epub-parser-io 0.1.6a

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

@@ -0,0 +1,170 @@
+require 'set'
+
+module EPUB
+  module Publication
+    class Package
+      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 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 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.selector {|refiner| refiner.property == voc}.first
+            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 generale 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 generale way to handle with identifier-type refiners
+          def isbn?
+            refiners.any? {|refiner|
+              refiner.property == 'identifier-type' and
+              refiner.scheme == 'onix:codelist5' and
+              %w[02 15].include? refiner.content
+            } or
+            scheme == 'ISBN' or
+            content.to_s.downcase.start_with? 'urn:isbn'
+          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)
+            @refines = refinee
+            refinee.refiners << self
+          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)
+            @refines = refinee
+            refinee.refiners << self
+          end
+        end
+      end
+    end
+  end
+end

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

@@ -0,0 +1,106 @@
+require 'set'
+
+module EPUB
+  module Publication
+    class Package
+      class Spine
+        include Inspector::PublicationModel
+        attr_accessor :package,
+                      :id, :toc, :page_progression_direction
+        attr_reader :itemrefs
+
+        def initialize
+          @itemrefs = []
+        end
+
+        # @return self
+        def <<(itemref)
+          itemref.spine = self
+          @itemrefs << itemref
+          self
+        end
+
+        # @yield [itemref]
+        # @yieldparam [Itemref] itemref
+        # @yieldreturn [Object] returns the last value of block
+        # @return [Object, Enumerator]
+        #   returns the last value of block when block given, Enumerator when not
+        def each_itemref
+          if block_given?
+            itemrefs.each {|itemref| yield itemref}
+          else
+            enum_for :each_itemref
+          end
+        end
+
+        # @return [Enumerator] Enumerator which yeilds {Manifest::Item}
+        #   referred by each of {#itemrefs}
+        def items
+          itemrefs.collector {|itemref| itemref.item}
+        end
+
+        class Itemref
+          PAGE_SPREAD_PROPERTIES = ['left'.freeze, 'right'.freeze].freeze
+          PAGE_SPREAD_PREFIX = 'page-spread-'.freeze
+
+          attr_accessor :spine,
+                        :idref, :linear, :id
+          attr_reader :properties
+
+          def initialize
+            @properties = Set.new
+          end
+
+          def properties=(props)
+            @properties = props.kind_of?(Set) ? props : Set.new(props)
+          end
+
+          # @return [true|false]
+          def linear?
+            !! linear
+          end
+
+          # @return [Package::Manifest::Item] item referred by this object
+          def item
+            @item ||= @spine.package.manifest[idref]
+          end
+
+          def item=(item)
+            self.idref = item.id
+            item
+          end
+
+          def ==(other)
+            [:spine, :idref, :id].all? {|meth|
+              self.__send__(meth) == other.__send__(meth)
+            } and
+              (linear? == other.linear?) and
+              (properties == other.properties)
+          end
+
+          # @return ["left", "right", nil]
+          def page_spread
+            property = properties.find {|prop| prop.start_with? PAGE_SPREAD_PREFIX}
+            property ? property.gsub(/\A#{Regexp.escape(PAGE_SPREAD_PREFIX)}/, '') : nil
+          end
+
+          # @param new_value ["left", "right", nil]
+          def page_spread=(new_value)
+            if new_value.nil?
+              properties.delete_if {|prop| prop.start_with? PAGE_SPREAD_PREFIX}
+              return new_value
+            end
+
+            raise "Unsupported page-spread property: #{new_value}" unless PAGE_SPREAD_PROPERTIES.include? new_value
+
+            props_to_be_deleted = (PAGE_SPREAD_PROPERTIES - [new_value]).map {|prop| "#{PAGE_SPREAD_PREFIX}#{prop}"}
+            properties.delete_if {|prop| props_to_be_deleted.include? prop}
+            new_property = "#{PAGE_SPREAD_PREFIX}#{new_value}"
+            properties << new_property unless properties.include? new_property
+            new_value
+          end
+        end
+      end
+    end
+  end
+end

data/lib/epub/publication/package.rb

@@ -0,0 +1,68 @@
+module EPUB
+  module Publication
+    class Package
+      include Inspector
+
+      CONTENT_MODELS = [:metadata, :manifest, :spine, :guide, :bindings]
+      RESERVED_VOCABULARY_PREFIXES = {
+        ''        => 'http://idpf.org/epub/vocab/package/#',
+        'dcterms' => 'http://purl.org/dc/terms/',
+        'marc'    => 'http://id.loc.gov/vocabulary/',
+        'media'   => 'http://www.idpf.org/epub/vocab/overlays/#',
+        'onix'    => 'http://www.editeur.org/ONIX/book/codelists/current.html#',
+        'xsd'     => 'http://www.w3.org/2001/XMLSchema#'
+      }
+
+
+      class << self
+        def define_content_model(model_name)
+          define_method "#{model_name}=" do |model|
+            current_model = __send__(model_name)
+            current_model.package = nil if current_model
+            model.package = self
+            instance_variable_set "@#{model_name}", model
+          end
+        end
+      end
+
+      attr_accessor :book, 
+                    :version, :prefix, :xml_lang, :dir, :id
+      attr_reader *CONTENT_MODELS
+      alias lang  xml_lang
+      alias lang= xml_lang=
+
+      CONTENT_MODELS.each do |model|
+        define_content_model model
+      end
+
+      def initialize
+        @prefix = {}
+      end
+
+      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
+
+EPUB::Publication::Package::CONTENT_MODELS.each do |f|
+  require_relative "package/#{f}"
+end

data/lib/epub/publication.rb

@@ -0,0 +1,2 @@
+require 'epub/publication/package'
+require 'epub/publication/fixed_layout'

data/lib/epub.rb

@@ -0,0 +1,14 @@
+require 'epub/inspector'
+require 'epub/ocf'
+require 'epub/publication'
+require 'epub/content_document'
+require 'epub/book/features'
+
+module EPUB
+  class << self
+    def included(base)
+      warn 'Including EPUB module is deprecated. Include EPUB::Book::Features instead.'
+      base.__send__ :include, EPUB::Book::Features
+    end
+  end
+end

data/man/epubinfo.1.ronn

@@ -0,0 +1,19 @@
+epubinfo(1) -- show metadata of an EPUB book
+============================================
+
+SYNOPSIS
+--------
+
+`epubinfo` <filename><br>
+`epubinfo` `-f`|`--format=<formnt>` <filename>
+
+DESCRIPTION
+-----------
+
+**epubinfo** shows metadata of an EPUB book.
+
+OPTIONS
+-------
+
+ * `-f`, `--format`:
+   Output format; currently "line"(for human reading), "json" and "yaml" are acceptable

data/schemas/epub-nav-30.rnc

@@ -0,0 +1,10 @@
+
+
+   default namespace = "http://www.w3.org/1999/xhtml"
+   namespace epub = "http://www.idpf.org/2007/ops"
+      
+   include "epub-xhtml-30.rnc" {   
+        html5.nav.content = html5.headings.class?, html5.ol     
+        html5.oli.content = html5.a.phrasing | ((html5.a.phrasing | html5.span), html5.ol)
+   }
+   

data/schemas/epub-nav-30.sch

@@ -0,0 +1,72 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<schema xmlns="http://purl.oclc.org/dsdl/schematron">
+
+    <ns uri="http://www.w3.org/1999/xhtml" prefix="html"/>
+    <ns uri="http://www.idpf.org/2007/ops" prefix="epub"/>
+
+    <pattern id="nav-ocurrence">
+        <rule context="html:body">
+            <assert test="count(.//html:nav[@epub:type='toc']) = 1">Exactly one 'toc' nav element
+                must be present</assert>
+            <assert test="count(.//html:nav[@epub:type='page-list']) &lt; 2">Multiple occurrences of
+                the 'page-list' nav element</assert>
+            <assert test="count(.//html:nav[@epub:type='landmarks']) &lt; 2">Multiple occurrences of
+                the 'landmarks' nav element</assert>
+        </rule>
+    </pattern>
+
+    <pattern id="span-no-sublist">
+        <rule context="html:body//html:nav//html:span">
+            <assert test="count(.//ol) = 0"> The span element must only be used as heading for flat
+                sublists (not hierarchical navigation structures) </assert>
+        </rule>
+    </pattern>
+
+    <pattern id="landmarks">
+        <rule context="html:nav[@epub:type='landmarks']//html:ol//html:a">
+            <assert test="@epub:type">Missing epub:type attribute on anchor inside 'landmarks' nav
+                element</assert>
+        </rule>
+    </pattern>
+
+    <pattern id="link-labels">
+        <rule context="html:nav//html:ol//html:a">
+            <assert test="string-length(normalize-space(string(.))) > 0">Anchors within nav elements
+                must contain text</assert>
+        </rule>
+    </pattern>
+
+    <pattern id="span-labels">
+        <rule context="html:nav//html:ol//html:span">
+            <assert test="string-length(normalize-space(string(.))) > 0">Spans within nav elements
+                must contain text</assert>
+        </rule>
+    </pattern>
+
+    <pattern id="req-heading">
+        <rule
+            context="html:nav[not(@epub:type = 'toc') and not (@epub:type = 'page-list') and not (@epub:type = 'landmarks')]">
+            <let name="fc" value="local-name(./*[1])"/>
+            <assert test="(starts-with($fc,'h') and string-length($fc) = 2) or ($fc = 'hgroup')">nav
+                elements other than 'toc', 'page-list' and 'landmarks' must contain a heading as the
+                first child</assert>
+        </rule>
+    </pattern>
+
+    <pattern id="heading-content">
+        <rule context="html:h1|html:h2|html:h3|html:h4|html:h5|html:h6|html:hgroup">
+            <assert test="string-length(normalize-space(string(.))) > 0">Heading elements must
+                contain text</assert>
+        </rule>
+    </pattern>
+
+
+    <!-- warnings mode <pattern id="page-list-flat">
+        <rule context="html:body//html:nav[@epub:type='page-list']">
+        <assert test="count(.//html:ol) = 1">The page-list navigation structure should be a
+        list, not a nested hierarchy</assert>
+        </rule>
+        </pattern> 
+    -->
+
+</schema>