libis-format 1.0.5 → 2.0.3

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

data/lib/libis/format/library.rb

@@ -0,0 +1,147 @@
+# froze_string_litteral: true
+# coding: utf-8
+require 'singleton'
+
+module Libis
+  module Format
+
+    class Library
+      include Singleton
+
+      class << self
+        def implementation=(impl)
+          instance.implementation = impl
+        end
+
+        def get_info(format)
+          instance.get_info(format)
+        end
+
+        def get_info_by(key, value)
+          instance.get_info_by(key, value)
+        end
+
+        def get_infos_by(key, value)
+          instance.get_infos_by(key, value)
+        end
+
+        def get_field(format, field)
+          instance.get_field(format, field)
+        end
+
+        def get_field_by(key, value, field)
+          instance.get_field_by(key, value, field)
+        end
+
+        def get_fields_by(key, value, field)
+          instance.get_fields_by(key, value, field)
+        end
+
+        def known?(key, value)
+          instance.known?(key, value)
+        end
+
+        def enrich(info, map_keys = {})
+          instance.enrich(info, map_keys)
+        end
+
+        def normalize(info, map_keys = {})
+          instance.normalize(info, map_keys)
+        end
+      end
+
+      def implementation=(impl)
+        @implementation = impl
+      end
+
+      def get_field(format, field)
+        get_field_by(:name, format, field)
+      end
+
+      def get_field_by(key, value, field)
+        info = get_info_by(key, value)
+        return nil unless info
+        case field
+        when :mimetype
+          info[:mimetypes]&.first
+        when :puid
+          info[:puids]&.first
+        when :extension
+          info[:extensions]&.first
+        else
+          info[field]
+        end
+      end
+
+      def get_fields_by(key, value, field)
+        get_infos_by(key, value)&.map { |info| info[field] }.compact
+      end
+
+      def get_info(format)
+        get_info_by(:name, format)
+      end
+
+      def get_info_by(key, value)
+        get_infos_by(key, value)&.first
+      end
+
+      def get_infos_by(key, value)
+        result = @implementation.query(key, value)
+        result.map(&:to_hash)
+      end
+
+      def known?(key, value)
+        !get_info_by(key, value).nil?
+      end
+
+      def enrich(info, map_keys = {})
+        info = normalize(info, map_keys)
+        mapper = Hash.new { |hash, key| hash[key] = key }.merge(map_keys)
+        unless (format = info[mapper[:name]]).nil?
+          lib_info = get_info(format)
+          mapper.keys.each do |key|
+            case key
+            when :mimetype
+              info[mapper[key]] = lib_info[:mimetypes].first if lib_info[:mimetypes].first
+            when :puid
+              info[mapper[key]] = lib_info[:puids].first if lib_info[:puids].first
+            when :extension
+              info[mapper[key]] = lib_info[:extensions].first if lib_info[:extensions].first
+            else
+              info[mapper[key]] = lib_info[key] if lib_info[key]
+            end
+          end
+        end
+        info
+      end
+
+      # Derive name from the available info
+      def normalize(info, map_keys = {})
+        return {} unless info.is_a? Hash
+        mapper = Hash.new { |hash, key| hash[key] = key }.merge(map_keys)
+        # fill format from looking up by puid
+        unless (puid = info[mapper[:puid]]).blank?
+          info[mapper[:name]] ||= get_field_by(:puid, puid, :name)
+        end
+        # fill format from looking up by mimetype
+        unless (mime = info[mapper[:mimetype]]).blank?
+          info[mapper[:name]] ||= get_field_by(:mimetype, mime, :name)
+        end
+        # finally complete the information from looking up by format name
+        unless (format = info[mapper[:name]]).nil?
+          info[mapper[:mimetype]] = get_field(format, :mimetype)
+          info[mapper[:category]] = get_field(format, :category)
+        end
+        info
+      end
+
+      private
+
+      def initialize
+        @implementation = eval(Libis::Format::Config[:format_library_implementation])
+      end
+
+    end
+
+  end
+end

data/lib/libis/format/tool.rb

@@ -12,10 +12,13 @@ module Libis
       autoload :OfficeToPdf, 'libis/format/tool/office_to_pdf'
       autoload :FFMpeg, 'libis/format/tool/ffmpeg'
       autoload :FopPdf, 'libis/format/tool/fop_pdf'
-      autoload :PdfCopy, 'libis/format/tool/pdf_copy'
       autoload :PdfMerge, 'libis/format/tool/pdf_merge'
+      autoload :PdfMetadata, 'libis/format/tool/pdf_metadata'
       autoload :PdfOptimizer, 'libis/format/tool/pdf_optimizer'
+      autoload :PdfProtect, 'libis/format/tool/pdf_protect'
+      autoload :PdfSelect, 'libis/format/tool/pdf_select'
       autoload :PdfSplit, 'libis/format/tool/pdf_split'
+      autoload :PdfWatermark, 'libis/format/tool/pdf_watermark'
       autoload :PdfToPdfa, 'libis/format/tool/pdf_to_pdfa'
       autoload :PdfaValidator, 'libis/format/tool/pdfa_validator'
 

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

@@ -37,14 +37,14 @@ module Libis
         def runner(*args)
 
           args.map do |file|
-            info = ::Libis::Format::TypeDatabase.ext_infos(File.extname(file)).first
+            info = ::Libis::Format::Library.get_info_by(:extension, File.extname(file))
             if info
               {
                   filepath: file,
-                  mimetype: (info[:MIME].first rescue nil),
-                  puid: (info[:PUID].first rescue nil),
+                  mimetype: (info[:mimetypes].first rescue nil),
+                  puid: (info[:puids].first rescue nil),
                   matchtype: 'extension',
-                  tool: :type_database
+                  tool: :format_library
               }
             end
           end.cleanup

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

@@ -7,7 +7,7 @@ require 'libis/tools/logger'
 require 'libis/tools/command'
 
 require 'libis/format/config'
-require 'libis/format/type_database'
+require 'libis/format/library'
 
 module Libis
   module Format
@@ -99,7 +99,7 @@ module Libis
           end
 
           # Normalize the mimetype
-          Libis::Format::TypeDatabase.normalize(result, PUID: :puid, MIME: :mimetype)
+          Libis::Format::Library.normalize(result)
 
           # Default score is 5
           result[:score] = 5
@@ -117,14 +117,14 @@ module Libis
             # Signature match increases score with 2
             when 'signature'
               result[:score] += 2
-            # typeinfo = ::Libis::Format::TypeDatabase.puid_typeinfo(result[:puid])
+            # typeinfo = ::Libis::Format::Library.get_info_by(:puid, result[:puid])
             # ext = File.extname(result[:filename])
             # result[:score] += 1 if typeinfo and typeinfo[:EXTENSIONS].include?(ext)
 
             # Container match increases score with 4
             when 'container'
               result[:score] += 4
-            # typeinfo = ::Libis::Format::TypeDatabase.puid_typeinfo(result[:puid])
+            # typeinfo = ::Libis::Format::Library.get_info_by(:puid, result[:puid])
             # ext = File.extname(result[:filename])
             # result[:score] += 1 if typeinfo and typeinfo[:EXTENSIONS].include?(ext)
 
@@ -152,11 +152,11 @@ 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, :mimetype) rescue nil
         end
 
         def get_puid(mimetype)
-          ::Libis::Format::TypeDatabase.mime_infos(mimetype).first[:PUID].first rescue nil
+          ::Libis::Format::Library.get_field_by(:mimetype, mimetype, :puid) rescue nil
         end
 
         attr_accessor :bad_mimetypes, :bad_puids

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

@@ -27,9 +27,9 @@ module Libis
           timeout = Libis::Format::Config[:timeouts][:pdf_merge]
           result = Libis::Tools::Command.run(
               Libis::Format::Config[:java_cmd],
-              '-cp', Libis::Format::Config[:pdf_tool],
-              'MergePdf',
-              '--file_output', target,
+              '-jar', Libis::Format::Config[:pdf_tool],
+              'merge',
+              '-o', target,
               *options,
               *source,
               timeout: timeout,

data/lib/libis/format/tool/{pdf_copy.rb → pdf_metadata.rb}

@@ -10,7 +10,7 @@ module Libis
   module Format
     module Tool
 
-      class PdfCopy
+      class PdfMetadata
         include ::Libis::Tools::Logger
 
         def self.run(source, target, options = [])
@@ -26,10 +26,10 @@ module Libis
           timeout = Libis::Format::Config[:timeouts][:pdf_copy]
           result = Libis::Tools::Command.run(
               Libis::Format::Config[:java_cmd],
-              '-cp', Libis::Format::Config[:pdf_tool],
-              'CopyPdf',
-              '--file_input', source,
-              '--file_output', target,
+              '-jar', Libis::Format::Config[:pdf_tool],
+              'metadata',
+              '-i', source,
+              '-o', target,
               *options,
               timeout: timeout,
               kill_after: timeout * 2

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

@@ -0,0 +1,47 @@
+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 PdfProtect
+        include ::Libis::Tools::Logger
+
+        def self.run(source, target, options = [])
+          self.new.run source, target, options
+        end
+
+        def run(source, target, options = [])
+
+          if OS.java?
+            # TODO: import library and execute in current VM. For now do exactly as in MRI.
+          end
+
+          timeout = Libis::Format::Config[:timeouts][:pdf_copy]
+          result = Libis::Tools::Command.run(
+              Libis::Format::Config[:java_cmd],
+              '-jar', Libis::Format::Config[:pdf_tool],
+              'protect',
+              '-i', source,
+              '-o', target,
+              *options,
+              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?
+
+          result
+        end
+      end
+
+    end
+  end
+end

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

@@ -0,0 +1,47 @@
+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 PdfSelect
+        include ::Libis::Tools::Logger
+
+        def self.run(source, target, options = [])
+          self.new.run source, target, options
+        end
+
+        def run(source, target, options = [])
+
+          if OS.java?
+            # TODO: import library and execute in current VM. For now do exactly as in MRI.
+          end
+
+          timeout = Libis::Format::Config[:timeouts][:pdf_select]
+          result = Libis::Tools::Command.run(
+              Libis::Format::Config[:java_cmd],
+              '-jar', Libis::Format::Config[:pdf_tool],
+              'select',
+              '-i', source,
+              '-o', target,
+              *options,
+              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?
+
+          result
+        end
+      end
+
+    end
+  end
+end

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

@@ -26,10 +26,10 @@ module Libis
           timeout = Libis::Format::Config[:timeouts][:pdf_split]
           result = Libis::Tools::Command.run(
               Libis::Format::Config[:java_cmd],
-              '-cp', Libis::Format::Config[:pdf_tool],
-              'SplitPdf',
-              '--file_input', source,
-              '--file_output', target,
+              '-jar', Libis::Format::Config[:pdf_tool],
+              'split',
+              '-i', source,
+              '-o', target,
               *options,
               timeout: timeout,
               kill_after: timeout * 2

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

@@ -0,0 +1,47 @@
+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 PdfWatermark
+        include ::Libis::Tools::Logger
+
+        def self.run(source, target, wm_type, options = [])
+          self.new.run source, target, wm_type, options
+        end
+
+        def run(source, target, wm_type, options = [])
+
+          if OS.java?
+            # TODO: import library and execute in current VM. For now do exactly as in MRI.
+          end
+
+          timeout = Libis::Format::Config[:timeouts][:pdf_watermark]
+          result = Libis::Tools::Command.run(
+              Libis::Format::Config[:java_cmd],
+              '-jar', Libis::Format::Config[:pdf_tool],
+              'watermark', wm_type,
+              '-i', source,
+              '-o', target,
+              *options,
+              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?
+
+          result
+        end
+      end
+
+    end
+  end
+end