xref: /third_party/cups-filters/README

README - OpenPrinting CUPS Filters v1.28.17 - 2023-01-24
--------------------------------------------------------

Looking for compile instructions?  Read the file "INSTALL.txt"
instead...


INTRODUCTION

    CUPS is a standards-based, open source printing system developed
    by Apple Inc. for Mac OS® X and other UNIX®-like operating
    systems.  CUPS uses the Internet Printing Protocol ("IPP") and
    provides System V and Berkeley command-line interfaces, a web
    interface, and a C API to manage printers and print jobs.

    This package contains backends, filters, and other software that
    was once part of the core CUPS distribution but is no longer
    maintained by Apple Inc. In addition it contains additional
    filters and software developed independently of Apple, especially
    filters for the PDF-centric printing workflow introduced by
    OpenPrinting and a daemon to browse broadcasts of remote CUPS
    printers and IPP printers to make them available locally.

    From CUPS 1.6.0 on, this package is required for using printer
    drivers (and also driverless printing) with CUPS under Linux. With
    CUPS 1.5.x and 1.4.x this package can be used optionally to switch
    over to PDF-based printing. In that case some filters are provided
    by both CUPS and this package. Then the filters of this package
    should be used.

    For compiling and using this package CUPS, libqpdf (8.3.0 or
    newer), libjpeg, libpng, libtiff, freetype, fontconfig, liblcms
    (liblcms2 recommended), libavahi-common, libavahi-client, libdbus,
    and glib are needed. It is highly recommended, especially if
    non-PDF printers are used, to have at least one of Ghostscript,
    Poppler, or MuPDF.

    The Poppler-based pdftoraster filter needs a C++ compiler which
    supports C++11 and Poppler being built with the "./configure"
    option "-DENABLE_CPP=ON" for building the C++ support library
    libpoppler-cpp. This is the case for most modern Linux
    distributions.

    If you use MuPDF as PDF renderer make sure to use at least version
    1.15, as the older versions have bugs and so some files get not
    printed correctly.

    For Apple Raster output (used by AirPrint printers) at least CUPS
    2.2.2 is required.

    For Braille embosser support (see below) you will also need at
    least liblouis, ImageMagick, and poppler-utils. Recommended is to
    also have liblouisutdml, antiword, docx2txt for more sophisticated
    Braille generation representing also the formatting of the input
    text. None of these is needed for compiling cups-filters.

    CUPS, this package, and Ghostscript contain some rudimentary
    printer drivers and especially the filters needed for driverless
    printing (currently PWG Raster, Apple Raster, PCLm, and PDF output
    formats, for printers supporting IPP Everywhere, AirPrint, Wi-Fi
    Direct, and other standards), see
    http://www.openprinting.org/drivers/ for a more comprehensive set
    of printer drivers for Linux.

    See
    
    https://wiki.linuxfoundation.org/openprinting/pdf_as_standard_print_job_format

    for information about the PDF-based printing workflow.

    Report bugs to

    https://github.com/OpenPrinting/cups-filters/issues

    or alternatively to

    https://bugs.linuxfoundation.org/

    Choose "OpenPrinting" as the product and "cups-filters" as the component.

    See the "COPYING" files for legal information.

IMAGE PRINTING DEFAULT CHANGED TO "SCALE TO FIT"

    Compared to the PostScript-based original CUPS filters there is a
    change of defaults: The imagetopdf and imagetoraster filters print
    in "scale-to-fit" mode (image is scaled to fill one page but
    nothing of the image being cut off) by default.

    This is done to support photo printing via AirPrint. The photo
    apps on Apple's iOS devices send print jobs as JPEG images and do
    not allow to set any options like "scaling" or the page size. With
    "scale-to-fit" mode set by default, the iOS photos come out on one
    page, as expected.

    To get back to the old behavior, supply one of the options
    "nofitplot" "filplot=Off", "nofit-to-page", or "fit-to-page=Off".

GHOSTSCRIPT RENDERING OF FILLED PATHS

    When Ghostscript is rendering PostScript or PDF files into a
    raster format the filled paths are ususally rendered with the
    any-part-of-pixel method as it is PostScript standard. On
    low-resolution printers, like label printers with 203 dpi,
    graphics output can get inaccurate and so for example bar codes do
    not work any more. This problem can be solved by letting
    Ghostscript use the center-of-pixel method.

    This can be done by either supplying the option "-o
    center-of-pixel" or "-o CenterOfPixel" on the command line when
    printing or by adding a "CenterOfPixel" option to the PPD file and
    set it to "true", for example by adding the following lines to the
    PPD file of the print queue (usually in /etc/cups/ppd/):

        *OpenUI *CenterOfPixel/Center Of Pixel: PickOne
        *OrderDependency: 20 AnySetup *CenterOfPixel
        *DefaultCenterOfPixel: true
        *CenterOfPixel true/true: ""
        *CenterOfPixel false/false: ""
        *CloseUI: *CenterOfPixel

   This option can be used when the print queue uses the gstoraster
   filter.

POSTSCRIPT PRINTING RENDERER AND RESOLUTION SELECTION

    If you use CUPS with this package and a PostScript printer then
    the included pdftops filter converts the print job data which is
    in PDF format into PostScript. By default, the PostScript is
    generated with Ghostscript's "ps2write" output device, which
    generates a DSC-conforming PostScript with compressed embedded
    fonts and compressed page content. This is resource-saving and
    leads to fast wire transfer of print jobs to the printer.

    Unfortunately, Ghostscript's PostScript output is not compatible
    with some printers due to interpreter bugs in the printer and in
    addition, processing (by Ghostscript or by the printer's
    interpreter) can get very slow with high printing resolutions when
    parts of the incoming PDF file are converted to bitmaps if they
    contain graphical structures which are not supported by
    PostScript. The bitmap problem especially occurs on input files
    with transparency, especially also the ones produced by Cairo
    (evince and many other GNOME/GTK applications) which unnecessarily
    introduces transparency even if the input PDF has no transparency.

    Therefore there are two possibilities to configure pdftops at
    runtime:

    1. Selection of the renderer: Ghostscript, Poppler, pdftocairo,
    Adobe Reader, or MuPDF

    Ghostscript has better color management and is generally optimized
    more for printing. Poppler produces a PostScript which is
    compatible with more buggy built-in PostScript interpreters of
    printers and it leads to a somewhat quicker workflow when
    graphical structures of the input PDF has to be turned into
    bitmaps. Adobe Reader is the PDF renderer from Adobe, the ones who
    created PDF and PostScript. pdftocairo is a good choice for the
    PDF output of Cairo (for example when printing from evince).  It
    is less resource-consuming when rasterizing graphical elements
    which cannot be represented in PostScript (like
    transparency). Note that pdftocairo only supports PDF input using
    DeviceRGB, DeviceGray, RGB or sGray and is not capable of
    generating PostScript level 1. So its support is only experimental
    and distributions should not choose it as default.

    The selection is done by the "pdftops-renderer" option, setting it
    to "gs", "pdftops", "pdftocairo", "acroread", "mupdf", or "hybrid":

    Per-job:           lpr -o pdftops-renderer=pdftops ...
    Per-queue default: lpadmin -p printer -o pdftops-renderer-default=gs
    Remove default:    lpadmin -p printer -R pdftops-renderer-default

    By default, pdftops uses Ghostscript if this does not get changed
    at compile time, for example by the Linux distribution vendor.

    Hybrid means Ghostscript for most printers, but Poppler's pdftops
    for Brother, Minolta, and Konica Minolta. Printer make and model
    information comes from the PPD or via the "make-and-model" option.

    2. Limitation of the image rendering resolution

    If graphical structures of the incoming PDF file have to be
    converted to bitmaps due to limitations of PostScript, the
    conversion of the file by pdftops or the rendering by the printer
    can get too slow if the bitmap resolution is too high or the
    printout quality can degrade if the bitmap resolution is too low.

    By default, pdftops tries to find out the actual printing
    resolution and sets the resolution for bitmap generation to the
    same value. If it cannot find the printing resolution, it uses 300
    dpi. It never goes higher than a limit of 1440 dpi. Note that this
    default limit can get changed at compile time, for example by the
    Linux distribution vendor.

    The resolution limit for bitmaps can be changed to a lower or
    higher value, or be set to unlimited. This is done by the option
    "pdftops-max-image-resolution", setting it to the desired value
    (in dpi) or to zero for unlimited. It can be used per-job or as
    per-queue default as the "pdftops-renderer" option described
    above.

    The "pdftops-max-image-resolution" option is ignored when Adobe
    Reader is selected as PDF renderer.

POSTSCRIPT PRINTING DEBUG MODE

    Sometimes a PostScript printer's interpreter errors, crashes, or
    somehow else misbehaves on Ghostscript's output. To find
    workarounds (currently we have already workarounds for Brother and
    Kyocera) it is much easier to work with uncompressed PostScript.
    To get uncompressed PostScript as output, send a job with the
    "psdebug" option, with commands like the following:

      lpr -P <printer> -o psdebug <file>
      lp -d <printer> -o psdebug <file>

    If you want to send your job out of a desktop application, run

      lpoptions -p <printer> -o psdebug

    to make "psdebug" a personal default setting for you.

    To extract the PostScript output for a developer to analyse it,
    clone your print queue to a one which prints into a file:

      cupsctl FileDevice=yes
      lpadmin -p test -E -v file:/tmp/printout \
      -P /etc/cups/ppd/<name of original queue>.ppd

    and print into this queue as described above. The PostScript
    output is in /tmp/printout after the job has completed.

    This option does not change anything if Poppler's pdftops is used
    as renderer.

HELPER DAEMON FOR BROWSING REMOTE CUPS PRINTERS AND IPP NETWORK PRINTERS

    From version 1.6.0 on in CUPS the CUPS broadcasting/browsing
    facility was dropped, in favour of Bonjour-based broadcasting of
    shared printers. This is done as Bonjour broadcasting of shared
    printers is a standard, established by the PWG (Printing Working
    Group, http://www.pwg.org/), and most other network services
    (shared file systems, shared media files/streams, remote desktop
    services, ...) are also broadcasted via Bonjour.

    Problem is that CUPS only broadcasts its shared printers but does
    not browse broadcasts of other CUPS servers to make the shared
    remote printers available locally without any configuration
    efforts. This is a regression compared to the old CUPS
    broadcasting/browsing. The intention of CUPS upstream is that the
    application's print dialogs browse the Bonjour broadcasts as an
    AirPrint-capable iPhone does, but it will take its time until all
    toolkit developers add the needed functionality, and programs
    using old toolkits or no toolkits at all, or the command line stay
    uncovered.

    The solution is cups-browsed, a helper daemon running in parallel
    to the CUPS daemon which listens to Bonjour broadcasts of shared
    CUPS printers on remote machines in the local network via Avahi,
    and can also listen for (and send) CUPS Browsing broadcasts. For
    each reported remote printer it creates a local raw queue pointing
    to the remote printer so that the printer appears in local print
    dialogs and is also available for printing via the command
    line. As with the former CUPS broadcasting/browsing with this
    queue the driver on the server is used and the local print dialogs
    give access to all options of the server-side printer driver.

    Note that CUPS broadcasting/browsing is available for legacy
    support, to let the local CUPS daemon work seamlessly together
    with remote CUPS daemons of version 1.5.x and older which only
    support CUPS broadcasting/browsing. In networks with only CUPS
    1.6.x servers (or Ubuntu or Fedora/Red Hat servers with CUPS
    1.5.x) please use the native Bonjour broadcasting of your servers
    and cups-browsed, configured for Bonjour browsing only on the
    clients.

    Also high availability with redundant print servers and load
    balancing is supported. If there is more than one server providing
    a shared print queue with the same name, cups-browsed forms a
    cluster locally with this name as queue name and printing through
    the "implicitclass" backend. Each job triggers cups-browsed to
    check which remote queue is suitable for the job, meaning that it
    is enabled, accepts jobs, and is not currently printing.  If none
    of the remote queues fulfills these criteria, we check again in 5
    seconds, until a printer gets free to accommodate the job. When we
    search for a free printer, we do not start at the first in the
    list, but always on the one after the last one used (as CUPS also
    does with classes), so that all printer get used, even if the
    frequency of jobs is low. This is also what CUPS formerly did with
    implicit classes. Optionally, jobs can be sent immediately into
    the remote queue with the lowest number of waiting jobs, so that
    no local queue of waiting jobs is built up.

    For maximum security cups-browsed uses IPPS (encrypted IPP)
    whenever possible.

    In addition, cups-browsed is also capable of discovering IPP
    network printers (native printers, not CUPS queues) with known
    page description languages (PWG Raster, Apple Raster, PDF,
    PostScript, PCL XL, PCL 5c/e) in the local network and auto-create
    print queues with auto-created PPD file or PPD-less for them (for
    the latter using a System V interface script to control the filter
    chain, only available for CUPS 2.1.0 and older, clients have to
    IPP-poll the capabilities of the printer and send option settings
    as standard IPP attributes then). This functionality is primarily
    for mobile devices running CUPS to not need a printer setup tool
    nor a collection of printer drivers and PPDs.

    cups-browsed can also be started on-demand, for example to save
    resources on mobile devices. For this, cups-browsed can be set
    into an auto shutdown mode so that it stops automatically when it
    has no remote printers to take care of any more, especially if an
    on-demand running avahi-daemon stops. Note that CUPS must stay
    running for cups-browsed removing its queues and so being able to
    shut down. Ideal is if CUPS stays running another 30 seconds after
    finishing its last job so that cups-browsed can take down the
    queue. For how to set up and control this mode via command line,
    configuration directives, or sending signals see the man pages
    cups-browsed(8) and cups-browsed.conf(5).

    The configuration file for cups-browsed is
    /etc/cups/cups-browsed.conf.  This file can include limited forms
    of the original CUPS BrowseRemoteProtocols, BrowseLocalProtocols,
    BrowsePoll, and BrowseAllow directives. It also can contain the
    new CreateIPPPrinterQueues to activate discovering of IPP network
    printers and creating PPD-less queues for them.

    Note that cups-browsed does not work with remote CUPS servers
    specified by a client.conf file. It always connects to the local
    CUPS daemon by setting the CUPS_SERVER environment variable and so
    overriding client.conf. If your local CUPS daemon uses a
    non-standard domain socket as only way of access, you need to
    specify it via the DomainSocket directive in
    /etc/cups/cups-browsed.conf.

    The "make install" process installs init scripts which make the
    daemon automatically started during boot. You can also manually
    start it with (as root):

    /usr/sbin/cups-browsed &

    or in debug mode with

    /usr/sbin/cups-browsed --debug

    Shut it down by sending signal 2 (SIGINT) or 15 (SIGTERM) to
    it. The queues which it has created get removed then (except a
    queue set as system default, to not loose its system default
    state).

    On systems using systemd use a
    /usr/lib/systemd/system/cups-browsed.service file like this:

        [Unit]
        Description=Make remote CUPS printers available locally
        After=cups.service avahi-daemon.service
        Wants=cups.service avahi-daemon.service

        [Service]
        ExecStart=/usr/sbin/cups-browsed

        [Install]
        WantedBy=multi-user.target

    On systems using Upstart use an /etc/init/cups-browsed.conf file like this:

        start on (filesystem
                  and (started cups or runlevel [2345]))
        stop on runlevel [016]

        respawn
        respawn limit 3 240

        pre-start script
            [ -x /usr/sbin/cups-browsed ]
        end script

        exec /usr/sbin/cups-browsed

    These files are included in the source distribution as
    utils/cups-browsed.service and utils/cups-browsed-upstart.conf.

    In the examples we start cups-browsed after starting
    avahi-daemon. This is not required. If cups-browsed starts first,
    then Bonjour/DNS-SD browsing kicks in as soon as avahi-daemon comes
    up. cups-browsed is also robust against any shutdown and restart
    of avahi-daemon.

    Here is some info on how cups-browsed works internally (first concept of a
    daemon which does only Bonjour browsing):

    - Daemon start
      o Wait for CUPS daemon if it is not running
      o Read out all CUPS queues created by this daemon (in former sessions)
      o Mark them unconfirmed and set timeout 10 sec from now
    - Main loop (use avahi_simple_poll_iterate() to do queue list maintenance
                 regularly)
      o Event: New printer shows up
        + Queue for printer is already created by this daemon -> Mark list
          entry confirmed, if discovered printer is ipps but existing queue ipp,
	  upgrade existing queue by setting URI to ipps. Set status to
	  to-be-created and timeout to now-1 sec to make the CUPS queue be
	  updated.
        + Queue does not yet exist -> Mark as to-be-created and set
	  timeout to now-1 sec.
      o Event: A printer disappears
        + If we have listed a queue for it, mark the entry as disappeared, set
          timeout to now-1 sec
      o On any of the above events and every 2 sec
        + Check through list of our listed queues
          - If queue is unconfirmed and timeout has passed, mark it as
            disappeared, set timeout to now-1 sec
          - If queue is marked disappered and timeout has passed, check whether
	    there are still jobs in it, if yes, set timeout to 10 sec from now,
	    if no, remove the CUPS queue and the queue entry in our list. If
	    removal fails, set timeout to 10 sec.
	  - If queue is to-be-created, create it, if succeeded set to
	    confirmed, if not, set timeout to 10 sec fron now. printer-is-shared
	    must be set to false.
    - Daemon shutdown
      o Remove all CUPS queues in our list, as long as they do not have jobs.

    Do not overwrite existing queues which are not created by us If
    the simple <remote_printer> name is already taken, try to create a
    <remote_printer>@<server> name, if this is also taken, ignore the
    remote printer. Do not retry, to avoid polling CUPS all the time.

    Do not remove queues which are not created by us. We do this by
    listing only our queues and remove only listed queues.

    Queue names: Use the name of the remote queue. If a queue with the
    same name from another server already exists, mark the new queue
    as duplicate and when a queue disappears, check whether it has
    duplicates and change the URI of the disappeared queue to the URI
    of the first duplicate, mark the queue as to-be-created with
    timeout now-1 sec (to update the URI of the CUPS queue) and mark
    the duplicate disappeared with timeout now-1 sec. In terms of
    high availability we replace the old load balancing of the
    implicit class by a failover solution. Alternatively (not
    implemented), if queue with same name but from other server
    appears, create new queue as <original name>@<server name without
    .local>. When queue with simple name is removed, replace the first
    of the others by one with simple name (mark old queue disappeared
    with timeout now-1 sec and create new queue with simple name).

    Fill description of the created CUPS queue with the Bonjour
    service name (= original description) and location with the server
    name without .local.

    stderr messages only in debug mode (command line options:
    "--debug" or "-d" or "-v").

    Queue identified as from this daemon by doing the equivalent of
    "lpadmin -p printer -o cups-browsed-default", this generates a
    "cups-browsed" attribute in printers.conf with value "true".

CUPS FILTERS FOR PDF AS STANDARD PRINT JOB FORMAT

    Here is documentation from the former CUPS add-on tarball with the filters
    for the PDF-based printing workflow: imagetopdf, texttopdf,
    pdftopdf, and pdftoraster

    The original filters are from http://sourceforge.jp/projects/opfc/

    NOTE: the texttops and imagetops filters shipping with this package
    are simple wrapper scripts for backward compatibility with third-party
    PPD files and custom configurations. There are not referred to in the
    cupsfilters.convs file and therefore not used by the default
    configuration. Direct conversion of text or images to PostScript is
    deprecated in the PDF-based printing workflow. So do not use these
    filters when creating new PPDs or custom configurations. The parameters
    for these filters are the same as for texttopdf and imagetopdf (see
    below) as the ...tops filter calls the ....topdf filter plus
    Ghostscript's pdf2ps.


IMAGETOPDF
==========

1. INTRODUCTION

This program is "imagetopdf". "imagetopdf" is a CUPS filter which reads
a single image file, converts it into a PDF file and outputs it to stdout.

This program accepts the following image file format;

  gif, png, jpeg, tiff, photocd, portable-anymap, portable-bitmap,
  portable-graymap, portable-pixmap, sgi-rgb, sun-raster, xbitmap,
  xpixmap, xwindowdump

xbitmap, xpixmap and xwindowdump images are converted into png images by
the "convert" command. Other kinds of image file format can be supported
if the "convert" command support them.

Output PDF file format conforms to PDF version 1.3 specification, and
input image is converted and contained in the output PDF file as a binary
format non-compression image.

"imagetopdf" may outputs multiple pages if the input image exceeds page
printable area.

2. LICENSE

"imagetopdf.c" is under the CUPS license. See the "COPYING" file.
For other files, see the copyright notice and license of each file.

3. COMMAND LINE

"imagetopdf" is a CUPS filter, and the command line arguments, environment
variables and configuration files are in accordance with the CUPS filter
interface.

imagetopdf <job> <user> <title> <num-copies> <options> [<filename>]

"imagetopdf" ignores <job> and <user>.
<title> is appended into the PDF dictionary as /Title.
<num-copies> specifies the number of document copies.
<options> is a CUPS option list.
<filename> is an input image file name.

When omit the <filename>, "imagetopdf" reads an image file from stdin.

4. ENVIRONMENT VARIABLES

This program refers the following environment variable;

   PPD:  PPD file name of the printer.

5. COMMAND OPTIONS

"imagetopdf" accepts the following CUPS standard options;

fitplot
mirror
PageSize
page-left, page-right, page-bottom, page-top
OutputOrder
Collate
sides
cupsEvenDuplex
position
scaling
ppi
natural-scaling
landscape
orientation-requested

See the CUPS documents for details of these options.

6. KNOWN PROBLEMS

Problem:
  PBM and SUN raster images can not be printed.
Solution:
  Due to the CUPS libcupsimage library's bug. Update the CUPS on your system.
  
7. INFORMATION FOR DEVELOPERS

Following information is for developers, not for driver users.

7.1 Options handled by a printer or "imagetopdf"

Following options are handled by a printer or "imagetopdf".
  Collate, Copies, Duplex, OutputOrder

Which handles these options depends on following options and attributes.
  Collate, Copies, Duplex, OutputOrder, cupsEvenDuplex, cupsManualCopies

"imagetopdf" judges whether a printer can handle these options according to
the followings option settings in a PPD file.

Collate:
  If Collate is defined, "imagetopdf" judges the printer supports Collate.
Copies:
  If cupsManualCopies is defined as True, "imagetopdf" judges the printer
  does not support Copies feature.
　
Duplex:
  If Duplex is defined, "imagetopdf" judges the printer supports Duplex.
  If cupsEvenDuplex is True, Number of pages must be even.
OutputOrder:
  If OutputOrder is defined, "imagetopdf" judges the printer supports
  OutputOrder.

If the printer cannot handle these options, "imagetopdf" handles it.

Following pseudo program describes how "imagetopdf" judges to handle
these options.

Variables

Copies : specified Copies
Duplex : specified Duplex 
Collate : specified Collate
OutputOrder : specified OutputOrder
EvenDuplex : specified cupsEvenDuplex
pages : number of pages
number_up : specified number-up

device_copies : Copies passed to the printer
device_duplex : Duplex passed to the printer
device_collate : Collate passed to the printer
device_outputorder : OutputOrder passed to the printer

soft_copies : copies by imagetopdf


device_copies = 1;
device_duplex = False;
device_collate = False;
device_outputorder = False;

if (Copies == 1) {
  /* Collate is not needed. */
  Collate = False;
}

if (!Duplex) {
  /* EvenDuplex is not needed */
  EvenDuplex = False;
}


if (Copies > 1 && the printer can handle Copies) device_copies = Copies;
if (Duplex && the printer can handle Duplex) {
       device_duplex = True;
} else {
   /* imagetopdf cannot handle Duplex */
}
if (Collate && the printer can handle Collate) device_collate = True;
if (OutputOrder == Reverse && the printer can handle OutputOrder)
             device_outputorder = True;

if (Collate && !device_collate) {
   /* The printer cannot handle Collate.
      So imagetopdf handle Copies */
              device_copies = 1;
}

if (device_copies != Copies /* imagetopdf handle Copies */ && Duplex)
    /* Make imagetopdf handle Collate, otherwise both paper side may have
       same page */
              Collate = True;
              device_collate = False;
}

if (Duplex && Collate && !device_collate) {
   /* Handle EvenDuplex, otherwise the last page has
      the next copy's first page in the other side of the paper. */
   EvenDuplex = True;
}

if (Duplex && OutputOrder == Reverse && !device_outputorder) {
   /* Handle EvenDuplex, otherwise the first page's other side of paper
      is empty. */
   EvenDuplex = True;
}

soft_copies = device_copies > 1 ? 1 : Copies;

7.2 JCL

When you print PDF files to a PostScript(PS) printer, you can specify
device options in PS. In this case, you can write PS commands in a PPD file
like as follows.

*OpenUI *Resolution/Resolution : PickOne
*DefaultResolution: 600
*Resolution 300/300 dpi: "<</HWResolution[300 300]>>setpagedevice"
*Resolution 600/600 dpi: "<</HWResolution[600 600]>>setpagedevice"
*CloseUI: *Resolution

However, if options cannot be described in PS file, you can write JCLs
as follows;

*JCLOpenUI *JCLFrameBufferSize/Frame Buffer Size: PickOne
*DefaultJCLFrameBufferSize: Letter
*OrderDependency: 20 JCLSetup *JCLFrameBufferSize
*JCLFrameBufferSize Off: '@PJL SET PAGEPROTECT = OFF<0A>'
*JCLFrameBufferSize Letter: '@PJL SET PAGEPROTECT = LTR<0A>'
*JCLFrameBufferSize Legal: '@PJL SET PAGEPROTECT = LGL<0A>'
*JCLCloseUI: *JCLFrameBufferSize

Because PDF cannot specify device options in a PDF file, you have to define
all the device options as JCLs.

When a printer does not support PS or PDF, you can use Ghostscript (GS).
In this case, you can specify device options like a PS printer.
If you want to use the same printer and same PPD file for both PDF and PS
printing, when you print a PS file, you can specify that GS handles it,
and when you print a PDF file, you can also specify that PDF filters handle
it in the same PPD file. However in this case, previous methods is not
appropriate to specify device options.

So, "imagetopdf" handles this case as follows;
(In following pseudo program, JCL option is an option specified with JCLOpenUI)

if (Both JCLBegin and JCLToPSInterpreter are specified in the PPD file) {
    output JCLs that marked JCL options.
}

if (pdftopdfJCLBegin attribute is specified in the PPD file) {
    output it's value
}

if (Copies option is specified in the PPD file) {
    mark Number of copies specified
} else if (pdftopdfJCLCopies is specified in the PPD file) {
    output JCL specified with JCLCopies
}

for (each marked options) {
    if (pdftopdfJCL<marked option's name> is specified in the PPD file) {
	output it's value as a JCL
    } else if (pdftopdfJCLBegin attributes is specified in the PPD file) {
	output "<option's name>=<marked choice>;" as a JCL
    }
}
output NEWLINE

Thus, if you want to use both PDF filters and GS by single PPD file,
what you should do is to add the following line in the PPD file;

*pdftopdfJCLBegin: "pdfto... jobInfo:"

Replace "pdfto..." by the name of the actual filter to be called after
pdftopdf.

Note:
  If you specify JCLBegin, you have to specify JCLToPSInterpreter as well.

Note:
  When you need to specify the value which is different from the choosen
  value based on the PPD into the jobInfo, you have to specify the values
  with the key started by "pdftopdfJCL" string.

  For example, if the page size is defined in a PPD file as following;

  *OpenUI *PageSize/Page Size: PickOne
  *DefaultPageSize: A4
  *PageSize A4/A4:
  *PageSize Letter/US Letter:
  *CloseUI: *PageSize

  if you choose the page size "Letter", the string "PageSize=Letter;" is
  added to jobInfo. On the other hand, if the driver requires the different
  value for the "Letter" size, for instance driver requires "PS=LT;"
  instead of "PageSize=Letter;" as the jobInfo value, the PPD file has to
  be defined as following;

  *OpenUI *PageSize/Page Size: PickOne
  *DefaultPageSize: A4
  *PageSize A4/A4:
  *pdftopdfJCLPageSize A4/A4: "PS=A4;"
  *PageSize Letter/US Letter:
  *pdftopdfJCLPageSize Letter/US Letter: "PS=LT;"
  *CloseUI: *PageSize

7.3 Temporally files location

"imagetopdf" creates temporally files if needed. Temporary files are created
in the location specified by TMPDIR environment variable. Default location
is "/tmp".


PDFTOPDF
========

The pdftopdf filter depends on libqpdf to read and write PDF files.

It replaces and imitates the pstops filter in the PDF-based workflow.
A similar filter (which can serve as behavior reference)
is called "cgpdftopdf" in OS X (not open source).

Command line
------------

pdftopdf follows the usual CUPS filter calling conventions, i.e.

  pdftopdf <job> <user> <title> <num-copies> <options> [<filename>]

together with the environment variables "PPD" and "CLASSIFICATION".

When omitting <filename>, "pdftopdf" reads a PDF file from stdin.
Internally this will write the data to a temporary file, because
the PDF format cannot be processed in a streaming fashion.

<options> are delimited by space; boolean type CUPS options can be set
by only adding the option key, other types are provided as
pairs of key and value, <key>=<value>.

pdftopdf processes the following standard command-line and/or PPD options:

Copies      # ppd will only override, when commandline parameter was 1
fitplot / fit-to-page / ipp-attribute-fidelity
landscape / orientation-requested
PageSize / page-size / MediaSize / media-size
page-left / page-right / page-bottom / page-top
media-top-margin / media-left-margin / media-right-margin / media-bottom-margin
Duplex / JCLDuplex / EFDuplex / JD03Duplex / sides
number-up / number-up-layout
page-border
OutputOrder / OutputBin / DefaultOutputOrder / page-delivery
page-label
page-set
page-ranges
MirrorPrint / mirror
emit-jcl
position
Collate / multiple-document-handling / sheet-collate
cupsEvenDuplex
cupsManualCopies  # via ppd

Additional (non-standard) options
---------------------------------

1) Booklet printing

  booklet=Off/On/Shuffle-Only

"On" also tries to set DuplexTumble (two-sided-short-edge) and forces number-up=2

  booklet-signature=(multiple of 4, or default: -1 to use "all pages")

2) Page autorotate

pdftopdf automatically rotates pages to the same orientation,
instead of (e.g. fitplot) scaling them down unrotated.
This behavier can be controlled by

  pdfAutorotate / nopdfAutorotate

Specifically, if a PDF file contains pages with page width greater than
page height (a landscape page), such pages are automatically rotated
anticlockwise by 90 degrees, unless the PPD file specifies
"*LandscapeOrientation: Minus90". In this case, clockwise rotation is used.
To turn off the feature on a job-by-job basis use

  lp -d <print_queue_name> -o nopdfAutorotate <document>

On a per-queue basis use

  -o nopdfAutorotate-default

as an option to lpadmin.

When the 'landscape' or 'orientation-requested=4' (or =5) option of CUPS is
given, the pdfAutorotate processing will adjust and accordingly rotate the
non-landscape pages are rotated instead.

Note: Some pages might end up 180 degree rotated (instead of 0 degree).
Those should probably be rotated manually before binding the pages together.

3) Method of flattening interactive PDF forms and annotations.

Some PDF files (like application forms) contain interactive forms
which the user can fill in inside a PDF viewer like evince. The filled
in data is not integrated in each page of the PDF file but stored in
an extra layer. Due to this the data gets lost when applying
manipulations like scaling or N-up to the pages. To prevent the loss
of the data pdftopdf flattens the form before doing the
manipulations. This means the PDF will be converted into a static PDF
file with the data being integral part of the pages.

The same flattening is needed for annotations in PDF files.

By default the actual flattening work is done by QPDF, as QPDF is also
doing everything else in pdftopdf. This way no external utilities need
to be called and so extra piping between processes and extra PDF
interpreter runs are avoided which makes the filtering process faster.

As we did not test the new QPDF-based form-flattening with thousands
of PDF files yet and it has not been available to actual users yet it
is possible that there are still some bugs. To give users a
possibility to work around possible bugs in QPDF's form flattening, we
have introduced an option to get back to the old flattening by the
external tools pdftocairo or Ghostscript.

The selection of the method is done by the "pdftopdf-form-flattening"
option, setting it to "auto", "qpdf", "pdftocairo", "ghostscript",
"gs", "internal" or "external":

Per-job:           lpr -o pdftopdf-form-flattening=pdftocairo ...
Per-queue default: lpadmin -p printer -o pdftopdf-form-flattening-default=gs
Remove default:    lpadmin -p printer -R pdftopdf-form-flattening-default

By default, pdftopdf uses QPDF if the option is not supplied, also the
settings "auto" and "internal" make QPDF being used. "external"
auto-selects from the two external utilities, trying pdftocairo at
first and on failure Ghostscript. If the selected utility fails, the
form stays unflattened and so the filled in data will possibly not get
printed.

Native PDF Printer / JCL Support
--------------------------------

Note that for most modern native PDF printers JCL is not needed any
more as they are controlled via IPP. For these the PPD files get
auto-generated by the support of CUPS and cups-filters for driverless
IPP printing.

pdftopdf will emit JCL when provided with a PPD file that includes the
"*JCLToPDFInterpreter:" keyword.

This enables for hardware copy generation and device collate; e.g. with PJL:

*JCLBegin:           "<1B>%-12345X@PJL JOB<0A>"
*JCLToPDFInterpreter: "@PJL ENTER LANGUAGE = PDF <0A>"
*JCLEnd:             "<1B>%-12345X@PJL EOJ <0A><1B>%-12345X"

For each marked option, the prefixed "pdftopdfJCL<option name>" keywords
can also be used to send raw JCL strings for that option.
These keywords also include *pdftopdfJCLBegin and *pdftopdfJCLCopies,
This allows the use of the same PPD for PDF- and PS-based workflows,
as pdftopdfJCL... will not be read in the PS case.

When the PPD contains the "Copies" keyword, pdftopdf will detect the use
of PJL and has special code which adds "@PJL SET COPIES=...",
or "@PJL SET QTY=...", respectively.

Other JCL code can be injected via "*JCLOpenUI: ..." ... "*JCLCloseUI: ...".

Special PDF comments
--------------------

pdftopdf adds comments to the pdf preamble that might esp. be of use
to subsequent filters, e.g.

  % This file was generated by pdftopdf
  %%PDFTOPDFNumCopies : 1
  %%PDFTOPDFCollate : false

The "NumCopies" and "Collate" values refer to the expected device/hardware
copies, i.e. when pdftopdf's soft-copy generation did not handle this options.

Limitations
-----------

pdftopdf does not support functions that are not related to printing
features, including interactive features and document interchange features.
Many of these operators and sections are just ignored.
Some of these may be output, but those functions are not assured.

Most notable is the use of AcroForms; their content will not be printed
if any non-trivial processing by pdftopdf is involved (e.g. "fitplot").
This only occurs when a file is printed directly, e.g. by "lpr".

Usual PDF viewer applications (xpdf, evince, acroread, ghostscript, ...)
will hardcopy the form content into printable pdf operations,
when choosing to print such a document.

Known issues
------------

- Borders, esp. in the "number-up=1 fitplot=false"-case might be drawn
  at incorrect locations.

- JCL documentation is sparse.
  The imagetopdf or old pdftopdf documentation contains a tad more information.

- Missing AcroForm-content might surprise users printing PDF files directly /
  from the command-line (see the Limitations section, above).

License
-------

pdftopdf is released under the MIT license.

The required libqpdf is available under version 2.0 of the Apache License,
e.g. here: https://github.com/qpdf/qpdf


TEXTTOPDF
=========

This implements a texttopdf filter, and is derived from cups' texttops.

To configure:
-------------

- texttopdf uses CUPS_DATADIR/charset/pdf.utf-8 for font configuration
  (when utf-8 was requested as charset). The font names given there are 
  used as fontconfig selectors; the best matching font, that is both 
  monospaced and in a supported format (TTC, TTF or OTF) will then be used.

- As a special exception, all fontnames that start with a '.' or '/' are
  considered filenames, and fontconfig is skipped; the name is used directly
  for loading the font file.

- Implementation note: TrueType Collections (.TTC) are internally handled
  by appending '/' and the index of the font inside the collection to 
  the filename (e.g. to use the second font of uming.ttc, the filename 
  uming.ttc/1 must be given to the fontembed-library).
  By appending the index-field returned from fontconfig, this is completely
  transparent to the user (but currently not widely tested).

- You may look at the two examples: pdf.utf-8.simple and pdf.utf-8.heavy.

To use:
-------

The filter is called just like any other cups filter. Have a
look at test.sh for example. 

Known Issues
------------

 - Text extraction does not work (at least for pdftotext from xpdf)
   for the resulting pdfs.

 - OTF(CFF) embedding currently does not subset the fonts.

 - Text wrapping in pretty-printing mode does not respect double-wide
   characters (CJK), and thus produce wrong results (wrap too late)
   for lines where they occur.  The fix is not trivial, since all the
   pretty-printing processing is done without knowledge of / prior to
   the font configuration (which is where single or double width
   code-ranges are specified).

 - The hebrew example in test5.pdf shows one of our limitations:
   Compose glyphs are not composed with the primary glyph but printed
   as separate glyphs.

Further Infos
-------------

Font embedding is handled by libfontembed in the filter/fontembed
subdirectory.

Please report all bugs to

https://github.com/OpenPrinting/cups-filters/issues

or to

https://bugs.linuxfoundation.org/

(product "OpenPrinting", component "cups-filters").


PDFTORASTER
===========

1. INTRODUCTION

"pdftoraster" is a filter for CUPS. It reads PDF files, convert it and
output CUPS raster.

"pdftoraster" does not support functions that are not related to printing
features, including interactive features and document interchange features.
Many of these operators and sections are just ignored.
Some of these may be output, but those functions are not assured.
Encryption feature is not supported.

2. LICENSE

Almost all source files are under MIT like license. However,
"pdftoraster" links some "poppler" libraries, and these files are
under GNU public license.  See copyright notice of each file for
details.

3. COMMAND LINE

"pdftoraster" is a CUPS filter, and the command line arguments, environment
variables and configuration files are in accordance with the CUPS filter
interface.

pdftoraster <job> <user> <title> <num-copies> <options> [<filename>]

"pdftoraster" ignores <job> and <user>.
<title> is appended into the PDF dictionary as /Title.
<num-copies> specifies the number of document copies.
<options> is a CUPS option list.
<filename> is an input PDF file name.

When omit the <filename>, "pdftoraster" reads a PDF file from the stdin,
and save it as a temporary file.

4. ENVIRONMENT VARIABLES

This program refers the following environment variable;
   PPD:  PPD file name of the printer.

5. COMMAND OPTIONS

See CUPS documents for details.

6. INFORMATION FOR DEVELOPERS

Following information is for developers, not for driver users.

6.1 Options handled by a printer or "pdftoraster"

"pdftopdf" outputs the following special comments from the 4th line in the
created PDF data.

%%PDFTOPDFNumCopies : <copies> --- <copies> specified Number of Copies
%%PDFTOPDFCollate : <collate> --- <collate> is true or false

"pdftoraster" overrides the command line options by above two option's values.
 
6.2 Temporally files location

"pdftoraster" creates temporally files if needed. Temporary files are created
in the location specified by TMPDIR environment variable. Default location
is "/tmp".


URFTOPDF
========

"urftopdf" is a filter to convert Apple's proprietary URF raster
format into PDF. URF raster is generated by some iOS applications when
printing via Airprint, so this filter provides a more complete support
for AirPrint clients. Note that it is not clear whether nowadays all
iOS applications send PDF and not URF any more. Also the filter does
not support all variants of URF format so the URF support is most
probably incomplete.

Apple does not provide any official documentation of the format but there is
already some reverse engineering done. A description of the format as far as it
got found out and two sample files can be found here:

https://github.com/AlanQuatermain/unirast

An actual implementation of an urftopdf filter is here:

https://github.com/superna9999/urftopdf

This original version uses libharu and to avoid an extra dependency
the filter coming with this package is converted to use libqpdf
instead (the same library as pdftopdf uses).

License: GNU General Public License version 3 or any newer version


TEXTTOTEXT
==========

This is a special filter for text-only printers (e. g. line printers,
daisy-wheel printers, POS printers, ...) or for using printers in
their text mode (e. g. dot-matrix printers or otherwise unsupported
printers). It takes plain text (UTF-8-encoded as this is standard with
CUPS) and not PDF as input.

The texttotext filter replaces the former textonly filter.

It is for the following use cases:

- Using text-only printers, like line printers or daisy-wheel
  printers. Note that only text can get printed in the way the printer
  is designed. No support for graphics printing tricks like ASCII art
  or printing pixels with the period character.

- Fast and less resource-consuming text printing with dot-matrix
  printers using the printer's text mode instead of converting the
  text to PDF and printing the PDF in the printer's graphics mode,
  which is slow, loud, and consumes much more ink.

- POS printing. POS printers often print only text on roll paper. This
  filter has a non-paginated mode which prints continuously, ignoring
  page height definitions.

The filter has the following features:

- Conversion of UTF-8 to most printer's encodings.

- To each page size a number of lines and columns is assigned, after that
  you only need to select the size of the paper in use.

- At end of page you can optionally send a Form Feed or let the filter fill
  up the rest of the page with blank lines.

- New lines can be initiated by Line Feed, Carriage Return, or both.

- Adjustable margins.

- Adjustable width for tab stops.

- Pagination can be turned off for roll paper or continuous printing in
  general.

- Wrapping or truncation of long lines

- Support for most of CUPS' page management options (only with
  pagination turned on): page-ranges, page-set, output-order, collate,
  multiple copies.

Setting up the printer

In the printer setup tool select the "Generic Text-Only Printer" (with
lpadmin use "-m drv:///cupsfilters.drv/textonly.ppd"), then under the
"Installable Options" adjust the following:

- Which page sizes to use and how many lines and columns the printer
  is capable to print on them. The default setting for lines and
  columns assume 6 lines per inch and 10 columns per inch.

- Whether to send a Form Feed character after each page. Sending a
  Form Feed is highly recommended to get the content of each page
  exactly onto the desired sheet. If the printer does not support Form
  Feed characters, turn them off and make sure that you have adjusted
  the correct number of lines for each page size, as the printer is
  advancing pages by filling up the rest of the paper with blank
  lines.

- How the printer advancs to a new line. Most printers require both
  Crriage Return and Line Feed (the DOS/Windows standard), but some
  would also work with either Carriage Return or Line Feed.

- The printer's encoding: Most text and dot-matrix printers (usually
  older devices) do not understand CUPS' standard encoding UTF-8 but
  instead, the use a simpler encoding (where each character is
  represented by one byte). ASCII should always work, but does not
  support letters with accents. So check the printer's manual what is
  supported. You cannot only use the encodings suggested by the PPD
  file, but any one-byte-per-character encoding which the "iconv"
  utility supports (see "iconv --list" for a list of encodings).

Also note that text-only and dot-matrix printers often have a DIP
switch block which allows for some hardware configuration, like
newline characters, length of page, input encoding, ...

Options of the texttotext filter:

To be usually used when sending a job:

PageSize: Paper format to be used. Make sure that the number of lines
and columns printable on each paper size are correctly adjusted with
the appropriate setup option. The page height is ignore when
pagination is turned off. Possible values: Letter, Legal, Tabloid,
Ledger, A4, A3, FanFoldGerman, FanFoldGermanLegal, 11x14Rotated,
LegalRotated, Custom1, Custom2, Custom3

OverLongLines: What to do with lines longer that the width of the
page: Truncate: Simply drop the extra characters; WrapAtWidth
(default): Continue the line in the next line on the paper; WordWrap:
As WrapAtWidth, but do not cut in the middle of a word.

TabWidth: Width of a tab stop. Can be any positive number.

Pagination: On: Text is divided in pages depending on the page size
selection, with each page having the user-selected margins,
recommended for sheet paper; Off: Text is printed continuously,
ignoring page breaks and the height and upper and lower margins of the
destination page size, recommended for roll paper, POS, long lists on
continuous paper, ... Note that with pagination turned off, multiple
copies, collate, page-ranges, page-set, and output-order are not
supported and therefore ignored.

page-left, page-right, page-top, page-bottom: Width of the margins
left blank, counted in lines or columns. Top and bottom margins are
ignored when pagination is turned off. Can be any positive number or
zero for no margin.

To be usually used when setting up the printer:

PrinterEncoding: The printer's character encoding (code page). Any
encoding which the iconv utility can generate (see "iconv --list") and
which uses only one byte per character can be used. This should
support practically any printer which is capable of printing
text. ASCII is the default setting. See the printer's manual for the
correct encoding to use.

NewlineCharacters: The characters sent on the end of a line, LineFeed
(LF), Crriage Return (CR), or both Carriage Return and Line Feed
(CRLF). Default is CRLF as most printers require this.

SendFF: On: Send a Form Feed after each page, so that printer changes
to the next sheet. Off: Do not send Form Feeds. To advance to the next
page blank lines are printed to fill up the page (requires the number
of limes for the selected page size correctly being set). When
pagination is off, Form Feeds are never sent.

LetterAvailable, LegalAvailable, TabloidAvailable, LedgerAvailable,
A4Available, A3Available, FanFoldGermanAvailable,
FanFoldGermanLegalAvailable, 11x14RotatedAvailable,
LegalRotatedAvailable, Custom1Available, Custom2Available,
Custom3Available: On: Paper of this size is available; Off: This paper
size is not available.

LetterNumLines, LegalNumLines, TabloidNumLines, LedgerNumLines,
A4NumLines, A3NumLines, FanFoldGermanNumLines,
FanFoldGermanLegalNumLines, 11x14RotatedNumLines,
LegalRotatedNumLines, Custom1NumLines, Custom2NumLines,
Custom3NumLines: Maximum number of text lines fitting on the paper
size. Default value is selected assuming 6 lines per inch. Can be any
positive number.

LetterNumColumns, LegalNumColumns, TabloidNumColumns,
LedgerNumColumns, A4NumColumns, A3NumColumns, FanFoldGermanNumColumns,
FanFoldGermanLegalNumColumns, 11x14RotatedNumColumns,
LegalRotatedNumColumns, Custom1NumColumns, Custom2NumColumns,
Custom3NumColumns: Maximum number of columns (characters) fitting on
the paper size. Default value is selected assuming 10 characters per
inch. Can be any positive number.

Standard CUPS options supported:

page-ranges, page set, output-order, collate

Note that these options and multiple copies are ignored when
pagination is turned off.


BEH - Backend Error Handler wrapper backend
===========================================

A wrapper for CUPS backends to make error handling more configurable

Usually, if a CUPS backend exits with an error status other than zero
(for example if a printer is not turned on or not reachable on the
network), CUPS disables the print queue and one can only print again
if a system administrator re-enables the queue manually. Even
restarting CUPS (or rebooting) does not re-enable disabled queues.

For system administrators this can get annoying, for newbie users who
are not aware of this problem it looks like that CUPS is severely
broken. They remove and re-install print queues, getting on the nerves
of distro install support, people, or even switch back to a
proprietary operating system.

Nowadays CUPS allows some configurability to avoid this, setting the
Error Policy to "retry-job", but this does not allow to retry for
infinitely many times and generally does not allow to change the
number of repetitions. It is also not possible to simply drop the job
without disabling the queue when CUPS gives up repeating the job.

This script makes the handling of such backend errors more
configurable, so that the problem can easily be worked around. The new
possibilities are:

 - Let queues simply not being disabled. Simple approach, but job gets
   lost.

 - Repeat a given number of times.

 - Repeat infinitely often, until the job gets finally through. This
   is the standard of LPRng, and it eliminates loss of the job.

 - The interval between two attempts to run the backend can also be
   configured.

 - Configuration is done independently for each print queue. So local
   printers and network printers can be treated differently.


Usage: 

Activate "beh" for your print queue(s) with command(s) like this:

lpadmin -p <queue name> -E -v beh:/<dd>/<att>/<delay>/<originaluri>

with 
  <queue name>:     The name of your print queue
  <dd>:             Don't Disable, if "1", beh always exits with zero
                    status, so the queue gets never disabled when the
                    original backend exits with an error. "0" carries
                    the error status of the last call of the backend
                    (after <att> retries) on to CUPS, so the queue
                    usually gets disabled.
  <att>:            Attempts, number of attempts to recall the backend
                    in case of an error. "0" means infinite retries. In
                    this case <dd> gets meaningless.
  <delay>:          Delay between two attempts to call the beckend, to
                    be given in seconds and as an integer number.
                    Meaningless if <att> is one.
  <originaluri>:    The original URI, which your queue had before. Can 
                    be determined with "lpstat -v".

All parameters, especially, <dd>, <att>, and <delay> have always to be
specified, even if one of them is meaningless due to the setting of
the others.

beh works with every backend except the "hp" backend of HPLIP, as the
"hp" backend repeats failed jobs by itself.

Example URIs:

beh:/1/3/5/socket://printer:9100

  On the network printer with host name "printer" it is tried to
  access 3 times with 5 second delays between the attempts. If the job
  still fails, the queue is not disabled (and the job discarded).

beh:/0/10/60/socket://printer:9100

  Retry 10 times in one minute intervals, disable the queue when still
  not succeeding.

beh:/1/0/60/usb://Brother/HL-5040%20series

  On a Brother HL-5040 on the USB try infinitely often until the
  printer comes back, in intervals of one minute. This way the job
  does not get lost when the printer is turned off and one can
  intendedly delay printing by simply switching off the printer. The
  ideal configuration for desktop printers and/or home users.

Originally this backend was written in Perl and part of the
foomatic-filters package. It was not overtaken into cups-filters
together with foomatic-rip to avoid the introduction of a dependency
on Perl. Now it has been re-written in C and so it can be part of
cups-filters without introducing new dependencies.


BRAILLE EMBOSSING
=================

cups-filters also provides filters and drivers for braille
embossers. It supports:

- Text on all kinds of embossers with generic support
- Text and graphics on the Index V3 embossers and above.

This is configured in CUPS just like any printer. Options can then be configured
in the standard printer panel, or passed as -o options to the lp command.


------------
Text support
------------

Text can be embossed either with no translation on the computer side (the
embosser will translate), or with translation on the computer side (thanks to
liblouis). It is a matter of running

lp file.txt

or even

lp file.html
lp file.odt
lp file.doc
lp file.rtf
lp file.docx
lp file.pdf

Important: it is really preferrable to directly print the document files
themselves, and not a pdf output, or printing from the application (which
would first convert to pdf). That way, the braille conversion will have the
proper document structure (paragraphs, titles, footnotes, etc.) to produce good
quality.


--------------------
Vector Image support
--------------------

Vector images can be embossed by converting them to braille dots.

This needs the inkscape package installed. Various input formats are then
supported: .svg, .fig, .wmf, .emf, .cgm, .cmx

The conversion assumes that the input is black-on-white. If it is
white-on-black, the -o Negate option can be used.

This image support is preferred over the generic image support described below,
which has to reconstruct lines to be embossed.


-------------
Image support
-------------

Images can be embossed by converting them to braille dots.

The orientation of the image can be controlled. By default it will be rotate to
fit the image orientation, i.e. it will be rotate by 90 degree if it is wider
than high and the paper is higher than wide, or if vice-versa. Other rotation
modes are provided.

By default, the image will be resized to fit the size of the paper. Disabling
the resize (fitplot set to No) will crop the image to the paper size. This is
useful for instance when a carefully-drawn image was designed especially for
embossing, and thus its pixels should exactly match with braille dots. In such
case, edge detection should very probably be disabled too.

The image can be processed for edge detection. When no processing is done (edge
detection is configured to "None"), the dark pixels are embossed if the Negate
option is off, or the light pixels are embossed if the Negate option is on. When
edge processing is done, only the edges of the images will be embossed. The
Basic and the Canny algorithms bring differing results. The Basic algorithm
can be tuned thanks to the edge factor only. The Canny algorithm can also be
tuned: increasing the Upper value will reduce the amount of detected edges (and
vice-versa), increasing the Lower value will reduce the lengths of the detected
edges (and vice-versa). The Radius and Sigma parameter control the blurring
performed before edge detection, to improve the result; the Radius parameter
controls how large blurring should be performed, setting it to zero requests
autodetection; the Sigma parameter determines how strongly blurring should be
performed.

A lot of images formats are support, so one can just run

lp file.png
lp file.gif
lp file.jpg
...

Here are complete examples for controlling the processing (all options can be
omitted, the default values are shown here):

Emboss the image without edge detection, as black on white or white on black:

lp -o "Edge=None" file.png
lp -o "Edge=None Negate" file.png

Emboss the image with edge detection, the default tuning parameters are set
here:

lp -o "Edge=Edge EdgeFactor=1" file.png
lp -o "Edge=Canny CannyRadius=0 CannySigma=1 CannyLower=10 CannyUpper=30" file.png

Emboss the image as it is, without any resize or edge detection, as black on
white or white on black:

lp -o "nofitplot Edge=None" file.png
lp -o "nofitplot Edge=None Negate" file.png

------------------------
Generic embosser support
------------------------

It should be possible to make all embossers use the generic driver. For this to
work, one has to:

- configure the embosser itself so that it uses an MIT/NABCC/BRF braille table
- add in CUPS a printer with the "Generic" manufacturer and "Braille embosser"
  model
- configure CUPS options according to the embosser settings, so that CUPS knows
  the page size, braille spacing, etc.

The generic driver can emboss text, as well as images, but images will probably
be distorted by the Braille interline spacing.


-----------------------
Index embossers support
-----------------------

Supported models:

Basic-S V3/4
Basic-D V3/4
Everest-D V3/4
4-Waves PRO V3
4X4 PRO V3
Braille Box V4

Index V3 embosser support has been well tested. It supports both text and
graphics mode.  Embossers with firmware 10.30 and above can be easily configured
from CUPS (paper dimension, braille spacing, etc.).

Index V4 embosser support has not been tested, but is very close to V3 support,
so it is probably working fine already.  Feedback would be very welcome.

To connect an Index embosser through Ethernet, gather its IP adress, select the
"AppSocket/HP JetDirect" network printer protocol, and set
socket://the.embosser.IP.address:9100 as Connection URL.

The density of dots for images can easily be chosen from the command line, for
instance:

lp -o "GraphicDotDistance=160" file.png

to select 1.6mm dots spacing

Troubleshooting: if your embosser starts every document with spurious
"TM0,BM0,IM0,OM0" or "TM0,BI0", your embosser is most probably still using an
old 10.20 firmware.  Please either reflash the embosser with a firmware version
10.30 or above, or select the 10.20 firmware version in the "index" panel of the
cups printer options.

----------------------
Braille output options
----------------------

The output can be finely tuned from the standard printing panel, or from the
command-line, the following example selects translation tables for French and
Greek, with 2.5mm dot spacing and 5mm line spacing. All options can be omitted,
the default values are shown here.

lp -o "LibLouis=fr-fr-g1 LibLouis2=gr-gr-g1 TextDotDistance=250 LineSpacing=500" file.txt


---------------------------------
Reworking output before embossing
---------------------------------

One may want to check and modify the .brf or .ubrl output before sending it to
the embosser.  This can be achieved by first generating the .brf file:

    /usr/sbin/cupsfilter -m application/vnd.cups-brf -p /etc/cups/ppd/yourprinter.ppd yourdocument.txt > ~/test.brf

One can choose a ppd file and additionally pass -o options to control the
generated output. One can then modify the .brf file with a text editor. One can
then emboss it:

    lp -o document-format=application/vnd.cups-brf ~/test.brf


The same can be achieved for images:

    /usr/sbin/cupsfilter -m image/vnd.cups-ubrl -p /etc/cups/ppd/yourprinter.ppd yourimage.png > ~/test.ubrl

    lp -o document-format=image/vnd.cups-ubrl ~/test.ubrl


---------------
BRF file output
---------------

One can generate BRF files by adding a virtual BRF printer.

When creating it in the cups interface, choose the CUPS-BRF local printer,
select the Generic maker, and choose the Generic Braille embosser model.

Printing to the resulting printer will generate a .brf file in a BRF
subdirectory of the home directory.


----------------
UBRL file output
----------------

One can generate Unicode braille files, not useful for embossing, but which can
be easily looked at by sighted people to check for the output.

In the cups interface, create a printer with the CUPS-BRF local printer,
the Generic maker, and choose the Generic UBRL generator model.

Printing to the resulting printer will generate a .brf file in a BRF
subdirectory of the home directory.


----------------------------
Remark about the source code
----------------------------

The file filter/braille/drivers/index/ubrlto4dot.c is used to generate
the translation table in
filter/braille/drivers/index/imageubrltoindexv[34]. It is included as
"source code" for these two files, even if actually running the
generation in the Makefile is more tedious than really useful.


----
TODO
----

- Test whether one wants to negate, e.g. to emboss as few dots as possible
- textubrltoindex when liblouis tools will be able to emit 8dot braille