PDF conversion ignores images and cropping

[Starko]

Before i put (text based) e-book PDF on my Kindle i crop all the margins/whitespace and headers (repeated chapter names and/or author) and footers (page numbers).
Basically I leave just a few pixels around the text. That way i get the best PDF reading experience. Today, for the first time, i tried converting a PDF to MOBI & EPUB with Calibre. While I was AMAZED by the quality of the conversion there are two main issues i noticed rightaway:
1. No Images. I wonder why?
2. All the header and footer texts are back. I hope that this can be fixed.

Kovid, if you need the PDF file i am using, let me know.


Cheers

[Starson17]

Originally Posted by Starko

All the header and footer texts are back.

That's because the "cropped" pdf is really just a pdf with the cutoff material hidden. You can tell your pdf editor to remove the cropped material and then it will be gone, or you can remove headers and footers during conversion.

[DoctorOhh]

Originally Posted by Starko

2. All the header and footer texts are back. I hope that this can be fixed.

That is because the cropped area was hidden not removed. I often crop PDFs using Adobe then export the cropped PDF as HTML. Once exported I then use calibre to convert the html to epub. All Images and most links work fine and the cropped area is gone completely.

[Starko]

I am fully aware that cropping does not actually remove anything from pdf, just resizes the view area.

Originally Posted by dwanthny

... crop PDFs using Adobe then export the cropped PDF as HTML. Once exported I then use calibre to convert the html to epub. All Images and most links work fine and the cropped area is gone completely.

So why not do it in one go with Calibre?

All PDF viewers i am aware off display cropped documents correctly. Some of those viewers support (cropping-aware) reflow. That means that apparently there is some API that provides PDF content depending on cropping. Which in turn, i hope, means that rather than being something impossible to implement, this is just a feature that is not implemented in Calibre yet.

So, once again "I wonder why?" & "I hope that this can be fixed."

[kovidgoyal]

Implementing cropping in a viewer is easy, you just draw the whole page and crop out the part of the page specified by the cropbox. Implementing cropping in a text extraction tool is not nearly that easy.

Patches welcome.

[Manichean]

Originally Posted by Starko

So why not do it in one go with Calibre?

You can, but not by cropping. There's a removal mechanism in place for headers and footers. Look in the structure detection part of the conversion settings.

[Starko]

Originally Posted by kovidgoyal

Implementing cropping in a text extraction tool is not nearly that easy.

Because Adobe and other programs can do it, i was hoping it would be a well known Open Source library. Do you think you'd ever be interested in implementing this and image extraction functionality?

[kovidgoyal]

Who knows, I don't make open ended commitments like that. If at some point I get interested in PDF, maybe. But most likely not, since I have no personal motivation to work on PDF.