Building ePub with PHP and Markdown

Share this article

The ePub format is a publishing standard built on top of XHTML, CSS, XML and more. And since PHP is well suited for working with HTML and friends, why not use it to build ebooks? In this article we’ll see what goes into building a tool for creating ePub packages starting from a set of content files. Maybe it’s your next best selling cyber-sci-fi novel or documentation for your latest code project… because we all write good documentation for our projects, don’t we? In order to make things easy on the writers, our tool will be able to parse the Markdown syntax and the resulting markup will be placed into HTML templates using the RainTPL library. The sample tool I wrote for this article is named md2epub and has become a real open source project on Github. It’s far from perfect at the moment, but feel free to fork and improve it!

Quick Start ePub Theory

The specifications for the ePub format are detailed by the IDPF (International Digital Publishing Forum), and we’ll refer to version 2.0.1 of the specs which is the most widely used today. Other useful resources are the Wikipedia ePub Page and the Epub Format Construction Guide by Harrison Ainsworth. Basically, an ePub book is a zipped archive with a well defined structure. The contents includes XHTML documents, CSS stylesheets, images (GIFs, JPEGs, PNGs, and SVGs), and Open Type fonts. There are other specific files used to describe the content of the book; these are the package and container files:
The first four files are specific for the ePub package, the others are our content. The mimetype file must contain application/epub+zip in ASCII without an end-of-line character. It must be the first file included in the archive and must not be compressed; we cannot do this with PHP, but I’ll show you a work-around for this later. The file META-INF/container.xml is fixed in name and location. The scope of this file is to tell where the OPF file (usually content.opf) is located relative to the ebook’s root directory. content.opf is an XML file that contains the book’s meta data, the references to all the content resources, and the the order in which the contents must be loaded by the reader application. The toc.ncx file is an optional but recommended XML navigation map of the ebook. There are also other optional files that are not taken in consideration for now. And I won’t tell you how to add DRM, sorry.

Our First eBook

So how do we prepare our content for ePub-lishing? I’ve set up a test book directory like this:
Most of the files are content files for the publication, markdown text and images. We then have a style.css file (the name and position should be fixed for now). Keep in mind that the ePub format understands only a subset of the CSS 2.1 specs. At last we have a special file called book.json. This file contains all the data that our md2epub tool will need to generate the standard package files. Most of the keys in this file have a direct match with the OPF and NCX files. The JSON file is structured like this:
    "id": "com.acme.books.MyUniqueBookID",
    "title": "Sample eBook Title",
    "language": "en",

    "authors": [
        { "name": "John Smith", "role": "aut"},
        { "name": "Jane Appleseed", "role": "dsr"}

    "description": "Brief description of the book",
    "subject": "List of keywords, pertinent to content, separated by commas",
    "publisher": "ACME Publishing Inc.",
    "rights": "Copyright (c) 2013, Someone",
    "date": "2013-02-27",
    "relation": "",

    "files": {
        "coverpage": "",
        "title-page": "",
        "include": [
            { "id":"ncx", "path":"toc.ncx" },
        "index": "",
        "exclude": []

    "spine": {
        "toc": "ncx",
        "items": [
The first three groups of keys from id to relation maps to the <metadata> section of the OPF file. They contain information about the book, the author, the publisher, and so on. The keys id, title, and language are required. All the others are optional and their meaning is self-explanatory, except maybe for relation which is used to link to another publication using an ISBN format or a site URL. The value of id must be unique, and I’ve chosen the reverse domain format com.publisher.series.BookID although you can use whatever format you like. The files key of the JSON maps to the <manifest> section of the OPF. This section contains references to every file used in the book, including fonts, stylesheets, images, and special files. Every item in this section needs a unique id, path to the resource (href), and its mimetype. md2epub will take care of the media type and will generate a suitable ID if it’s not provided. I also wanted it to be able to specify bulk file inclusions using wildcards such as *.md and media/*, although the exclude list is not parsed at the moment either. The spine JSON key maps to <spine> in the OPF, it describes the order in which the various chapters should appear in the book. Starting with these data our md2epub will:
  • create a well organized temporary book directory
  • copy all the referenced files
  • convert and copy all the markdown file to XHTML
  • generate the ePub required files
  • create the zip/epub archive and cleanup
The final file can (and should) be validated by the official EPUB Validator which you can also download and run locally as a Java console application.

Md2Epub – The Making of

In order to stay on topic in this article, I’ll discuss only the most epub-specific code and leave the “utility code” for you to explore. The complete program consists of a console PHP script md2epub.php that collects the user’s input, creates the working directory, and passes all of the data to the EBook class. The md2epub file is a shell wrapper. The EBook class performs all of the operations and checks. It uses the RainTPL library for template parsing and the PHP Markdown Extra library
as a content filter. An interesting thing about this last component is that the Markdown() function is provided to the EBook class as an external filter to be applied to *.md files. In this way we can inject any other text filters such as Textile or a wiki syntax parser. The share directory contains common files, in particular the XHTML and XML template files; you can edit the markup of these files to customize your specific book type. The file is part of the work-around I’ve mentioned earlier – since the PHP Zip library does not allow us to specify a compression level when creating an archive, I’ve created this archive stub using the command line zip program:
zip -0 -D -X mimetype
PHP will copy this basic archive and add the other files in sequence.

Exploring the EBook Object

The constructor of the EBook class takes two arguments: the source directory of the book and an optional array of parameters. The parameters specify the name of the JSON file to parse (which defaults to book.json) and the directory that will contain the content files of the compiled book. I chose OEBPS, which stands for Open eBook Publication Structure, as it’s a convention used by many of the ebooks I’ve explored. The role of the parseFiles() method is to normalize the files section loaded from the JSON. The wildcard paths are expanded using glob() and the IDs and media types are generated. At the end, it gives us an associative array where each file is represented in this way:
'cover' => Array (
    'path' => 'cover.jpg',
    'type' => 'image/jpeg'
Each file must not be listed more than once. The path is relative to the content directory, the directory that will contain the OPF file. Markdown files are of type text/plain in this step and they will be converted later. The parseSpine() method does a similar job. It sets a NCX/TOC file and compiles a list of declared IDs. The IDs can be regular expressions. For example, I’ve specified the pattern ^cd{1,2}-.*$ that matches the IDs for the chapters (c01-first-chapter, c02-second-chapter, etc). The beginning “c” is important since XML id attributes cannot start with a number. The powerful makeEpub() method then starts the transformation process for our ebook. The data needed is:
  • the target .epub file path
  • the temporary directory in which compile and copy all the resources
  • the source directory for template files
  • the optional content filters, in this case Markdown linked to the md extension
It first checks the input parameters to be sure it has the right filesystem permissions, and then calls four helper methods in sequence, corresponding to the required steps:
  • create the META-INF directory and data
  • export all the content files, compile templates and apply filters
  • create the OPF manifest file
  • create the NCX navigation file (if needed)
  • create the compressed epub archive
The createMetaInf() method is the simplest and helps us to understand the template logic behind the others:
protected function createMetaInf($workDir)
    // create destination directory
    if (!mkdir("$workDir/META-INF")) {
        throw new Exception('Unable to create content META-INF directory');

    // compile file
    $tpl = $this->initTemplateEngine(
            'tpl_dir'   => "{$this->params['templates_dir']}/",
            'tpl_ext'   => 'xml',
            'cache_dir' => sys_get_temp_dir() . '/'
    $tpl->assign('ContentDirectory', $this->params['content_dir']);
    $container = $tpl->draw('book/META-INF/container', true);

    // write compiled file to destination
    if (file_put_contents("$workDir/META-INF/container.xml", $container) === false) {
        throw new Exception("Unable to create content META-INF/container.xml");
The container.xml template has only one variable to set which is the path of the content.opf file relative to the base directory of the ebook. The syntax is {$VariableName}.
<?xml version="1.0"?>
<container version="1.0" xmlns="urn:oasis:names:tc:opendocument:xmlns:container">
  <rootfile full-path="{$ContentDirectory}/content.opf"
First the RainTPL library is initialized with the templates directory, the extension of the template files, and a temporary cache directory. Then the template variable is assigned ($tpl->assign()) and the draw() method parses the provided template file. Its second parameter tells draw() whether to return the result. The compiled file is then saved to the destination path. The createOpf() method has the same logic and some more variables to assign. I chose the pattern Book<Varname> so the optional meta data is assigned in the loop:
$optParams = array(

foreach ($optParams as $p) {
    if (isset($this->$p)) {
        $tpl->assign('Book' . ucfirst($p), $this->$p);
And they are inserted in the template with a conditional syntax:
The files and spine variables are formatted in the method and are inserted in the template with a loop:
    {loop="$BookFiles"}<item id="{$key}" href="{$value.path}" media-type="{$value.type}"/>
The processBookFiles() is called before the createOpf() and runs a big loop with all the referenced resources ($this->files). For each resource it can choose between two actions:
  • process and filter the file
  • copy the file path “as is” in the destination directory
The filter applied to a file must be a callable PHP function or class method that takes one parameter: the content. The file content is loaded from the original source using file_get_contents() and the filter is applied using the native call_user_func():
$content = call_user_func($filters[$ext], $content);
The processed content is then injected in an XHTML template, similar to the previous methods. In this case there is also the possibility to use a custom template. If there is a file named .xhtml in the templates directory, that file will be used overriding the default page.xhtml. Also, if a style.css file is present then the {$BookStyle} template variable is set and the CSS is linked to the page. The resulting file is then written to the destination directory and the manifest entry for that file is updated with the new path and media type, ready for createOpf(). The createNcx() is a little tricky because it has to look into the items of the spine section and extract the content of the H1 and H2 tags. The toc.ncx file has this structure, first a standardized header:
<?xml version="1.0" encoding="UTF-8"?>
<ncx xmlns="" version="2005-1">
  <meta name="dtb:uid" content="{$BookID}"/>

  <!-- 1 = chapters only, 2 = subchapters, 3 = third level section -->
  <meta name="dtb:depth" content="2"/>

  <!-- Both required but unused, can be 0 -->
  <meta name="dtb:totalPageCount" content="0"/>
  <meta name="dtb:maxPageNumber" content="0"/>

The customized dtb:uid meta indicates our book ID, and the dtb:depth meta tag says how deep our navigation is. I choose a fixed value of 2 here so in the parsing code I have to extract the first two level headers. After the header there’s the <navMap> that is generated using a template loop.
 <navPoint id="{$key}" playOrder="{$playOrder++}">
  <content src="{$value.path}"/>
  <navPoint id="{$key}" playOrder="{$playOrder++}">
   <content src="{$value.path}"/>
Each H1 becomes a <navPoint>, the id is the corresponding resource ID (ex c01-first-chapter, etc), the H1 content becomes the label, and the playOrder attribute indicates the global reading order. In our example, each nav point can have nested nav points for H2 subtitles. In the method, a double loop generates an array structured like this:
$chapters['c01-first-chapter'] = array(
    'title' => 'Title for chapter 1',
    'path' => '01-first-chapter.xhtml',
    'sections' => array(
        'section-1-1' => array(
            'title' => 'Title for subsection 1.1',
            'path' => '01-first-chapter.xhtml#section-1-1',
In the external loop, each item is loaded as a SimpleXML object:
$doc = simplexml_load_file("$workDir/{$this->params['content_dir']}/{$this->files[$item]['path']}");
if ($doc && isset($doc->body->h1)) {
    // chapter title
    $chapters[$item] = array(
        'title' => $doc->body->h1,
        'path'  => $this->files[$item]['path']

    // subchapter title
    foreach ($doc->body->h2 as $section) {
        if (!empty($section['id'])) {
            $section_id = (string) $section['id'];
            $chapters[$item]['sections'][$section_id] = array(
                'title' => $section,
                'path' => $this->files[$item]['path'] . '#' . $section['id']
If the document has a H1 header, a first level item is inserted into the $chapters array. Then the second level headers are queried, each header that has a non-empty id attribute is included in the sections for each chapter. The $chapters array becomes {$BookChapters} in the template. The template is then rendered and saved to its destination. Now we have a destination working directory somewhere at /tmp/path/com.publisher.BookID that can be packed by the final step createArchive():
protected function createArchive($workDir, $epubFile)
    $excludes = array('.DS_Store', 'mimetype');

    $mimeZip = "{$this->params['templates_dir']}/";
    $zipFile = sys_get_temp_dir() . '/';

    if (!copy($mimeZip, $zipFile)) {
        throw new Exception("Unable to copy temporary archive file");

    $zip = new ZipArchive();
    if ($zip->open($zipFile, ZipArchive::CREATE) != true) {
        throw new Exception("Unable open archive '$zipFile'");

    $files = new RecursiveIteratorIterator(new RecursiveDirectoryIterator($workDir), RecursiveIteratorIterator::SELF_FIRST);
    foreach ($files as $file) {
        if (in_array(basename($file), $excludes)) {
        if (is_dir($file)) {
            $zip->addEmptyDir(str_replace("$workDir/", '', "$file/"));
        } elseif (is_file($file)) {
                str_replace("$workDir/", '', $file),

    rename($zipFile, $epubFile);
The $excludes array contains special and system files that we don’t want in the archive. A temporary path for an empty $zipFile is created, and then the $zipFile path is filled with a copy of our basic that is loaded in a ZipArchive object. At this point the content of the working directory is recursively loaded into the $files array using the RecursiveIteratorIterator and RecursiveDirectoryIterator objects. In the following loop the structure of the working directory is replicated in the zip archive. As a finishing touch, the archive is moved to the destination path indicated by the caller.


When all is said and done, we have our shiny new eBook. We have a good chance that it will be validated by the EpubCheck app and, most importantly, we have another useful tool to add to our PHP-super-hero belt! And this is only a starting point, you can play with the code to add filters, features, themes and styles and more if you like. Keep calm, have fun, and write those eBooks! Image via Fotolia

Frequently Asked Questions (FAQs) on Building EPUB with PHP and Markdown

What is the significance of using PHP and Markdown for building EPUB?

PHP and Markdown are powerful tools for building EPUB files. PHP, a server-side scripting language, is known for its flexibility and efficiency in handling complex tasks. It can be used to automate the process of EPUB creation, making it faster and more efficient. Markdown, on the other hand, is a lightweight markup language that allows you to write using an easy-to-read and easy-to-write plain text format. It simplifies the process of content creation, making it easier to format text and include elements like headers, links, and lists. When combined, PHP and Markdown can streamline the process of creating EPUB files, making it more accessible and manageable.

How can I convert Markdown to HTML using PHP?

Converting Markdown to HTML using PHP can be done using a parser. A parser like Parsedown or Michelf’s PHP Markdown can be used. You simply need to include the parser in your PHP script and then call the appropriate function to convert the Markdown text to HTML. Here’s a basic example using Parsedown:

require 'Parsedown.php';
$Parsedown = new Parsedown();
echo $Parsedown->text('Hello _Parsedown_!'); // prints: <p>Hello <em>Parsedown</em>!

How can I create an EPUB file using PHP?

Creating an EPUB file using PHP involves several steps. First, you need to create the content in XHTML or DTBook format. Then, you need to create a .opf file that describes the content. Next, you need to create a .ncx file that provides navigation for the content. Finally, you need to zip these files together in a specific order to create the EPUB file. There are libraries available, like PHPePub, that can simplify this process by providing functions to handle these steps.

What are the benefits of using PHPePub for creating EPUB files?

PHPePub is a PHP class that simplifies the process of creating EPUB files. It provides functions for creating and adding content to the EPUB file, setting metadata, and building the final EPUB file. This can save you a lot of time and effort compared to creating an EPUB file manually. Additionally, PHPePub supports both EPUB 2 and EPUB 3, making it a versatile tool for EPUB creation.

How can I add images to an EPUB file using PHP?

Adding images to an EPUB file using PHP can be done using the PHPePub library. The library provides a function for adding images to the EPUB file. You simply need to specify the path to the image file and the ID you want to use for the image in the EPUB file. Here’s a basic example:

$book = new EPub();
$book->addImage("cover.jpg", file_get_contents("cover.jpg"));

In this example, “cover.jpg” is the ID for the image, and file_get_contents(“cover.jpg”) is the image data.

How can I add CSS to an EPUB file using PHP?

Adding CSS to an EPUB file using PHP can be done using the PHPePub library. The library provides a function for adding CSS to the EPUB file. You simply need to specify the CSS rules you want to add. Here’s a basic example:

$book = new EPub();
$book->addCSSFile("styles.css", "css1", file_get_contents("styles.css"));

In this example, “styles.css” is the ID for the CSS file, “css1” is the CSS class, and file_get_contents(“styles.css”) is the CSS data.

How can I validate an EPUB file?

Validating an EPUB file can be done using an EPUB validation tool like EPUBCheck. EPUBCheck is a tool that can be used to validate EPUB files against the EPUB specification. It can check for common errors and issues that can prevent the EPUB file from being correctly displayed in an EPUB reader. You can use EPUBCheck online or download it and run it locally.

How can I read an EPUB file using PHP?

Reading an EPUB file using PHP can be done using a library like PHPePub. The library provides functions for opening and reading the content of an EPUB file. However, it’s important to note that PHP is not designed to be an EPUB reader. While you can extract and read the content of an EPUB file using PHP, displaying that content in a way that mimics an EPUB reader would require additional programming and is beyond the scope of PHP.

Can I use PHP and Markdown to create other types of eBooks?

Yes, you can use PHP and Markdown to create other types of eBooks. While this guide focuses on creating EPUB files, the principles can be applied to other eBook formats. For example, you could use PHP and Markdown to create a MOBI file for Kindle. However, the specific steps and tools required may vary depending on the eBook format.

What are some common issues I might encounter when creating an EPUB file with PHP and Markdown?

Some common issues you might encounter when creating an EPUB file with PHP and Markdown include formatting issues, problems with image or CSS inclusion, and validation errors. Formatting issues can often be resolved by carefully checking your Markdown syntax and ensuring that it’s being correctly converted to HTML. Problems with image or CSS inclusion can often be resolved by ensuring that you’re correctly specifying the path to the file and that the file is being correctly included in the EPUB file. Validation errors can often be resolved by using a tool like EPUBCheck to identify and fix issues with the EPUB file.

Vito TardiaVito Tardia
View Author

Vito Tardia (a.k.a. Ragman), is a web designer and full stack developer with 20+ years experience. He builds websites and applications in London, UK. Vito is also a skilled guitarist and music composer and enjoys writing music and jamming with local (hard) rock bands. In 2019 he started the BlueMelt instrumental guitar rock project.

Share this article
Read Next
Get the freshest news and resources for developers, designers and digital creators in your inbox each week
Loading form