epub-parser 0.2.5 → 0.2.6

data/lib/epub/constants.rb

@@ -7,7 +7,8 @@ module EPUB
     'epub'  => 'http://www.idpf.org/2007/ops',
     'm'     => 'http://www.w3.org/1998/Math/MathML',
     'svg'   => 'http://www.w3.org/2000/svg',
-    'smil'  => 'http://www.w3.org/ns/SMIL'
+    'smil'  => 'http://www.w3.org/ns/SMIL',
+    'metadata' => 'http://www.idpf.org/2013/metadata'
   }
 
   module MediaType

data/lib/epub/inspector.rb

@@ -25,7 +25,6 @@ module EPUB
     end
 
     module PublicationModel
-      TEMPLATE = "#<%{class}:%{object_id} @package=%{package} %{attributes}>"
       class << self
         def included(mod)
           mod.__send__ :include, Inspector
@@ -33,13 +32,19 @@ module EPUB
       end
 
       def inspect
-        TEMPLATE % {
+        template % {
           :class      => self.class,
-          :package    => package.inspect_simply,
+          :package    => (package && package.inspect_simply),
           :object_id  => inspect_object_id,
           :attributes => inspect_instance_variables(exclude: [:@package])
         }
       end
+
+      def template
+        t = "#<%{class}:%{object_id}"
+        t << " @package=%{package}" if package
+        t << " %{attributes}>"
+      end
     end
   end
 end

data/lib/epub/metadata.rb

@@ -0,0 +1,178 @@
+require 'set'
+
+module EPUB
+  class Metadata
+    include Inspector::PublicationModel
+
+    DC_ELEMS = [:identifiers, :titles, :languages] +
+               [:contributors, :coverages, :creators, :dates, :descriptions, :formats, :publishers,
+                :relations, :rights, :sources, :subjects, :types]
+    attr_accessor :package, :unique_identifier, :metas, :links,
+                  *(DC_ELEMS.collect {|elem| "dc_#{elem}"})
+    DC_ELEMS.each do |elem|
+      alias_method elem, "dc_#{elem}"
+      alias_method "#{elem}=", "dc_#{elem}="
+    end
+
+    def initialize
+      (DC_ELEMS + [:metas, :links]).each do |elem|
+        __send__ "#{elem}=", []
+      end
+    end
+
+    def release_identifier
+      "#{unique_identifier}@#{modified}"
+    end
+    alias package_identifier release_identifier
+
+    def title
+      return extended_title unless extended_title.empty?
+      compositted = titles.select {|title| title.display_seq}.sort.join("\n")
+      return compositted unless compositted.empty?
+      return main_title unless main_title.empty?
+      titles.sort.join("\n")
+    end
+
+    %w[main short collection edition extended].each do |type|
+      define_method "#{type}_title" do
+        titles.select {|title| title.title_type.to_s == type}.sort.join(' ')
+      end
+    end
+
+    def subtitle
+      titles.select {|title| title.title_type.to_s == 'subtitle'}.sort.join(' ')
+    end
+
+    def description
+      descriptions.join(' ')
+    end
+
+    def date
+      dates.first
+    end
+
+    def language
+      languages.first
+    end
+
+    def modified
+      metas.find {|meta|
+        meta.property == 'dcterms:modified' &&
+          meta.refiners.empty?
+      }
+    end
+
+    def to_h
+      DC_ELEMS.inject({}) do |hsh, elem|
+        hsh[elem] = __send__(elem)
+        hsh
+      end
+    end
+
+    def primary_metas
+      metas.select {|meta| meta.primary_expression?}
+    end
+
+    module Refinee
+      PROPERTIES = %w[alternate-script display-seq file-as group-position identifier-type meta-auth role title-type]
+
+      attr_writer :refiners
+
+      def refiners
+        @refiners ||= Set.new
+      end
+
+      PROPERTIES.each do |voc|
+        met = voc.gsub(/-/, '_')
+        attr_writer met
+        define_method met do
+          refiners.find {|refiner| refiner.property == voc}
+        end
+      end
+    end
+
+    class DCMES
+      include Refinee
+
+      attr_accessor :content, :id, :lang, :dir
+
+      def to_s
+        content.to_s
+      end
+    end
+
+    class Identifier < DCMES
+      # @note This is ad-hoc
+      # @todo Define and include OPF module for opf:scheme attribute
+      # @todo Define general way to handle with identifier-type refiners
+      attr_accessor :scheme
+
+      # @note This is ad-hoc
+      # @todo Define and include OPF module for opf:scheme attribute
+      # @todo Define general way to handle with identifier-type refiners
+      def isbn?
+        scheme == 'ISBN' or
+        content.to_s.downcase.start_with? 'urn:isbn' or
+        refiners.any? {|refiner|
+          refiner.property == 'identifier-type' and
+          refiner.scheme == 'onix:codelist5' and
+          %w[02 15].include? refiner.content
+        }
+      end
+    end
+
+    class Title < DCMES
+      include Comparable
+
+      def <=>(other)
+        return 1 if other.display_seq.nil?
+        return -1 if display_seq.nil?
+        display_seq.to_s.to_i <=> other.display_seq.to_s.to_i
+      end
+    end
+
+    class Meta
+      include Refinee
+
+      attr_accessor :property, :id, :scheme, :content
+      attr_reader :refines
+
+      def refines=(refinee)
+        refinee.refiners << self
+        @refines = refinee
+      end
+
+      def refines?
+        ! refines.nil?
+      end
+      alias subexpression? refines?
+
+      def primary_expression?
+        ! subexpression?
+      end
+
+      def inspect
+        ivs = instance_variables.map {|iv|
+          [iv, instance_variable_get(iv).inspect].join('=')
+        }.join(' ')
+        '<#%s:%#0x %s>' % [self.class, __id__, ivs]
+      end
+
+      def to_s
+        content.to_s
+      end
+    end
+
+    class Link
+      include Refinee
+
+      attr_accessor :href, :rel, :id, :media_type
+      attr_reader :refines
+
+      def refines=(refinee)
+        refinee.refiners << self
+        @refines = refinee
+      end
+    end
+  end
+end

data/lib/epub/ocf/container.rb

@@ -15,7 +15,8 @@ module EPUB
       end
 
       class Rootfile
-        attr_accessor :full_path, :media_type
+        attr_accessor :full_path, :media_type,
+                      :package
 
         # @param full_path [Addressable::URI|nil]
         # @param media_type [String]

data/lib/epub/ocf/metadata.rb

@@ -1,6 +1,7 @@
 module EPUB
   class OCF
-    class Metadata
+    class UnknownFormatMetadata
+      attr_accessor :content
     end
   end
 end

data/lib/epub/ocf/physical_container.rb

@@ -1,3 +1,4 @@
+require 'monitor'
 require 'epub/ocf/physical_container/archive_zip'
 require 'epub/ocf/physical_container/unpacked_directory'
 require 'epub/ocf/physical_container/unpacked_uri'
@@ -11,6 +12,14 @@ module EPUB
       @adapter = ArchiveZip
 
       class << self
+        def find_adapter(adapter)
+          return adapter if adapter.instance_of? Class
+          if adapter == :Zipruby && ! const_defined?(adapter)
+            require 'epub/ocf/physical_container/zipruby'
+          end
+          const_get adapter
+        end
+
         def adapter
           raise NoMethodError, "undefined method `#{__method__}' for #{self}" unless self == PhysicalContainer
           @adapter
@@ -18,8 +27,7 @@ module EPUB
 
         def adapter=(adapter)
           raise NoMethodError, "undefined method `#{__method__}' for #{self}" unless self == PhysicalContainer
-          @adapter = adapter.instance_of?(Class) ? adapter : const_get(adapter)
-          adapter
+          @adapter = find_adapter(adapter)
         end
 
         def open(container_path)
@@ -43,6 +51,7 @@ module EPUB
 
       def initialize(container_path)
         @container_path = container_path
+        @monitor = Monitor.new
       end
     end
   end

data/lib/epub/ocf/physical_container/archive_zip.rb

@@ -12,11 +12,13 @@ module EPUB
 
         def open
           Archive::Zip.open @container_path do |archive|
-            @archive = archive
-            begin
-              yield self
-            ensure
-              @archive = nil
+            @monitor.synchronize do
+              @archive = archive
+              begin
+                yield self
+              ensure
+                @archive = nil
+              end
             end
           end
         end

data/lib/epub/ocf/physical_container/unpacked_directory.rb

@@ -8,6 +8,10 @@ module EPUB
 
         def read(path_name)
           ::File.read(::File.join(@container_path, path_name))
+        rescue ::Errno::ENOENT => error
+          no_entry = NoEntry.new(error.message)
+          no_entry.set_backtrace error.backtrace
+          raise no_entry
         end
       end
     end

data/lib/epub/ocf/physical_container/unpacked_uri.rb

@@ -19,6 +19,10 @@ module EPUB
 
         def read(path_name)
           (@container_path + path_name).read
+        rescue ::OpenURI::HTTPError => error
+          no_entry = NoEntry.new(error.message)
+          no_entry.set_backtrace error.backtrace
+          raise no_entry
         end
       end
     end

data/lib/epub/ocf/physical_container/zipruby.rb

@@ -6,11 +6,17 @@ module EPUB
       class Zipruby < self
         def open
           Zip::Archive.open @container_path do |archive|
-            begin
-              @archive = archive
-              yield self
-            ensure
-              @archive = nil
+            @monitor.synchronize do
+              begin
+                @archive = archive
+                yield self
+              rescue ::Zip::Error => error
+                no_entry = NoEntry.new(error.message)
+                no_entry.set_backtrace error.backtrace
+                raise no_entry
+              ensure
+                @archive = nil
+              end
             end
           end
         end
@@ -21,6 +27,12 @@ module EPUB
           else
             open {|container| container.read(path_name)}
           end
+        rescue ::Zip::Error => error
+          no_entry = NoEntry.new(error.message)
+          no_entry.set_backtrace error.backtrace
+          raise no_entry
+        ensure
+          @archive = nil
         end
       end
     end

data/lib/epub/parser.rb

@@ -46,22 +46,29 @@ module EPUB
                      options[:container_adapter] == :UnpackedURI or
                      EPUB::OCF::PhysicalContainer.adapter == EPUB::OCF::PhysicalContainer::UnpackedURI)
 
-      raise "File #{filepath} not readable" if
-        !path_is_uri and !File.readable_real?(filepath)
+      raise "File #{filepath} not found" if
+        !path_is_uri and !File.exist?(filepath)
 
       @filepath = path_is_uri ? filepath : File.realpath(filepath)
       @book = create_book(options)
+      if File.directory? @filepath
+        @book.container_adapter = :UnpackedDirectory
+      end
       @book.epub_file = @filepath
       if options[:container_adapter]
-        adapter = options[:container_adapter]
-        @book.container_adapter = adapter
+        @book.container_adapter = options[:container_adapter]
       end
     end
 
     def parse
       @book.container_adapter.open @filepath do |container|
         @book.ocf = OCF.parse(container)
-        @book.package = Publication.parse(container, @book.rootfile_path)
+        @book.ocf.container.rootfiles.each {|rootfile|
+          package = Publication.parse(container, rootfile.full_path.to_s)
+          rootfile.package = package
+          @book.packages << package
+          package.book = @book
+        }
       end
 
       @book

data/lib/epub/parser/metadata.rb

@@ -0,0 +1,67 @@
+module EPUB
+  class Parser
+    module Metadata
+      def parse_metadata(elem, unique_identifier_id, default_namespace)
+        metadata = EPUB::Publication::Package::Metadata.new
+        id_map = {}
+
+        metadata.identifiers = extract_model(elem, id_map, './dc:identifier', :Identifier, ['id']) {|identifier, e|
+          identifier.scheme = extract_attribute(e, 'scheme', 'opf')
+          metadata.unique_identifier = identifier if identifier.id == unique_identifier_id
+        }
+        metadata.titles = extract_model(elem, id_map, './dc:title', :Title)
+        metadata.languages = extract_model(elem, id_map, './dc:language', :DCMES, %w[id])
+        %w[contributor coverage creator date description format publisher relation source subject type].each do |dcmes|
+          metadata.__send__ "#{dcmes}s=", extract_model(elem, id_map, "./dc:#{dcmes}")
+        end
+        metadata.rights = extract_model(elem, id_map, './dc:rights')
+        metadata.metas = extract_refinee(elem, id_map, "./#{default_namespace}:meta", :Meta, %w[property id scheme])
+        metadata.links = extract_refinee(elem, id_map, "./#{default_namespace}:link", :Link, %w[id media-type]) {|link, e|
+          link.href = extract_attribute(e, 'href')
+          link.rel = Set.new(extract_attribute(e, 'rel').split(nil))
+        }
+
+        id_map.values.each do |hsh|
+          next unless hsh[:refiners]
+          next unless hsh[:metadata]
+          hsh[:refiners].each {|meta| meta.refines = hsh[:metadata]}
+        end
+
+        metadata
+      end
+
+      def extract_model(elem, id_map, xpath, klass=:DCMES, attributes=%w[id lang dir])
+        models = elem.xpath(xpath, EPUB::NAMESPACES).collect do |e|
+          model = EPUB::Publication::Package::Metadata.const_get(klass).new
+          attributes.each do |attr|
+            model.__send__ "#{attr.gsub(/-/, '_')}=", extract_attribute(e, attr)
+          end
+          model.content = e.content unless klass == :Link
+
+          yield model, e if block_given?
+
+          model
+        end
+
+        models.each do |model|
+          id_map[model.id] = {metadata: model} if model.respond_to?(:id) && model.id
+        end
+
+        models
+      end
+
+      def extract_refinee(elem, id_map, xpath, klass, attributes)
+        extract_model(elem, id_map, xpath, klass, attributes) {|model, e|
+          yield model, e if block_given?
+          refines = extract_attribute(e, 'refines')
+          if refines && refines[0] == '#'
+            id = refines[1..-1]
+            id_map[id] ||= {}
+            id_map[id][:refiners] ||= []
+            id_map[id][:refiners] << model
+          end
+        }
+      end
+    end
+  end
+end