[Plugin] ePub3-itizer - epub3 output plugin for Sigil

[KevinH]

Hi,

I tested this in Sigil pre 0.9.0 master and your epub3 is left unchanged. That said, it does not display properly in the QWebView at all. I will look into fixing that.

Take care,
Kevin

[KevinH]

Hi,

I will add the list of mathml tags to the next Sigil 0.8.6 tidy set of new tags in an attempt to prevent tidy from removing the tags when cleaning is done.

Kevin

[RbnJrg]

After converting an epub to epub3, svg images that looked ok under ADE 4 (as epub2), they look very tiny or they are not showed at all (as epub3). Also, after conversion to epub3, ADE 4 doesn't support widows, orphans and page-break-after: avoid (but the properties are supported if I read the epub -on ADE 4- as epub2). So, it would seem that is a problem of the plugin; or not?

[DiapDealer]

Also, after conversion to epub3, ADE 4 doesn't support widows, orphans and page-break-after: avoid (but the properties are supported if I read the epub -on ADE 4- as epub2). So, it would seem that is a problem of the plugin; or not?

Not necessarily. The Epub3 support of ADE4 could be being provided by an entirely different rendering engine. If the exported epub3 validates (epubCheck), I'd suspect ADE4's epub3 rendering engine is the culprit.

If it doesn't validate and/or those properties are actually missing/altered after using the epub3 plugin, then perhaps it might be an issue with the plugin.

[RbnJrg]

Originally Posted by DiapDealer

Not necessarily. The Epub3 support of ADE4 could be being provided by an entirely different rendering engine. If the exported epub3 validates (epubCheck), I'd suspect ADE4's epub3 rendering engine is the culprit.

If it doesn't validate and/or those properties are actually missing/altered after using the epub3 plugin, then perhaps it might be an issue with the plugin.

Hi Diap;

I followed your advice and I have validate the epub (epubCheck) as epub2 and I get the message: "Congratulations! No problems were found in..."

But as epub3, I get the following error messages:

-value of attribute "height" is invalid; must be an integer.
-XML document structures must start and end within the same entity.
-The string "--" is not permitted within comments.
-External DTD entities are not allowed. Remove the DOCTYPE.

Regarding my issues with svg images, the respective error message is:

-XML document structures must start and end within the same entity.

So, what must we think?

Regards
Rubén

[Doitsu]

@RbnJrg: I've tested the plugin with many ePub2 files and never had any validation problems with the generated ePub3 files.

The only message that I consistently got is:

External DTD entities are not allowed. Remove the DOCTYPE.

However, this is a known bug in epubcheck 3.0.1 related to the .ncx file.

As for ADE 4. They released a new version with MathML support some time ago. If you haven't already installed it, re-check your file with the latest version. For good measure also check the converted ePub3 file with Readium.

Please also post an ePub2 test file with svg images for KevinH.

[DiapDealer]

Originally Posted by Doitsu

@RbnJrg:Please also post an ePub2 test file with svg images for KevinH.

@RbnJrg:

Also, if at all possible, make sure the test epub2 file you post contains the other errors you mentioned after converting and validating (ignoring the external DTD one for now).

[RbnJrg]

Hi Doitsu and DiapDealer;

Here I attach both, the epub2 and epub3 ebook. As you'll be able to see, as epub2 is showed ok under ADE 4 but as epub3, the cover is very tiny and the svg images are not showed. Also the epub2 validates in EpubCheck but not as epub3 (however, this time I couldn't reproduce "The string "--" is not permitted within comments" error message).

@Doitsu: I'm using the last version of ADE (release 4.0.3.114137).

Regards
Rubén

[Doitsu]

Hi Rubén,

I had a look at your file and it seems that some of the problems are caused by missing spaces before and after comments in caratula.xhtml and titulo.xhtml. After inserting spaces after the initial two hyphens and before the final two hyphens the book exports fine with the plugin and looks fine in Readium but not in ADE 4.
IMHO, this means that ADE 4 is not 100% ePub3 compatible when it comes to svg rendering.

When I checked the exported ePub3 file with the latest ePubcheck 4.x Alpha version I only got the following error messages:

WARNING(OPF-007): epub3.epub/OEBPS/content.opf(2,146): Re-declaration of reserved prefix 'rendition'.
ERROR(RSC-005): epub3.epub/OEBPS/content.opf(36,9): Error while parsing file 'element "guide" incomplete; missing required element "reference"'.
ERROR(RSC-005): epub3.epub/OEBPS/Text/autor.xhtml(14,74): Error while parsing file 'value of attribute "width" is invalid; must be an integer'.

The first warning seems to have been caused the following attribute, which other ePub3 books lack:

prefix="rendition: http://www.idpf.org/vocab/rendition/#"

The other two errors are most likely epubcheck4 bugs, because the first one is caused by an empty guide section and the last one complains about a perfectly valid attribute value.

[roger64]

Hi

1. He just forgot to suppress all the comments ( this <!-- before <rect )

<!--<rect y="5" width="590px" style="fill: none; stroke-width:5; stroke: black" x="5" height="890px"/>

2. then for "autor":

width="175"

instead of "25%"

3. This kind of Doctype is no more allowed (toc.ncx)

<!DOCTYPE ncx PUBLIC "-//NISO//DTD ncx 2005-1//EN"
 "http://www.daisy.org/z3986/2005/ncx-2005-1.dtd">

Only (example)

<?xml version="1.0" encoding="UTF-8" ?>
<ncx version="2005-1" xml:lang="fr" xmlns="http://www.daisy.org/z3986/2005/ncx/">

[KevinH]

FWIW, this appears to NOT be a plugin issue.

That DOCTYPE on the ncx is actually legal as Doitsu pointed out, that is a bug in epubcheck 3 that has already been fixed.

KevinH

[Doitsu]

Originally Posted by KevinH

FWIW, this appears to NOT be a plugin issue.

True, but there are some cosmetic issues that you could fix by simply suppressing some output that epubcheck 4 doesn't like:

1. Since it doesn't like prefix="rendition: http://www.idpf.org/vocab/rendition/#", I'd delete it.

2. Since it doesn't like empty <guide /> tags in .opf files, I'd recommend removing empty guide tags from the .opf file, even though they're perfectly valid.

[RbnJrg]

Originally Posted by Doitsu

Hi Rubén,

I had a look at your file and it seems that some of the problems are caused by missing spaces before and after comments in caratula.xhtml and titulo.xhtml. After inserting spaces after the initial two hyphens and before the final two hyphens the book exports fine with the plugin and looks fine in Readium but not in ADE 4.
IMHO, this means that ADE 4 is not 100% ePub3 compatible when it comes to svg rendering.

When I checked the exported ePub3 file with the latest ePubcheck 4.x Alpha version I only got the following error messages:

WARNING(OPF-007): epub3.epub/OEBPS/content.opf(2,146): Re-declaration of reserved prefix 'rendition'.
ERROR(RSC-005): epub3.epub/OEBPS/content.opf(36,9): Error while parsing file 'element "guide" incomplete; missing required element "reference"'.
ERROR(RSC-005): epub3.epub/OEBPS/Text/autor.xhtml(14,74): Error while parsing file 'value of attribute "width" is invalid; must be an integer'.

The first warning seems to have been caused the following attribute, which other ePub3 books lack:

prefix="rendition: http://www.idpf.org/vocab/rendition/#"

The other two errors are most likely epubcheck4 bugs, because the first one is caused by an empty guide section and the last one complains about a perfectly valid attribute value.

Thank you very much Doitsu! Yes, you are right; the problem with .svg was the lack of a space after "<!--" and before "-->! in my coments. However, is like you say, ADE 4 is not 100% ePub3 compatible when it comes to svg rendering (it's incredible, because for epub2 works fine).

@Roger; I didn't forget to supress "<!--", in fact, I didn't want the rectangle

Many thanks again
Rubén

[KevinH]

Hi Doitsu,
The first is just a warning and is not an issue given the current spec. The second is correct as it stands again according to my reading of the spec. We are trying to hit the spec, not epubcheck 3 or 4.

Perhaps once this plugin becomes part of Sigil internally, I will revisit this again, otherwise I would rather focus on things that violate the actual spec.
Thanks,
KevinH

Originally Posted by Doitsu

True, but there are some cosmetic issues that you could fix by simply suppressing some output that epubcheck 4 doesn't like:

1. Since it doesn't like prefix="rendition: http://www.idpf.org/vocab/rendition/#", I'd delete it.

2. Since it doesn't like empty <guide /> tags in .opf files, I'd recommend removing empty guide tags from the .opf file, even though they're perfectly valid.

[roger64]

Originally Posted by KevinH

FWIW, this appears to NOT be a plugin issue.

That DOCTYPE on the ncx is actually legal as Doitsu pointed out, that is a bug in epubcheck 3 that has already been fixed.

KevinH

http://www.idpf.org/epub/30/spec/epu...emovals-dtbook

4.1 Features Removed from EPUB 3
› 4.1.1 DTBook

DAISY DTBook [Z3986-2005] was an alternative syntax to XHTML 1.1 for Content Documents in OPS 2.0.1 [OPS2] in order to provide an option for more semantic, and thus more accessible, content. As HTML5 includes intrinsic semantic markup capabilities of a similar nature to DTBook, DTBook is no longer an alternative syntax in EPUB 3.

I do not know how one has to interpret this. What's the point of keeping a Doctype which references a removed feature? (Z3986). Without it, it's still possible to provide a toc.ncx for retro-compatibility purposes.