libis-format 1.2.1 → 2.0.0

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

@@ -5,13 +5,13 @@ 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
@@ -19,84 +19,34 @@ module Libis
         end
 
         def self.output_types(format = nil)
-          return [] unless input_types.include?(format)
+          return [] unless input_types.include?(format) if format
           [:PDF, :PDFA]
         end
 
-        def pdf_convert(_)
-          #force usage of this converter
+        def title(v)
+          @options[:title] = 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 [:title, :author, :creator, :keywords, :subject].include?(k)
-            @options["md_#{k}"] = v
-          end
+        def author(v)
+          @options[:author] = 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 creator(v)
+          @options[:creator] = 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 keywords(v)
+          @options[:keywords] = 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) === setting
+        def subject(v)
+          @options[:subject] = 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.blank? ? nil : selection
         end
 
         def convert(source, target, format, opts = {})
@@ -104,12 +54,6 @@ 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
@@ -120,24 +64,17 @@ module Libis
             result = pdf_to_pdfa(source, target)
           end
 
-          { 
-            files: [result],
-            converter: self.class.name
-          }
+          result
 
         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
+        OPTIONS_TABLE = {
+            title: 'md_title',
+            author: 'md_author',
+            creator: 'md_creator',
+            keywords: 'md_keywords',
+            subject: 'md_subject'
+        }
 
         def convert_pdf(source, target)
 
@@ -148,8 +85,9 @@ module Libis
                   if v.nil?
                     nil
                   else
+                    k = OPTIONS_TABLE[k] || k
                     ["--#{k}", (v.is_a?(Array) ? v : v.to_s)]
-                  end}.flatten
+                  end}.compact.flatten
             )
             unless result[:err].empty?
               error("Pdf conversion encountered errors:\n%s", result[:err].join(join("\n")))
@@ -164,16 +102,11 @@ module Libis
 
           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

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/spreadsheet_converter.rb

@@ -3,7 +3,7 @@
 require_relative 'base'
 
 require 'libis/format/tool/spreadsheet_to_ods'
-require 'libis/format/type_database'
+require 'libis/format/library'
 
 module Libis
   module Format
@@ -20,7 +20,7 @@ module Libis
         end
 
         def self.output_types(format = nil)
-          return [] unless input_types.include?(format)
+          return [] unless input_types.include?(format) if format
           [:OO_CALC]
         end
 
@@ -31,7 +31,9 @@ module Libis
         def convert(source, target, format, opts = {})
           super
 
-          Format::Tool::SpreadsheetToOds.run(source, target)
+          return nil unless Format::Tool::SpreadsheetToOds.run(source, target)
+
+          target
 
         end
 

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

@@ -1,5 +1,5 @@
 require_relative 'base'
-require 'libis/format/tool/ff_mpeg'
+require 'libis/format/tool/ffmpeg'
 
 require 'fileutils'
 
@@ -14,7 +14,7 @@ module Libis
         end
 
         def self.output_types(format = nil)
-          return [] unless input_types.include?(format)
+          return [] unless input_types.include?(format) if format
           [:GIF, :WEBM, :MP4, :MPG, :MKV, :MJP2, :QTFF, :AVI, :OGGV, :WMV, :DV, :FLV, :SWF]
         end
 
@@ -179,10 +179,7 @@ module Libis
 
           end
 
-          {
-            files: [target],
-            converter: self.class.name
-          }
+          target
 
         end
 

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

@@ -1,17 +1,17 @@
-require_relative "base"
+require_relative 'base'
 
 module Libis
   module Format
     module Converter
+
       class XsltConverter < Libis::Format::Converter::Base
+
         def self.input_types
           [:XML]
         end
 
         def self.output_types(format = nil)
-          if format
-            return [] unless input_types.include?(format)
-          end
+          return [] unless input_types.include?(format) if format
           [:XML, :HTML, :TXT]
         end
 
@@ -28,20 +28,20 @@ module Libis
           end
 
           unless @options[:xsl_file]
-            error "No xsl_file supplied"
+            error 'No xsl_file supplied'
             return nil
           end
 
           FileUtils.mkpath(File.dirname(target))
 
           if RUBY_PLATFORM == "java"
-            require "saxon-xslt"
+            require 'saxon-xslt'
             xsl = Saxon.XSLT(File.open(@options[:xsl_file]))
             xml = Saxon.XML(File.open(source))
             result = xsl.transform(xml)
-            File.write(target, result.to_s)
+            File.open(target, 'w') {|f| f.write(result.to_s)}
           else
-            require "nokogiri"
+            require 'nokogiri'
 
             doc = nil
             begin
@@ -65,7 +65,7 @@ module Libis
             xsl = nil
 
             begin
-              fp = File.open(file, "r")
+              fp = File.open(file, 'r')
               xsl = Nokogiri::XSLT(fp) do |config|
                 config.options = Nokogiri::XML::ParseOptions::STRICT | Nokogiri::XML::ParseOptions::NOBLANKS
               end
@@ -80,7 +80,7 @@ module Libis
 
             begin
               target_xml = xsl.transform(doc)
-              fp = File.open(target, "w")
+              fp = File.open(target, 'w')
               fp.write(target_xml)
             rescue Exception => e
               error "Error transforming '#{source}' with '#{file}': #{e.message} @ #{e.backtrace[0]}"
@@ -89,14 +89,13 @@ module Libis
               fp.close unless fp.nil? or fp.closed?
             end
 
-            {
-              command: {status: 0},
-              files: [target]
-            }
-
+            target
           end
+
         end
+
       end
+
     end
   end
 end

data/lib/libis/format/identifier.rb

@@ -9,7 +9,7 @@ require 'libis/tools/extend/string'
 require 'libis/tools/extend/empty'
 require 'nori/core_ext/object'
 
-require 'libis/format/type_database'
+require 'libis/format/library'
 
 require_relative 'config'
 require_relative 'tool/fido'
@@ -132,7 +132,7 @@ module Libis
             if doc.validates_against?(xsd_file)
               log_msg result, :debug, "XML file validated against XML Schema: #{xsd_file}"
               info = {mimetype: mime, tool_raw: file_result[:tool], tool: :xsd_validation, match_type: 'xsd_validation', format_version: '', }
-              file_result.merge! Libis::Format::TypeDatabase.enrich(info, PUID: :puid, MIME: :mimetype, NAME: :format_name)
+              file_result.merge! Libis::Format::Library.enrich(info, puid: :puid, mimetype: :mimetype, name: :format_name)
             end
           rescue => e
             # Do nothing - probably Nokogiri chrashed during validation. Could have many causes
@@ -208,7 +208,7 @@ module Libis
       end
 
       def get_mimetype(puid)
-        ::Libis::Format::TypeDatabase.puid_typeinfo(puid)[:MIME].first rescue nil
+        ::Libis::Format::Library.get_field_by(:puid, puid, :mimetypes)
       end
 
       def get_best_result(results)

data/lib/libis/format/info.rb

@@ -0,0 +1,27 @@
+module Libis
+  module Format
+    class Info
+      attr_reader :name, :category, :description, :puids, :mimetypes, :extensions
+
+      def initialize(name:, category:, description: '', puids: [], mimetypes: [], extensions: [])
+        @name = name
+        @category = category
+        @description = description
+        @puids = puids
+        @mimetypes = mimetypes
+        @extensions = extensions
+      end
+
+      def to_hash
+        {
+            name: name,
+            description: description.dup,
+            category: category,
+            puids: puids.dup,
+            mimetypes: mimetypes.dup,
+            extensions: extensions.dup
+        }
+      end
+    end
+  end
+end