[PDF → ePUB] Alternative to Calibre?

[Shohreh]

Hello,

Like everyone here, I tried using Calibre to convert a PDF to ePUB… with mixed results. It's mostly OK, but there are small issues every few pages.

"Heuristic Processing" and regexes didn't work to remove headers/footers and other unwanted parts.

I also tried a few online sites… with no better result.

Is Calibre the only open-source solution?

Is there any tool to turn PDF → HTML, to see if I could clean up the HTML myself before turning it into ePUB?

Thank you.

--
Edit: I just read that an ePUB is actually a zip file with HTML, CSS, PNG etc. Renaming the extension from .epub to .zip is all that is required to check the files.

[DNSB]

Originally Posted by Shohreh

Edit: I just read that an ePUB is actually a zip file with HTML, CSS, PNG etc. Renaming the extension from .epub to .zip is all that is required to check the files.

You could also use calibre's builtin editor or Sigil to check and edit the files. Very likely easier than extracting files from the epub to edit them. I use both since there are some tasks for which one is better/easier to use than the other.

[Shohreh]

Thanks for the tips.

Apparently, Sigil is not available for Windows as binaries.

Besides those below, is there a good, recent article about how to use Calibre to generate an EPUB with the least amount of errors, and, if need be, use its editor to clean up and finalize an EPUB? For instance, it seems not to like "inserts", tables, and headers/footers.

• https://www.mobileread.com/forums/sh...d.php?t=118605
• https://manual.calibre-ebook.com/faq...rious-problems
• https://manual.calibre-ebook.com/con...#pdfconversion
• https://dearauthor.com/ebooks/calibr...nversion-tips/

[itimpi]

Apparently, Sigil is not available for Windows as binaries.

Why do you say this? I see both 64-bit and 32-bit versions available for Windows

[Shohreh]

Where?

https://sigil-ebook.com/get/ → https://github.com/Sigil-Ebook/Sigil/releases

Another couple of questions while I experiment:
1. In the Search & replace, although the regexes work fine when testing with the wizard, the strings still end up in the EPUB. Any idea why?

2. Is there a way to tell Calibre to ignore such and such pages in the input file, such as those that contain the ToC or tables, since I know they'll turn into garbage anyway? I might try and replace the tables later in the EPUB as PNG instead.

[DNSB]

Originally Posted by Shohreh

Where?

https://sigil-ebook.com/get/ → https://github.com/Sigil-Ebook/Sigil/releases

Disregard the Sigil 1.2.1 source only release --as it says, for an issue with Python and Linux.

Browse down to the 1.20 release, click on the arrow by Assets. See attached image.

One question out of curiosity is in your regex: why are you looking for a period instead of an underscore? Your search string contains simmons\.qxd while you are searching for c12_simmons_qxd.

[Shohreh]

Thanks, I didn't know the files were hidden in the Assets section.

You're right for the regex.

---
Edit: It says "Note that currently Sigil only provides binaries that work for Windows x86 and x64 and will only run on Vista or newer releases.", but only Sigil-1.2.0-Windows-x64-Setup.exe is available; The last 32 bit release is 0.9.14 (Jun 11, 2019).

[DNSB]

Originally Posted by Shohreh

Thanks, I didn't know the files were hidden in the Assets section.

You're right for the regex.

---
Edit: It says "Note that currently Sigil only provides binaries that work for Windows x86 and x64 and will only run on Vista or newer releases.", but only Sigil-1.2.0-Windows-x64-Setup.exe is available; The last 32 bit release is 0.9.14 (Jun 11, 2019).

There are issues with some of the binaries used by Sigil not being supplied in 32 bit versions. This requires that not only compiling Sigil but also custom compiling those binaries.

This has been discussed in the Sigil forum and is probably best continued there.

I will admit the last 32 bit machine I ran was close to a decade ago so the disappearance of the x86 version is not a big deal to me.

[Shohreh]

My e-reader reads PDFs relatively well; It just doesn't support changing the font size, making it a bit harder to read than EPUB. Maybe it's a limitation of PDF, not this reader particularly.

So it looks like, because of the nature of PDF, Calibre and other tools simply cannot handle more complex parts such as tables or insets.

Incidently, considering how many e-reader users need to access PDFs, I might not be the only person curious to know more about PDF without delving too deeply either, and understand how it works (Postscript, etc.) under the hood and why it's such a pain to turn into EPUB.

Are there other sites/books you would recommend?

• https://en.wikipedia.org/wiki/PDF
• Understanding the Portable Document Format (PDF)
• Understanding the PDF File Format: Overview by Leon Atherton
• "Conversion of PDF books in ePub format" by Simone Marinai, Emanuele Marino, and Giovanni Soda
• Adobe Acrobat X PDF Bible by Ted Padova
• Acrobat PDF Bible Paperback by Ted Padova
• PDF Hacks: 100 Industrial-Strength Tips & Tools by Sid Steward
• PDF Explained: The ISO Standard for Document Exchange by John Whitington

[Shohreh]

Is Caliber using pdftohtml before doctoring the HTML and turning it into an EPUB file?

I notice that…

1. pdftohtml hasn't been updated since 2006-08-03 (!)
2. mutool, which is open-source like pdftohtml, does an excellent job (graphically speaking) converting a PDF into a single HTML file with pictures embedded as base64.

Why doesn't Caliber use mutool instead?

Do you know of a good tool to turn the HTML file from mutool into EPUB?

[kovidgoyal]

that is non-reflowable HTML, it is just as useless as the original PDF file. And calibre uses poppler, which took over pdftohtml from xpdf over a decade ago.

[Shohreh]

Good to know. The HTML files in the EPUB still mention pdtohtml, hence the confusion:

<meta name="generator" content="pdftohtml 0.36"/>"

--
Edit: Is there a good article somewhere that goes through the options, and generally speaking, explains how to best use Calibre to convert a PDF into EPUB?

[DNSB]

Originally Posted by Shohreh

Good to know. The HTML files in the EPUB still mention pdtohtml, hence the confusion:

<meta name="generator" content="pdftohtml 0.36"/>"

--
Edit: Is there a good article somewhere that goes through the options, and generally speaking, explains how to best use Calibre to convert a PDF into EPUB?

The best way to convert PDF is not to do so (IMO). In the real world, be prepared to spend some quality time with an epub editor to fix up the issues from converting a page oriented format to a reflowable format.

[Shohreh]

Thank you.

Turns out my e-reader isn't bad at displaying PDF after changing a couple of settings, including a… "Reflow text" option.

Since PDFs are so common, when buying an e-reader, it's a good idea to check how well it supports PDFs without bothering trying to turn them into EPUBs.

Out of curiosity, I'll see if I can find infos about what all the settings in Caliber's Convert dialog mean.

Am I correct in understanding that Caliber first lets poppler convert the PDF into HTML + PNG/JPG, and then goes to work on the HTML based on the settings in the Conversion dialog?

[kovidgoyal]

Originally Posted by Shohreh

Am I correct in understanding that Caliber first lets poppler convert the PDF into HTML + PNG/JPG, and then goes to work on the HTML based on the settings in the Conversion dialog?

yes .