path: root/pdftool.pl

#! /usr/bin/perl -w


use Getopt::Long qw(:config posix_default no_ignore_case gnu_compat
                            bundling auto_version auto_help); 
use Pod::Usage;
use IPC::Open2;
use IPC::Open3;
use POSIX qw(floor);
use strict;


=head1 NAME

pdftool.pl - a PDF swiss army knife

=head1 SYNOPSIS

B<pdftool.pl> [-s I<pages>] [-w I<width> ] [-h I<heigth>] [-p I<paper>]
[-W I<width>] [-H I<heigth>] [-P I<paper>] [-p I<paper>] [-m I<margin>]
[-b I<border>] [-c] [--book] [--column] [-n I<nup>] [-t] [-q]
[I<infile> [I<outfile>]]

=head1 DESCRIPTION

I<PDFTool> combines the tools in the PSUtils bundle in a nice way. The
input should be either a Portable Document Format (PDF) file, or a
PostScript file. The output format is the same as the input one.

If no input file is given, or if a single hyphen-minus (B<->) is given as
file name, I<PDFTool> will read the PDF or PostScript data from the standard
input. In that case, and if the input data is in PDF format, an
auxiliary file will be created (since the conversion from PDF to PS
requires random access to the data), and removed afterwards. Also, if the crop
option (B<-c>) is set, an auxiliary file will be created, and removed
afterwards.

If no output file is given, or if a single hyphen-minus (B<->) is given as
file name, I<PDFTool> will send the data (of the same type as the input) to
the standard output.

The document will be treated as follows:

=over 8

=item Conversion from PDF to PostScript (if necessary),

=item Selection of the page range,

=item Rearranging pages for printing books or booklets,

=item Putting multiple pages per sheets,

=item Conversion from PS to PDF (if necessary).

=back

=head1 OPTIONS

=over 8

=item B<-s, --select>

Specifies the pages which are to be selected.
I<Pages> is a comma separated list of page ranges, each of which
may be a page number, or a page range of the form
I<first>-I<last>. If I<first> is omitted, the first page is assumed,
and if I<last> is omitted,  the last page is assumed.

The prefix character `_' indicates that the page number is relative
to the end of the document, counting backwards. If just this
character with no page number is used, a blank page will be inserted.

=item B<-w, --width>

Specify the width of the output file. If the height is not specified as
well, it will be ignored. The known units are B<pt>, B<in>, B<cm> and
B<mm>. The default unit is B<pt>.

=item B<-h, --height>

Specify the height of the output file. If the width is not specified as
well, it will be ignored. The known units are B<pt>, B<in>, B<cm> and
B<mm>. The default unit is B<pt>.

=item B<-p, --paper>

Specify the paper size of the output file, as an alternative to B<-w>
and B<-h>. Can be set to B<a0>, B<a1>, B<a2>, B<a3>, B<a4>, B<a5>, B<b5>,
B<letter>, B<legal>, B<ledger>, B<tabloid>, B<statement>, B<executive>,
B<folio>, B<quarto>, or B<10x14>. The  default output paper size is B<a4>.

=item B<-W, --Width>

Same as the option B<-w>, but for the input file. This option is useless if
the crop option (B<-c>) is set.


=item B<-H, --Height>

Same as the option B<-h>, but for the input file. This option is useless if
the crop option (B<-c>) is set.


=item B<-P, --Paper>

Same as the option B<-p>, but for the input file. By default,
I<PDFTool> will try to guess this value from the header of the file,
and fail if the information is missing. This option is useless if the
crop option (B<-c>) is set.

=item B<-b, --border>

Add a margin around each logical page on a sheet. Possible units are B<pt>,
B<in>, B<cm> and B<mm>. The default unit is B<pt>. The default border is
B<1cm> if the crop option (B<-c>) is set, and B<0> otherwise.

=item B<-m, --margin>

Add a margin around the whole page. Possible units are B<pt>, B<in>, B<cm>
and B<mm>. The default unit is B<pt>. The default margin is B<0>.

=item B<-c, --crop>

If this option is set, the PostScript code will interpreted to calculate the
maximal effective bounding box. This operation may take time and be quite
demanding for the CPU.

=item B<--book>

Rearrange pages for printing books or booklets.

=item B<-n, --nup>

Puts multiple logical pages onto each physical sheet of paper. The inner
margin might be same as the outer one (depending on the booklet option
B<-b>), see B<-m> for details.

If I<nup> is less than 10, the option B<->I<nup> may be used as an
alternative.

=item B<-t, --tumble>

By default, B<PDFTool> ensures that your pdf will be printable using the
DuplexNoTumble mode, hence even pages of landscape files are rotated by
180°. This option removes this behavior; also, landscape files are rotated
in order to make it easy to read on the computer.

=item B<--column>

Changes the order to `column-major', where successive pages are placed in
columns down the paper.

=item B<-q, --quiet>

I<PDFTool> normally prints the page numbers of the pages output; this
option suppresses this.

=item B<--help>

Display a brief help.

=item B<--version>

Display the version number of the I<PDFTool> program.

=item B<--man>

Display the manual page.

=back

=head1 EXAMPLES

The following comand can be used to remotely crop a PDF file, convert it
to A4 paper, and rearrange the pages to print a booklet: 

ssh remote pdftool.pl -cpA4 --book -2 -b2cm -m-1cm < in.pdf > out.pdf

=head1 REQUIRE

Requires PSUtils installed and available in the command line
http://www.tardis.ed.ac.uk/~ajcd/psutils/.

=head1 AUTHOR

Public domain, (c) Guilhem Moulin.

=head1 VERSION

Version: 0.3, 27 December 2010

=cut

# TODO: inline it in the header
$main::VERSION = "0.3, 27 December 2010";



my $tmpdir   = '/tmp';

#
# Options & arguments 
#

my $select;
my ($outwidth,$outheight, $inwidth,$inheight);
my ($margin, $border);
my $crop;
my $book;
my $nup = 1;
my $column;
my $quiet;
my $man;
my $tumble;
my $landscape;

GetOptions( "select|s=s" => \$select,
            "w|width=s"  => \$outwidth,
            "h|height=s" => \$outheight,
            "p|paper=s"  => sub { &papersize ($_[1],\$outwidth,\$outheight) },
            "W|Width=s"  => \$inwidth,
            "H|Height=s" => \$inheight,
            "P|Paper=s"  => sub { &papersize ($_[1],\$inwidth,\$inheight) },
            "margin|m=s" => \$margin,
            "border|b=s" => \$border,
            "crop|c"     => \$crop,
            "book"       => \$book,
            "nup|n=i"    => \$nup,
            "t|tumble"   => \$tumble,
            "1"          => sub { $nup = 1 },
            "2"          => sub { $nup = 2 },
            "3"          => sub { $nup = 3 },
            "4"          => sub { $nup = 4 },
            "5"          => sub { $nup = 5 },
            "6"          => sub { $nup = 6 },
            "7"          => sub { $nup = 7 },
            "8"          => sub { $nup = 8 },
            "9"          => sub { $nup = 9 },
            "column"     => \$column,
            "q|quiet"    => \$quiet,
            "man"        => \$man )
    or  pod2usage(2);

pod2usage(2)  if  ($#ARGV > 1);
pod2usage(-exitstatus => 0, -verbose => 2)  if  defined $man;

# Input and output files
my ($infile, $outfile) = @ARGV;



#
# Default values
#

# Default margin & border
$margin = 0 unless defined $margin;
unless (defined $border) {
    if (defined $crop) {
        $border = '1cm';
    } else {
        $border = 0;
    }
}

# Default output papersize
&papersize ("a4", \$outwidth, \$outheight)
    unless  (defined $outwidth  and  defined $outheight);

# Default unit: PostScript point
map {&topoints ($_)} ( \$outwidth, \$outheight,
                       \$inwidth, \$inheight,
                       \$margin, \$border );
die "Margins are too big" if $outwidth <= $margin*2 or $outheight <= $margin*2;


# Open input and output files
my $infile_display;
if (defined $infile && $infile ne "-") {
    open FIN, '<', "$infile"  or  die "Can't read `$infile': $!";
    $infile_display = $infile;
} else {
    undef $infile;
    *FIN = *STDIN;
    $infile_display = "(stdin)";
}
# After the pipe, it won't be detected as seekable
my $inseek = (seek FIN, 0, 1) ? 1 : undef; 


if (defined $outfile && $outfile ne "-") {
    open FOUT, '>', "$outfile"  or  die "Can't create `$outfile': $!";
} else {
    *FOUT = *STDOUT;
}

*LOG = *STDERR;



#
# Detect filetype
#
# To avoid to seek into FIN, it gonna be copied from WRITE to READ in
# background, once the filetype has been read

# TODO: read specification to detect filetype properly
my $filetype;
pipe *READ, *WRITE  or  die "Can't pipe: $!";
while (not (defined $filetype) && defined (my $l = <FIN>)) {
    print WRITE $l  or  die "Can't close: $!";

    if (defined $l && $l =~ /^%!PS/) {
        $filetype = "PS";
    } elsif (defined $l && $l =~ /^%PDF/) {
        $filetype = "PDF";
    } 
}

die "Cannot recognise FileType"  unless  defined $filetype;

unless (my $pid = fork) {
    # Child: cat FIN > WRITE in background
    die "Can't fork: $!" unless defined $pid;
    close READ  or  die "Can't close: $!";
        
    while (<FIN>) {
        print WRITE  or  die "Can't print: $!";
    }
    exit;
}
# Parent

close WRITE  or  die "Can't close: $!";
close FIN  or  die "Can't close: $!";
        
open *FIN, "<&READ"  or  die "Can't dup: $!";



# Auxiliary files, to remove
my @auxfiles;

# Pids, to waid for
my @pids;



#
# Conversion from PDF to PS, if necessary 
#
my @cmd;
if ($filetype eq "PDF") {
    unless (defined $infile && $inseek) {
        # Need to copy the whole input to an auxiliary file, since
        # conversion from PDF to PS requires random access to the data
        
        $infile = "$tmpdir/pdftool-stdin-$$" . lc ".$filetype";

        open FINAUX, '>', "$infile"
            or die "Can't write into `$infile': $!";
        push @auxfiles, "$infile";

        # cat > $infile
        while (<FIN>) {
            print FINAUX  or  die "Can't print: $!";
        }
        close FINAUX;
    }

    my ($first, $last);
    if (defined $select) {
        # Convert to PS only the pages we are interested in
        ($first, $last) = (1<<16,-(1<<16));
        for (split / *, */, $select) {
            my ($rmin, $rmax) = split (/ *- */, $_);
            undef $first if defined $rmin && not $rmin;
            undef $last if defined $rmax && not $rmax;
            if (defined $rmin && $rmin) {
                $first = $rmin if defined $first && $rmin < $first; 
                $last = $rmin if defined $last && $rmin > $last; 
            }
            if (defined $rmax && $rmax) {
                $first = $rmax if defined $first && $rmax < $first; 
                $last = $rmax if defined $last && $rmax > $last; 
            }
        }

        # Calculate the new page range
        my @newselect;
        for (split / *, */, $select) {
            my ($rmin, $rmax) = split / *- */, $_;
            $rmin -= $first-1 if defined $first && defined $rmin && $rmin;
            $rmax -= $first-1 if defined $first && defined $rmax && $rmax;
            my $r = "";
            $r .= $rmin if defined $rmin;
            $r .= "-";
            $r .= $rmax if