epub-parser 0.1.5 → 0.1.6

data/lib/epub/book.rb

@@ -2,6 +2,6 @@ require 'epub'
 
 module EPUB
   class Book
-    include EPUB
+    include EPUB::Book::Features
   end
 end

data/lib/epub/book/features.rb

@@ -0,0 +1,85 @@
+module EPUB
+  class Book
+    module Features
+      modules = [:ocf, :package]
+      attr_reader *modules
+      attr_accessor :epub_file
+      modules.each do |mod|
+        define_method "#{mod}=" do |obj|
+          instance_variable_set "@#{mod}", obj
+          obj.book = self
+        end
+      end
+
+      Publication::Package::CONTENT_MODELS.each do |model|
+        define_method model do
+          package.__send__(model)
+        end
+      end
+
+      %w[ title main_title subtitle short_title collection_title edition_title extended_title description date unique_identifier ].each do |met|
+        define_method met do
+          metadata.__send__(met)
+        end
+      end
+
+      %w[nav].each do |met|
+        define_method met do
+          manifest.__send__ met
+        end
+      end
+
+      # @overload each_page_on_spine(&blk)
+      #   iterate over items in order of spine when block given
+      #   @yieldparam item [Publication::Package::Manifest::Item]
+      # @overload each_page_on_spine
+      #   @return [Enumerator] which iterates over {Publication::Package::Manifest::Item}s in order of spine when block not given
+      def each_page_on_spine(&blk)
+        enum = package.spine.items
+        if block_given?
+          enum.each &blk
+        else
+          enum
+        end
+      end
+
+      def each_page_on_toc(&blk)
+        raise NotImplementedError
+      end
+
+      # @overload each_content(&blk)
+      #   iterate all items over when block given
+      #   @yieldparam item [Publication::Package::Manifest::Item]
+      # @overload each_content
+      #   @return [Enumerator] which iterates over all {Publication::Package::Manifest::Item}s in EPUB package when block not given
+      def each_content(&blk)
+        enum = manifest.items
+        if block_given?
+          enum.each &blk
+        else
+          enum.to_enum
+        end
+      end
+
+      def other_navigation
+        raise NotImplementedError
+      end
+
+      # @return [Array<Publication::Package::Manifest::Item>] All {Publication::Package::Manifest::Item}s in EPUB package
+      def resources
+        manifest.items
+      end
+
+      # Syntax sugar
+      def rootfile_path
+        ocf.container.rootfile.full_path.to_s
+      end
+
+      # Syntax sugar
+      def cover_image
+        manifest.cover_image
+      end
+
+    end
+  end
+end

data/lib/epub/constants.rb

@@ -12,7 +12,9 @@ module EPUB
     }
 
     module MediaType
+      # @deprecated Use {UnsupportedMediaType} instead
       class UnsupportedError < StandardError; end
+      class UnsupportedMediaType < StandardError; end
 
       EPUB = 'application/epub+zip'
       ROOTFILE = 'application/oebps-package+xml'

data/lib/epub/content_document/navigation.rb

@@ -46,16 +46,34 @@ module EPUB
         navigations.first
       end
 
+      module Hidable
+        attr_accessor :hidden, :parent
+
+        def hidden?
+          if @hidden.nil?
+            @parent ? @parent.hidden? : false
+          else
+            true
+          end
+        end
+      end
+
       class Item
-        attr_accessor :items, :text, :hidden,
+        include Hidable
+
+        attr_accessor :items, :text,
                       :content_document, :href, :item
 
         def initialize
-          @items = []
+          @items = ItemList.new
+          @items.parent = self
         end
 
-        def hidden?
-          !! hidden
+        def traverse(depth=0, &block)
+          block.call self, depth
+          items.each do |item|
+            item.traverse depth + 1, &block
+          end
         end
       end
 
@@ -72,6 +90,15 @@ module EPUB
         alias heading text
         alias heading= text=
       end
+
+      class ItemList < Array
+        include Hidable
+
+        def <<(item)
+          super
+          item.parent = self
+        end
+      end
     end
   end
 end

data/lib/epub/content_document/xhtml.rb

@@ -17,7 +17,7 @@ module EPUB
       # @return [String] Returns the value of title element.
       #                  If none, returns empty string
       def title
-        title_elem = Nokogiri.XML(read).search('title').first
+        title_elem = nokogiri.search('title').first
         if title_elem
           title_elem.text
         else

data/lib/epub/inspector.rb

@@ -1,9 +1,10 @@
 module EPUB
   module Inspector
     INSTANCE_VARIABLES_OPTION = {:exclude => []}
+    SIMPLE_TEMPLATE = "#<%{class}:%{object_id}>"
 
     def inspect_simply
-      "#<%{class}:%{object_id}>" % {
+      SIMPLE_TEMPLATE % {
         :class => self.class,
         :object_id => inspect_object_id
       }
@@ -24,6 +25,7 @@ module EPUB
     end
 
     module PublicationModel
+      TEMPLATE = "#<%{class}:%{object_id} @package=%{package} %{attributes}>"
       class << self
         def included(mod)
           mod.__send__ :include, Inspector
@@ -31,12 +33,12 @@ module EPUB
       end
 
       def inspect
-          "#<%{class}:%{object_id} @package=%{package} %{attributes}>" % {
-            :class      => self.class,
-            :package    => package.inspect_simply,
-            :object_id  => inspect_object_id,
-            :attributes => inspect_instance_variables(exclude: [:@package])
-          }
+        TEMPLATE % {
+          :class      => self.class,
+          :package    => package.inspect_simply,
+          :object_id  => inspect_object_id,
+          :attributes => inspect_instance_variables(exclude: [:@package])
+        }
       end
     end
   end

data/lib/epub/parser.rb

@@ -49,7 +49,7 @@ module EPUB
     def parse
       Zip::Archive.open @filepath do |zip|
         @book.ocf = OCF.parse(zip)
-        @book.package = Publication.parse(zip, @book.rootfile_path)
+        @book.package = Publication.parse(zip, @book.ocf.container.rootfile.full_path.to_s)
       end
 
       @book

data/lib/epub/parser/content_document.rb

@@ -47,8 +47,12 @@ module EPUB
       def parse_navigation(element)
         nav = EPUB::ContentDocument::Navigation::Navigation.new
         nav.text = find_heading(element)
+        hidden = extract_attribute(element, 'hidden')
+        nav.hidden = hidden.nil? ? nil : true
         nav.type = extract_attribute(element, 'type', 'epub')
-        nav.items = element.xpath('./xhtml:ol/xhtml:li', EPUB::NAMESPACES).map {|elem| parse_navigation_item(elem)}
+        element.xpath('./xhtml:ol/xhtml:li', EPUB::NAMESPACES).map do |elem|
+          nav.items << parse_navigation_item(elem)
+        end
 
         nav
       end

data/lib/epub/parser/ocf.rb

@@ -9,7 +9,7 @@ module EPUB
       include Utils
 
       DIRECTORY = 'META-INF'
-      EPUB::OCF::MODULES.each {|m| self.const_set "#{m.upcase}_FILE", "#{m}.xml"}
+      EPUB::OCF::MODULES.each {|m| self.const_set "#{m.upcase}_FILE", "#{m}.xml"} # Deprecated
 
       class << self
         def parse(zip_archive)
@@ -25,7 +25,7 @@ module EPUB
       def parse
         EPUB::OCF::MODULES.each do |m|
           begin
-            file = @zip.fopen(File.join(DIRECTORY, self.class.const_get("#{m.upcase}_FILE")))
+            file = @zip.fopen(File.join(DIRECTORY, "#{m}.xml"))
             @ocf.__send__ "#{m}=", __send__("parse_#{m}", file.read)
           rescue Zip::Error
           end

data/lib/epub/parser/publication.rb

@@ -24,12 +24,9 @@ module EPUB
       end
 
       def parse
-        parse_package
-        parse_metadata
-        parse_manifest
-        parse_spine
-        parse_guide
-        parse_bindings
+        ([:package] + EPUB::Publication::Package::CONTENT_MODELS).each do |model|
+          __send__ "parse_#{model}"
+        end
 
         @package
       end
@@ -51,80 +48,21 @@ module EPUB
         elem = @doc.xpath('/opf:package/opf:metadata', EPUB::NAMESPACES).first
         id_map = {}
 
-        metadata.identifiers = elem.xpath('./dc:identifier', EPUB::NAMESPACES).collect do |e|
-          identifier = EPUB::Publication::Package::Metadata::DCMES.new
-          identifier.content = e.content
-          identifier.id = id = extract_attribute(e, 'id')
-          metadata.unique_identifier = identifier if id == @unique_identifier_id
-
-          identifier
-        end
-        metadata.identifiers.each {|i| id_map[i.id] = {metadata: i} if i.respond_to?(:id) && i.id}
-
-        metadata.titles = elem.xpath('./dc:title', EPUB::NAMESPACES).collect do |e|
-          title = EPUB::Publication::Package::Metadata::Title.new
-          %w[ id lang dir ].each do |attr|
-            title.__send__("#{attr}=", extract_attribute(e, attr))
-          end
-          title.content = e.content
-
-          title
-        end
-        metadata.titles.each {|t| id_map[t.id] = {metadata: t} if t.respond_to?(:id) && t.id}
-
-        metadata.languages = elem.xpath('./dc:language', EPUB::NAMESPACES).collect do |e|
-          language = EPUB::Publication::Package::Metadata::DCMES.new
-          language.content = e.content
-          language.id = e['id'] if e['id']
-
-          language
-        end
-        metadata.languages.each {|l| id_map[l.id] = {metadata: l} if l.respond_to?(:id) && l.id}
-
+        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=", collect_dcmes(elem, "./dc:#{dcmes}")
-          metadata.__send__("#{dcmes}s").each {|d| id_map[d.id] = {metadata: d} if d.respond_to?(:id) && d.id}
+          metadata.__send__ "#{dcmes}s=", extract_model(elem, id_map, "./dc:#{dcmes}")
         end
-
-        metadata.rights = collect_dcmes(elem, './dc:rights')
-        metadata.rights.each {|r| id_map[r.id] = {metadata: r} if r.respond_to?(:id) && r.id}
-
-        metadata.metas = elem.xpath('./opf:meta', EPUB::NAMESPACES).collect do |e|
-          meta = EPUB::Publication::Package::Metadata::Meta.new
-          %w[property id scheme].each do |attr|
-            meta.__send__ "#{attr}=", extract_attribute(e, attr)
-          end
-          meta.content = e.content
-          refines = extract_attribute(e, 'refines')
-          if refines && refines[0] == '#'
-            id = refines[1..-1]
-            id_map[id] ||= {}
-            id_map[id][:refiners] ||= []
-            id_map[id][:refiners] << meta
-          end
-
-          meta
-        end
-        metadata.metas.each {|m| id_map[m.id] = {metadata: m} if m.respond_to?(:id) && m.id}
-
-        metadata.links = elem.xpath('./opf:link', EPUB::NAMESPACES).collect do |e|
-          link = EPUB::Publication::Package::Metadata::Link.new
-          %w[ id media-type ].each do |attr|
-            link.__send__ (attr.gsub(/-/, '_') + '='), extract_attribute(e, attr)
-          end
+        metadata.rights = extract_model(elem, id_map, './dc:rights')
+        metadata.metas = extract_refinee(elem, id_map, './opf:meta', :Meta, %w[property id scheme])
+        metadata.links = extract_refinee(elem, id_map, './opf:link', :Link, %w[id media-type]) {|link, e|
           link.href = Addressable::URI.parse(extract_attribute(e, 'href'))
-          link.rel = extract_attribute(e, 'rel').strip.split
-          refines = extract_attribute(e, 'refines')
-          if refines && refines[0] == '#'
-            id = refines[1..-1]
-            id_map[id] ||= {}
-            id_map[id][:refiners] ||= []
-            id_map[id][:refiners] << link
-          end
-
-          link
-        end
-        metadata.links.each {|l| id_map[l.id] = {metadata: l} if l.respond_to?(:id) && l.id}
+          link.rel = Set.new(extract_attribute(e, 'rel').split(nil))
+        }
 
         id_map.values.each do |hsh|
           next unless hsh[:refiners]
@@ -150,7 +88,7 @@ module EPUB
           fallback = extract_attribute(e, 'fallback')
           fallback_map[fallback] = item if fallback
           properties = extract_attribute(e, 'properties')
-          item.properties = properties ? properties.split(' ') : []
+          item.properties = properties.split(' ') if properties
           manifest << item
         end
         fallback_map.each_pair do |id, from|
@@ -174,7 +112,7 @@ module EPUB
           end
           itemref.linear = (extract_attribute(e, 'linear') != 'no')
           properties = extract_attribute(e, 'properties')
-          itemref.properties = properties ? properties.split(' ') : []
+          itemref.properties = properties.split(' ') if properties
           spine << itemref
         end
 
@@ -225,16 +163,37 @@ module EPUB
         prefixes
       end
 
-      def collect_dcmes(elem, selector)
-        elem.xpath(selector, EPUB::NAMESPACES).collect do |e|
-          md = EPUB::Publication::Package::Metadata::DCMES.new
-          md.content = e.content
-          %w[ id lang dir ].each do |attr|
-            md.__send__ "#{attr}=", extract_attribute(e, attr)
+      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
-          yield(md, e) if block_given?
-          md
+          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

data/lib/epub/parser/version.rb

@@ -1,5 +1,5 @@
 module EPUB
   class Parser
-    VERSION = "0.1.5"
+    VERSION = "0.1.6"
   end
 end

data/lib/epub/publication/fixed_layout.rb

@@ -124,8 +124,7 @@ module EPUB
             base.__send__ :define_method, :page_spread do
               property = page_spread_without_fixed_layout
               return property if property
-              property = properties.find {|prop| prop == prefixed_page_spread_property}
-              property ? PAGE_SPREAD_PROPERTY : nil
+              properties.include?(prefixed_page_spread_property) ? PAGE_SPREAD_PROPERTY : nil
             end
 
             base.__send__ :define_method, :page_spread= do |new_value|
@@ -162,7 +161,7 @@ module EPUB
             values_to_be_deleted = (values - [new_value]).map {|value| "#{rendition_property_prefix}#{value}"}
             properties.delete_if {|prop| values_to_be_deleted.include? prop}
             new_property = "#{rendition_property_prefix}#{new_value}"
-            properties << new_property unless properties.any? {|prop| prop == new_property}
+            properties << new_property unless properties.include? new_property
             new_value
           end
         end