Generated TOC links go to wrong place

[brendontowle]

I'm told that this is user error (https://bugs.launchpad.net/calibre/+bug/2074284), so I'm checking here.

I'm trying to generate a TOC using Calibre 7.10, for an ePub that has very well-formed chapters and chapter/section headings. In the tutorial at https://manual.calibre-ebook.com/con...le-of-contents, I found these instructions:

Then, we set the options as:

Level 1 TOC : //h:h1
Level 2 TOC : //h:h2

This will result in an automatically generated two level Table of Contents that looks like:

Chapter 1

Section 1.1
Section 1.2

Chapter 2

Section 2.1

Using those instructions, I do get a two level ToC that looks like that, but: all of the section links, when clicked on, send you to the start of the chapter, not the section itself. And, sure enough, when I look at the data in the generated toc.ncx file, I see this:

<navPoint id="uIXA2jfX6ChloBsjmNnSAg9" playOrder="2" class="chapter">
<navLabel>
<text>1月</text>
</navLabel>
<content src="OPS/chapter-2_split_000.xhtml"/>
<navPoint id="uqfPktaL93cTQmqLLO4idQ3" playOrder="2" class="chapter">
<navLabel>
<text>1月1日</text>
</navLabel>
<content src="OPS/chapter-2_split_000.xhtml"/>
</navPoint>
<navPoint id="utq9DOX7bB9u69KAIm5i7y2" playOrder="2" class="chapter">
<navLabel>
<text>1月2日</text>
</navLabel>
<content src="OPS/chapter-2_split_000.xhtml"/>
</navPoint>

Note that the src link for each of the contained navLabels points to ...split_000.xhtml. But, the actual header elements for the contained navLabels are not in ...split_000; they're in ...split_001, ...split_002, etc.

So. What do I need to do to convince Calibre that I actually want the section links to point to the sections themselves, and not to the chapter?

[DNSB]

You will need to either split the files at the sections or use IDs to locate the section headers. All your current links are pointing to the same file which suggests that your sections are in the same file as the chapter header with no way to differentiate them.

  <h1>Chapter 1</h1>

  <h2 id="section0001">Section 1</h2>

  <h2 id="section0002">Section 2</h2>

  <h2 id="section0003">Section 3</h2>

  <h2 id="section0004">Section 4</h2>

  <h2 id="section0005">Section 5</h2>

  <h2 id="section0006">Section 6</h2>

<navMap>
  <navPoint id="navPoint1">
    <navLabel>
      <text>Chapter 1</text>
    </navLabel>
    <content src="Text/Section0001.xhtml" />
    <navPoint id="navPoint2">
      <navLabel>
        <text>Section 1</text>
      </navLabel>
      <content src="Text/Section0001.xhtml#section0001" />
    </navPoint>
    <navPoint id="navPoint3">
      <navLabel>
        <text>Section 2</text>
      </navLabel>
      <content src="Text/Section0001.xhtml#section0002" />
    </navPoint>
    <navPoint id="navPoint4">
      <navLabel>
        <text>Section 3</text>
      </navLabel>
      <content src="Text/Section0001.xhtml#section0003" />
    </navPoint>
    <navPoint id="navPoint5">
      <navLabel>
        <text>Section 4</text>
      </navLabel>
      <content src="Text/Section0001.xhtml#section0004" />
    </navPoint>
    <navPoint id="navPoint6">
      <navLabel>
        <text>Section 5</text>
      </navLabel>
      <content src="Text/Section0001.xhtml#section0005" />
    </navPoint>
    <navPoint id="navPoint7">
      <navLabel>
        <text>Section 6</text>
      </navLabel>
      <content src="Text/Section0001.xhtml#section0006" />
    </navPoint>
  </navPoint>
</navMap>

[JSWolf]

Before posting an error report, make sure you are using the latest calibre in case the error has been fixed.

[brendontowle]

Originally Posted by DNSB

You will need to either split the files at the sections or use IDs to locate the section headers. All your current links are pointing to the same file which suggests that your sections are in the same file as the chapter header with no way to differentiate them.

The whole point of my post is that they're not in the same file; all of them are in different files. See what I said about ...split_000, split_001, etc.

[brendontowle]

Originally Posted by JSWolf

Before posting an error report, make sure you are using the latest calibre in case the error has been fixed.

I just confirmed that the same issue is present in 7.15.

[brendontowle]

I've attached the .epub I'm trying to convert if that's helpful to anyone.

[DNSB]

For what it may be worth, your ePub did not contain any section files with names such as whatever_split_0001, see the attached image. I took a couple of seconds to open the file in Sigil and use it's ToC tool to regenerate the ePub3 nav document and the ncx ePub2 nav document for the h1/h2 header items. It seems to move to the section headers now in Sigil and calibre's viewer.

[kovidgoyal]

Your EPUB already has a table of contents, did you check the option to "Force use of auto-generated table of contents" without that the existing table of contents will be used.

[brendontowle]

Originally Posted by kovidgoyal

Your EPUB already has a table of contents, did you check the option to "Force use of auto-generated table of contents" without that the existing table of contents will be used.

Yes, I did check that option.

[brendontowle]

Originally Posted by DNSB

For what it may be worth, your ePub did not contain any section files with names such as whatever_split_0001, see the attached image. I took a couple of seconds to open the file in Sigil and use it's ToC tool to regenerate the ePub3 nav document and the ncx ePub2 nav document for the h1/h2 header items. It seems to move to the section headers now in Sigil and calibre's viewer.

Interesting. The only way I know of to add the ToC in Calibre is to convert the book, so I was converting from ePub to ePub; the one I attached was the original, not the converted. Apparently, in that process Calibre split the files, but maybe forgot to adjust the links so that they pointed back to the split files?

[kovidgoyal]

It's because your EPUB file is using chinese characters in id attributes. These break with some EPUB renderers like adobe digital editions. If you look at the conversion log it will be full of messgaes like

Removing fragment identifier '4月10日' from TOC as Adobe Digital Editions cannot handle it

[DNSB]

Kovid is both correct and incorrect. You can use UTF-8 characters in the IDs including Kanji but the IDs may not start with a number. If you edit the IDs to add say, 'x' after the id=" and then regenerate the table of contents (both the ePub3 nav document and the ePub2 ncx document), your ePub is valid. The three lines below are the XHTML code, the reference from nav.xhtml and the reference from toc.ncx
.

<h2 class="section-title" id="x2月1日">x1月1日</h2>

<a href="chapter-3.xhtml#x2月1日">x2月1日</a>

<content src="Text/chapter-3.xhtml#x2月1日" />

On a different note, Adobe Digital Editions is not happy with that ToC and I am not enthused enough to find out why. Sigil, calibre viewer and Thorium on my computer and my Kobo ereaders are happy with it.

And here's another copy of your file where I changed the ID to use month/day which seems to make ADE 4.5 happy with using the ToC though the heading may end up part way down the page.

[kovidgoyal]

I didnt say you cant use them, I said some renderers dont like them, principally Adobe Digital Editions. If you are using a halfway decent renderer, then they work fine, but conversion will strip them out because it tried to produce EPUBs that work on the widest variety of places possible. Having them in the ToC causes ADE to actually refuse to open the file.

[DNSB]

And my disagreement was that since the IDs started with a numeral, they were not valid XML ids.

[brendontowle]

Thanks, guys -- this is very helpful. I'm using Ulysses to generate the ePub, and apparently it's using the section header text directly as the ID; I'll talk to the developers there. If that doesn't pan out, I can use regex to do a one-time change of the IDs.

Thanks again for your help!