Why can't you just purchase/license a 3rd party tool that can do this
for you? There are a variety of them available for various OS
platforms.
Leonard
On Oct 4, 2007, at 8:32 PM, Ted Chen wrote:
I ended up using pdfcopy by adding imported pages, each page
twice. Then define cropbox on each page accordingly. The result
pdf is almost as large as the original one, even though the # of
pages are doubled. This is better than my original solution, but
still has unnecessary references for the cropped out slides.
It's not as ideal as Mark's proposal, but works for now. :)
Thanks again.
Ted
----- Original Message ----
From: Ted Chen <[EMAIL PROTECTED]>
To: Post all your questions about iText here <itext-
[EMAIL PROTECTED]>
Sent: Thursday, October 4, 2007 3:51:31 PM
Subject: Re: [iText-questions] remove content from a page
Thanks a bunch, Mark, for the detailed analysis. It's very clear
and helpful for a pdf novice like me.
To make things worse, I not only need to parse this pdf, but would
need to run the same code against many docs. The good news is that
these docs are all produced by the same pdf printer. I guess in
order to reach my goal, I'll need to analyze all the docs and
figure out the common pattern. would be fun. :)
Although it might be a bit too fun for the current short time
period I have, I'll definitely get back to it, as I don't like my
current solution of cropping out the invisibles.
I appreciate your help.
Ted
----- Original Message ----
From: Mark Storer <[EMAIL PROTECTED]>
To: Post all your questions about iText here <itext-
[EMAIL PROTECTED]>
Sent: Thursday, October 4, 2007 2:15:21 PM
Subject: Re: [iText-questions] remove content from a page
No bueno.
Each of those slides is made up of a collection of resources &
drawing commands. To determine where one ends and the other starts
(and which one is which), you'd need to parse the page's content
stream.
!trivial.
iText includes a basic content tokenizer, but beyond that you're in
for a medium-large amount of research and work to pull this off.
I'll talk you through this particular page's contents to let you
know what you're in for:
q // push the current graphic state onto the state stack
0.1 0 0 0.1 0 0 cm // Concatenate Matrix to 1/10th scale (from the
default 1 0 0 1 0 0)
/R7 gs // use an extended graphic state
q // push
0.101562 0 6119.8 7920 re W n // clip to the given REctangle
1 1 1 rg // set the rgb foreground color to white
0.101562 0 6119.8 7920 re // rect
f // fill
0.199951 0.199951 0.199951 rg // new fill color
387.102 4136.9 4201.8 8.20312 re // rect
f // fill
387.102 4136.9 7.79688 3155.2 re //yada yada
f // blah blah
387.102 7283.9 4201.8 8.20312 re
f
4581.1 4136.9 7.79688 3155.2 re
f
Q // pop the graphic state
q // push
395.102 7284.1 m //move to
395.102 4144.9 l // line to
4580.9 4144.9 l //line
4580.9 7284.1 l //line
h // close the path
W n // create a new clipping region
1 1 1 rg // white fill
395.102 4144.9 4185.8 3139.2 re
f
q // push
4186 0 0 3139.5 394.5 4144.5 cm // big numbers due to prior 1/10th
matrix
/R8 Do // draw the R8 resource
Q //pop
Q //pop
q //push
419.102 7260.1 m // move, line line line, close, clip
419.102 4290.9 l
2397.9 4290.9 l
2397.9 7260.1 l
h
W n
q//push
1979.5 0 0 2969 418.5 4291 cm // we'd popped back to the 1/10th again
/R9 Do // draw R9
Q // pop
Q //pop
q //push
// lots more of the same till we get down to some text operators
// I'll leave in the intervening content to let you know what
// you're up against.
2333.1 7147.1 m
2333.1 5712.9 l
3927.9 5712.9 l
3927.9 7147.1 l
h
W n
q 1595.5 0 0 1434.5 2332.5 5712.5 cm
/R10 Do
Q
Q
q
1862.1 5769.1 m
1862.1 4303.9 l
3858.9 4303.9 l
3858.9 5769.1 l
h
W n
q 1997 0 0 1465 1861.5 4304 cm
/R11 Do
Q
Q
q
0.101562 0 6119.8 7920 re W n
0.199951 0.199951 0.199951 rg
387.102 336.898 4201.8 8.20312 re
f
387.102 336.898 7.79688 3156.2 re
f
387.102 3484.9 4201.8 8.20312 re
f
4581.1 336.898 7.79688 3156.2 re
f
Q
q
395.102 3485.1 m
395.102 344.898 l
4580.9 344.898 l
4580.9 3485.1 l
h
W n
1 1 1 rg
395.102 344.898 4185.8 3140.2 re
f
q 4186 0 0 3139.5 394.5 345 cm
/R12 Do
Q
Q
q
419.102 3461.1 m
419.102 1101.9 l
1678.9 1101.9 l
1678.9 3461.1 l
h
W n
q 1260 0 0 2358.5 418.5 1102 cm
/R13 Do
Q
Q
q
395.102 3485.1 m
395.102 344.898 l
4580.9 344.898 l
4580.9 3485.1 l
h
W n
1 1 1 RG // sets the foreground color to white
1 1 1 rg
// ******** And here comes the text ********
q
10 0 0 10 0 0 cm // 1/10th * 10. An odd way to go about things
BT // begin a text object
/R14 17 Tf // set the current font
1 0 0 1 242.9 249.85 Tm // set the Text Matrix
// my content editor turns odd binary values into {hex hex}
// so bear with me:
({01}{02}{03}{04}{03}{04}{05}{03}{06}{07}{08} \n{0b}{0c})Tj //
draw some characters
-41.45 -20.8 Td //
(\r{03}{0b}{0e}{07}{0f}{03}{0b}{07}\n{08}{07}\n{0b}{07}{08} {03}
{07}{10}\n{08}{08}{10}{03}{07} {11}{08}{03}{10}{07}\n{12})Tj
84.4 -20.8 Td // move the start of the next line
({13}{14}{05}{15}\n{16})Tj // draw more text
ET // end text object
// and then we're back to more of the same
Q
Q
q
419.102 344.898 4161.8 779.203 re W n
q 4186 0 0 802 418.5 321.5 cm
/R16 Do
Q
Q
Q
---------------------------
So given all that STUFF, you need to parse it, interpret it, and
figure out which resource is used where. You can then determine
the resource names of things used outside the page's media box (or
whichever page pox you're altering), and remove them from the
page's resource dictionary... then do the whole removeUnusedObjects
() thing.
Have fun. :\
--
--Mark Storer
Professional Geek
Catch up on fall's hot new shows on Yahoo! TV. Watch previews, get
listings, and more!
Shape Yahoo! in your own image. Join our Network Research Panel today!
----------------------------------------------------------------------
---
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration files using AJAX and a
browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
iText-questions mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/itext-questions
Buy the iText book: http://itext.ugent.be/itext-in-action/
-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
iText-questions mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/itext-questions
Buy the iText book: http://itext.ugent.be/itext-in-action/
