Download: sample-epub.zip

Horizontal: sample-yoko.epub
Vertical: sample-tate.epub

This sample book consists of the title page and three chapters. The first two chapters are for testing major HTML constructs such as a figure, table, list and link. The last chapter is for testing vertical rendering mode introduced in Epub 3.0.

I show how the sample looks like with the following viewers:

• Adobe Digital Edition 1.8.3 Preview 4
• iBooks Reader (iPad)

Digital Edition

Here are display examples in Digital Edition 1.8.3 under Windows:

Click one of the screen images to see only that image.

Horizontal rendering seems OK with both Epub 2.0 compliant Digital Edition 1.7.2 (latest official version as of 12-Jul-09) and preview version 1.8.3 said to support Epub 3.0 partially. However, in 1.8.3, strangely enough, the first page always starts from the right regardless of vertical or horizontal rendering (see Page Progression Direction for detail).

Vertical rendering is OK with 1.8.3.

iBooks

Here are display examples in iPad's iBooks reader:

Click one of the screen images to see only that image.

There seems no problem in horizontal rendering.

For vertical rendering, as opposed to Digital Edition' case above, the first page starts from left (detail described in Page Progression Direction).

Epub is one of open standards for electronic books (eBook); named after the extension of the file containg the book. International Digital Publishing Forum (IDPF) defined this standard.

The most notable characteristic is, unlike PDF, there is no fixed page boundary. The page boundaries move depending on the screen size and font size chosen at one time. This enabled book reading on a small screen of a mobile device possible.

In summary, an Epub 2.0 book:

• Is a collection of files in a ZIP file whose filename extension is .epub,
• Has some meta files in XML format such as list of files or table of contents,
• The book contents are written in a subet of XHTML 1.1 plus CSS 2.0, PNG and JPEG files thus conforming to Web standards.

Epub defines how to package files, specify page navigation order and write table of contents, but it does not define the content document format. Epub adopts HTML and CSS as its document format and defines subset of them. Therefore, some Epub-compliant eBook readers support HTML and CSS standards beyond the subset defined by Epub.

The latest Web standards (CSS3 and HTML5) that Epub specification depends include support of traditional way of rendering of the document written in the far east countries including Japan. These are adopted in Epub 3.0.

With this sample, I tested vertical writing, ruby (small characters by side), horizontal writing among vertical writing, emphasis dots and line break rule.

The following revisions are made in Epub 3.0:

• XHTML5 is adopted.
• Format of table of contents is changed from NCX to NAV. It is a plain XHTML file and can coexist with 2.0 NCX.
• Subset of MathML is adopted.
• DIASY DTBook is removed

By the way, what determines the version of an Epub file?
That is, version attribute of package tag in content.opf file:

  <?xml version="1.0" encoding="UTF-8"?>
  <package version="3.0" xmlns="http://www.idpf.org/2007/opf" unique-identifier="BookId">
    <metadata xmlns:dc="http://purl.org/dc/elements/1.1/"
       ...

However, it is a viewer that determines what to do:

• An intelligent viewer will check this version and performs operation depending on the version.
• Check this version and treat it as an error if the version is not the one the viewer can support.
• Don't check the version and reads the file as if it is the version it can support.

Epubcheck is a validation tool for an Epub file.

http://code.google.com/p/epubcheck/downloads/list

This is originally developed by Adobe and currently developed as an open source product. For example, the iBooks authoring guide requests a publisher to test its Epub file with epubcheck before uploading to the iBooks store.

As of 2012 July, two versions of Epubcheck are provided for different Epub versions:

Epubcheck 1.2

Latest stable version for Epub 2.0

Epubcheck 3.0b5

Beta version that also supports Epub 3.0

I checked the sample file with the 3.0-enabled beta version.

The following are major corrections I made to the sample this time in order to pass epubcheck.

• Removed DOCTYPE.
• Deleted meta http-equiv that specifies a character encoding.
• Specified character encoding by adding xml declaration and meta charset.
• Added a name space "http://www.w3.org/1999/xhtml".
• Removed rb and rp tags from ruby tag; only rt tag is used in HTML5.

I could remove all but one error of this sample:

     [java] Epubcheck Version 3.0b5
     [java] Validating against EPUB version 3.0
     [java] ERROR: epub/sample.epub/OEBPS/navdoc.html(12,31): attribute "epub:type" not allowed here; expected attribute "accesskey", ...
     [java] ERROR: epub/sample.epub/OEBPS/navdoc.html(6,7): assertion failed: Exactly one 'toc' nav element must be present
     [java] Check finished with warnings or errors
     [java] Java Result: 1

This error relates to a nav tag in an Epub 3.0 table of contents file (navdoc.html). I am not sure why this happens.

Then I found the page below, followed the same way described there (moved the xmlns:epub declaration from <html> to <nav> element) and received no error.

Epub Tutorial (in Japanese)

I just recently knew that there are two levels of specifications in order to implement vertical rendering in Epub 3.0.

Page progression direction of an entire book

Specifying "rtl" (right to left) for page-progression-direction attribute of spine element in content.opf file.

Line rendering direction within a file or part of a file

Specifying "vertical-rl" for CSS writing-mode attribute.

First, take an example of iBooks reader. As of 2012-Jul-07, if you read sample-tate.epub with iBooks Reader, the first page appears on the left side of the book. If you turn pages, lines of each chapter are displayed from right to left.

• As seen from entire book level, pages advance from left to right, which is the same direction of a latin literature and opposite to usual far east vertical reading direction.
• As seen from chapter (or file) level, lines advance from right to left, correct direction for vertical reading.
• Not all content of a chapter is displayed.

This indicates file-level specification is effective but book-level specification is not effective. It seems lack of Epub 3.0 page-progression-direction support in iBooks.

In digital edition 1.8.3, in an opposite way, page direction is right to left even in horizontal rendering although the default value for page-progression-direction is defined as "ltr" (left to right) suitable for horizontal rendering. I deliberately specified "ltr" for the page-progression-direction attribute but the result was the same.

I hope either will be fixed in a near future.

You can find the following files if you unzip the sample zip file (sample-epub.zip).

	epub
	  sample-yoko.epub
	  sample-tate.epub
	  sample/
	    mimetype
	    META-INF/
	      container.xml
	    OEBPS/
	      content.opf
	      toc.ncx
	      navdoc.html
	      title_page.xhtml
	      chap01.xhtml
	      chap02.xhtml
	      chap03.xhtml
	      stylesheet.css
	      images/
	        koma.png
	  stylesheet-yoko.css
	  stylesheet-tate.css
	  content-yoko.opf
	  content-tate.opf

sample-yoko.epub and sample-tate.epub are Epub format ZIP files. One for horizontal- and the other for vertical-reading. Its source contents are in sample folder. sample-xxxx.epub is a ZIP file containing the contents of the sample folder. If you rename the file extension of sample-xxxx.epub to "zip" and unzip it, you can get the original contents of the folder.

The following files are meta files conforming to Epub standard.

The following files are files that make up the Epub book conent. They are displayable with a web browser.

	title_page.xhtml
	chap01.xhtml
	chap02.xhtml
	chap03.xhtml
	stylesheet.css
	koma.png

Finally, the following two pairs of files have different content depending on the rendering direction of horizontal (yoko) and vertical (tate). They are stylesheet and meta file. Files with "-yoko" are for horizontal rendering. Files with "-tate" are for vertical rendering. sample-yoko.epub and sample-tate.epub are created by copying these files to stylesheet.css and content.opf in the sample folder before zipping.

	stylesheet-yoko.css
	stylesheet-tate.css
	content-yoko.opf
	content-tate.opf

You need a ZIP tool to build an Epub-format ZIP file from the source.

But there is a troublesome condition. mimetype must be the first file in the ZIP file and must be uncompressed.

I introduce you three ways to do that.

	cd sample
	zip -0 -X ../sample.epub mimetype
	zip -r ../sample.epub * -x mimetype
	cd ..

  c:\Temp\sample>unzip -v test.zip
  Archive:  test.zip
   Length   Method    Size  Cmpr    Date    Time   CRC-32   Name
  --------  ------  ------- ---- ---------- ----- --------  ----
        20  Stored       20   0% 2012/07/05 21:00 2cab616f  mimetype
       241  Defl:N      163  32% 2012/07/05 21:00 d3c8670a  META-INF/container.xml
      2006  Defl:N     1156  42% 2012/07/05 23:12 3f5fefff  OEBPS/chap01.xhtml
      2329  Defl:N     1281  45% 2012/07/05 23:11 7a375863  OEBPS/chap02.xhtml
      1938  Defl:N     1070  45% 2012/07/06 01:20 4e103098  OEBPS/chap03.xhtml
      1734  Defl:N      749  57% 2012/07/06 01:59 696c89c3  OEBPS/content.opf
         0  Stored        0   0% 2012/07/12 05:24 00000000  OEBPS/images/
      1914  Defl:N     1197  38% 2012/07/05 21:00 2760b6de  OEBPS/images/koma.png
      1002  Defl:N      567  43% 2012/07/06 01:59 b8cdba9a  OEBPS/navdoc.html
      1078  Defl:N      443  59% 2012/07/06 02:00 8325797e  OEBPS/stylesheet.css
       752  Defl:N      461  39% 2012/07/06 02:00 f7272379  OEBPS/title_page.xhtml
      1187  Defl:N      482  59% 2012/07/06 02:00 9cf9d99c  OEBPS/toc.ncx
  --------          -------  ---                            -------
     14201             7589  47%                            12 files

    <target name="zip-epub" description="create zipped epub file">
        <delete file="${epub.file}"/>
        <!-- merge non-compressed mimetype and rest of the files in epub folder -->
        <zip destfile="${epub.dir}/mimetype.zip" basedir="${epub.dir}" compress="false" includes="mimetype" />
        <zip destfile="${epub.dir}/others.zip" basedir="${epub.dir}" level="9" compress="true" includes="OEBPS/** META-INF/**" />
        <zip destfile="${epub.file}" update="true" keepcompression="true" encoding="UTF-8">
            <zipfileset src="${epub.dir}/mimetype.zip" />
            <zipfileset src="${epub.dir}/others.zip" />
        </zip>
        <!-- delete temp files -->
        <delete file="${epub.dir}/mimetype.zip"/>
        <delete file="${epub.dir}/others.zip"/>
    </target>

Other related links:

• Testing InDesign CS6 Epub Export function
• Japanese DocBook Sample. Auto-generation of Epub and Kindle.
• Japanese Epub Sample Converted to Amazon Kindle Format Updated!
• An English Epub Book titled "Java Programming with Mifare and ACR120" is on sale through Apple iBooks Store and Amazon Kindle Store.

Kobu.Com's related businesses are:

• Batch Processing of Large Document
• Digital book publishing for tablets and smartphones
• On-site training courses for engineers and designers

Kobu.Com welcomes questions and comments about this Epub sample. Please contact us if you need some help with text- and XML-based document creation and digital publishing in general.