Re: chain of filters

Peter van Schijndel wrote:
> I have my Laserjet 5N running nicely with Cups and Foo-matic. I however
> want to implement a simple kind of accounting. In order to do this
> i need to run an extra *cupsFilter. This filter will not do anything
> with the printjob itself, it will just log the user. Based on the 7
> arguments filters are called with. 
> I however do not seem to able to have the job traverse more than one
> filter. Is it possible at all, to setup such a chain of filters. Ie
> have multiple filters react on the same mime-type? 
> Regards, 
> Peter
> 


Create two separate files:

   /etc/cups/petersch.types
   /etc/cups/petersch.convs

The separate files inoculate your setup in case of updates, which would
overwrite "/etc/cups/mime.types" and "/etc/cups/mime.convs". At cupsd
startup, all *.types and *.convs files are parsed and taken into account
for the setup of any filtering chains.

Now in "/etc/cups/petersch.types" define the "rules", for the MIME types
you want to either auto-recognize, or to process in the middle of your
filtering chain. Your definitions could be anything you like. See the
"mime.types" file for instuctions. Your mime.convs conversion rules must
match your definitions, of course.

I'll assume, that you want to plug a few of your own filters after CUPS'
own "pstops" filter, and that your filters don't change the fact that the
processed files are all PostScript



/etc/cups/petersch.types:
-------------------------

  application/vnd.cups-postscript_post_1   contains(0,128,<1B>%-12345X) + 
                                (contains(0,1024,"LANGUAGE=POSTSCRIPT")   
                                 contains(0,1024,"LANGUAGE = Postscript") 
                                 contains(0,1024,"LANGUAGE = POSTSCRIPT"))

  application/vnd.cups-postscript_post_2   contains(0,128,<1B>%-12345X) + 
                                (contains(0,1024,"LANGUAGE=POSTSCRIPT")   
                                 contains(0,1024,"LANGUAGE = Postscript") 
                                 contains(0,1024,"LANGUAGE = POSTSCRIPT"))

  application/vnd.cups-postscript_post_3   contains(0,128,<1B>%-12345X) + 
                                (contains(0,1024,"LANGUAGE=POSTSCRIPT")   
                                 contains(0,1024,"LANGUAGE = Postscript")  
                                 contains(0,1024,"LANGUAGE = POSTSCRIPT"))



/etc/cups/petersch.convs:
-------------------------

application/vnd.cups-postscript  application/vnd.cups-postscript_post_1 66    
petersch-filter.sh1

application/vnd.cups-postscript_post_1  application/vnd.cups-postscript_post_2 
66    petersch-filter.sh1

application/vnd.cups-postscript_post_2  application/vnd.cups-postscript_post_3 
66      petersch-filter.sh2

application/vnd.cups-postscript_post_3  application/vnd.cups-postscript 66      
petersch-filter.sh3


This invokes a filtering chain of

   pstops --> petersch-filter.sh1 --> petersch-filter.sh2 --> 
petersch-filter.sh3


Now it is your responsibility to program your various "petersch-filter.sh*".... 
;-)

Your first filter could for example do the page-counting, The second one 
could print a watermark onto each page. The third one could do an "imposition", 
laying out several pages on one sheet (like you need for pamphlet output.). 

Of course, you can also do all these steps inside one filter. The setup with 
3 different filters is artificial, just to show how it could be done, if you 
need various custom filters concatenated....

Of course, you can also do much more "wild" things, with your very own MIME 
types defined, and the target MIME type (the one that is the last generated 
by the filtering chain, going to the "backend", which finally sends the job 
to the printer), defined in a "*cupsFilter:" line in the PPD.

Cheers,
Kurt