Error in epub file generated by Sigil

[Mazarin07]

I have tried to upload an epub file generated by Sigil on Google Books Partner Center and I get the following error message (excerpt):

Required: Fix ePub and resubmit
The ePub you submitted has the following errors. These will prevent the book from going live. See the Help Center for more information. Please correct and resubmit.
index_split_003.html:55, 248: Error while parsing file: attribute "role" not allowed here; expected attribute "accesskey", "charset", "coords", "dir", "hreflang", "lang", "rel", "rev", "shape", "style", "tabindex", "target", "type" or "xml:lang"
...
index_split_009.html:175, 288: Error while parsing file: attribute "data-docx-vert" not allowed here; expected attribute "accesskey", "charset", "class", "coords", "dir", "href", "hreflang", "id", "lang", "rel", "rev", "shape", "style", "tabindex", "target", "title", "type" or "xml:lang"

I have found a similar question on StackExchange and wasn’t able to figure out the solution there.

[DiapDealer]

Sigil will not generate role or data-docx-vert attributes. Those had to come from somewhere else. The last looks like it might be xml from a Word document. But regardless, I don't think they're epub spec compliant. The solution is to remove the two attributes and their values. Use spans and superscript css to accomplish what data-docx-vert="sup" seems to be trying to achieve.

Somewhere along the the line, a docx to epub conversion is leaving non-compliant markup behind.

[KevinH]

Sigil itself does not auto generate error free epubs as it depends on the errors introduced by the user in his/her xhtml. The *user* of Sigil can generate error free epubs using Sigil to track down and fix errors. Have you tried using Sigil's epubcheck plugin to help you track down and fix your errors?

[Doitsu]

@Mazarin07 The role and data-* attributes were introduced in the HTML5 specs. I.e., they can only be used in epub3 books.
A relatively easy fix would be to convert your epub2 book to an epub3 book with KevinH's ePub3-itizer plugin.

[KevinH]

And data attributes are generally used to add non-visible to the reader additional information that is typically accessed via javascript. And javascript is only available in epub3 readers and is technically not allowed in epub2 epubs.

[Mazarin07]

Originally Posted by Doitsu

@Mazarin07 The role and data-* attributes were introduced in the HTML5 specs. I.e., they can only be used in epub3 books.
A relatively easy fix would be to convert your epub2 book to an epub3 book with KevinH's ePub3-itizer plugin.

Thank you, this plugin of KevinH's worked fine. I am not sure if epub3 files are accepted by my PocketBook Basic reader, but will create a separate epub2 file for it if needed.

[DNSB]

Originally Posted by Mazarin07

Thank you, this plugin of KevinH's worked fine. I am not sure if epub3 files are accepted by my PocketBook Basic reader, but will create a separate epub2 file for it if needed.

Depending on what ePub3 features you use, you may have to add fallback code for ePub2. In most cases, simply adding the toc.ncx ePub2 navigation document is all that is needed since an ePub2 renderer will simply disregard ePub3 specific items. An example is that drop caps or small caps first phrases on the first paragraph may not show but the ebook will still be readable.