class Softcover::Builders::Epub

Public Instance Methods

build!(options={}) click to toggle source
# File lib/softcover/builders/epub.rb, line 145
def build!(options={})
  @preview = options[:preview]
  Softcover::Builders::Html.new.build!
  if manifest.markdown?
    opts = options.merge({ source: :polytex, origin: :markdown })
    self.manifest = Softcover::BookManifest.new(opts)
  end
  remove_html
  remove_images
  create_directories
  write_mimetype
  write_container_xml
  write_ibooks_xml
  copy_image_files
  write_html(options)
  write_contents(options)
  create_style_files(options)
  write_toc
  write_nav
  make_epub(options)
  move_epub
end
chapter_name(n) click to toggle source
# File lib/softcover/builders/epub.rb, line 603
def chapter_name(n)
  n == 0 ? language_labels["frontmatter"] : strip_html(chapters[n].menu_heading)
end
chapter_template(title, content) click to toggle source

Returns the HTML template for a chapter.

# File lib/softcover/builders/epub.rb, line 647
def chapter_template(title, content)
  %(<?xml version="1.0" encoding="utf-8"?>
  <!DOCTYPE html>

  <html xmlns="http://www.w3.org/1999/xhtml">
  <head>
    <title>#{title}</title>
    <link rel="stylesheet" href="styles/pygments.css" type="text/css" />
    <link rel="stylesheet" href="styles/softcover.css" type="text/css" />
    <link rel="stylesheet" href="styles/epub.css" type="text/css" />
    <link rel="stylesheet" href="styles/custom.css" type="text/css"/>
    <link rel="stylesheet" href="styles/custom_epub.css" type="text/css"/>
    <link rel="stylesheet" type="application/vnd.adobe-page-template+xml" href="styles/page-template.xpgt" />
  </head>

  <body>
    #{content}
  </body>
  </html>)
end
chapters() click to toggle source

Returns the chapters to write.

# File lib/softcover/builders/epub.rb, line 221
def chapters
  preview? ? manifest.preview_chapters : manifest.chapters
end
clean_book_id(filename) click to toggle source

Removes the '#book' CSS id. For some reason, EPUB books hate the book ids in the stylesheet (i.e., such books fail to validate), so remove them.

# File lib/softcover/builders/epub.rb, line 443
def clean_book_id(filename)
  File.read(filename).gsub(/#book /, '')
end
container_xml() click to toggle source
# File lib/softcover/builders/epub.rb, line 513
      def container_xml
%(<?xml version="1.0"?>
<container version="1.0" xmlns="urn:oasis:names:tc:opendocument:xmlns:container">
    <rootfiles>
        <rootfile full-path="OEBPS/content.opf" media-type="application/oebps-package+xml"/>
   </rootfiles>
</container>)
      end
content_opf(options={}) click to toggle source

Returns the content configuration file.

# File lib/softcover/builders/epub.rb, line 532
def content_opf(options={})
  man_ch = chapters.map do |chapter|
             %(<item id="#{chapter.slug}" href="#{xhtml(chapter.fragment_name)}" media-type="application/xhtml+xml"/>)
           end
    toc_ch = chapters.map do |chapter|
               %(<itemref idref="#{chapter.slug}"/>)
             end
  image_files = Dir['epub/OEBPS/images/**/*'].select { |f| File.file?(f) }
  images = image_files.map do |image|
             ext = File.extname(image).sub('.', '')   # e.g., 'png'
             ext = 'jpeg' if ext == 'jpg'
             # Strip off the leading 'epub/OEBPS'.
             sep  = File::SEPARATOR
             href = image.split(sep)[2..-1].join(sep)
             # Define an id based on the filename.
             # Prefix with 'img-' in case the filname starts with an
             # invalid character such as a number.
             label = File.basename(image).gsub('.', '-')
             id = "img-#{label}"
             %(<item id="#{id}" href="#{href}" media-type="image/#{ext}"/>)
           end

  manifest.html_title
  content_opf_template(manifest.html_title, manifest.copyright,
                       manifest.author, manifest.uuid, cover_id(options),
                       toc_ch, man_ch, images)
end
copy_image_files() click to toggle source

Copies the image files from the HTML version of the document.

# File lib/softcover/builders/epub.rb, line 448
def copy_image_files
  # Copy over all images to guarantee the same directory structure.
  FileUtils.cp_r(File.join('html', 'images'),
                 File.join('epub', 'OEBPS'))
  # Parse the full HTML file with Nokogiri to get images actually used.
  html = File.read(manifest.full_html_file)
  html_image_filenames = Nokogiri::HTML(html).css('img').map do |node|
                           node.attributes['src'].value
                         end
  # Form the corresponding EPUB image paths.
  used_image_filenames = html_image_filenames.map do |filename|
                           "epub/OEBPS/#{filename}"
                         end.to_set
  # Delete unused images.
  Dir.glob("epub/OEBPS/images/**/*").each do |image|
    next if File.directory?(image)
    rm image unless used_image_filenames.include?(image)
  end
end
cover_id(options) click to toggle source
# File lib/softcover/builders/epub.rb, line 576
def cover_id(options)
  cover?(options) ? "img-#{cover_img.sub('.', '-')}" : nil
end
cover_page() click to toggle source
# File lib/softcover/builders/epub.rb, line 560
      def cover_page
%(<?xml version="1.0" encoding="utf-8"?>
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
  <title>Cover</title>
</head>
<body>
  <div id="cover">
     <img width="573" height="800" src="images/#{cover_img}" alt="cover" />
  </div>
</body>
</html>
)
      end
create_directories() click to toggle source
# File lib/softcover/builders/epub.rb, line 185
def create_directories
  mkdir('epub')
  mkdir(path('epub/OEBPS'))
  mkdir(path('epub/OEBPS/styles'))
  mkdir(path('epub/META-INF'))
  mkdir(images_dir)
  mkdir('ebooks')
end
create_style_files(options) click to toggle source
# File lib/softcover/builders/epub.rb, line 422
def create_style_files(options)
  html_styles = File.join('html', 'stylesheets')
  epub_styles = File.join('epub', 'OEBPS', 'styles')

  FileUtils.cp(File.join(html_styles, 'pygments.css'), epub_styles)
  File.write(File.join(epub_styles, 'softcover.css'),
             clean_book_id(path("#{html_styles}/softcover.css")))

  # Copy over the EPUB-specific CSS.
  template_dir = Softcover::Utils.template_dir(options)
  epub_css     = File.join(template_dir, epub_styles, 'epub.css')
  FileUtils.cp(epub_css, epub_styles)

  # Copy over custom CSS.
  File.write(File.join(epub_styles, 'custom.css'),
             clean_book_id(path("#{html_styles}/custom.css")))
end
html_with_math(chapter, images_dir, texmath_dir, pngs, options={}) click to toggle source

Returns HTML for HTML source that includes math. As a side-effect, html_with_math creates PNGs corresponding to any math in the given source. The technique involves using PhantomJS to hit the HTML source for each page containing math to create SVGs for every math element. Since ereader support for SVGs is spotty, they are then converted to PNGs using Inkscape. The filenames are SHAs of their contents, which arranges both for unique filenames and for automatic disk caching.

# File lib/softcover/builders/epub.rb, line 284
def html_with_math(chapter, images_dir, texmath_dir, pngs, options={})
  content = File.read(File.join("html",
                                "#{chapter.slug}.#{html_extension}"))
  pagejs = "#{File.dirname(__FILE__)}/utils/page.js"
  url = "file://#{Dir.pwd}/html/#{chapter.slug}.#{html_extension}"
  cmd = "#{phantomjs} #{pagejs} #{url}"
  silence { silence_stream(STDERR) { system cmd } }
  # Sometimes in tests the phantomjs_source.html file is missing.
  # It shouldn't ever happen, but it does no harm to skip it.
  return nil unless File.exist?('phantomjs_source.html')
  raw_source = File.read('phantomjs_source.html')
  source = strip_attributes(Nokogiri::HTML(raw_source))
  rm 'phantomjs_source.html'
  # Remove the first body div, which is the hidden MathJax SVGs.
  if (mathjax_svgs = source.at_css('body div'))
    mathjax_svgs.remove
  else
    # There's not actually any math, so return nil.
    return nil
  end
  # Remove all the unneeded raw TeX displays.
  source.css('script').each(&:remove)
  # Remove all the MathJax preview spans.
  source.css('MathJax_Preview').each(&:remove)
  # Suck out all the SVGs
  svgs   = source.css('div#book svg')
  frames = source.css('span.MathJax_SVG')
  svgs.zip(frames).each do |svg, frame|
    # Save the SVG file.
    svg['viewBox'] = svg['viewbox']
    svg.remove_attribute('viewbox')
    # Workaround for bug in Inkscape 0.91 on MacOS X:
    # extract height/width from svg attributes and move them to style attr
    svg_height = svg['height']  # in ex
    svg_width  = svg['width']   # in ex
    svg['style'] += ' height:'+svg_height+';' + ' width:'+svg_width+';'
    svg.remove_attribute('height')
    svg.remove_attribute('width')
    # /Workaround
    first_child = frame.children.first
    first_child.replace(svg) unless svg == first_child
    output = svg.to_xhtml
    svg_filename = File.join(texmath_dir, "#{digest(output)}.svg")
    svg_abspath  = File.join("#{Dir.pwd}", svg_filename)
    File.write(svg_filename, output)
    # Convert to PNG named:
    png_filename = svg_filename.sub('.svg', '.png')
    png_abspath  = svg_abspath.sub('.svg', '.png')
    pngs << png_filename
    #
    # Settings for texmath images in ePub / mobi
    ex2em_height_scaling = 0.51     # =1ex/1em for math png height
    ex2em_valign_scaling = 0.481482 # =1ex/1em for math png vertical-align
    ex2px_scale_factor   = 20       # =1ex/1px scaling for SVG-->PNG conv.
    # These are used a three-step process below: Compute, Convert, Replace
    # STEP1: compute height and vertical-align in `ex` units
    svg_height_in_ex = Float(svg_height.gsub('ex',''))
    # MathJax sets SVG height in `ex` units but we want `em` units for PNG
    png_height = (svg_height_in_ex * ex2em_height_scaling).to_s + 'em'
    # Extract vertical-align css proprty for inline math equations:
    if svg.parent.parent.attr('class') == "inline_math"
      vertical_align = svg['style'].scan(/vertical-align: (.*?);/)
      vertical_align = vertical_align.flatten.first
      if vertical_align
        valign_in_ex = Float(vertical_align.gsub('ex',''))
        png_valign = (valign_in_ex * ex2em_valign_scaling).to_s + 'em'
      else
        png_valign = "0em"
      end
    else # No vertical align for displayed math
      png_valign = nil
    end
    # STEP2: Generate PNG from each SVG (unless PNG exists already).
    unless File.exist?(png_filename)
      unless options[:silent] || options[:quiet]
        puts "Creating #{png_filename}"
      end
      # Generate png from the MathJax_SVG using Inkscape
      # Use the -d option to get a sensible size:
      #   Resolution for bitmaps and rasterized filters
      cmd = "#{inkscape} #{svg_abspath} -o #{png_abspath} -d 2"
      if options[:silent]
        silence { silence_stream(STDERR) { system cmd } }
      else
        puts cmd
        silence_stream(STDERR) { system cmd }
      end
    end
    rm svg_filename
    # STEP 3: Replace svg element with an equivalent png.
    png = Nokogiri::XML::Node.new('img', source)
    png['src']   = File.join('images', 'texmath',
                             File.basename(png_filename))
    png['alt']   = png_filename.sub('.png', '')
    png['style'] = 'height:' + png_height + ';'
    if png_valign
      png['style'] += ' vertical-align:' + png_valign + ';'
    end
    svg.replace(png)
  end
  # Make references relative.
  source.css('a.hyperref').each do |ref_node|
    ref_node['href'] = ref_node['href'].sub('.html',
                                            xhtml('_fragment.html'))
  end
  source.at_css('div#book').children.to_xhtml
end
ibooks_xml() click to toggle source
# File lib/softcover/builders/epub.rb, line 522
      def ibooks_xml
%(<?xml version="1.0" encoding="UTF-8"?>
<display_options>
  <platform name="*">
    <option name="specified-fonts">true</option>
  </platform>
</display_options>)
      end
inkscape() click to toggle source

Returns the Inkscape executable (if available).

# File lib/softcover/builders/epub.rb, line 398
def inkscape
  @inkscape ||= executable(dependency_filename(:inkscape))
end
make_epub(options={}) click to toggle source

Make the EPUB, which is basically just a zipped HTML file.

# File lib/softcover/builders/epub.rb, line 469
def make_epub(options={})
  filename = manifest.filename
  zfname = filename + '.zip'
  base_file = "#{zip} -X0 #{zfname} mimetype"
  fullzip = "#{zip} -rDXg9"
  meta_info = "#{fullzip} #{zfname} META-INF -x \*.DS_Store -x mimetype"
  main_info = "#{fullzip} #{zfname} OEBPS    -x \*.DS_Store \*.gitkeep"
  rename = "mv #{zfname} #{filename}.epub"
  commands = [base_file, meta_info, main_info, rename]
  command = commands.join(' && ')
  Dir.chdir('epub') do
    if Softcover.test? || options[:quiet] || options[:silent]
      silence { system(command) }
    else
      system(command)
    end
  end
end
math?(string) click to toggle source

Returns true if a string appears to have LaTeX math. We detect math via opening math commands: (, [, and begin{equation} This gives a false positive when math is included in verbatim environments and nowhere else, but it does little harm (requiring only an unnecessary call to page.js).

# File lib/softcover/builders/epub.rb, line 418
def math?(string)
  !!string.match(/(?:\\\(|\\\[|\\begin{equation})/)
end
move_epub() click to toggle source

Move the completed EPUB book to the `ebooks` directory. Note that we handle the case of a preview book as well.

# File lib/softcover/builders/epub.rb, line 494
def move_epub
  origin = manifest.filename
  target = preview? ? origin + '-preview' : origin
  FileUtils.mv(File.join('epub',   "#{origin}.epub"),
               File.join('ebooks', "#{target}.epub"))
end
nav_html() click to toggle source

Returns the nav HTML content.

phantomjs() click to toggle source

Returns the PhantomJS executable (if available).

# File lib/softcover/builders/epub.rb, line 393
def phantomjs
  @phantomjs ||= executable(dependency_filename(:phantomjs))
end
preview?() click to toggle source

Returns true if generating a book preview.

# File lib/softcover/builders/epub.rb, line 169
def preview?
  !!@preview
end
remove_html() click to toggle source

Removes HTML. All the HTML is generated, so this clears out any unused files.

# File lib/softcover/builders/epub.rb, line 175
def remove_html
  FileUtils.rm(Dir.glob(path('epub/OEBPS/*.html')))
  FileUtils.rm(Dir.glob(path('epub/OEBPS/*.xhtml')))
end
remove_images() click to toggle source

Removes images in case they are stale.

# File lib/softcover/builders/epub.rb, line 181
def remove_images
  rm_r images_dir
end
section_names_and_ids(article) click to toggle source

Returns a list of the section names and CSS ids. Form is [['Beginning', 'sec-beginning'], ['Next', 'sec-next']]

# File lib/softcover/builders/epub.rb, line 635
def section_names_and_ids(article)
  # Grab section names and ids from the article.
  filename = File.join('epub', 'OEBPS', xhtml(article.fragment_name))
  doc = Nokogiri::HTML(File.read(filename))
  names = doc.css('div.section>h2').map do |s|
            s.children.children.last.content
          end
  ids = doc.css('div.section').map { |s| s.attributes['id'].value }
  names.zip(ids)
end
strip_attributes(doc) click to toggle source

Strip attributes that are invalid in EPUB documents.

# File lib/softcover/builders/epub.rb, line 403
def strip_attributes(doc)
  attrs = %w[data-tralics-id data-label data-number data-chapter
             role aria-readonly target]
  doc.tap do
    attrs.each do |attr|
      doc.xpath("//@#{attr}").remove
    end
  end
end
strip_html(text) click to toggle source

Strip HTML elements from the given text.

# File lib/softcover/builders/epub.rb, line 608
def strip_html(text)
  Nokogiri::HTML.fragment(text).content
end
toc_ncx() click to toggle source

Returns the Table of Contents for the spine.

# File lib/softcover/builders/epub.rb, line 581
def toc_ncx
  chapter_nav = []

  if article?
    article = chapters.first
    section_names_and_ids(article).each_with_index do |(name, id), n|
      chapter_nav << %(<navPoint id="#{id}" playOrder="#{n+1}">)
      chapter_nav << %(    <navLabel><text>#{escape(name)}</text></navLabel>)
      chapter_nav << %(    <content src="#{xhtml(article.fragment_name)}##{id}"/>)
      chapter_nav << %(</navPoint>)
    end
  else
    chapters.each_with_index do |chapter, n|
      chapter_nav << %(<navPoint id="#{chapter.slug}" playOrder="#{n+1}">)
      chapter_nav << %(    <navLabel><text>#{chapter_name(n)}</text></navLabel>)
      chapter_nav << %(    <content src="#{xhtml(chapter.fragment_name)}"/>)
      chapter_nav << %(</navPoint>)
    end
  end
  toc_ncx_template(manifest.html_title, manifest.uuid, chapter_nav)
end
write_container_xml() click to toggle source

Writes the container XML file. This is required by the EPUB standard.

# File lib/softcover/builders/epub.rb, line 202
def write_container_xml
  File.write(path('epub/META-INF/container.xml'), container_xml)
end
write_contents(options={}) click to toggle source

Writes the content.opf file. This is required by the EPUB standard.

# File lib/softcover/builders/epub.rb, line 216
def write_contents(options={})
  File.write(path('epub/OEBPS/content.opf'), content_opf(options))
end
write_html(options={}) click to toggle source

Writes the HTML for the EPUB. Included is a math detector that processes the page with MathJax (via page.js) so that math can be included in EPUB (and thence MOBI).

# File lib/softcover/builders/epub.rb, line 228
def write_html(options={})
  texmath_dir = File.join(images_dir, 'texmath')
  mkdir images_dir
  mkdir texmath_dir
  if cover?(options)
    File.write(path("epub/OEBPS/#{cover_filename}"), cover_page)
  end

  pngs = []
  chapters.each_with_index do |chapter, i|
    target_filename = path("epub/OEBPS/#{xhtml(chapter.fragment_name)}")
    File.open(target_filename, 'w') do |f|
      content = File.read(path("html/#{chapter.fragment_name}"))
      doc = strip_attributes(Nokogiri::HTML(content))
      # Use xhtml in references.
      doc.css('a.hyperref').each do |ref_node|
        ref_node['href'] = ref_node['href'].sub('.html', xhtml('.html'))
      end
      body = doc.at_css('body')
      if body.nil?
        $stderr.puts "\nError: Document not built due to empty chapter"
        $stderr.puts "Chapters must include a title using the Markdown"
        $stderr.puts "    # This is a chapter"
        $stderr.puts "or the LaTeX"
        $stderr.puts "    \\chapter{This is a chapter}"
        exit(1)
      end
      inner_html = body.children.to_xhtml
      if math?(inner_html)
        html = html_with_math(chapter, images_dir, texmath_dir, pngs,
                              options)
        html ||= inner_html # handle case of spurious math detection
      else
        html = inner_html
      end
      f.write(chapter_template("Chapter #{i}", html))
    end
  end
  # Clean up unused PNGs.
  png_files = Dir[path("#{texmath_dir}/*.png")]
  (png_files - pngs).each do |f|
    if File.exist?(f)
      puts "Removing unused PNG #{f}" unless options[:silent]
      FileUtils.rm(f)
    end
  end
end
write_ibooks_xml() click to toggle source

Writes iBooks-specific XML. This allows proper display of monospace fonts in code samples, among other things.

# File lib/softcover/builders/epub.rb, line 209
def write_ibooks_xml
  xml_filename = 'com.apple.ibooks.display-options.xml'
  File.write(path("epub/META-INF/#{xml_filename}"), ibooks_xml)
end
write_mimetype() click to toggle source

Writes the mimetype file. This is required by the EPUB standard.

# File lib/softcover/builders/epub.rb, line 196
def write_mimetype
  File.write(path('epub/mimetype'), 'application/epub+zip')
end
write_nav() click to toggle source

Writes the navigation file. This is required by the EPUB standard.

# File lib/softcover/builders/epub.rb, line 509
def write_nav
  File.write("epub/OEBPS/#{nav_filename}", nav_html)
end
write_toc() click to toggle source

Writes the Table of Contents. This is required by the EPUB standard.

# File lib/softcover/builders/epub.rb, line 503
def write_toc
  File.write('epub/OEBPS/toc.ncx', toc_ncx)
end
zip() click to toggle source
# File lib/softcover/builders/epub.rb, line 488
def zip
  @zip ||= executable(dependency_filename(:zip))
end