#!/usr/bin/perl -w
#
# Organize a folder full of picture files into subfolders by name.
#
# To do:
#    load all _* dir names, and nab all files that contain a dir name.
#    on those dir names, try inverting around _, and dropping _, too.

use strict;

if ($ARGV[0] eq "-h") {
    warn "Moves files into sub-folders based on common parts of their names, ignoring\n";
    warn "    any final digits.\n";
    warn "Also normalizes extensions jpg/jpeg/JPG/JPEG, renames to lose odd characters.\n";
    exit;
}

print "Number of files: " . `ls | wc -l` . "\n";

# Normalize file extensions
#
my $ls = `ls`;
my @files = split('\n',$ls);

foreach my $x (@files) {
   my $newName = $x;
   $newName =~ s/\.JPG/.jpg/;
   $newName =~ s/\.JPEG/.jpg/;
   $newName =~ s/\.jpeg/.jpg/;
   if ($x ne $newName) {
       print "moving '$x' to '$newName'\n";
       rename($x,$newName);
  }
}


# Clean up the character set
#
$ls = `ls \*.jpg`;
@files = split('\n',$ls);
foreach $x (@files) {
   $newName = $x;
   # $newName =~ s/[- ~!@#\$%^&*()=+{}|;:<>,?\\\/\'\"]/_/g;   
   $newName =~ s/[^a-zA-Z0-9_.]/_/g;   
   $newName =~ s/__+/_/g;
   if ($x ne $newName) {
      print "normalizing filename '$x' to '$newName'\n";
       rename($x,$newName);
   }
}
 

# Strip extensions and serial numbers to find common base names.
# But if there's A1.jpg and Amarillo1.jpg, don't put all under A.
#
my $prefixes = 
    `ls [a-zA-Z]*[0-9]*.jpg [0-9]*[a-zA-Z_]*[0-9]*.jpg | sed 's/index//' | sed 's/[0-9]*\.jpg//' | sort | uniq`;
my @plist = split('\n', $prefixes);
foreach $x (@plist) {
    next unless ($x ne "");
    my @group = glob("$x*");
    my $nfiles = scalar(@group);
    if ($nfiles > 1) {
        # print "Globbed $x and got $nfiles items.\n";
        my $i = 0;
        while ($i < scalar(@group)) {
	    if (@group[$i] !~ m/$x[0-9]+\.jpg/ ||
		-d @group[$i]) {
     	        delete @group[$i];
	    }
	    $i++;
        }
        $nfiles = scalar(@group);
        if ($nfiles <= 1) {
		   #print "    but none have just numbers after '$x'\n";
       } else {
	   $totaldirs++;
		   print "Moving $nfiles files for '$x*' to '_$x/'\n";
		   if (-e "_$x" == 0) {
			   mkdir("_$x");
		   }
		   foreach $f (@group) {
			   rename($f, "_$x/$f");
			   $total++;
		   }
       }
   }
}

print "Moving all-numeric files to _numeric/...\n";

system "mkdir -p _numeric";
my @ns = `ls [0-9]*[0-9].jpg`;
$totaldirs++;
for $f (@ns) {
    chomp $f;
    if ($f =~ m/^\d+\.jpg$/) {
	rename($f, "_numeric/$f");
	$total++;
    }
}
   
print "Done, $total files grouped into $totaldirs directories.\n";
my $left = `ls -F | grep -v '/' | wc -l` - 0;
print "    ($left files left ungrouped)\n";

exit;