How do I find the Chapter and or Page break in a PDF

[rachalmers]

Real newbie to Calibre here.,..
I'm trying to convert a pdf to MOBI. Mostly it works fine. but no matter what I do I can't get it to insert a page break between Chapters or anywhere else that I've got a chapter or even a page break.
In Debug output, the Input shows a <hr/> and displays a line in the webpage output, but after that, it's ignored.
What am I doing wrong?

There is nothing special in the document at all. No fancy formatting nothing. Just can't get it to insert page breaks at chapters. Which I believe it can do?
So where do I start please?

[jhowell]

See Read this before Posting PDF Questions, especially the section "My PDF has a table of contents or links/bookmarks, but they weren't used during conversion".

[Shohreh]

Hello,

Indeed and sadly, adding bookmarks to the source PDF doesn't help Calibre splitting chapters right when converting to EPUB.

Is there no way to help Calibre with this task, without having to mess with the HTML in the EPUB output?

[DNSB]

Pretty much no. See the numerous comments about PDF being the worst format to convert from.

[Shohreh]

I know, but it's odd that Calibre can't just use the PDF's bookmarks to know where new chapters start.

At this point, the conversion went pretty well after I removed the headers in the source PDF with a bit of Python.

I just need to figure out how to have Calibre start a new page with each new chapter.

--
Edit: If Calibre really is unable to split chapters by relying on the PDF's bookmarks, what about splitting the PDF into multiple files (one chapter = one PDF; For this, check qpdf, cpdf, mutool, etc.), have Calibre convert them into EPUB files, and then merge them into a single EPUB?
--
Edit: Is there an option to prevent ebook-convert.exe from adding a "Document Outline" at the end of the EPUB?

[Quoth]

If you knew how PDFs work you'd not think it odd.
Only actual paper, vellum and stone tablets are a worse conversion source than PDF, and some scanned PDFs are so bad that photographing the paper with a phone is better.

It's a waste otr time, because what you get working for one PDF (if you ever get it to 'work') may not work on the next PDF. My TCL NxtPaper 11 arrived today which is my latest solution to reading PDFs. Cheaper than Scribe, Elipsa, Boox etc.

[Shohreh]

Out of curiosity, why can't Calibre use a PDF's bookmarks to know where each chapter starts instead of guessing while reading the XHTML generated by pdftohtml?

[kovidgoyal]

Let's count the reasons:

1) there is no guarantee bookmarks entries are chapter starts
2) PDF consists of a bunch of font glyphs placed at absolute co-ordinates on the page. A bookmark or any link really is also just another co-ordinate on a page. There is no way to map that to some semantic element reliably. One has to use heuristics.

[Quoth]

3) Other mad stuff too.

[Shohreh]

Thanks for the infos.

Suggestion: If it's the user who added the bookmark into the PDF, it's reliable, and Calibre could make it an option to use that info to find chapters.

Anyway, an alternative, simpler solution than adding+removing bookmarks is to… 1) open the PDF in a reader eg. SumatraPDF, 2) make a list of pages/slices that make up chapters, 3) use it to split the source PDF into sub-PDFs (eg. cpdf; one chapter = one PDF) 4) run Calibre to convert them into EPUBs, and 5) finally join them into a single EPUB.

It does nothing to help converting PDFs into clean EPUBs, but at least, Calibre will know where each chapter starts and ends.