scripts/pdf_to_bitmap.pl

   1 #!/usr/bin/perl
   2
   3 use strict;
   4
   5 # usage:
   6 #   pdf_to_bitmap.pl foo.pdf
   7 #   pdf_to_bitmap.pl foo.pdf jpg
   8 #   pdf_to_bitmap.pl foo.pdf png
   9 #   pdf_to_bitmap.pl foo.pdf png 1
  10 # default format is png
  11 # quality for jpg is set to imagemagick's default of 92
  12 # If the file foo.jpg or foo.png exists and is newer than foo.pdf, then nothing is done, and this is not an error.
  13 #   [currently I have the date checking commented out...why?]
  14 # If the file foo.jpg or foo.png exists and foo.pdf doesn't exist, then nothing is done, and this is not an error.
  15 # Adding the 1 as the third command-line arg forces rendering even if it seems up to date.
  16
  17 my $pdf = $ARGV[0];
  18 my $dest_fmt = $ARGV[1];
  19 if ($dest_fmt=='') {$dest_fmt='png'}
  20 my $force = ($ARGV[2] eq '1');
  21
  22 my @temp_files = ();
  23
  24 unless ($pdf =~ /\.pdf$/) {finit("Error in pdf_to_bitmap.pl, input file doesn't have extension .pdf")}
  25
  26 my $rendered = $pdf;
  27 $rendered =~ s/\.pdf$/.$dest_fmt/;
  28
  29
  30 if ((!$force) && -e $rendered && (! (-e $pdf))) {exit(0)}
  31 ###if (-e $rendered && -M $pdf > -M $rendered) {exit(0)} # -M finds age in days
  32
  33 print "rendering $pdf to $rendered\n";
  34
  35 # Don't use inkscape --export-png, because as of april 2013, it messes up on transparency.
  36 # Can convert pdf directly to bitmap of the desired resolution using imagemagick, but it messes up on some files (e.g., huygens-1.pdf), so
  37 # go through pdftoppm first.
  38 my $ppm = 'z-1.ppm'; # only 1 page in pdf
  39 push @temp_files,$ppm;
  40 if (system("pdftoppm -r 300 $pdf z")!=0) {finit("Error in pdf_to_bitmap.pl, pdftoppm")}
  41
  42 # Work around misfeature in ImageMagick's convert. See notes in computer/apps. It's possible that ca. 2015
  43 # ImageMagick's color handling will change, and I will need to get rid of this code.
  44 # Detect whether or not it's color:
  45 my $stats = `identify -colorspace HSL -verbose $ppm`;
  46 $stats =~ s/\n//g; # strip newlines
  47 my $is_color = !($stats=~quotemeta("Green:      min: 0 (0)      max: 0 (0)"));
  48   # when ImageMagick's identify utility is given the -colorspace HSL, saturation is in the green channel
  49 my $gamma_correction = 1.0;
  50 if (!$is_color) {$gamma_correction=0.44} # grayscale images are much too dark otherwise
  51
  52 if (system("convert $ppm -density 300 -evaluate Pow $gamma_correction -units PixelsPerInch $rendered")!=0) {finit("Error in pdf_to_bitmap.pl, ImageMagick's convert")}
  53 # ... uses default quality of 92
  54
  55 # It seems that some tool on my toolchain has had inconsistencies/regressions in its behavior with respect to
  56 # how it handles gamma. I think this was probably inkscape, although imagemagick has also had some flaky changes
  57 # in how it handles color management.
  58 # Currently I have this:
  59 #   $ inkscape --version
  60 #   Inkscape 0.48.4 r9939 (Jan 22 2014)
  61 #   $ convert --version
  62 #   Version: ImageMagick 6.7.7-10 2014-03-06 Q16 http://www.imagemagick.org
  63 #   $ pdftoppm -v
  64 #   pdftoppm version 0.24.5
  65 # With these versions, it seems OK if I set gamma_correction to 1.0. With some other versions, previously,
  66 # I needed 0.44.
  67
  68 print "\n";
  69 finit();
  70
  71 sub finit {
  72   my $message = shift;
  73   tidy();
  74   if ($message eq '') {
  75     exit(0);
  76   }
  77   else {
  78     die $message;
  79   }
  80 }
  81
  82 sub tidy {
  83   foreach my $f(@temp_files) {
  84     unlink($f) or die "error deleting $f, $!";
  85   }
  86 }