Welcome to Eric Hameleers (Alien BOB)'s Wiki pages.

If you want to support my work, please consider a small donation:

This is an old revision of the document!

Printing with CUPS

FIXME this page is a work in progress FIXME

CUPS (the Common UNIX Printing System) is now the default printing subsystem in Slackware, after years of faithful service by the lprng package which is no longer part of Slackware. The commandline lpr program that some of you use is in fact part of CUPS, but it emulates the old lpr program that has been known in Linux/UNIX for ages and which was the way to print if you had lprng installed in Slackware. But a set of commandline tools is not all that CUPS installs on your system.

CUPS has a browser-based graphical user interface (check out yours at http://localhost:631/) for administering the service, like creating print queues, monitoring, deleting and restarting print jobs and such. CUPS servers in a network can be configured to talk to each other, so that printing from one CUPS machine (the client) to another CUPS machine (a CUPS server with attached printer) works automagically.
Not everything can be done through the graphical interface though. The basic initial setup of the CUPS service is still a matter of hand-editing certain configuration files.

Setting up the CUPS service

Slackware comes with a functional CUPS server out of the box. All you have to do is make the file /etc/rc.d/rc.cups executable and start the server. After that you can point your browser to http://localhost:631/ for the administratitive interface, and start adding printer queues.

chmod +x /etc/rc.d/rc.cups
/etc/rc.d/rc.cups start

If localhost does not work for you when you type the URL in the browser, try the IP address for localhost which is The URL would then look like this: But in any case, this would indicate that you messed up your /etc/hosts file which should contain (among others) the line       localhost

If that line is not present, you should add it. It's absence will break quite many network services.

In a later section I will show you how you can prevent other people on the network from accessing your computer's CUPS administrative interface. It would be silly if all of a sudden your print queue would be disabled by a joker. Access to localhost is safe - no one else but you on your computer can talk to the localhost address (which is tied to the loopback interface or ”lo” interface).


Creating print queues

Securing the admin web interface

Configuring CUPS clients


Making CUPS work with Samba


A PDF printer for CUPS

Usually, the CUPS print queues will print to paper of course. But there are cases when you want an electronic image of some document, or web page, that you do not need on paper. The PDF file format is the ideal candidate for this, since this format is supported on virtually all operating systems and architectures.
So, we need to add a print queue to CUPS which does not print our submitted jobs to paper sheets but instead generates a PDF file from our input, and makes this file available to us in some way.
This section describes exactly how to set this up, leaving the choice up to you whether you want your PDF files delivered to you via email, or have them dumped into a directory of your choice.
This setup uses no proprietary software to create the PDF files - all you need is already present on your Slackware machine. For those with several computers in a network, having a CUPS server to generate the PDF's has the additional advantage that it does not matter what the client computer is. Your Linux and Windows workstation, or your Mac, will be equally able to use this pdfprinter as long as your CUPS service allows job submissions from the network.

If you are running a network server with Samba, your CUPS pdfprinter will be integrated into the Samba configuration if you tell it to use CUPS for printing support (which is the default behaviour anyway). Windows clients will be able to generate PDF's by using a Samba print queue.

I came across an implementation of a PDF printer that supports CUPS and Samba. It can be found at http://www.linux-als-server.de/html/special-pdfprinter.php. My recipe below is based on the instructions found at that URL. Since the original text is in german, my article will help you non-german speaking persons setting this up I hope :-)

Required software

You will need to meet the following software requirements on your CUPS machine:

  • CUPS installed and configured (naturally)
  • GhostScript installed

Only if you want to deliver PDF files by email, you'll need these too:

  • Sendmail configured to deliver emails
  • Perl installed, with the additional module MIME::Entity. If you don't have MIME::Entity installed, you can compile it yourself (it is part of MIME::tools, which in turn requires the IO::stringy and Mail perl modules), or use the program cpan2tgz to create Slackware packages for it yourself, or download my precompiled Slackware packages (in my repository).

Installing a new CUPS backend

CUPS uses so-called backends to support output to printing devices. Examples of these backends are ipp, lpd and socket; they show up in the CUPS URI that you define when creating a new CUPS print queue, for instance printing via IPP: ”ipp://”.
We are going to create a new backend, which supports printing to PDF files and delivering these files to the correct end location (mailbox or directory path). The backend is going to be called pdf since that name is not yet being used.

  • First off, let's start with creating the necessary directories (as root). The PDF files are going to be output to a temporary location (the spool directory) before being emailed to their final destination. We'll be using a new directory ”/var/spool/pdf” for that, and make sure that everyone has write access there. If you do not want to use email delivery, this spool directory must be accessible by everyone who wants to use your pdfprinter. That means if you offer the pdfprinter as a network service to other computers in your network, you will have to export this spool directory using Samba or NFS. It is not within the scope of this article to describe how you would do that.

    The executables (in the form of Shell and Perl scripts) for the pdf backend are going to be stored in another new directory called /usr/lib/cups/pdf:
    mkdir -p /var/spool/pdf
    chmod 1777  /var/spool/pdf
    mkdir /usr/lib/cups/pdf
    chmod 775 /usr/lib/cups/pdf

    The chmod 1777 /var/spool/pdf makes /var/spool/pdf writable for everyone, plus sets the sticky bit which means that other users will be unable to delete files that they do not own themselves.

  • Next, we are going the create the file that provides or “pdf backend” functionality. Defining a new backend is really no more than creating an executable file called in CUPS's backends directory /usr/lib/cups/backend - this file's name will become the name of the backend. So, for the “pdf” backend we create the file /usr/lib/cups/backend/pdf and make this a shellscript, the contents of which you can find in the section below - PDF Printer Scripts. Do not forget to make this new “pdf” script executable:
    chmod 755 /usr/lib/cups/backend/pdf

    This backend will in turn call two other scripts which we shall place in the library directory /usr/lib/cups/pdf - see next bullet point.

  • All right, we will now create the two scripts that will do all the hard work for us. Install the two files ”ps2pdf.cups” and ”sendpdf” into the cups library directory which we just created: /usr/lib/cups/pdf. You can find these scripts in the section below called PDF Printer Scripts. Do not forget to make these files executable:
    chmod 755 /usr/lib/cups/pdf/ps2pdf.cups
    chmod 755 /usr/lib/cups/pdf/sendpdf

    The first script, ps2pdf.cups is seemingly derived from the ps2pdf script that came with an old version of GhostScript. It converts the PostScript input that CUPS generated for us into PDF output. The sendpdf script takes this PDF file and delivers it. By default, the version of the script as found in PDF Printer Scripts uses the sendmail program to deliver the PDF file to your inbox. It uses your logon name to determine the destination email address.

    If you have not setup local email delivery, or don't want email delivery of your PDF's, you can change the behaviour of the /usr/lib/cups/backend/pdf' script. Edit the last part of that script, commenting out the email delivery part and uncommenting the directory part.
    That piece of script would then look like this: <code> ## Normally the PDF file will be emailed to the creating user. ## Alternatively, you can decide not to email it, ## but leave the file on the server and restrict access by others: #if [ “$2” != ”” ]; then # $MAILBIN $2 $OUTPUTFILENAME # rm -f $OUTPUTFILENAME #fi if [ “$2” != ”” ]; then chown $2 $OUTPUTFILENAME chmod 700 $OUTPUTFILENAME fi exit 0 </code> The thing to remember here, is that the output directory for the PDF files in this case is the directory name you are going to use with the
    lpadmin command when you create the print queue - see the last bullet point. Remember to make that directory writeable by everyone and also set the sticky bit by running <code>chmod 1777 /your/pdf/output_dir</code> just like I did for my example directory /var/spool/pdf. * CUPS uses PPD files (PostScript Printer Definition files) for outputting correctly formatted print data. For our PDF backend, we will need a PPD file that can output color information in PostScript format (if you prefer grey-scale PDF files, you should of course get an appropriate non-colour PPD file).
    Adobe wrote a PPD file
    distiller.ppd which you can freely download at their web site. The Adobe site is often inaccessible, but fortunately there are many mirrors of that PPD file. Here is one: destiller.ppd; note that this file is called “destiller.ppd” instead.

    Another (really free) PPD file can be found on the cups-pdf web site, where you can find another implementation of a CUPS PDF print solution which I have not yet tried. The URL to this PPD file is PostscriptColor.ppd

    Save either of those files in the CUPS ppd-collection directory, under the name of
    /usr/share/cups/model/pdfcolor.ppd - the name does not really matter, you can keep the original filename if you wish, but I am using that name pdfcolor.ppd in the rest of this example as well. CUPS PPD files are gzipped - probably to save space. It is not a requirement to do so for our own PPD file, but we will do it anyway: <code> gzip /usr/share/cups/model/pdfcolor.ppd</code> This creates the file /usr/share/cups/model/pdfcolor.ppd.gz which we will use in the lpadmin command further down. * Finally, we need to restart the CUPS service to activate our new pdf backend, so we can create our pdfprinter queue . <code> /etc/rc.d/rc.cups restart </code> * Using the commandline tool lpadmin we create a new CUPS print queue for our PDF backend. <code> lpadmin -p pdfprinter -v pdf:/var/spool/pdf/ -D “Generate PDF files” -E -P /usr/share/cups/model/pdfcolor.ppd.gz </code> This creates the print queue called pdfprinter using the pdf backend, with the directory /var/spool/pdf to generate PDF files in, and it will use /usr/share/cups/model/pdfcolor.ppd.gz as the PPD file for this printer definition.

    If you want to add the printer queue through the CUPS web interface, you should pick ”PDF Creator” as the device and ”Adobe | Adobe Distiller” as the Model/Type of the printer. We're ready! The PDF printer setup is complete. ==== Using the PDF printer ==== Client computers (or you on your local computer) can create a PDF file of any printable data, by just printing to the CUPS print queue ”pdfprinter”. Linux client computers on the network which have CUPS configured as a client, will pick up the CUPS server automatically (CUPS services talk among each other on the network) so this will all work completely transparently. If you have a network with Windows computers, the best thing is to setup a Samba server on the CUPS server and let Samba use CUPS for printing. That way, our “pdfprinter” will automatically become available to these Windows computers as a network printer. Windows machines should install this network printer queue using a Generic PostScript driver. ===== PDF Printer Scripts ===== * Script
    /usr/lib/cups/backend/pdf <file> #!/bin/sh # ——————————————————————- # ”/usr/lib/cups/backend/pdf”: # ——————————————————————- # PDFBIN=/usr/lib/cups/pdf/ps2pdf.cups MAILBIN=/usr/lib/cups/pdf/sendpdf FILENAME= # filename of the PDF File PRINTTIME=`date +%Y-%m-%d_%H.%M.%S` # no argument, prints available URIs if [ $# -eq 0 ]; then if [ ! -x “$PDFBIN” ]; then exit 0 fi echo “direct pdf \”Unknown\” \”PDF Creator\”” exit 0 fi # case of wrong number of arguments if [ $# -ne 5 -a $# -ne 6 ]; then echo “Usage: pdf job-id user title copies options [file]” exit 1 fi # get PDF directory from device URI, and check write status PDFDIR=${DEVICE_URI#pdf:} if [ ! -d “$PDFDIR” -o ! -w “$PDFDIR” ]; then echo “ERROR: directory $PDFDIR not writable” exit 1 fi # generate output filename OUTPUTFILENAME= if [ “$3” = ”” ]; then OUTPUTFILENAME=“$PDFDIR/unknown.pdf” else if [ “$2” != ”” ]; then OUTPUTFILENAME=“$PDFDIR/$2-$PRINTTIME.pdf” else OUTPUTFILENAME=“$PDFDIR/$PRINTTIME.pdf” fi echo “PDF file: $OUTPUTFILENAME placed in: $PDFDIR” » $LOGFILE fi # run ghostscript if [ $# -eq 6 ]; then $PDFBIN $6 $OUTPUTFILENAME >& /dev/null else $PDFBIN - $OUTPUTFILENAME >& /dev/null fi # Make the file visible (but read-only except for owner); # This is only needed when the username ($2) is not set, # for instance when printing a test page from the web interface. chmod 644 $OUTPUTFILENAME ## Normally the PDF file will be emailed to the creating user. ## Alternatively, you can decide not to email it, ## but leave the file on the server and restrict access by others: if [ “$2” != ”” ]; then $MAILBIN $2 $OUTPUTFILENAME rm -f $OUTPUTFILENAME fi #if [ “$2” != ”” ]; then # chown $2 $OUTPUTFILENAME # chmod 700 $OUTPUTFILENAME #fi exit 0 # EOF # ——————————————————————- </file> * Script /usr/lib/cups/pdf/ps2pdf.cups <file> #!/bin/sh # ——————————————————————- # ”/usr/lib/cups/pdf/ps2pdf.cups”: # Convert PostScript to PDF. # ——————————————————————- OPTIONS=”” while true do case “$1” in -*) OPTIONS=“$OPTIONS $1” ;; *) break ;; esac shift done if [ $# -lt 1 -o $# -gt 2 ]; then echo “Usage: `basename $0` [options…] input.ps [output.pdf]” 1>&2 exit 1 fi infile=$1; if [ $# -eq 1 ] then outfile=$1 else outfile=$2 fi # Doing an initial 'save' helps keep fonts from being flushed between pages. exec gs -q -dNOPAUSE -dBATCH -sDEVICE=pdfwrite \ -sOutputFile=$outfile $OPTIONS -c save pop -f $infile # EOF # ——————————————————————- </file> * Script /usr/lib/cups/pdf/sendpdf''

    # -------------------------------------------------------------------
    # "/usr/lib/cups/pdf/sendpdf":
    # Parameter: Username, PDF File
    # -------------------------------------------------------------------
    use MIME::Entity;
    # Get the ARGS
    $to = $ARGV[0];
    $pdffile = $ARGV[1];
    # Set some variables
    $mailpipe = '| /usr/sbin/sendmail -t -oi';
    $from = "pdfprinter";
    $subject ="PDF File";
    $content = "Your PDF print";
    # create the mail
    my $mail = MIME::Entity->build( Type    => 'text/plain',
                                    From    => $from,
                                    To      => $to,
                                    Subject => $subject,
                                    Data    => $content
    # Attach the pdf file
    $mail->attach( Path     => $pdffile,
                   Type     => 'application/pdf',
                   Encoding => 'base64'
    # Open mailpipe and send the mail
    open MAIL, "$mailpipe" or die "Could not open mailpipe \"$mailpipe\" !\n";
    close MAIL;
    # EOF
    # -------------------------------------------------------------------

Personal Tools