epub-parser-io 0.1.6a

data/docs/Publication.markdown

@@ -0,0 +1,54 @@
+{file:docs/Home.markdown} > **{file:docs/Publication.markdow}**
+
+Publication(Information about EPUB book)
+========================================
+
+EPUB Publications is information about EPUB books.
+
+EPUB Parser represents it as {EPUB::Publication} module and classes under the namespace and you can access them such like `EPUB::Parser.parse("path/to/book.epub").package`
+
+Let
+
+    book = EPUB::Parser.parse("path/to/book.epub")
+
+for continuing.
+
+Five Models
+-----------
+
+`book.package` is a package document, a root of information tree about the book, and it has attributes to access five major models of the publication; {EPUB::Publication::Package::Metadata Metadata}, {EPUB::Publication::Package::Manifest Manifest}, {EPUB::Publication::Package::Spine Spine}, {EPUB::Publication::Package::Guide Guide} and {EPUB::Publication::Package::Bindings Bindings}.
+
+Each of them has information the book in the way its own.
+
+Metadata
+--------
+
+{EPUB::Publication::Package::Metadata Metadata} is literally metadata of the book, including identifiers, titles, languages, links and so on.
+
+You can access them by:
+
+    md = book.package.metadata # => EPUB::Publication::Package::Metadata
+    md.titles # => [#<EPUB::Publication::Package::Metadata::Title...>, #<EPUB::Publication::Package::Metadata::Title...>, ...]
+    # ...
+
+Manifest
+--------
+
+Spine
+-----
+
+Guide
+-----
+
+Bindings
+--------
+
+Package
+-------
+
+References
+----------
+
+* [EPUB Publications 3.0][publications] on IDPF site
+
+[publications]: http://www.idpf.org/epub/30/spec/epub30-publications.html

data/epub-parser.gemspec

@@ -0,0 +1,49 @@
+# -*- encoding: utf-8 -*-
+$:.push File.expand_path("../lib", __FILE__)
+require "epub/parser/version"
+
+Gem::Specification.new do |s|
+  s.name        = "epub-parser-io"
+  s.version     = EPUB::Parser::VERSION
+  s.authors     = ["KITAITI Makoto, Brian Glusman"]
+  s.email       = ["KitaitiMakoto@gmail.com", "brian@glusman.me"]
+  s.homepage    = "https://github.com/bglusman/epub-parser"
+  s.summary     = %q{EPUB 3 Parser}
+  s.description = %q{Parse EPUB 3 book loosely}
+  s.license     = 'MIT'
+
+  # s.rubyforge_project = "epub-parser"
+
+  s.files         = `git ls-files`.split("\n")
+                    .push('test/fixtures/book/OPS/ルートファイル.opf')
+                    .push('test/fixtures/book/OPS/日本語.xhtml')
+                    .push(Dir['docs/*.md'])
+  s.files.reject! do |fn|
+    ['"test/fixtures/book/OPS/\343\203\253\343\203\274\343\203\210\343\203\225\343\202\241\343\202\244\343\203\253.opf"', '"test/fixtures/book/OPS/\346\227\245\346\234\254\350\252\236.xhtml"'].include? fn
+  end
+  s.test_files    = s.files & Dir['{test,spec,features}/**/*.{rb,feature}']
+  s.executables   = `git ls-files -- bin/*`.split("\n").map{ |f| File.basename(f) }
+  s.require_paths = ["lib"]
+  s.has_rdoc = 'yard'
+
+  s.add_development_dependency 'rake'
+  s.add_development_dependency 'pry'
+  s.add_development_dependency 'pry-doc'
+  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'
+  s.add_development_dependency 'gem-man'
+  s.add_development_dependency 'ronn'
+  s.add_development_dependency 'epzip'
+  s.add_development_dependency 'epubcheck'
+  s.add_development_dependency 'epub_validator'
+  s.add_development_dependency 'aruba'
+
+  s.add_runtime_dependency 'enumerabler'
+  s.add_runtime_dependency 'zipruby'
+  s.add_runtime_dependency 'nokogiri', '~> 1.6'
+  s.add_runtime_dependency 'addressable', '>= 2.3.5'
+end

data/features/epubinfo.feature

@@ -0,0 +1,6 @@
+Feature: We can see information about EPUB file
+
+Scenario: See info about existing EPUB file
+  Given the file "test/fixtures/book.epub" exists
+  When I successfully run `bundle exec epubinfo /home/ikeda/ruby/projects/epub-parser/test/fixtures/book.epub`
+  Then the stdout should contain "The New French Cuisine Masters"

data/features/step_definitions/epubinfo_steps.rb

@@ -0,0 +1,5 @@
+Given /^the file "([^"]*)" exists$/ do |file_name|
+  unless File.exist? File.join(File.dirname(__FILE__), '..', '..', file_name)
+    raise "File #{file_name} does not exist"
+  end
+end

data/features/support/env.rb

@@ -0,0 +1 @@
+require 'aruba/cucumber'

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/book.rb

@@ -0,0 +1,7 @@
+require 'epub'
+
+module EPUB
+  class Book
+    include EPUB::Book::Features
+  end
+end

data/lib/epub/constants.rb

@@ -0,0 +1,48 @@
+module EPUB
+  module Constants
+    NAMESPACES = {
+      'dc'    => 'http://purl.org/dc/elements/1.1/',
+      'ocf'   => 'urn:oasis:names:tc:opendocument:xmlns:container',
+      'opf'   => 'http://www.idpf.org/2007/opf',
+      'xhtml' => 'http://www.w3.org/1999/xhtml',
+      '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'
+    }
+
+    module MediaType
+      # @deprecated Use {UnsupportedMediaType} instead
+      class UnsupportedError < StandardError; end
+      class UnsupportedMediaType < StandardError; end
+
+      EPUB = 'application/epub+zip'
+      ROOTFILE = 'application/oebps-package+xml'
+      IMAGE = %w[
+        image/gif
+        image/jpeg
+        image/png
+        image/svg+xml
+      ]
+      APPLICATION = %w[
+        application/xhtml+xml
+        application/x-dtbncx+xml
+        application/vnd.ms-opentype
+        application/font-woff
+        application/smil+xml
+        application/pls+xml
+      ]
+      AUDIO = %w[
+        audio/mpeg
+        audio/mp4
+      ]
+      TEXT = %w[
+        text/css
+        text/javascript
+      ]
+      CORE = IMAGE + APPLICATION + AUDIO + TEXT
+    end
+  end
+
+  include Constants
+end

data/lib/epub/content_document/navigation.rb

@@ -0,0 +1,104 @@
+module EPUB
+  module ContentDocument
+    class Navigation < XHTML
+      attr_accessor :navigations
+
+      def initialize
+        @navigations = []
+        super
+      end
+
+      def toc
+        navigations.selector {|nav| nav.type == Navigation::Type::TOC}.first
+      end
+
+      def page_list
+        navigations.selector {|nav| nav.type == Nagivation::Type::PAGE_LIST}.first
+      end
+
+      def landmarks
+        navigations.selector {|nav| nav.type == Navigation::Type::LANDMARKS}.first
+      end
+
+      # Enumerator version of toc
+      #  Usage: nagivation.enum_for(:contents)
+      def contents
+      end
+
+      # Enumerator version of page_list
+      #  Usage: navigation.enum_for(:pages)
+      def pages
+      end
+
+      # iterator for #toc
+      def each_content
+      end
+
+      # iterator for #page_list
+      def each_page
+      end
+
+      # iterator for #landmark
+      def each_landmark
+      end
+
+      def navigation
+        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
+        include Hidable
+
+        attr_accessor :items, :text,
+                      :content_document, :href, :item
+
+        def initialize
+          @items = ItemList.new
+          @items.parent = self
+        end
+
+        def traverse(depth=0, &block)
+          block.call self, depth
+          items.each do |item|
+            item.traverse depth + 1, &block
+          end
+        end
+      end
+
+      class Navigation < Item
+        module Type
+          TOC       = 'toc'
+          PAGE_LIST = 'page_list'
+          LANDMARKS = 'landmarks'
+        end
+
+        attr_accessor :type
+        alias navigations items
+        alias navigations= items=
+        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

@@ -0,0 +1,41 @@
+module EPUB
+  module ContentDocument
+    class XHTML
+      attr_accessor :item
+
+      # @return [String] Returns the content string.
+      def read
+        item.read
+      end
+      alias raw_document read
+
+      # @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.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/content_document.rb

@@ -0,0 +1,2 @@
+require 'epub/content_document/xhtml'
+require 'epub/content_document/navigation'

data/lib/epub/inspector.rb

@@ -0,0 +1,45 @@
+module EPUB
+  module Inspector
+    INSTANCE_VARIABLES_OPTION = {:exclude => []}
+    SIMPLE_TEMPLATE = "#<%{class}:%{object_id}>"
+
+    def inspect_simply
+      SIMPLE_TEMPLATE % {
+        :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
+      TEMPLATE = "#<%{class}:%{object_id} @package=%{package} %{attributes}>"
+      class << self
+        def included(mod)
+          mod.__send__ :include, Inspector
+        end
+      end
+
+      def inspect
+        TEMPLATE % {
+          :class      => self.class,
+          :package    => package.inspect_simply,
+          :object_id  => inspect_object_id,
+          :attributes => inspect_instance_variables(exclude: [:@package])
+        }
+      end
+    end
+  end
+end