force usage of this converter
+          #force usage of this converter
         end
 
         def convert(source, target, format, opts = {})
           super
 
-          Format::Tool::SpreadsheetToOds.run(source, target)
+          return nil unless Format::Tool::SpreadsheetToOds.run(source, target)
+
+          target
+
         end
+
       end
+
     end
   end
 end