How to write a (pre-)filter

[Goffioul Michael]

> Hi,
> 
> I'm trying to get the KDE printing system to use a hand-made 
> script as a virtual printer, but I can't figure out how.
> 
> The complete task is: I want to print documents from my 
> laptop on my maths department's printers.
> 
> The situation: I use a laptop with Kubuntu (Edgy), hooking it 
> up to the math department's network, which consists of Sun 
> workstations. I have a login for those Suns. Due to 
> historical reasons, it is not possible to send a print job to 
> the department's print server from anywhere else than these Suns.
> 
> Work-around at the moment: I have written a script, which 
> automates the process of taking a .ps file (or converting 
> .pdf and .dvi to .ps), copying it to a temporary file on a 
> Sun via scp, logging in on the Sun, issueing the print 
> command, and finally deleting the temporary file. This script 
> needs a .ps or .pdf or .dvi file as an argument, i.e. it 
> can't just read the data to be printed from stdin. This 
> script is called "miprint".
> 
> What I would like: Calling the miprint script from the 
> command line is quite tedious, so it would be nice to have a 
> KDE printer that does the same: Print to a temp file, and 
> then get the tempfile printed on the Suns. 
> 
> I've tried to solve this problem by defining a "special 
> printer" with a filter filter, but how does this exactly 
> work? Is a filter the right tool to print a document to a 
> temporary .ps file, and then give this temp file to a script? 
> If yes, how does the syntax work, e.g. what exactly is "%in" 
> (input file? ps made from input file? data stream?), how is 
> it connected with "%filterinput" 
> and what is this "pipe" thing one can also define in this 
> context? I've looked at the XML files of the pre-defined 
> filters but they didn't help too much.
> 
> If a filter is the wrong approach anyway, what would be the right one?

You don't need a filter for this. If your script can be called
as simply as "miprint file_to_print", then create a pseudo/special
printer and define its command to "miprint %in". The script will
be fed with a temporary postscript file, kdeprint will take care
of the temporary file creation and removal.

> Anyway, when KDEPrint prints something for e.g. user "anna", with what
> user ID is this done? (This could be important for things like reading
> ssh keys etc.)

Scripts associated with pseudo printers are executed in the user
space (they are independent of CUPS); more specifically within the
kded daemon. So the script is executed with the current user ID.

Michael.