epub-parser 0.1.4 → 0.1.5

data/epub-parser.gemspec

@@ -29,7 +29,9 @@ Gem::Specification.new do |s|
   s.add_development_dependency 'rake'
   s.add_development_dependency 'pry'
   s.add_development_dependency 'pry-doc'
-  s.add_development_dependency 'test-unit-full'
+  s.add_development_dependency 'test-unit'
+  s.add_development_dependency 'test-unit-rr'
+  s.add_development_dependency 'test-unit-notify'
   s.add_development_dependency 'simplecov'
   s.add_development_dependency 'thin'
   s.add_development_dependency 'yard'
@@ -42,6 +44,7 @@ Gem::Specification.new do |s|
 
   s.add_runtime_dependency 'enumerabler'
   s.add_runtime_dependency 'zipruby'
-  s.add_runtime_dependency 'nokogiri', '1.5.9'
+  s.add_runtime_dependency 'nokogiri', '~> 1.6'
   s.add_runtime_dependency 'addressable'
+  s.add_runtime_dependency 'method_decorators', '0.9.3'
 end

data/lib/epub.rb

@@ -1,3 +1,5 @@
+require 'method_decorators/deprecated'
+require 'epub/inspector'
 require 'epub/ocf'
 require 'epub/publication'
 require 'epub/content_document'
@@ -37,6 +39,11 @@ module EPUB
     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?
@@ -50,6 +57,11 @@ module EPUB
     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?
@@ -63,13 +75,14 @@ module EPUB
     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
+    ocf.container.rootfile.full_path.to_s
   end
 
   # Syntax sugar

data/lib/epub/content_document.rb

@@ -1,6 +1,2 @@
+require 'epub/content_document/xhtml'
 require 'epub/content_document/navigation'
-
-module EPUB
-  module ContentDocument
-  end
-end

data/lib/epub/content_document/navigation.rb

@@ -1,5 +1,3 @@
-require 'epub/content_document/xhtml'
-
 module EPUB
   module ContentDocument
     class Navigation < XHTML
@@ -11,15 +9,15 @@ module EPUB
       end
 
       def toc
-        items.selector {|nav| nav.type == Navigation::Type::TOC}.first
+        navigations.selector {|nav| nav.type == Navigation::Type::TOC}.first
       end
 
       def page_list
-        items.selector {|nav| nav.type == Nagivation::Type::PAGE_LIST}.first
+        navigations.selector {|nav| nav.type == Nagivation::Type::PAGE_LIST}.first
       end
 
       def landmarks
-        items.selector {|nav| nav.type == Navigation::Type::LANDMARKS}.first
+        navigations.selector {|nav| nav.type == Navigation::Type::LANDMARKS}.first
       end
 
       # Enumerator version of toc

data/lib/epub/content_document/xhtml.rb

@@ -3,15 +3,39 @@ module EPUB
     class XHTML
       attr_accessor :item
 
+      # @return [String] Returns the content string.
       def read
         item.read
       end
       alias raw_document read
 
-      # referenced directly from spine?
+      # @return [true|false] Whether referenced directly from spine or not.
       def top_level?
         !! item.itemref
       end
+
+      # @return [String] Returns the value of title element.
+      #                  If none, returns empty string
+      def title
+        title_elem = Nokogiri.XML(read).search('title').first
+        if title_elem
+          title_elem.text
+        else
+          warn 'title element not found'
+          ''
+        end
+      end
+
+      # @return [REXML::Document] content as REXML::Document object
+      def rexml
+        require 'rexml/document'
+        @rexml ||= REXML::Document.new(raw_document)
+      end
+
+      # @return [Nokogiri::XML::Document] content as Nokogiri::XML::Document object
+      def nokogiri
+        @nokogiri ||= Nokogiri.XML(raw_document)
+      end
     end
   end
 end

data/lib/epub/inspector.rb

@@ -0,0 +1,43 @@
+module EPUB
+  module Inspector
+    INSTANCE_VARIABLES_OPTION = {:exclude => []}
+
+    def inspect_simply
+      "#<%{class}:%{object_id}>" % {
+        :class => self.class,
+        :object_id => inspect_object_id
+      }
+    end
+
+    def inspect_object_id
+      (__id__ << 1).to_s(16)
+    end
+
+    def inspect_instance_variables(options={})
+      options = INSTANCE_VARIABLES_OPTION.merge(options)
+      exclude = options[:exclude]
+
+      (instance_variables - exclude).map {|name|
+        value = instance_variable_get(name)
+        "#{name}=#{value.inspect}"
+      }.join(' ')
+    end
+
+    module PublicationModel
+      class << self
+        def included(mod)
+          mod.__send__ :include, Inspector
+        end
+      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])
+          }
+      end
+    end
+  end
+end

data/lib/epub/ocf/container.rb

@@ -17,6 +17,8 @@ module EPUB
       class Rootfile
         attr_accessor :full_path, :media_type
 
+        # @param full_path [Addressable::URI|nil]
+        # @param media_type [String]
         def initialize(full_path=nil, media_type=EPUB::MediaType::ROOTFILE)
           @full_path, @media_type = full_path, media_type
         end

data/lib/epub/parser.rb

@@ -50,8 +50,6 @@ module EPUB
       Zip::Archive.open @filepath do |zip|
         @book.ocf = OCF.parse(zip)
         @book.package = Publication.parse(zip, @book.rootfile_path)
-        # @book.content_document =??? parse_content_document
-        # ...
       end
 
       @book

data/lib/epub/parser/content_document.rb

@@ -32,8 +32,6 @@ module EPUB
         # parse_content_document(document)
         if @item.nav?
           content_document.navigations = parse_navigations(document)
-        else
-          raise NotImplementedError
         end
         content_document
       end
@@ -71,12 +69,12 @@ module EPUB
               when 'canvas'
               when 'embed'
               when 'iframe'
-                item.text = (extract_attribute(embedded_content, 'name') || extract_attribute(embedded_content, 'srcdoc')).to_s
+                item.text = extract_attribute(embedded_content, 'name') || extract_attribute(embedded_content, 'srcdoc')
               when 'img'
-                item.text = extract_attribute(embedded_content, 'alt').to_s
+                item.text = extract_attribute(embedded_content, 'alt')
               when 'math'
               when 'object'
-                item.text = extract_attribute(embedded_content, 'name').to_s
+                item.text = extract_attribute(embedded_content, 'name')
               when 'svg'
               when 'video'
               else

data/lib/epub/parser/ocf.rb

@@ -39,10 +39,8 @@ module EPUB
         doc = Nokogiri.XML(xml)
         doc.xpath('/ocf:container/ocf:rootfiles/ocf:rootfile', EPUB::NAMESPACES).each do |elem|
           rootfile = EPUB::OCF::Container::Rootfile.new
-          %w[full-path media-type].each do |attr|
-            value = extract_attribute(elem, attr)
-            rootfile.__send__(attr.gsub(/-/, '_') + '=', value)
-          end
+          rootfile.full_path = Addressable::URI.parse(extract_attribute(elem, 'full-path'))
+          rootfile.media_type = extract_attribute(elem, 'media-type')
           container.rootfiles << rootfile
         end
 

data/lib/epub/parser/publication.rb

@@ -41,10 +41,7 @@ module EPUB
         end
         @unique_identifier_id = elem['unique-identifier']
         @package.prefix = parse_prefix(extract_attribute(elem, 'prefix'))
-        if @package.prefix.key? 'rendition'
-          require 'epub/publication/fixed_layout'
-          EPUB::Publication.__send__ :include, EPUB::Publication::FixedLayout
-        end
+        EPUB::Publication.__send__ :include, EPUB::Publication::FixedLayout if @package.prefix.key? EPUB::Publication::FixedLayout::PREFIX_KEY
 
         @package
       end
@@ -76,7 +73,11 @@ module EPUB
         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|
-          e.content
+          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}
 
@@ -199,8 +200,7 @@ module EPUB
         @doc.xpath('/opf:package/opf:bindings/opf:mediaType', EPUB::NAMESPACES).each do |elem|
           media_type = EPUB::Publication::Package::Bindings::MediaType.new
           media_type.media_type = extract_attribute(elem, 'media-type')
-          items = @package.manifest.items
-          media_type.handler = items.detect {|item| item.id == extract_attribute(elem, 'handler')}
+          media_type.handler = @package.manifest[extract_attribute(elem, 'handler')]
           bindings << media_type
         end
 

data/lib/epub/parser/version.rb

@@ -1,5 +1,5 @@
 module EPUB
   class Parser
-    VERSION = "0.1.4"
+    VERSION = "0.1.5"
   end
 end

data/lib/epub/publication.rb

@@ -1 +1,2 @@
 require 'epub/publication/package'
+require 'epub/publication/fixed_layout'

data/lib/epub/publication/package.rb

@@ -1,6 +1,8 @@
 module EPUB
   module Publication
     class Package
+      include Inspector
+
       CONTENT_MODELS = [:metadata, :manifest, :spine, :guide, :bindings]
       RESERVED_VOCABULARY_PREFIXES = {
         ''        => 'http://idpf.org/epub/vocab/package/#',
@@ -23,7 +25,7 @@ module EPUB
         end
       end
 
-      attr_accessor :book,
+      attr_accessor :book, 
                     :version, :prefix, :xml_lang, :dir, :id
       attr_reader *CONTENT_MODELS
       alias lang  xml_lang
@@ -40,6 +42,23 @@ module EPUB
       def unique_identifier
         @metadata.unique_identifier
       end
+
+      def inspect
+        "#<%{class}:%{object_id} %{attributes} %{models}>" % {
+          :class      => self.class,
+          :object_id  => inspect_object_id,
+          :attributes => inspect_instance_variables(exclude: CONTENT_MODELS.map {|model| :"@#{model}"}),
+          :models     => inspect_models
+        }
+      end
+
+      def inspect_models
+        CONTENT_MODELS.map {|name|
+          model = __send__(name)
+          representation = model.nil? ? model.inspect : model.inspect_simply
+          "@#{name}=#{representation}"
+        }.join(' ')
+      end
     end
   end
 end

data/lib/epub/publication/package/bindings.rb

@@ -2,10 +2,14 @@ module EPUB
   module Publication
     class Package
       class Bindings
+        include Inspector::PublicationModel
         attr_accessor :package
 
+        def initialize
+          @media_types = {}
+        end
+
         def <<(media_type)
-          @media_types ||= {}
           @media_types[media_type.media_type] = media_type
         end
 

data/lib/epub/publication/package/guide.rb

@@ -4,6 +4,7 @@ module EPUB
   module Publication
     class Package
       class Guide
+        include Inspector::PublicationModel
         attr_accessor :package
 
         def references

data/lib/epub/publication/package/manifest.rb

@@ -6,12 +6,17 @@ module EPUB
   module Publication
     class Package
       class Manifest
+        include Inspector::PublicationModel
+
         attr_accessor :package,
                       :id
 
+        def initialize
+          @items = {}
+        end
+
         # @return self
         def <<(item)
-          @items ||= {}
           item.manifest = self
           @items[item.id] = item
           self
@@ -29,6 +34,12 @@ module EPUB
           items.selector {|i| i.properties.include? 'cover-image'}.first
         end
 
+        def each_item
+          @items.each_value do |item|
+            yield item
+          end
+        end
+
         def items
           @items.values
         end
@@ -38,12 +49,15 @@ module EPUB
         end
 
         class Item
+          include Inspector
+
           # @!attribute [rw] manifest
           #   @return [Manifest] Returns the value of manifest
           # @!attribute [rw] id
           #   @return [String] Returns the value of id
           # @!attribute [rw] href
-          #   @return [Addressable::URI] Returns the value of href
+          #   @return [Addressable::URI] Returns the value of href,
+          #                              which is relative path from rootfile(OPF file)
           # @!attribute [rw] media_type
           #   @return [String] Returns the value of media_type
           # @!attribute [rw] properties
@@ -55,19 +69,31 @@ module EPUB
           attr_accessor :manifest,
                         :id, :href, :media_type, :fallback, :properties, :media_overlay
 
+          def initialize
+            @properties = []
+          end
+
           # @todo Handle circular fallback chain
           def fallback_chain
             @fallback_chain ||= traverse_fallback_chain([])
           end
 
+          # full path in archive
+          def entry_name
+            rootfile = manifest.package.book.ocf.container.rootfile.full_path
+            Addressable::URI.unescape(rootfile + href.normalize.request_uri)
+          end
+
           def read
-            rootfile = Addressable::URI.parse(manifest.package.book.ocf.container.rootfile.full_path)
             Zip::Archive.open(manifest.package.book.epub_file) {|zip|
-              path = Addressable::URI.unescape(rootfile + href.normalize.request_uri)
-              zip.fopen(path).read
+              zip.fopen(entry_name).read
             }
           end
 
+          def xhtml?
+            media_type == 'application/xhtml+xml'
+          end
+
           def nav?
             properties.include? 'nav'
           end
@@ -101,6 +127,15 @@ module EPUB
             manifest.package.spine.itemrefs.find {|itemref| itemref.idref == id}
           end
 
+          def inspect
+            "#<%{class}:%{object_id} %{manifest} %{attributes}>" % {
+              :class      => self.class,
+              :object_id  => inspect_object_id,
+              :manifest   => "@manifest=#{@manifest.inspect_simply}",
+              :attributes => inspect_instance_variables(exclude: [:@manifest])
+            }
+          end
+
           protected
 
           def traverse_fallback_chain(chain)