[Booki-dev] questions

[Douglas Bagnall]

Raj wrote:

> If you go to a book scanned in our Scribe system, you can generate a
> link that will produce an epub. The link will have the
> form /download/id/id.epub


Thanks!  I've downloaded a few to test against.  The full list is here
(with archive.org ones prefixed 'ia-'):

http://booki-dev.flossmanuals.net/git?p=espri.git;a=tree;f=tests/epub-examples

I think I found a couple of bugs.

There are some jpg images listed in the spine in most of the examples,
which I don't think the OPF specification allows unless the images have
xhtml fallbacks (and they don't).  (My test would actually fail even if
you did have fallbacks, because espri doesn't deal with them yet).  It
seems the common workaround for image-only pages is to have an xhtml
page that only has an <img/> tag in it.

Also, you have used the ISO-639-2 code "eng" instead of the ISO-639-1
code "en", which goes against RFC 3066 which the specifications refer to
(section 2.3, point 2).

The trickiest thing about these epubs, from bookis point of view, is
that they tend not to have meaningful chapters, nor an easy way to find
the chapters (e.g., no <h*> tags, just <p>).


cheers,

Douglas