Cropping PDFs

[harryo]

I was thinking that it would be nice if iRex allowed us to add some cropping information to the manifest file.

It would be great if, on a per document basis, we could add something like (hope this comes through OK) ...

<croptop>20</croptop><cropbottom>20</cropbottom>
<cropleft>50</cropleft><cropright>50</cropright>

The PDF renderer could generate a larger page that is then cropped by the specified number of pixels before displaying.

This would allow the user to reduce the amount of white space around the pages by as much as they prefer.

I'm suggesting this for the case where one only has the PDF (I have a lot of reference books I've purchased in PDF format that I don't have the source for).

[ali]

I found that all PDF's I tried so far could be cropped manually by editing the MediaBox.

For example, an A4 PDF will contain something like

/MediaBox [ 0 0 595 842 ]

in one or several places. I just adjusted the four values - to crop 50 points from each border, I'd change that to

/MediaBox [ 50 50 545 792 ]

Both acroread and xpdf will honor that. That way, you can even crop page numbers / chapter headings and such away.

Under Linux, I'd do it using sed:

sed 's/0 0 595 842/50 50 545 792/g' <book.pdf >cropped.pdf

Under Windows, you'd have to open the pdf in notepad or something (don't know if that harms binary streams in the pdf)

[yokos]

Originally Posted by harryo

I was thinking that it would be nice if iRex allowed us to add some cropping information to the manifest file.

It would be great if, on a per document basis, we could add something like (hope this comes through OK) ...

<croptop>20</croptop><cropbottom>20</cropbottom>
<cropleft>50</cropleft><cropright>50</cropright>

Nice idea! Yepp, not all iLiad users have access to commercial software like Acrobat to crop pdf files.

Cropping of pdf files was dicussed earlier in this forum, I don't know, what the result was [existence of an OS program].

Uh, just change the mediabox tag. This is so easy? Cool!

[harryo]

Originally Posted by ali

I found that all PDF's I tried so far could be cropped manually by editing the MediaBox.

For example, an A4 PDF will contain something like

/MediaBox [ 0 0 595 842 ]

in one or several places. I just adjusted the four values - to crop 50 points from each border, I'd change that to

/MediaBox [ 50 50 545 792 ]

Both acroread and xpdf will honor that. That way, you can even crop page numbers / chapter headings and such away.

Fantastic! I'll try whipping up a Ruby script to do this automatically for me.

Thanks for the information.

[yokos]

Well, I tried this; I didn't worked for me. Adobe Reader can't read this file anymore. Perhaps I did it the wrong way.

[ali]

Originally Posted by yokos

Well, I tried this; I didn't worked for me. Adobe Reader can't read this file anymore. Perhaps I did it the wrong way.

What did you do, and how?

[harryo]

Well, I had a certain amount of success with the MediaBox hacking approach.

However, for some of my PDFs, while this change seems to have had the desired effect when looking at them in Acrobat, after transfer to the Iliad, there seems to be extra white space on the left and right, somewhat negating the desired result.

I've not worked out yet how the ones that work well differ from those that exhibit the extra white space, but I'll keep working at it.

At least I've been able to get rid of extraneous white space at top and bottom in every document, which does make a difference.

[ElaHuguet]

The iLiad fits the pdf into its screen respecting the document's proportions, so maybe that's why you're still seeing white side margins.

edit: ie. cropping the side margins on a 15cm long pdf will make absolutely no difference.

[harryo]

Originally Posted by ali

What did you do, and how?

I don't know how ali did it, but I used this piece of Ruby code

cropLeft   = ARGV[0].to_i
cropRight  = ARGV[1].to_i
cropTop    = ARGV[2].to_i
cropBottom = ARGV[3].to_i

originalPath = ARGV[4]
croppedPath  = originalPath.sub(/\.pdf$/, " (cropped).pdf")

File.open(originalPath, "rb") do |input|
  pdf = input.read

  pdf.gsub!(/\/MediaBox \[\s*(\d*)\s*(\d*)\s*(\d*)\s*(\d*)\s*\]/) do
      n1 = $1.to_i + cropLeft
      n2 = $2.to_i + cropTop
      n3 = $3.to_i - cropRight
      n4 = $4.to_i - cropBottom

      "/MediaBox [#{n1} #{n2} #{n3} #{n4}]"
  end

  File.open(croppedPath, "wb") { |output| output.write pdf }
end

I don't have a reference for the /MediaBox parameters, so I did it by trial and error. In reality, I didn't take much care, just ran the program and fiddled the four values until I got what looked good for each document.

I just wanted to get something working. I'll clean it up later, once I have the details worked out a bit better.

[ath]

Originally Posted by ali

I found that all PDF's I tried so far could be cropped manually by editing the MediaBox.

PDF files conforming to older versions of the format can be done that way ... as long as you do not add or delete any characters! -- that will screw up the object offset table, and it's not certain that recovery will work. (If you have some other Box nearby, containing the same data, you may be able to remove it to get the extra character space: if I remember, all Boxes default to MediaBox if they aren't specified.)

Of course, if your editor changes the original newlines behind your back, it will mess up thing even more. PDF files should be treated as binary files.

[ali]

Originally Posted by ath

Of course, if your editor changes the original newlines behind your back, it will mess up thing even more. PDF files should be treated as binary files.

That explains the "bad xref table" errors I was getting. Xpdf complains "Error (0): PDF file is damaged - attempting to reconstruct xref table..." but displays pages correctly afterwards.

"Save a copy" in acroread doesn't fix it... Hmmmmmmmmmmmmmmmm Have to think about this.

[arivero]

Also a pdfcrop utility exist in the TeX packages, but it generates files bigger.

[k2r]

I don't know if anybody mentioned if, but on MacOSX you can just open your PDF in preview.app, select an area on a page, see if the content-area on the other pages are complete and have preview crop all pages using the same stencil.
Then use "save as..." to save a cropped copy.
5 mouseclicks.

k2r

[harryo]

Originally Posted by ElaHuguet

The iLiad fits the pdf into its screen respecting the document's proportions, so maybe that's why you're still seeing white side margins.

edit: ie. cropping the side margins on a 15cm long pdf will make absolutely no difference.

Ah, yes. That could explain it. It may be the ones where I didn't need to crop top and bottom much that are showing more white space on the left and right than I'd expected.

[harryo]

Originally Posted by arivero

Also a pdfcrop utility exist in the TeX packages, but it generates files bigger.

That sounds like the perfect way to do it.

I'm not overly worried about the files being larger, unless we're talking an order of magnitude. The 100 or so MB in the Iliad is enough to hold a lot of reference material. My largest book is only about 8MB and that's over 800 pages worth.

I'll give that a go tonight.