cmppdf

#!/usr/bin/perl

# compare pdfs to find ones that are likely the same even if not identical
# -samy kamkar
#
# todo - bucketize by number of pages
#

use strict;
use y;;;

my %o = optusage('[-t(hreshold) <% (default 99)>]', '<dir ...>');
my @DIRS = @{$o{args}};
my @files = grep { file() =~ /^PDF/ } ls(@DIRS, 1); # recursively get files
my $DIFFPDF = "diff-pdf";
my %d;

# s4args -t 99 /Users/samy/Dropbox/Mac/Documents/Home/542nvistast90036/ladbs/

# go through pdf files
for my $f1 (@files)
{
  for my $f2 (@files)
  {
    #print "$f1\n$f2\n\n";
    next if $d{px}{$f1}{$f2} || $d{px}{$f2}{$f1} || $f1 eq $f2;

    diffpdf($f1, $f2);
  }
}

my %px;
sub diffpdf
{
  my ($f1, $f2) = @_;
  my $out = runenv($DIFFPDF, "-v", $f1, $f2);
  #print "$f1 -> $f2:\n$out\n\n";
  $d{px}{$f1}{$f2} += $1 while $out =~ /(\d+) pixels? that differ/gs;
  #$d{pg}{$f1} = $1 if $out =~ /of (\d+) pages?/;
  if ($out =~ /pages count differs: (\d+) vs (\d+)/)
  {
    @{$d{pg}}{$f1, $f2} = ($1, $2) if $out =~ /pages count differs: (\d+) vs (\d+)/;
  }
  else
  {
    @{$d{pg}}{$f1, $f2} = ($1, $1) if $out =~ /of (\d+) pages?/;
  }

  print filename($f1) . " -> " . filename($f2) . "\n";
  print "$_\n" for grep { /pixels that differ/ && !/ 0 pixels/ } split("\n", $out);
  print "\n";
}
__DATA__