peregrin 1.1.1 → 1.1.4

data/README.md

@@ -146,3 +146,8 @@ it's being kept as simple as possible.
 
 All this rhyming on "ook" put me in mind of the Took family. There is no
 deeper meaning.
+
+
+## History
+
+* 1.1.4 - Basic EPUB3 and EPUB fixed-layout read support (thanks @klacointe!)

data/lib/formats/epub.rb

@@ -7,7 +7,8 @@ class Peregrin::Epub
     :opf => { 'opf' => 'http://www.idpf.org/2007/opf' },
     :dc => { 'dc' => 'http://purl.org/dc/elements/1.1/' },
     :ncx => { 'ncx' => 'http://www.daisy.org/z3986/2005/ncx/' },
-    :svg => { 'svg' => 'http://www.w3.org/2000/svg' }
+    :svg => { 'svg' => 'http://www.w3.org/2000/svg' },
+    :nav => { 'nav' => 'http://www.w3.org/1999/xhtml'}
   }
   OCF_PATH = "META-INF/container.xml"
   HTML5_TAGNAMES = %w[section nav article aside hgroup header footer figure figcaption] # FIXME: Which to divify? Which to leave as-is?
@@ -85,13 +86,13 @@ class Peregrin::Epub
         docs = load_config_documents(zipfile)
         extract_properties(docs[:opf])
         extract_components(zipfile, docs[:opf], docs[:opf_root])
-        extract_chapters(zipfile, docs[:ncx])
+        extract_chapters(zipfile, {:ncx => docs[:ncx], :nav => docs[:nav]})
         extract_cover(zipfile, docs)
       }
       @book.read_resource_proc = lambda { |resource|
         media_path = from_opf_root(docs[:opf_root], resource.src)
         media_path = URI.unescape(media_path)
-        Zip::Archive.open(epub_path) { |zipfile| zipfile.read(media_path) }
+        Zip::Archive.open(epub_path) { |zipfile| zipfile.content(media_path) }
       }
     end
 
@@ -99,7 +100,8 @@ class Peregrin::Epub
     def load_config_documents(zipfile)
       # The OCF file.
       begin
-        docs = { :ocf => Nokogiri::XML::Document.parse(zipfile.read(OCF_PATH)) }
+        ocf_content = zipfile.content(OCF_PATH)
+        docs = { :ocf => Nokogiri::XML::Document.parse(ocf_content) }
       rescue
         raise FailureLoadingOCF
       end
@@ -111,12 +113,17 @@ class Peregrin::Epub
           NAMESPACES[:ocf]
         )['full-path']
         docs[:opf_root] = File.dirname(docs[:opf_path])
-        docs[:opf] = Nokogiri::XML::Document.parse(zipfile.read(docs[:opf_path]))
+        opf_content = zipfile.content(docs[:opf_path])
+        docs[:opf] = Nokogiri::XML::Document.parse(opf_content)
       rescue
         raise FailureLoadingOPF
       end
 
+      # Extract Epub version
+      @book.version = docs[:opf].at_xpath('//opf:package', NAMESPACES[:opf])['version'].to_f
+
       # The NCX file.
+      # Must be present only with Ebook < 3.0 but can be use for forward compatibility
       begin
         spine = docs[:opf].at_xpath('//opf:spine', NAMESPACES[:opf])
         ncx_id = spine['toc'] ? spine['toc'] : 'ncx'
@@ -126,10 +133,25 @@ class Peregrin::Epub
         )
 
         docs[:ncx_path] = from_opf_root(docs[:opf_root], item['href'])
-        ncx_content = zipfile.read(docs[:ncx_path])
+        ncx_content = zipfile.content(docs[:ncx_path])
         docs[:ncx] = Nokogiri::XML::Document.parse(ncx_content)
       rescue => e
-        raise FailureLoadingNCX
+        # Only raise an exeption for Ebook with version lower than 3.0
+        raise FailureLoadingNCX if @book.version < 3
+      end
+
+      # The NAV file. (Epub3 only)
+      if @book.version >= 3
+        begin
+          docs[:nav_path] = from_opf_root(
+            docs[:opf_root],
+            docs[:opf].at_xpath("//opf:manifest/opf:item[contains(concat(' ', normalize-space(@properties), ' '), ' nav ')]", NAMESPACES[:opf])['href']
+          )
+          nav_content = zipfile.content(docs[:nav_path])
+          docs[:nav] = Nokogiri::XML::Document.parse(nav_content)
+        rescue => e
+          raise FailureLoadingNAV
+        end
       end
 
       docs
@@ -153,10 +175,14 @@ class Peregrin::Epub
         end
         atts = elem.attributes.inject({}) { |acc, pair|
           key, attr = pair
-          acc[key] = attr.value  unless ["name", "content"].include?(key)
+          if !["name", "content", "property"].include?(key)
+            acc[key] = attr.value
+          elsif key == "property"
+            @book.add_property(attr.value, elem.text)
+          end
           acc
         }
-        @book.add_property(name, content, atts)
+        @book.add_property(name, content, atts) unless name.nil?
       }
     end
 
@@ -175,10 +201,10 @@ class Peregrin::Epub
           href = item['href']
           linear = iref['linear'] != 'no'
           begin
-            content = zipfile.read(from_opf_root(opf_root, href))
+            content = zipfile.content(from_opf_root(opf_root, href))
           rescue
             href = URI.unescape(href)
-            content = zipfile.read(from_opf_root(opf_root, href))
+            content = zipfile.content(from_opf_root(opf_root, href))
           end
           @book.add_component(
             href,
@@ -205,8 +231,16 @@ class Peregrin::Epub
       }
     end
 
+    def extract_chapters(zipfile, docs)
+      if @book.version >= 3 && !docs[:nav].nil?
+        extract_nav_chapters(zipfile, docs[:nav])
+      else
+        extract_ncx_chapters(zipfile, docs[:ncx])
+      end
+    end
 
-    def extract_chapters(zipfile, ncx_doc)
+    # Epub < 3.0 only
+    def extract_ncx_chapters(zipfile, ncx_doc)
       curse = lambda { |point|
         chp = Peregrin::Chapter.new(
           point.at_xpath('.//ncx:text', NAMESPACES[:ncx]).content,
@@ -225,6 +259,32 @@ class Peregrin::Epub
       }
     end
 
+    # Epub >= 3.0 only
+    def extract_nav_chapters(zipfile, nav_doc)
+      curse = lambda { |point, position|
+        chp = Peregrin::Chapter.new(
+          point.at_xpath('.//nav:a', NAMESPACES[:nav]).content,
+          position,
+          point.at_xpath('.//nav:a', NAMESPACES[:nav])['href']
+        )
+        ol = point.at_xpath('.//nav:ol', NAMESPACES[:nav])
+        ol.children.each { |pt|
+          next  unless pt.element? && pt.name == "li"
+          position += 1
+          position, chapter = curse.call(pt, position)
+          chp.children.push chapter
+        } if ol
+        [position, chp]
+      }
+      position = 0
+      nav_doc.at_xpath("//nav:nav/nav:ol", NAMESPACES[:nav]).children.each { |pt|
+        next  unless pt.element? && pt.name == "li"
+        position += 1
+        position, chapter = curse.call(pt, position)
+        @book.chapters.push chapter
+      }
+    end
+
 
     def extract_cover(zipfile, docs)
       @book.cover = nil
@@ -252,7 +312,7 @@ class Peregrin::Epub
       else
         path = from_opf_root(docs[:opf_root], res.src)
         begin
-          doc = Nokogiri::XML::Document.parse(zipfile.read(path))
+          doc = Nokogiri::XML::Document.parse(zipfile.content(path))
           src = nil
           if img = doc.at_css('img')
             src = img['src']
@@ -549,5 +609,6 @@ class Peregrin::Epub
   class FailureLoadingOCF < ValidationError; end
   class FailureLoadingOPF < ValidationError; end
   class FailureLoadingNCX < ValidationError; end
+  class FailureLoadingNAV < ValidationError; end
 
 end

data/lib/formats/zhook.rb

@@ -28,7 +28,7 @@ class Peregrin::Zhook
       raise MissingCoverPNG.new(path)
     end
 
-    doc = Nokogiri::HTML::Document.parse(zf.read(INDEX_PATH), nil, 'UTF-8')
+    doc = Nokogiri::HTML::Document.parse(zf.content(INDEX_PATH), nil, 'UTF-8')
     raise IndexHTMLRootHasId.new(path)  if doc.root['id']
 
   ensure
@@ -42,16 +42,14 @@ class Peregrin::Zhook
     validate(path)
     book = Peregrin::Book.new
     Zip::Archive.open(path) { |zf|
-      book.add_component(INDEX_PATH, zf.read(INDEX_PATH))
+      book.add_component(INDEX_PATH, zf.content(INDEX_PATH))
       zf.each { |entry|
         ze = entry.name
         book.add_resource(ze)  unless ze == INDEX_PATH || entry.directory?
       }
     }
     book.read_resource_proc = lambda { |resource|
-      Zip::Archive.open(path) { |zipfile|
-        zipfile.read(resource.src)
-      }
+      Zip::Archive.open(path) { |zipfile| zipfile.content(resource.src) }
     }
 
     extract_properties_from_index(book)

data/lib/peregrin/book.rb

@@ -19,6 +19,10 @@ class Peregrin::Book
   # A Resource that is used for the book cover.
   attr_accessor :cover
 
+  # The current version of document specifications
+  # Only used for Epub for now
+  attr_accessor :version
+
   # A proc that copies a resource to the given destination.
   attr_writer :read_resource_proc
 

data/lib/peregrin/version.rb

@@ -1,5 +1,5 @@
 module Peregrin
 
-  VERSION = "1.1.1"
+  VERSION = "1.1.4"
 
 end

data/lib/peregrin/zip_patch.rb

@@ -1,9 +1,10 @@
 class Zip::Archive
 
-  def read(path)
+  def content(path)
     fopen(path) { |f| f.read }
   end
 
+
   def find(path)
     detect { |f| f.name == path }
   end

data/test/formats/epub_test.rb

@@ -89,6 +89,91 @@ class Peregrin::Tests::EpubTest < Test::Unit::TestCase
     assert_equal("cover.png", epub.to_book.cover.src)
   end
 
+  def test_extracting_epub3_fixed_layout_properties
+    epub = Peregrin::Epub.read("test/fixtures/epubs/epub3_fixed_layout.epub")
+    book = epub.to_book
+    assert_equal("2012-05-09T08:58:00Z", book.property_for('dcterms:modified'))
+    assert_equal("pre-paginated", book.property_for('rendition:layout'))
+    assert_equal("auto", book.property_for('rendition:orientation'))
+    assert_equal("both", book.property_for('rendition:spread'))
+  end
+
+  def test_extracting_version
+    epub = Peregrin::Epub.read("test/fixtures/epubs/epub3_fixed_layout.epub")
+    assert_equal(3.0, epub.to_book.version)
+
+    epub = Peregrin::Epub.read("test/fixtures/epubs/strunk.epub")
+    assert_equal(2.0, epub.to_book.version)
+  end
+
+  def test_extracting_chapters_from_ocx
+    epub = Peregrin::Epub.read("test/fixtures/epubs/strunk.epub")
+    assert_equal(9, epub.to_book.chapters.count)
+    assert_equal("Title", epub.to_book.chapters.first.title)
+    assert_equal("title.xml", epub.to_book.chapters.first.src)
+    assert_equal(1, epub.to_book.chapters.first.position)
+    assert_equal("Recommendations", epub.to_book.chapters.last.title)
+    assert_equal("similar.xml", epub.to_book.chapters.last.src)
+    assert_equal(27, epub.to_book.chapters.last.position)
+  end
+
+  def test_extracting_chapters_from_nav
+    epub = Peregrin::Epub.read("test/fixtures/epubs/epub3_fixed_layout.epub")
+    assert_equal(3, epub.to_book.chapters.count)
+    assert_equal("Images and Text", epub.to_book.chapters.first.title)
+    assert_equal("page01.xhtml", epub.to_book.chapters.first.src)
+    assert_equal(1, epub.to_book.chapters.first.position)
+    assert_equal("Dragons", epub.to_book.chapters.last.title)
+    assert_equal("page04.xhtml", epub.to_book.chapters.last.src)
+    assert_equal(3, epub.to_book.chapters.last.position)
+  end
+
+  def test_extracting_nested_chapters_from_nav
+    epub = Peregrin::Epub.read("test/fixtures/epubs/epub3_nested_nav.epub")
+    assert_equal(11, epub.to_book.chapters.count)
+    assert_equal(
+      ["EPUB 3.0 Specification",
+       "EPUB 3 Specifications - Table of Contents",
+       "Terminology",
+       "EPUB 3 Overview",
+       "EPUB Publications 3.0",
+       "EPUB Content Documents 3.0",
+       "EPUB Media Overlays 3.0",
+       "Acknowledgements and Contributors",
+       "References",
+       "EPUB Open Container Format (OCF) 3.0",
+       "EPUB 3 Changes from EPUB 2.0.1"],
+      epub.to_book.chapters.map(&:title)
+    )
+    assert_equal(
+      [1, 2, 3, 4, 30, 85, 184, 230, 231, 232, 265],
+      epub.to_book.chapters.map(&:position)
+    )
+    assert_equal(
+      ["1. Introduction",
+       "2. Features",
+       "3. Global Language Support",
+       "4. Accessibility"],
+      epub.to_book.chapters[3].children.map(&:title)
+    )
+    assert_equal(
+      [5, 8, 22, 29],
+      epub.to_book.chapters[3].children.map(&:position)
+    )
+    assert_equal(
+      ["3.1. Metadata",
+       "3.2. Content Documents",
+       "3.3. CSS",
+       "3.4. Fonts",
+       "3.5. Text-to-speech",
+       "3.6. Container"],
+      epub.to_book.chapters[3].children[2].children.map(&:title)
+    )
+    assert_equal(
+      [23, 24, 25, 26, 27, 28],
+      epub.to_book.chapters[3].children[2].children.map(&:position)
+    )
+  end
 
   def test_read_epub_to_write_epub
     epub = Peregrin::Epub.read("test/fixtures/epubs/strunk.epub")

metadata

@@ -1,12 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: peregrin
 version: !ruby/object:Gem::Version 
-  prerelease: false
+  hash: 27
+  prerelease: 
   segments: 
   - 1
   - 1
-  - 1
-  version: 1.1.1
+  - 4
+  version: 1.1.4
 platform: ruby
 authors: 
 - Joseph Pearson
@@ -14,16 +15,17 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-06-28 00:00:00 +10:00
-default_executable: 
+date: 2012-05-14 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: nokogiri
   prerelease: false
   requirement: &id001 !ruby/object:Gem::Requirement 
+    none: false
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
+        hash: 3
         segments: 
         - 0
         version: "0"
@@ -33,9 +35,11 @@ dependencies:
   name: zipruby
   prerelease: false
   requirement: &id002 !ruby/object:Gem::Requirement 
+    none: false
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
+        hash: 3
         segments: 
         - 0
         version: "0"
@@ -45,9 +49,11 @@ dependencies:
   name: mime-types
   prerelease: false
   requirement: &id003 !ruby/object:Gem::Requirement 
+    none: false
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
+        hash: 3
         segments: 
         - 0
         version: "0"
@@ -57,9 +63,11 @@ dependencies:
   name: rake
   prerelease: false
   requirement: &id004 !ruby/object:Gem::Requirement 
+    none: false
     requirements: 
     - - ">="
       - !ruby/object:Gem::Version 
+        hash: 3
         segments: 
         - 0
         version: "0"
@@ -98,7 +106,6 @@ files:
 - test/utils/outliner_test.rb
 - README.md
 - MIT-LICENSE
-has_rdoc: true
 homepage: http://ochook.org/peregrin
 licenses: []
 
@@ -111,23 +118,27 @@ rdoc_options:
 require_paths: 
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement 
+  none: false
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
+      hash: 3
       segments: 
       - 0
       version: "0"
 required_rubygems_version: !ruby/object:Gem::Requirement 
+  none: false
   requirements: 
   - - ">="
     - !ruby/object:Gem::Version 
+      hash: 3
       segments: 
       - 0
       version: "0"
 requirements: []
 
 rubyforge_project: nowarning
-rubygems_version: 1.3.6
+rubygems_version: 1.8.24
 signing_key: 
 specification_version: 3
 summary: Peregrin - ebook conversion