Epub to Epub: issue with Multiple headings and page breaks

[erion]

Hello all,
I'm trying to restructure an Epub file, splitting chapters into separate html files.
The book has a proper ToC, so there is no problem on that side. However when it comes to splitting the files, the following happens:
The book has page breaks in <div> tags, such as

<div class="mbppagebreak"></div>

where Calibre splits accordingly. However, when a <h2> tag is followed by another <h2> tag, there is a file split between the two.
Using the command line converter (ebook-convert), I've tried turning heuristics on (--enable-heuristics), resulting in the whole chapter wrapped in <h1> tags. I also tried specifying '--chapter-mark none', which should not apply page breaks before detected chapters (i.e. h2 tags).
Then came the XPath way, setting it to heading 6 which is not present in the document: '--page-breaks-before //h:h6'.
Sadly none of these seem to help. The chapter title is either in the previous html file, then in the next the second h2 tag with the actual text, or the first h2 tag is in a separate file and in the next file there's the second h2 tag with the chapter text.
It's also worth mentioning that sometimes there is only one h2 tag, sometimes there is two, when the chapter has a date.
Could anyone please point out what could be wrong here?
Edit: h2 tag meaning <h2> and a closing </h2>

Erion

[theducks]

Originally Posted by erion

Hello all,
I'm trying to restructure an Epub file, splitting chapters into separate html files.
The book has a proper ToC, so there is no problem on that side. However when it comes to splitting the files, the following happens:
The book has page breaks in <div> tags, such as

<div class="mbppagebreak"></div>

where Calibre splits accordingly. However, when a <h2> tag is followed by another <h2> tag, there is a file split between the two.
Using the command line converter (ebook-convert), I've tried turning heuristics on (--enable-heuristics), resulting in the whole chapter wrapped in <h1> tags. I also tried specifying '--chapter-mark none', which should not apply page breaks before detected chapters (i.e. h2 tags).
Then came the XPath way, setting it to heading 6 which is not present in the document: '--page-breaks-before //h:h6'.
Sadly none of these seem to help. The chapter title is either in the previous html file, then in the next the second h2 tag with the actual text, or the first h2 tag is in a separate file and in the next file there's the second h2 tag with the chapter text.
It's also worth mentioning that sometimes there is only one h2 tag, sometimes there is two, when the chapter has a date.
Could anyone please point out what could be wrong here?
Edit: h2 tag meaning <h2> and a closing </h2>

Erion

Nothing wrong. You just found a condition that needs human help.

There is a Convert ((Heuristics) preference that re-assigns (downwards) a second Consecutive H tag.

[erion]

Hi,

Originally Posted by theducks

Nothing wrong. You just found a condition that needs human help.

There is a Convert ((Heuristics) preference that re-assigns (downwards) a second Consecutive H tag.

The problem is the renumbering does not actually happen. Even if I manually modify the second h2 tag of the first chapter (with the date), the chapter title with a <h2> tag will be in a separate file (probably because of the page break right before it), and the date with a <h3> tag will be in the next file with the chapter's text itself.
The original .epub's first html file ends with the chapter title of the first chapter, and in the next file is the date and the text.
This is just a wild guess, but could it be that for some odd reason Calibre honours the original file splits as well?

Erion

[Serpentine]

Use Sigil to do this kind of work, you just insert "SGF Chapter Markers" (hr tags with a specific class), where needed. You then use the "split on chapter markers" function.

So you can try automate it to only insert at the first h tags, or else it's usually quick enough to just do by hand.

[theducks]

Originally Posted by erion

Hi,



The problem is the renumbering does not actually happen. Even if I manually modify the second h2 tag of the first chapter (with the date), the chapter title with a <h2> tag will be in a separate file (probably because of the page break right before it), and the date with a <h3> tag will be in the next file with the chapter's text itself.
The original .epub's first html file ends with the chapter title of the first chapter, and in the next file is the date and the text.
This is just a wild guess, but could it be that for some odd reason Calibre honours the original file splits as well?

Erion

If it still breaks on H3 and H2 as well, look and see if you changed the chapter detection (normally H1 and H2) to include h3.

look at the stylesheet and see if a page-break-before is assigned to the class of those headers.

[erion]

Originally Posted by theducks

If it still breaks on H3 and H2 as well, look and see if you changed the chapter detection (normally H1 and H2) to include h3.

look at the stylesheet and see if a page-break-before is assigned to the class of those headers.

Thanks, this seems to be the problem. The second h2, i.e. the date has a page-break-before: always assigned to it. It does not make any sense to break there, but that's a different story.
Huge thanks to everyone who replied or even thought about it!

Erion