CUPS 1.2.7: definitions in .convs and.types notworking (SOLVED)

[Michael Leimann]

> Michael Leimann wrote:
> >> (Please remember that this isn't a commercial support forum, and
> >>   any responses you get are from users and developers in their free
> >>   time.  Not seeing a response over a weekend is not unusual...)
> >>
> >> Michael Leimann wrote:
> >>> ...
> >>> That is what we hoped to find with
> >>>
> >>> contains(0,900000,"<23466F726D3A") or
> >>> contains(0,900000,"23466F726D3A") or
> >>> contains(0,900000,23466F726D3A) or
> >>> contains(0,900000,<23466F726D3A>) or
> >>> contains(0,900000,"#Form:")or
> >>> ..
> >>>
> >>> What are we doing wrong here? Isn't is possible to search within a Postscript source?
> >> contains() currently only allows ranges of up to 4096 bytes.  This
> >> currently isn't documented anywhere but the source, but is a trade-
> >> off we made years ago for efficiency's sake (otherwise in your rule
> >> we'd have to read the first 900000 bytes of every print file, which
> >> will have serious a performance impact...)
> >>
> >> --
> >> ______________________________________________________________________
> >> Michael R Sweet                        Senior Printing System Engineer
> >>
> >
> > Thank you very much for this valuable information! I did not want to urge you to answer me, I only tried to promote my problem ;-).
> >
> > I agree, you are quite right regarding the performance impact. I did not want to do this (not knowing about the 4k limitation), but to make 100% sure that the "range" value is not my problem. I have created a one letter document (+"#Form:...") with OpenOffice Writer and printed it into a file:
> >
> > ll -h test.ps
> > -rw-r--r-- 1 mleimann dummy 37K 20. Jan 10:26 test.ps
> >
> > grep -b "<23466F726D3A544C424631>" test.ps
> > 36909:<23466F726D3A544C424631>
> >
> > This shows that I with ordinary methods a user is not able to set a searchable tag in a Postscript file.
> >
> > Any suggestion to help me out of this misery?
>
> If you are always printing from a particular application, you can
> probably look for that application in the PostScript header, and
> then have your filter always get used.  The filter can then do any
> substitutions it finds.
>
> --
> ______________________________________________________________________
> Michael R Sweet                        Senior Printing System Engineer
>

Thanks!

That's what I already planed to do. It is not nice but works. Maybe sooner or later an idea comes across how to steer that via an insertion into the DSC comments.

I do not know whether or not someone is interested in my weired ideas. For those of you who are, I post them hereafter:

-rw-r--r-- 1 root root       /etc/cups/form-ps.types
THIS ONE HAS TO BE AVAILABLE ON ALL CLIENTS AS WELL!

########################################################################
#
# Application-generated files...
#

application/form-ps		(string(0,%!) + \
                                contains(0,200,"Creator: (OpenOffice.org"))

-rw-r--r-- 1 root root       /etc/cups/form-ps.convs

########################################################################
#
# PostScript filters
#

application/form-ps     application/postscript	33      withform

-rwxr-xr-x 1 root root   /usr/lib/cups/filters/withform

#! /bin/bash
# see http://localhost:631/spm.html#WRITING_FILTERS
# debug info in /var/log/cups/error_log
set -x
# set inputfile to where the input comes from
inputfile="-"
[ -n "$6" ] && inputfile="$6"
# does /tmp/print exist?
outputfile=`ls -al /tmp | grep "^d"  | cut -d ":" -f 2 | cut -b 4- | grep "print"`
if [ "$outputfile" == "" ]; then
   mkdir /tmp/print
fi
# define outputfile
outputfile=`date "+%Y%m%d%H%M%S"`
# pipe input to file to make it searchable
cat $inputfile >/tmp/print/$outputfile-input.ps
# now set new inputfile
inputfile="/tmp/print/$outputfile-input.ps"
# '#Form:' tag given?
form=`grep -m 1 "<23466F726D3A" $inputfile`
if [ "$form" != "" ]; then
   # we have to do some work!
   #
   # piping printjob in one file per page
   p=1
   match="Yes"
   while [ $match = "Yes" ]
   do
     page=$(echo $((p+1000)) | cut -b 2-4)
     match=`psselect -p$p $inputfile /tmp/print/$outputfile-$page.ps 2>&1 | cut -d "[" -f 2 | cut -d "]" -f 1`
     if [ $match == $p ]; then
        p=$((p+1))
        match="Yes"
        else
           match="No"
           rm /tmp/print/$outputfile-$page.ps
     fi
   done
   #gs -q -dBATCH -dNOPAUSE -sDEVICE=pswrite -sOutputFile=/tmp/print/$outputfile-%03d.ps $inputfile
   # removing inputfile
   rm $inputfile
   inputfile=""
   #combining pages with overlay
   for inputfile in /tmp/print/$outputfile-*
   do
     page=`echo $inputfile | cut -d "-" -f 2 | cut -d "." -f 1`
     form=`grep -m 1 "<23466F726D3A" $inputfile`
     case $form in
        \<23466F726D3A544C424631\>)
                 overlay="/usr/lib/cups/filter/TOP-LAMP-Briefbogen1.eps";
                 sed -e 's/<23466F726D3A544C424631>/<                      >/g' $inputfile >/tmp/print/$outputfile-tmp.ps;
                 rm $inputfile;
                 mv /tmp/print/$outputfile-tmp.ps $inputfile;;
        \<23466F726D3A544C424632\>)
                 overlay="/usr/lib/cups/filter/TOP-LAMP-Briefbogen2.eps";
                 sed -e 's/<23466F726D3A544C424632>/<                      >/g' $inputfile >/tmp/print/$outputfile-tmp.ps;
                 rm $inputfile;
                 mv /tmp/print/$outputfile-tmp.ps $inputfile;;
        *)
                 overlay="";;
     esac
     # printing, EPS comes first, because it clears the page!
     # can -PAPERSIZE=a4 deleted to make it work on all sorts of paper?
     gs -q -dBATCH -dNOPAUSE -sPAPERSIZE=a4 -sDEVICE=pswrite -sOutputFile=/tmp/print/tmp-$outputfile-$page.ps $overlay $inputfile
     rm $inputfile
     # this is the trick!
     # offset of -0.035cm is only for specific printer!
     /usr/bin/pstops "2:0(0,-0.035cm)+1(0,0)" /tmp/print/tmp-$outputfile-$page.ps /tmp/print/final-$outputfile-$page.ps > /dev/null 2>&1
     rm /tmp/print/tmp-$outputfile-$page.ps
   done
   inputfile=""
   # this is a litte funny but seems to be necessary (merge all into one file for later cat to stdout)
   gs -q -dBATCH -dNOPAUSE -sDEVICE=pswrite -sOutputFile=/tmp/print/print-$outputfile.ps /tmp/print/final-$outputfile-*.ps
   rm /tmp/print/final-$outputfile-*.ps
   else
      # nothing left to do!
      mv $inputfile /tmp/print/print-$outputfile.ps
      inputfile=""
fi
cat /tmp/print/print-$outputfile.ps
rm /tmp/print/print-$outputfile.ps

There is only one question remaining:

Why do we have to have form-ps.types on all clients joining our network where we have two (to make it failover) CUPS-Servers?

Kind regards

Michael