remove-duplicate-files.php

<?php

$files = glob('*');

$sizes = array_map('filesize', $files);
$count_sizes = array_count_values($sizes);

$doubles = count($sizes) - count($count_sizes);
// var_dump($doubles);

$double_sizes = array_filter($count_sizes, function($num) {
	return $num > 1;
});
// print_r($double_sizes);
// var_dump(count($double_sizes));

$double_files = array();
foreach ( $sizes as $index => $size ) {
	if ( isset($double_sizes[$size]) ) {
		$double_files[$index] = $files[$index];
	}
}

// print_r($double_files);
// var_dump(count($double_files));
// exit;

function superhash($file) {
	return filesize($file) . ':' . sha1_file($file);
}

$file_hashes = array_map('superhash', $double_files);
$double_files = array();
foreach ( $file_hashes as $index => $hash ) {
	$double_files[$hash][] = $files[$index];
}

$double_files = array_filter($double_files, function($files) {
	return count($files) > 1;
});

$sorted_double_files = array_map(function($files) {
	usort($files, function($a, $b) {
		$a_rb = strpos($a, 'random-') !== FALSE;
		$b_rb = strpos($b, 'random-') !== FALSE;

		if ( $a_rb XOR $b_rb ) {
			return $a_rb ? -1 : 1;
		}

		return strlen($a) - strlen($b);
	});
	return $files;
}, $double_files);

print_r($sorted_double_files);
var_dump(count($sorted_double_files));

foreach ( $sorted_double_files as $hash => $files ) {
	print_r(array_map('unlink', array_slice($files, 1)));
}