Create an EPUB by authoring your pages as simple markdown files and then letting a script do the rest.

(Note that you could also use XHTML or templates -- see the FAQ.)

This is an 11ty project template, configured to build an EPUB 3 fileset. It also validates with EPUBCheck and runs accessibility checks with Ace.

Make sure you have already installed a reasonably recent version of nodejs.

Clone this repo and start editing:

$ mkdir my-epub-project && cd my-epub-project
$ npx degit marisademeglio/eleventypub
$ npm install

All the files you will be editing are in the src directory.

E.g. page.md:

---
title: Chapter One
---
# Chapter One

"Laborum culpa mollit id," she thought. "Do eu cillum eu magna? 
Dolor voluptate laborum reprehenderit esse labore occaecat."

Consectetur id ad sit excepteur commodo id pariatur ipsum voluptate 
et do pariatur. Et id nisi enim veniam ea non in dolor. Elit eu 
pariatur magna veniam consectetur sit deserunt excepteur.

"Hey! Non ullamco proident!"

{
  "dc": {
    "title": "eleventypub-demo",
    "creator": "Marisa DeMeglio"
  },
  "properties": {
    "dcterms:modified": "2019-03-13T00:00:00Z",
    "schema:accessibilityHazard": "none",
    "schema:accessMode": ["textual", "visual"]
  }
}

• Specify the cover image src and alt properties (you can leave page as-is)
• List pages in order in src/pub.json by listing the file slugs (in their simplest form, these are the markdown filenames w/o extensions).

{
  "cover": {
    "page": "cover/index.xhtml",
    "src": "resources/images/under-construction.jpg",
    "alt": "Cartoon bear with construction tools"
  },
  "readingOrder": [
    "intro", "chapter1", "end"
  ],
  ...
}

In the terminal:

$ npm run all

The following output is created:

1. build/epub: the expanded EPUB fileset
2. build/<title>.epub: the EPUB file itself
3. build/report: an Ace report about accessibility

You will see any EPUBCheck validation issues in the terminal.

If your page has scripting in it, add scripted to the block of front-matter data, e.g.

---
scripted: true
title: Chapter One
---

to make sure it gets noted as such in the package document.

If your page template (e.g. page.njk) has scripting, you can put the scripted property there instead, in the same kind of block, e.g.

---
scripted: true
---
{% extends 'base.njk' %}
...

Also note that the gulp plugin gulp-pretty-data that is automatically included as part of the npm run all build process does not like CDATA, often used in XHTML script blocks. So, you can just take it out of the command chain by editing package.json and removing pretty-xml from the list:

...
"all": "./node_modules/.bin/npm-run-all build save ace rename",
...

This part assumes you are familiar with 11ty. It discusses project structure and options.

You have as much freedom as you have with 11ty to mix and match templates and options. It's up to you to make sure the resulting output passes EPUB 3 validation.

This project uses the following fileset conventions:

Chapter files live in this directory.

In the pages-wide configuration file, pages.json, it says, among other things:

tags: pages

Don't change this.

This becomes the package document. Don't touch this file.

Table of contents. The filename isn't special, just reference the output correctly from pub.json and use a TOC template (in this case, toc.njk).

Cover. The filename isn't special, just reference the output correctly from pub.json.

The data in this directory ends up in the package document.

• metadata.json: as many dc terms, properties, and links as you want. Use arrays for metadata properties that should appear more than once, e.g.

"schema:accessibilityFeature": ["alternativeText", "readingOrder"]

will create this in package.opf:

<meta property="schema:accessibilityFeature">alternativeText</meta>
<meta property="schema:accessibilityFeature">readingOrder</meta>      

• pub.json: Says which files are the toc and cover, describes the cover image, and lists the readingOrder

This directory is for fonts, CSS, images, etc; basically, anything you want listed in the manifest and copied over.

This directory is for all the layout templates, which create XHTML, OPF, and navigation documents. You can edit and replace these with your own, using any markup syntax supported by 11ty. If you use your own templates, it's up to you to make sure you're producing valid EPUB 3 output.

$ npm run [something from the list below]

• all: build, prettify, validate+save, and check accessibility
• all-no-stylelint: same as the above, but without running stylelint on the CSS

You may want to run just a single step of the larger process, for which you can use these commands:

• build: create the EPUB fileset in build/epub
• epubcheck: run EPUBCheck on the output in build/epub
• save: run EPUBCheck on build/epub and if valid, save an EPUB file as build/<title>.epub
• ace: run the Ace accessibility checker on the files in build/epub. The report will be in build/report

Yes absolutely! You can do whatever's possible with 11ty, which is a lot.

The templates in this project are nunjucks and markdown. The configuration I use for markdown includes having it output XHTML. Then, in the postbuild step, all html files are renamed to xhtml. Even so, there are surely still ways to produce EPUB-invalid XHTML with this setup.

So, whatever you try, it's up to you to make sure that your output is valid.

Reference your TOC file from pub.json. Use the name of the output file, e.g. toc/index.html, not the source toc.md.

{
  "cover": {
    "page": "cover/index.xhtml",
    "src": "resources/images/under-construction.jpg",
    "alt": "Cartoon bear with construction tools"
  },
  "toc": {
    "page": "toc/index.xhtml"
  },
  "readingOrder": [
    "intro", "chapter1", "end"
  ]
}

In this basic example, the TOC is the file generated by src/toc.md. Because it is set to autogenerate (auto: true), the file is intentionally empty after the front matter data. It gets filled in by the automated process, which makes one entry per markdown file in pages.

If I wanted to write a TOC manually, I would write something like:

---
layout: toc.njk
title: Table of contents
EPUBRoot: ".."
auto: false
---
1. [Introduction]({{EPUBRoot}}/pages/intro/index.html)
1. [Chapter 1]({{EPUBRoot}}/pages/chapter1/index.html)
1. [The end]({{EPUBRoot}}/pages/end/index.html)

auto: true the TOC will be autogenerated. Each chapter file will get an entry with its title, in a flat hierarchy.

auto: false: you provide everything that's going to go between <nav> and </nav> in your TOC file. It's up to you to make sure these contents are EPUB 3 compliant (e.g. a valid XHTML ordered list).

If you're providing your own TOC, triple-check the file references to make sure the relative path is correct (use EPUBRoot to help). Also, make sure you point to xhtml files, as that's what they'll be in the end, as a result of our postbuild step.

You can add a readingOrder in pub.json. It's an array of file slugs of the files in order. If readingOrder is not specified, a reading order gets generated automatically in alphabetical order based on what's in pages/.

All file slugs in a custom reading order have to be for files in pages/.