Custom Epub Testing Area


This page has been set up to aid in testing epub files for DP's custom epub experiment. The goal is to permit PPers to create handcrafted epubs that will look good, validate, and be editable/regeneratable by PG in the future, while at the same time being able to retain a full-featured HTML file.

The experiment makes the following assumptions:


There are a number of open questions that we hope to answer with this experiment:

Software Issues

At the moment, PG uses a program created by Marcello Perathoner, Epubmaker, to automatically generate epubs. Epubmaker runs in Python, which the PPer would have to install first before installing Marcello's program. There are apparently some issues with installation at this point; Roger Frank was able to install it only on a Linux system. The Epubmaker files can be found here.

Roger has proposed that perhaps we could instead use an online version of Epubmaker to generate the epub. The advantage of this is that PPers would be using exactly the same software as PG. The disadvantage is that the PPer would have to submit the source file multiple times to the PG server while tweaking the source file in their own editing software to get it to look right.

To my mind, the ideal solution would be to allow PPers to use whatever software they like to create the epub. The open question, however, is whether epubs are "fungible," i.e., will an epub created in one program always be readable and updatable in Marcello's program.

There are a number of epub converter/editors available. If we are to make recommendations other than Marcello's Epubmaker, the software will have to be:

Experiments with various epub applications are set forth below.

Experiment 1 - Sigil

Sigil is an epub editor with the following features:

Sigil's ability to create valid epubs that will be readable/updatable by Epubmaker is yet to be determined.

Observations with Raw HTML

My first experiment involved taking an existing HTML that has already been posted to PG, Down Town Brooklyn, and running it through Sigil to create an epub. The HTML file has the following formatting:

A raw conversion, with no changes to the HTML, resulted in the following (as viewed in the Firefox epub reader extension):

The resulting epub (with the external link stripped out) can be downloaded here.

A look at the automatically generated epub at PG revealed similar results, except that the page numbers and the external link had been stripped out (but the [Enlarge] text remained), and there was no box around the title page.

Observations after HTML Editing

Using Sigil, I tweaked the HTML & CSS to address some of the problems noted above, as well as some problems noted in the forums. Specifically:

The resulting epub is here.

Experiment 2 - Calibre

Calibre is an epub converter, viewer, and manager - it is not a full-featured editor. In fact, the Calibre people recommend using Sigil or some other epub editor for actual epub editing.

[More to come]

Experiment 3 - KindleGen

KindleGen is a command-line app provided by Amazon to convert HTML and epub to mobi files, which are readable on the Kindle. A question was raised at PG as to whether KindleGen can handle custom epubs.

Using KindleGen, I successfully converted the custom epub discussed above to a mobi file, which I was able to view on my Kindle. Although some of the formatting wasn't properly rendered (e.g., dropcaps are not properly displayed; it ignored bold in image captions CSS, and ignored display: none in page number CSS), on the whole the book looked pretty decent.

The resulting mobi file is here. More work is needed to resolve the formatting issues.