Bug or feature in "Edit Table Of Contents"

[DiapDealer]

Originally Posted by JSWolf

It is actually  . I tested posting an actual non-breaking space and the forum converted it to an *.

That's exactly what I was NOT saying Jon. I KNOW Mobileread will butcher entities. What I am trying to determine (and you are interfering with) is whether it was Sigil that was butchering that particular entity. Which would be a bug. Please, please, please stop interfering when I am in the process of debugging something. You always tend to introduce unhelpful noise whenever you do.

[KevinH]

And I can not recreate it in Sigil.

When I add and & #160; (without the space after the ampersand) to a nav entry and then tell it to generate an ncx, I get the following:

  <navPoint id="navPoint4">
    <navLabel>
      <text>Chapter 2: The Pool & #160; of Tears</text>
    </navLabel>
    <content src="Chapter-02.xhtml" />
  </navPoint>

[Again the above without the space after the ampersand].

That is exactly correct.

Then I replaced the nbsp in the nav with the actual nbsp char then recreated the ncx from it and ( if Preserve Entities has that nbsp included) you get the correct entity in the ncx.

Next I replaced the nbsp in the nav with the actual nbsp char, then changed the Preserve Entities to *not* preserve the nbsp, and then recreated the ncx, it was generated with the correct nbsp character (no entity).

So no bug here at all.

This is working exactly as designed.

[DiapDealer]

Sounds good. I can't reproduce it either. So we'll assume that's the same thing the OP is seeing, and that Mobileread is just butchering what they pasted in (as usual).

[jd1004]

Hi guys

I can reproduce the error with Sigil 2.8.6 on Windows 10. But you have to follow steps 1) to 4) exactly as described.

It starts with image "01" in the NAV. Then after step 3) you get image "02" ... all is still OK. But after step 4) you get image "03".

[DiapDealer]

I have confirmed this behavior on Linux. I DO have   in my preserve entities list.

[KevinH]

So something in SourceUpdates when updating links in the NCX or in the Nav is most probably generating that error. I will look into it. As NCX generation with correct Nav works just fine.

[DiapDealer]

It is not until step four that the entity in the ncx gets borked. Is the non-breaking space decimal entity not valid in the nav or ncx document? By my thinking, everything in the nav/ncx is valid when the final Restructure to Sigil Norm is run. But I could be wrong.

[KevinH]

In xml, numeric entities are always valid even without a special namespace added as is done in epub2 xhtml for named entities.

So it is something in SourceUpdates that updates links in the ncx or parses the ncx that is probably the issue.

[DiapDealer]

I'm trying to nail it down to a shorter recreation process.

[KevinH]

https://github.com/Sigil-Ebook/Sigil...ement.py#L1341

I think the bug is in this routine as it does not handle other entities.

So to test that.

Start with an epub2 with ncx
1. Put numeric non-breaking space entity in one of the text fields of the ncx
2. Use Sigil BookBrowser Move to move the ncx to some other folder.

That should run the SourceUpdates on that NCX that in turn parses the ncx with sigil_bs4 to update links, then decode it back to xml which hits the routine above.

All of the above is a guess, all I have access to at the moment is an ipad.

[KevinH]

I think that snippet of code was meant to disable all non-primary named entities, but it also hits valid numeric entities and disables them as well.