libis-format 1.0.8 → 1.2.0

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

@@ -13,6 +13,12 @@ module Libis
       class PdfSplit
         include ::Libis::Tools::Logger
 
+        def self.installed?
+          result = Libis::Tools::Command.run(Libis::Format::Config[:java_cmd], "-version")
+          return false unless result[:status] == 0
+          File.exist?(Libis::Format::Config[:pdf_tool])
+        end
+
         def self.run(source, target, options = [])
           self.new.run source, target, options
         end
@@ -38,7 +44,11 @@ module Libis
           raise RuntimeError, "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
           raise RuntimeError, "#{self.class} errors: #{result[:err].join("\n")}" unless result[:status] == 0 && result[:err].empty?
 
-          result
+          {
+            command: result,
+            files: [ target ] # TODO: collect the files
+          }
+
         end
       end
 

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

@@ -1,83 +1,99 @@
-require 'tempfile'
-require 'csv'
-require 'fileutils'
+require "tempfile"
+require "csv"
+require "fileutils"
+require 'pdfinfo'
 
-require 'libis/tools/extend/string'
-require 'libis/tools/logger'
-require 'libis/tools/command'
-require 'libis/tools/temp_file'
+require "libis/tools/extend/string"
+require "libis/tools/logger"
+require "libis/tools/command"
+require "libis/tools/temp_file"
 
-require 'libis/format'
+require "libis/format"
 
 module Libis
   module Format
     module Tool
-
       class PdfToPdfa
         include ::Libis::Tools::Logger
 
+        def self.installed?
+          result = Libis::Tools::Command.run(Libis::Format::Config[:ghostscript_cmd])
+          result == 0
+        end
+
         def self.run(source, target = nil, options = {})
-          self.new.run source, target, options
+          new.run source, target, options
         end
 
         def run(source, target = nil, options = nil)
-
-          tmp_target = Tools::TempFile.name(File.basename(source, '.*'), '.pdf')
+          tmp_target = Tools::TempFile.name(File.basename(source, ".*"), ".pdf")
           target ||= tmp_target
 
+          metadata = get_metadata(source)
+
           icc_info = icc_options(options[:colorspace])
 
-          icc_file = Tools::TempFile.name(icc_info[:icc_name], '.icc')
+          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])
+          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("[**ICC profile**]", icc_file)
+              .gsub("[**ICC reference**]", icc_info[:icc_ref])
+              .gsub("[**METADATA**]", metadata)
           end
 
           timeout = Libis::Format::Config[:timeouts][:pdf_to_pdfa]
           result = Libis::Tools::Command.run(
-              Libis::Format::Config[:ghostscript_cmd],
-              '-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,
-              timeout: timeout,
-              kill_after: timeout * 2
+            Libis::Format::Config[:ghostscript_cmd],
+            "-q",
+            "-dBATCH", "-dNOPAUSE", "-dNOOUTERSAVE", "-dNOSAFER",
+            # "-dNOPLATFONTS", "-dUseCIEColor=true",
+            # "-sColorConversionStrategy=/UseDeviceIndependentColor",
+            "-sProcessColorModel=#{icc_info[:device]}",
+            "-sOutputICCProfile=#{icc_file}",
+            "-dCompatibilityLevel=1.4",
+            "-sDEVICE=pdfwrite", "-dPDFA=1", "-dPDFACompatibilityPolicy=1",
+            "-o", File.absolute_path(target),
+            def_filename,
+            source,
+            timeout: timeout,
+            kill_after: timeout * 2
           )
 
-          raise RuntimeError, "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
+          raise "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
+          raise "#{self.class} failed with error #{result[:status]}: \n#{(result[:out] + result[:err]).join("\n")}" if result[:status] != 0
 
           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
+          {
+            command: result,
+            files: [target]
+          }
         end
 
-
         private
 
+        def get_metadata(source)
+          info = Pdfinfo.new(source)
+          metadata = "/Title (#{info.title})"
+          metadata += "\n  /Author (#{info.author})" if info.author
+          metadata += "\n  /Subject (#{info.subject})" if info.subject
+          metadata += "\n  /Keywords (#{info.keywords})" if info.keywords
+          metadata += "\n  /Creator (#{info.creator})" if info.creator
+          metadata
+        end
+        
         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'}
+          when "cmyk"
+            {icc_name: "ISOcoated_v2_eci", icc_ref: "FOGRA39L", device: "DeviceCMYK"}
+          else
+            {icc_name: "AdobeRGB1998", icc_ref: "sRGB", device: "DeviceRGB"}
           end
         end
-
       end
-
     end
   end
 end

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

@@ -1,24 +1,22 @@
-require 'fileutils'
+require "fileutils"
 
-require 'libis/tools/extend/string'
-require 'libis/tools/logger'
-require 'libis/tools/command'
+require "libis/tools/extend/string"
+require "libis/tools/logger"
+require "libis/tools/command"
 
-require 'libis/format/config'
+require "libis/format/config"
 
 module Libis
   module Format
     module Tool
-
       class PdfaValidator
         include ::Libis::Tools::Logger
 
         def self.run(source)
-          self.new.run source
+          new.run source
         end
 
         def run(source)
-
           src_file = File.absolute_path(source)
 
           timeout = Libis::Format::Config[:timeouts][:pdfa_validator]
@@ -28,46 +26,41 @@ module Libis
             Dir.chdir(Dir.tmpdir)
 
             result = Libis::Tools::Command.run(
-                pdfa,
-                '--noxml',
-                '--level', 'B',
-                '--verb', '0',
-                src_file,
-                timeout: timeout,
-                kill_after: timeout * 2
+              pdfa,
+              "--noxml",
+              "--level", "B",
+              "--verb", "0",
+              src_file,
+              timeout: timeout,
+              kill_after: timeout * 2
             )
 
-            raise RuntimeError, "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
-            raise RuntimeError, "#{self.class} errors: #{result[:err].join("\n")}" unless result[:status] == 0 && result[:err].empty?
+            raise "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
+            raise "#{self.class} errors: #{result[:err].join("\n")}" unless result[:status] == 0 && result[:err].empty?
 
             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
+            out, err = result[:out].partition { |line| line =~ /^VLD-\[PASS\]/ }
+            result[:out] = out
+            result[:err] += err
+
+            result
           else
             jar = Libis::Format::Config[:preflight_jar]
             result = Libis::Tools::Command.run(
-                Libis::Format::Config[:java_cmd],
-                '-jar', jar,
-                src_file,
-                timeout: timeout,
-                kill_after: timeout * 2
+              Libis::Format::Config[:java_cmd],
+              "-jar", jar,
+              src_file,
+              timeout: timeout,
+              kill_after: timeout * 2
             )
-            raise RuntimeError, "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
 
-            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
+            raise "#{self.class} took too long (> #{timeout} seconds) to complete" if result[:timeout]
+
+            result
           end
-          true
         end
       end
-
     end
   end
 end

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

@@ -13,6 +13,11 @@ module Libis
       class SpreadsheetToOds
         include ::Libis::Tools::Logger
 
+        def self.installed?
+          result = Libis::Tools::Command.run(Libis::Format::Config[:soffice_cmd], "--version")
+          result == 0
+        end
+
         def self.run(source, target, options = {})
           self.new.run source, target, options
         end
@@ -48,10 +53,14 @@ module Libis
 
           FileUtils.copy tgt_file, target, preserve: true
 
+          {
+            command: result,
+            files: [ target ]
+          }
+
         ensure
           FileUtils.rmtree workdir rescue nil
 
-          result[:out]
         end
       end
 

data/lib/libis/format/tool.rb

@@ -10,7 +10,7 @@ module Libis
       autoload :FileTool, 'libis/format/tool/file_tool'
 
       autoload :OfficeToPdf, 'libis/format/tool/office_to_pdf'
-      autoload :FFMpeg, 'libis/format/tool/ffmpeg'
+      autoload :FFMpeg, 'libis/format/tool/ff_mpeg'
       autoload :FopPdf, 'libis/format/tool/fop_pdf'
       autoload :PdfCopy, 'libis/format/tool/pdf_copy'
       autoload :PdfMerge, 'libis/format/tool/pdf_merge'

data/lib/libis/format/version.rb

@@ -1,5 +1,5 @@
 module Libis
   module Format
-    VERSION = '1.0.8'
+    VERSION = '1.2.0'
   end
 end

data/libis-format.gemspec

@@ -39,4 +39,6 @@ Gem::Specification.new do |spec|
   spec.add_runtime_dependency 'deep_dive', '~> 0.3'
   spec.add_runtime_dependency 'chromaprint', '~> 0.0.2'
   spec.add_runtime_dependency 'naturally', '~> 2.1'
+  spec.add_runtime_dependency 'pdfinfo', '~> 1.4'
+
 end