Documentation

1 files changed, 292 insertions, 117 deletions

diff --git a/pdftool.pl b/pdftool.pl
index fb1785a..69c0484 100755
--- a/pdftool.pl
+++ b/pdftool.pl
@@ -1,98 +1,178 @@
-#! /usr/bin/perl
+#! /usr/bin/perl -w
 
-use Getopt::Long; 
+
+use Getopt::Long qw(:config posix_default no_ignore_case bundling); 
+use Pod::Usage;
 use IPC::Open2;
-use warnings;
+use POSIX qw(floor);
 use strict;
 
-Getopt::Long::Configure ("bundling");
 
-# Give an array if the command has least one argument.
-my @pdf2ps   = 'pdftops';
-my @pscrop   = ('psnup2.pl', '-s1', '-l1');
-my @psresize = 'psresize';
-my @psnup    = 'psnup';
-my @psbook   = 'psbook';
-my @ps2pdf   = 'ps2pdf';
+=head1 NAME
+
+pdftool.pl - a PDF swiss army knife
+
+=head1 SYNOPSIS
+
+B<pdftool.pl> [-s I<pages>] [-p I<paper>] [-m I<margin>] [-c] [-b]
+[-n I<nup>] [-q] [I<infile> [I<outfile>]]
+
+=head1 DESCRIPTION
+
+I<PDFTool>
+
+=head1 OPTIONS
+
+=over 8
+
+=item B<-s, --select>
+
+=item B<-p, --paper>
+
+=item B<-m, --margin>
+
+=item B<-c, --crop>
+
+=item B<-b, --book>
+
+=item B<-n, --nup>
+
+=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<--man>
+
+Display the manual page.
+
+=back
+
+=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: 0.1, 25 September 2010
+
+=cut
+
+
 
 my $tmpdir   = '/tmp';
 
+#
+# Options & arguments 
+#
 
-my $nup     = 1;
-my $margin  = 1;
-#TODO: units
+my $select;
+my $paper;
+my $margin;
 my $crop;
 my $book;
-my $papersize;
-
-GetOptions( "nup|n=i"       => \$nup,
-            "1"             => sub { $nup = 1 },
-            "2"             => sub { $nup = 2 },
-            "4"             => sub { $nup = 4 },
-            "8"             => sub { $nup = 8 },
-            "crop|c"        => \$crop,
-            "book|b"        => \$book,
-            "papersize|p=s" => \$papersize,
-            "margin|m=s"    => \$margin );
-
-die "I can handle only one file at the same time :P"
-    if $#ARGV > 0;
-
-die "How should I put a non-positive number of pages per sheet?"
-    if $nup <= 0;
-
-unless (defined $papersize) {
-    $papersize = `paperconf` or die "Can't guess papersize";
-    chomp $papersize;
+my $nup = 1;
+my $quiet;
+my $man;
+my $help;
+
+# TODO: select pages
+# TODO: choose the output type
+GetOptions( "select|s=s" => \$select,
+            "paper|p=s"  => \$paper,
+            "margin|m=s" => \$margin,
+            "crop|c"     => \$crop,
+            "book|b"     => \$book,
+            "nup|n=i"    => \$nup,
+            "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 },
+            "q|quiet"    => \$quiet,
+            "help"       => \$help,
+            "man"        => \$man )
+    or  pod2usage(2);
+
+pod2usage(1)  if  (defined $help  or $#ARGV > 1);
+pod2usage(-exitstatus => 0, -verbose => 2)  if  defined $man;
+
+# Input and output files
+my ($infile, $outfile) = @ARGV;
+
+
+
+#
+# Default values
+#
+
+# Default output papersize
+$paper = "a4"  unless  defined $paper;
+
+
+# Default margin
+unless (defined $margin) {
+    $margin = 0;
+    $margin = "1cm" if defined $crop;
 }
 
-for (my $n = $nup; $n > 1; $n /= 2) {
-    die "nup should be a power of two"
-        unless $n % 2 == 0;
-}
+# Default unit: PostScript point
+&topoints (\$margin);
 
-my $filename = $ARGV[0];
-$filename = "-" unless defined $filename;
-my $filename2 = $filename;
-$filename2 = "(stdin)" if $filename eq "-";
+# Inner and outer margins
+my ($mresize, $mnup) = (0,0);
+if ($nup > 1 && not defined $book) {
+    $mresize = $margin/2;
+    $mnup    = $mresize;
+} else {
+    $mresize = $margin;
+}
 
 
-my ($mresize, $mcrop, $mnup) = (0,0,0);
 
-if (defined $crop) {
-    $mresize = 0;
-    if (not defined $book && $nup > 1) {
-        $mcrop = $margin/2;
-        $mnup  = $mcrop;
-    } else {
-        $mcrop = $margin;
-        $mnup  = 0;
-    }
+# 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 {
-    $mcrop = 0;
-    if (not defined $book && $nup > 1) {
-        $mresize = $margin/2;
-        $mnup    = $mresize;
-    } else {
-        $mresize = $margin;
-        $mnup    = 0;
-    }
+    undef $infile;
+    *FIN = *STDIN;
+    $infile_display = "(stdin)";
 }
 
+if (defined $outfile && $outfile ne "-") {
+    open FOUT, '>', "$outfile"  or  die "Can't create `$outfile': $!";
+} else {
+    *FOUT = *STDOUT;
+}
 
 
 
-open(FILE, $filename)
-    or die "Can't read `$filename'";
-
+#
+# Detect filetype
+#
 my $filetype;
-my $fstline  = <FILE>;
-if (defined $fstline && $fstline =~ /%!PS.*/) {
+my $fstline  = <FIN>;
+if (defined $fstline && $fstline =~ /^%!PS.*/) {
     $filetype = "PS";
-} elsif (defined $fstline && $fstline =~ /%PDF.*/) {
+} elsif (defined $fstline && $fstline =~ /^%PDF.*/) {
     $filetype = "PDF";
 } else {
-    die "Can't recognize the type of `$filename2'";
+    die "Can't recognize the type of `$infile_display'";
 }
 
 # Auxiliary files, to remove
@@ -101,80 +181,127 @@ my @auxfiles;
 # Auxiliary file descriptors, to close
 my @auxfds;
 
-
 # Pids, to waid for
 my @pids;
 my $pid;
 
-if ($filename eq "-") {
-    # Need to copy the whole input to an auxiliary file, since
-    # conversion from pdf to ps requires random access to the data
-    
-    # TODO: only for pdfs
-    $filename = "$tmpdir/pdftool-stdin-" . int(rand 2**16) . lc ".$filetype";
-    open AUXFD, '>', "$filename"
-        or die "Can't read write into `$filename'";
-    push @auxfiles, "$filename";
-    
-    # cat > $filename
-    seek STDIN, 0, 0;
-    while (<FILE>) {
-        print AUXFD $_;
+
+
+#
+# Conversion from PDF to PS, if necessary 
+#
+my @cmd;
+if ($filetype eq "PDF") {
+    unless (defined $infile) {
+        # 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-" . 
+                      int(rand 2**16) . lc ".$filetype";
+
+        open FIN2, '>', "$infile"
+            or die "Can't write into `$infile': $!";
+        push @auxfiles, "$infile";
+
+        # cat > $filename
+        # TODO: useless seek
+        seek FIN, 0, 0  or die $!;
+        while (<FIN>) {
+            print FIN2 $_  or  die "Can't print: $!";
+        }
+        close FIN2;
     }
-    close AUXFD;
+    close FIN;
+
+    # Convert to PS
+    @cmd = ('pdftops', "$infile", '-');
+    push @cmd, '-q'  if  defined $quiet;
+    $pid = open *PSIN, "-|", @cmd
+               or die "Can't run: `" . &printcmd (@cmd) . "'";
+    push @pids, $pid;
+} else {
+    # TODO: useless seek
+    seek FIN, 0, 0  or die $!;
+    *PSIN = *FIN;
 }
-close FILE;
 
 
-# Convert to ps, or read the input file
-if ($filetype eq "PDF") {
-    $pid = open *INPS, "-|", @pdf2ps, "$filename", '-'
-               or die "Error with `@pdf2ps $filename -'";
+#
+# Select, if necessary
+#
+# TODO: preselection, during the conversion from pdf?
+if (defined $select) {
+    @cmd = ('psselect', "-p$select");
+    push @cmd, '-q'  if  defined $quiet;
+    $pid = open2 *PSSELECT, "<&PSIN", @cmd
+               or die "Can't run: `" . &printcmd (@cmd) . "'";
     push @pids, $pid;
 } else {
-    open *INPS, '<', "$filename"
-        or die "Can't open `$filename'";
+    *PSSELECT = *PSIN;
 }
-push @auxfds, fileno INPS;
 
 
-# Resize file to our papersize
-$pid = open2 *PSRESIZE, "<&INPS", @psresize, "-p$papersize"
-           or die "Error with `@psresize -p$papersize'";
-# Note: open2 closes the filehandles for us :)
+
+#
+# Resize file to our paper
+#
+@cmd = ('./psresize2.pl', "-p$paper", "-m$mresize");
+push @cmd, "-c" if defined $crop;
+push @cmd, '-q'  if  defined $quiet;
+
+$pid = open2 *PSRESIZE, "<&PSSELECT", @cmd
+           or die "Can't run: `" . &printcmd (@cmd) . "'";
 push @pids, $pid;
+# Note: open2 closes the filehandles for us :)
+
 
 
-# Pscrop
-if (defined $crop) {
-    $pid = open2 *PSCROP, "<&PSRESIZE", @pscrop, "-p$papersize", "-m${mnup}"
-               or die "Error with `@pscrop -p$papersize -m${mnup}cm'";
+#
+# PSBook
+#
+if (defined $book) {
+    @cmd = ('psbook');
+    push @cmd, '-q'  if  defined $quiet;
+    $pid = open2 *PSBOOK, "<&PSRESIZE", @cmd
+               or die "Can't run: `" . &printcmd (@cmd) . "'";
     push @pids, $pid;
 } else {
-    *PSCROP = *PSRESIZE;
+    *PSBOOK = *PSRESIZE;
 }
 
-# Psbook
-if (defined $book) {
-    $pid = open2 *PSBOOK, "<&PSCROP", "@psbook"
-               or die "Error with `@psbook";
+
+
+#
+# PSNup
+# 
+if ($nup > 1) {
+    @cmd = ('psnup', "-p$paper", "-m$mnup", "-$nup");
+    push @cmd, '-q'  if  defined $quiet;
+    $pid = open2 *PSOUT, "<&PSBOOK", @cmd
+               or die "Can't run: `" . &printcmd (@cmd) . "'";
     push @pids, $pid;
 } else {
-    *PSBOOK = *PSCROP;
+    *PSOUT = *PSBOOK;
 }
 
-# Psnup
-# TODO: sometimes unecessary
-$pid = open2 *PSNUP, "<&PSBOOK", @psnup, "-p$papersize", "-m${mnup}cm", "-$nup"
-           or die "Error with `@psnup -p$papersize -m${mnup}cm -$nup'";
-push @pids, $pid;
 
-# Convert back to pdf
-# TODO: not always stdout
-# TODO: return same type as input
-$pid = open2 ">&STDOUT", "<&PSNUP", @ps2pdf, "-sPAPERSIZE=$papersize", '-', '-'
-           or die "Error with `@ps2pdf -sPAPERSIZE=$papersize - -'";
-push @pids, $pid;
+
+#
+# Final file
+#
+if ($filetype eq "PDF") {
+    # Convert back to PDF
+
+    @cmd = ('ps2pdf', "-dEmbedAllFonts=true", "-sPAPERSIZE=$paper", '-', '-');
+    $pid = open2 ">&FOUT", "<&PSOUT", @cmd
+           or die "Can't run: `" . &printcmd (@cmd) . "'";
+    push @pids, $pid;
+} else {
+    # cat > FOUT
+    while (<PSOUT>) {
+        print FOUT $_  or  die "Can't print: $!";
+    }
+}
 
 
 # Avoid zombies
@@ -183,7 +310,55 @@ map {waitpid $_, 0} @pids;
 
 # Close auxiliary filehandles
 map {close $_} @auxfds;
+close FIN;
+close FOUT;
 
 
 # Delete auxiliary files
 unlink @auxfiles;
+
+
+
+
+# =========================================================
+
+#
+# In-place convert the given length to PostScript points
+#
+sub topoints {
+    my $l = $_;
+    return unless defined $$l;
+ 
+    $$l =~ /^([+-]?\d*\.?\d+)(\w*)$/  or  die "Unable to parse `$$l'";
+
+    my $r = $1;
+    if ($2 eq "" or $2 eq "pt") {
+        # nothing
+    } elsif ($2 eq "in") {
+        $r *= 72;
+    } elsif ($2 eq "cm") {
+        $r *= 72/2.54;
+    } elsif ($2 eq "mm") {
+        $r *= 72/25.4;
+    } else {
+        die "Unknown unit: `$2'";
+    }
+    $$l = floor ($r + .5);
+}
+
+
+#
+# Print a command just like you'd do in a shell
+#
+sub printcmd {
+    my @cmd;
+
+    for (@_) {
+        my $s = $_;
+        $s =~ s/"/\\"/;
+        $s = "\"$s\""  if  $s =~ /[ ()';#{}*?~&|`]/;
+        push @cmd, $s;
+    }
+
+    join ' ', @cmd; 
+}