libis-format 0.9.41 → 0.9.44

data/lib/libis/format/tool/pdf_copy.rb

@@ -0,0 +1,42 @@
+require 'os'
+
+require 'libis/tools/extend/string'
+require 'libis/tools/logger'
+require 'libis/tools/command'
+
+require 'libis/format/config'
+
+module Libis
+  module Format
+    module Tool
+
+      class PdfCopy
+        include ::Libis::Tools::Logger
+
+        def self.run(source, target, options = [])
+          self.new.run source, target, options
+        end
+
+        def run(source, target, options = [])
+          tool_dir = File.absolute_path(File.join(File.dirname(__FILE__), '..', '..', '..', 'tools'))
+          jar_file = File.join(tool_dir, 'PdfTool.jar')
+
+          if OS.java?
+            # TODO: import library and execute in current VM. For now do exactly as in MRI.
+          end
+
+          Libis::Tools::Command.run(
+              Libis::Format::Config[:java_path],
+              '-cp', jar_file,
+              'CopyPdf',
+              '--file_input', source,
+              '--file_output', target,
+              *options
+          )
+
+        end
+      end
+
+    end
+  end
+end

data/lib/libis/format/tool/pdf_merge.rb

@@ -0,0 +1,43 @@
+require 'os'
+
+require 'libis/tools/extend/string'
+require 'libis/tools/logger'
+require 'libis/tools/command'
+
+require 'libis/format/config'
+
+module Libis
+  module Format
+    module Tool
+
+      class PdfMerge
+        include ::Libis::Tools::Logger
+
+        def self.run(source, target, options = [])
+          self.new.run source, target, options
+        end
+
+        def run(source, target, options = [])
+          tool_dir = File.absolute_path(File.join(File.dirname(__FILE__), '..', '..', '..', 'tools'))
+          jar_file = File.join(tool_dir, 'PdfTool.jar')
+          source = [source] unless source.is_a?(Array)
+
+          if OS.java?
+            # TODO: import library and execute in current VM. For now do exactly as in MRI.
+          end
+
+          Libis::Tools::Command.run(
+              Libis::Format::Config[:java_path],
+              '-cp', jar_file,
+              'MergePdf',
+              '--file_output', target,
+              *options,
+              *source,
+          )
+
+        end
+      end
+
+    end
+  end
+end

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

@@ -0,0 +1,38 @@
+require 'os'
+
+require 'libis/tools/extend/string'
+require 'libis/tools/logger'
+require 'libis/tools/command'
+
+require 'libis/format/config'
+
+module Libis
+  module Format
+    module Tool
+
+      class PdfOptimizer
+        include ::Libis::Tools::Logger
+
+        def self.run(source, target, quality)
+          self.new.run source, target, quality
+        end
+
+        def run(source, target, quality)
+
+          Libis::Tools::Command.run(
+              'gs',
+              '-sDEVICE=pdfwrite',
+              '-dCompatibilityLevel=1.4',
+              "-dPDFSETTINGS=/#{quality}",
+              '-dNOPAUSE',
+              '-dBATCH',
+              "-sOutputFile=#{target}",
+              "#{source}"
+          )
+
+        end
+      end
+
+    end
+  end
+end

data/lib/libis/format/tool/pdf_split.rb

@@ -0,0 +1,41 @@
+require 'os'
+
+require 'libis/tools/extend/string'
+require 'libis/tools/logger'
+require 'libis/tools/command'
+
+require 'libis/format/config'
+
+module Libis
+  module Format
+    module Tool
+
+      class PdfSplit
+        include ::Libis::Tools::Logger
+
+        def self.run(source, target, options = [])
+          self.new.run source, target, options
+        end
+
+        def run(source, target, options = [])
+          tool_dir = File.absolute_path(File.join(File.dirname(__FILE__), '..', '..', '..', 'tools'))
+          jar_file = File.join(tool_dir, 'PdfTool.jar')
+          if OS.java?
+            # TODO: import library and execute in current VM. For now do exactly as in MRI.
+          end
+
+          Libis::Tools::Command.run(
+              Libis::Format::Config[:java_path],
+              '-cp', jar_file,
+              'SplitPdf',
+              '--file_input', source,
+              '--file_output', target,
+              *options
+          )
+
+        end
+      end
+
+    end
+  end
+end

data/lib/libis/format/tool/pdf_to_pdfa.rb

@@ -0,0 +1,78 @@
+require 'tempfile'
+require 'csv'
+require 'fileutils'
+
+require 'libis/tools/extend/string'
+require 'libis/tools/logger'
+require 'libis/tools/command'
+require 'libis/tools/temp_file'
+
+require 'libis/format'
+
+module Libis
+  module Format
+    module Tool
+
+      class PdfToPdfa
+        include ::Libis::Tools::Logger
+
+        def self.run(source, target = nil, options = {})
+          self.new.run source, target, options
+        end
+
+        def run(source, target = nil, options = nil)
+
+          tmp_target = Tools::TempFile.name(File.basename(source, '.*'), '.pdf')
+          target ||= tmp_target
+
+          icc_info = icc_options(options[:colorspace])
+
+          icc_file = Tools::TempFile.name(icc_info[:icc_name], '.icc')
+          FileUtils.cp(File.join(Libis::Format::DATA_DIR, "#{icc_info[:icc_name]}.icc"), icc_file)
+
+          def_filename = Tools::TempFile.name('PDFA_def', '.ps')
+          File.open(def_filename, 'w') do |f|
+            f.puts File.read(File.join(Libis::Format::DATA_DIR, 'PDFA_def.ps')).
+                gsub('[** Fill in ICC profile location **]', icc_file).
+                gsub('[** Fill in ICC reference name **]', icc_info[:icc_ref])
+          end
+
+          result = Libis::Tools::Command.run(
+              Libis::Format::Config[:ghostscript_path],
+              '-dBATCH', '-dNOPAUSE', '-dNOOUTERSAVE',
+              '-sColorConversionStrategy=/UseDeviceIndependentColor',
+              "-sProcessColorModel=#{icc_info[:device]}",
+              '-sDEVICE=pdfwrite', '-dPDFA', '-dPDFACompatibilityPolicy=1',
+              "-sOutputICCProfile=#{icc_file}",
+              '-o', File.absolute_path(target),
+              def_filename,
+              source
+          )
+
+          FileUtils.rm [icc_file, def_filename].compact, force: true
+
+          unless Format::Tool::PdfaValidator.run(target)
+            result[:status] = -999
+            result[:err] << 'Failed to validate generated PDF/A file.'
+          end
+
+          result
+        end
+
+
+        private
+
+        def icc_options(colorspace)
+          case colorspace.to_s.downcase
+            when 'cmyk'
+              {icc_name: 'ISOcoated_v2_eci', icc_ref: 'FOGRA39L', device: 'DeviceCMYK'}
+            else
+              {icc_name: 'eciRGB_v2', icc_ref: 'sRGB', device: 'DeviceRGB'}
+          end
+        end
+
+      end
+
+    end
+  end
+end

data/lib/libis/format/tool/pdfa_validator.rb

@@ -0,0 +1,63 @@
+require 'fileutils'
+
+require 'libis/tools/extend/string'
+require 'libis/tools/logger'
+require 'libis/tools/command'
+
+require 'libis/format/config'
+
+module Libis
+  module Format
+    module Tool
+
+      class PdfaValidator
+        include ::Libis::Tools::Logger
+
+        def self.run(source)
+          self.new.run source
+        end
+
+        def run(source)
+
+          src_file = File.absolute_path(source)
+
+          if (pdfa = Libis::Format::Config[:pdfa_path])
+            # Keep it clean: tool generates fontconfig/ cache dir in current working dir
+            previous_wd = Dir.getwd
+            Dir.chdir(Dir.tmpdir)
+
+            result = Libis::Tools::Command.run(
+                pdfa,
+                '--noxml',
+                '--level', 'B',
+                '--verb', '0',
+                src_file
+            )
+
+            Dir.chdir(previous_wd)
+
+            unless result[:out].any? {|line| line =~ /^VLD-\[PASS\]/}
+              warn "Validator failed to validate the PDF file '%s' against PDF/A-1B constraints:\n%s", source,
+                   result[:out].join("\n")
+              return false
+            end
+          else
+            jar = File.join(Libis::Format::ROOT_DIR, 'tools', 'pdfbox', 'preflight-app-1.8.10.jar')
+            result = Libis::Tools::Command.run(
+                Libis::Format::Config[:java_path],
+                '-jar', jar,
+                src_file
+            )
+            unless result[:status] == 0
+              warn "Validator failed to validate the PDF file '%s' against PDF/A-1B constraints:\n%s", source,
+                   result[:out].join("\n")
+              return false
+            end
+          end
+          true
+        end
+      end
+
+    end
+  end
+end

data/lib/libis/format/tool.rb

@@ -0,0 +1,23 @@
+# code utf-8
+
+module Libis
+  module Format
+    module Tool
+
+      autoload :Droid, 'libis/format/tool/droid'
+      autoload :ExtensionIdentification, 'libis/format/tool/extension_identification'
+      autoload :Fido, 'libis/format/tool/fido'
+      autoload :FileTool, 'libis/format/tool/file_tool'
+
+      autoload :OfficeToPdf, 'libis/format/tool/office_to_pdf'
+      autoload :FFMpeg, 'libis/format/tool/ffmpeg'
+      autoload :PdfCopy, 'libis/format/tool/pdf_copy'
+      autoload :PdfMerge, 'libis/format/tool/pdf_merge'
+      autoload :PdfOptimizer, 'libis/format/tool/pdf_optimizer'
+      autoload :PdfSplit, 'libis/format/tool/pdf_split'
+      autoload :PdfToPdfa, 'libis/format/tool/pdf_to_pdfa'
+      autoload :PdfaValidator, 'libis/format/tool/pdfa_validator'
+
+    end
+  end
+end

data/lib/libis/format/version.rb

@@ -1,5 +1,5 @@
 module Libis
   module Format
-    VERSION = '0.9.41'
+    VERSION = '0.9.44'
   end
 end

data/lib/libis/format.rb

@@ -5,22 +5,8 @@ module Libis
     autoload :Config, 'libis/format/config'
     autoload :TypeDatabase, 'libis/format/type_database'
     autoload :Identifier, 'libis/format/identifier'
-    autoload :Identifier, 'libis/format/identifier'
-
-    autoload :FileTool, 'libis/format/file_tool'
-    autoload :Fido, 'libis/format/fido'
-    autoload :Droid, 'libis/format/droid'
-    autoload :ExtensionIdentification, 'libis/format/extension_identification'
-
-    autoload :OfficeToPdf, 'libis/format/office_to_pdf'
-    autoload :PdfCopy, 'libis/format/pdf_copy'
-    autoload :PdfMerge, 'libis/format/pdf_merge'
-    autoload :PdfOptimizer, 'libis/format/pdf_optimizer'
-    autoload :PdfSplit, 'libis/format/pdf_split'
-    autoload :PdfToPdfa, 'libis/format/pdf_to_pdfa'
-    autoload :PdfaValidator, 'libis/format/pdfa_validator'
-    autoload :FFMpeg, 'libis/format/ffmpeg'
 
+    autoload :Tool, 'libis/format/tool'
     autoload :Converter, 'libis/format/converter'
 
     ROOT_DIR = File.absolute_path(File.join(File.dirname(__FILE__), '..', '..'))

data/libis-format.gemspec

@@ -25,10 +25,9 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency 'bundler', '~> 1.6'
   spec.add_development_dependency 'rake', '~> 10.3'
   spec.add_development_dependency 'rspec', '~> 3.1'
-  spec.add_development_dependency 'simplecov', '~> 0.9'
   spec.add_development_dependency 'awesome_print'
 
-  spec.add_runtime_dependency 'libis-tools', '~> 0.9.52'
+  spec.add_runtime_dependency 'libis-tools', '~> 0.9.57'
   spec.add_runtime_dependency 'os', '= 0.9.6'
   spec.add_runtime_dependency 'mini_magick', '~> 4.3'
   spec.add_runtime_dependency 'deep_dive', '~> 0.3'

data/spec/converter_audio_spec.rb

@@ -0,0 +1,66 @@
+# encoding: utf-8
+require 'spec_helper'
+
+require 'libis/format/converter/audio_converter'
+
+describe 'Converters' do
+
+  let(:repository) {Libis::Format::Converter::Repository}
+  let(:file_dir) {File.dirname(__FILE__)}
+
+  before(:all) {
+    Libis::Tools::Config.logger.level = 'off'
+  }
+
+  context 'Audio Converter' do
+
+    let(:converter) {Libis::Format::Converter::AudioConverter.new}
+    extensions = %w'aac aiff au flac m4a mka mp3 ra voc wav wma'
+    confidence = {
+        aac: 0.86,
+        aiff: 0.99,
+        au: 0.99,
+        flac: 0.99,
+        m4a: 0.93,
+        mka: 0.94,
+        mp3: 0.95,
+        ra: 0.92,
+        voc: 0.99,
+        wav: 0.99,
+        wma: 0.9
+    }
+    targets = %w'mp3 flac wav'
+    sources = %w'carlin_disappointed greensleeves king_nonviolence top_gun_secret'
+    quality = {
+        carlin_disappointed: 1.0,
+        greensleeves: 0.95,
+        king_nonviolence: 1.0,
+        top_gun_secret: 0.95
+    }
+
+    let(:data_dir) {File.join(file_dir, 'data', 'audio')}
+
+    context 'converts' do
+      sources.each do |source|
+        extensions.each do |ext|
+          next unless (File.exists?(File.join(File.dirname(__FILE__), 'data', 'audio', "#{source}.#{ext}")))
+          (targets - [ext]).each do |tgt|
+            it "#{source} #{ext} to #{tgt}" do
+              src_file = File.join(data_dir, "#{source}.#{ext}")
+              ref_file = File.join(data_dir, "#{source}.#{tgt}")
+              tgt_file = File.join('', 'tmp', "test.#{source}.#{ext}.#{tgt}")
+              FileUtils.remove tgt_file, force: true
+              FileUtils.mkdir_p File.dirname(tgt_file)
+              result = converter.convert(src_file, tgt_file, tgt.upcase.to_sym)
+              expect(result).to eq tgt_file
+              expect(result).to sound_like ref_file, confidence[ext.to_sym] * quality[source.to_sym], 11025, 1
+              FileUtils.remove tgt_file, force: true
+            end
+          end
+        end
+      end
+    end
+
+  end
+
+end

data/spec/converter_image_spec.rb

@@ -0,0 +1,166 @@
+# encoding: utf-8
+require 'spec_helper'
+
+require 'libis/format/converter/image_converter'
+require 'libis/format/converter/jp2_converter'
+
+describe 'Converters' do
+
+  let(:repository) {Libis::Format::Converter::Repository}
+  let(:file_dir) {File.dirname(__FILE__)}
+
+  before(:all) {
+    Libis::Tools::Config.logger.level = 'off'
+  }
+
+  context 'Image Converter' do
+
+    let(:converter) {Libis::Format::Converter::ImageConverter.new}
+    let(:diff_file) {File.join('', 'tmp', 'diff.jpg')}
+
+    it 'converts TIFF to JPEG' do
+      src_file = File.join(file_dir, 'data', 'test.tif')
+      ref_file = File.join(file_dir, 'data', 'test.jpg')
+      tgt_file = File.join('', 'tmp', 'test.jpg')
+      FileUtils.mkdir_p File.dirname(tgt_file)
+      converter.delete_date
+      result = converter.convert(src_file, tgt_file, :JPG)
+      expect(result).to eq tgt_file
+      compare = MiniMagick::Tool::Compare.new
+      compare << ref_file << tgt_file
+      compare.metric << 'MAE'
+      compare.fuzz << '1%'
+      compare << diff_file
+      compare.call {|_, _, status| expect(status).to be 0}
+      FileUtils.rm tgt_file, force: true
+      FileUtils.rm diff_file, force: true
+    end
+
+    it 'converts TIFF to PNG' do
+      src_file = File.join(file_dir, 'data', 'test.tif')
+      ref_file = File.join(file_dir, 'data', 'test.png')
+      tgt_file = File.join('', 'tmp', 'test.png')
+      FileUtils.mkdir_p File.dirname(tgt_file)
+      converter.delete_date
+      converter.page(0)
+      result = converter.convert(src_file, tgt_file, :PNG)
+      expect(result).to eq tgt_file
+      compare = MiniMagick::Tool::Compare.new
+      compare << ref_file << tgt_file
+      compare.metric << 'MAE'
+      compare << diff_file
+      compare.call {|_, _, status| expect(status).to be 0}
+      FileUtils.rm tgt_file, force: true
+      FileUtils.rm diff_file, force: true
+    end
+
+    it 'converts PDF to TIFF' do
+      src_file = File.join(file_dir, 'data', 'test.pdf')
+      ref_file = File.join(file_dir, 'data', 'test.pdf.tif')
+      tgt_file = File.join('', 'tmp', 'test.pdf.tif')
+      FileUtils.mkdir_p File.dirname(tgt_file)
+      converter.delete_date
+      result = converter.convert(src_file, tgt_file, :TIFF)
+      expect(result).to eq tgt_file
+      compare = MiniMagick::Tool::Compare.new
+      compare << ref_file << tgt_file
+      compare.metric << 'AE'
+      compare.fuzz << '100%'
+      compare << diff_file
+      compare.call {|_, _, status| expect(status).to be 0}
+      FileUtils.rm tgt_file, force: true
+      FileUtils.rm diff_file, force: true
+    end
+
+    it 'converts TIFF to PNG with many options' do
+      src_file = File.join(file_dir, 'data', 'test.tif')
+      ref_file = File.join(file_dir, 'data', 'test-options.png')
+      tgt_file = File.join('', 'tmp', 'test-options.png')
+      FileUtils.mkdir_p File.dirname(tgt_file)
+      converter.watermark(text: 'RSPEC', size: 5, opacity: 0.1, rotation: 15, gap: 0.5, composition: 'modulate')
+      converter.delete_date
+      result = converter.convert(src_file, tgt_file, :PNG, options: {scale: '150%'})
+      expect(result).to eq tgt_file
+      compare = MiniMagick::Tool::Compare.new
+      compare << ref_file << tgt_file
+      compare.metric << 'AE'
+      compare.fuzz << '100%'
+      compare << diff_file
+      compare.call do |_stdin, _stdout, status|
+        expect(status).to be 0
+      end
+      FileUtils.rm tgt_file, force: true
+      FileUtils.rm diff_file, force: true
+    end
+
+    it 'converts only first page of multipage TIFF to JP2' do
+      src_file = File.join(file_dir, 'data', 'multipage.tif')
+      ref_file = File.join(file_dir, 'data', 'multipage.tif.jp2')
+      tgt_file = File.join('', 'tmp', 'test.jp2')
+      FileUtils.mkdir_p File.dirname(tgt_file)
+      converter.delete_date
+      converter.quiet(true)
+      converter.page(0)
+      result = converter.convert(src_file, tgt_file, :JP2)
+      expect(result).to eq tgt_file
+      expect(File.exist?(tgt_file)).to be_truthy
+      compare = MiniMagick::Tool::Compare.new
+      compare << ref_file << tgt_file
+      compare.metric << 'MAE'
+      compare.fuzz << '10%'
+      compare << diff_file
+      compare.call {|_, _, status| expect(status).to be 0}
+      FileUtils.rm tgt_file, force: true
+      FileUtils.rm diff_file, force: true
+    end
+
+    it 'converts TIFF to JP2' do
+      src_file = File.join(file_dir, 'data', 'test.tif')
+      tgt_file = File.join('', 'tmp', 'test.jp2')
+      FileUtils.mkdir_p File.dirname(tgt_file)
+      converter.delete_date
+      result = converter.convert(src_file, tgt_file, :JP2)
+      expect(result).to eq tgt_file
+      expect(File.exist?(tgt_file)).to be_truthy
+      FileUtils.rm tgt_file, force: true
+    end
+
+
+  end
+
+  context 'JP2 Converter', if: File.exists?(Libis::Format::Config[:j2kdriver]) do
+
+    let(:converter) {Libis::Format::Converter::Jp2Converter.new}
+    let(:diff_file) {File.join('', 'tmp', 'diff.jpg')}
+
+    it 'converts TIFF to JP2' do
+      src_file = File.join(file_dir, 'data', 'test.tif')
+      tgt_file = File.join('', 'tmp', 'test.jp2')
+      FileUtils.mkdir_p File.dirname(tgt_file)
+      result = converter.convert(src_file, tgt_file, :JP2)
+      expect(result).to eq tgt_file
+      expect(File.exist?(tgt_file)).to be_truthy
+      FileUtils.rm tgt_file, force: true
+    end
+
+    it 'converts only first page of multipage TIFF to JP2' do
+      src_file = File.join(file_dir, 'data', 'multipage.tif')
+      ref_file = File.join(file_dir, 'data', 'multipage.tif.jp2')
+      tgt_file = File.join('', 'tmp', 'test.jp2')
+      FileUtils.mkdir_p File.dirname(tgt_file)
+      result = converter.convert(src_file, tgt_file, :JP2)
+      expect(result).to eq tgt_file
+      expect(File.exist?(tgt_file)).to be_truthy
+      compare = MiniMagick::Tool::Compare.new
+      compare << ref_file << tgt_file
+      compare.metric << 'MAE'
+      compare.fuzz << '10%'
+      compare << diff_file
+      compare.call {|_, _, status| expect(status).to be 0}
+      FileUtils.rm tgt_file, force: true
+      FileUtils.rm diff_file, force: true
+    end
+
+  end
+
+end