deduplicate.php

<?php

if (!isset($argv[1])) {
    die('Usage: deduplicate.php "Dropbox/Camera Uploads/2014"' . PHP_EOL);
}
$path = $argv[1];
if (!is_dir($path)) {
    die("$path is not a dir!" . PHP_EOL);
}

$files = scandir($path);
$hashes = array();
$fileCount = count($files);
foreach ($files as $i => $file) {
    if (substr($file, 0, 1) === '.') {
        continue;
    }
    if ($i % 100 === 0) {
        echo "MD5 $i / $fileCount" . PHP_EOL;
    }

    $filePath = $path . DIRECTORY_SEPARATOR . $file;

    $md5 = md5_file($filePath);

    if (isset($hashes[$md5])) {
        $file1 = $path . DIRECTORY_SEPARATOR . $hashes[$md5];
        $file2 = $filePath;
        if (strstr($file1, '_')) {
            echo "mv $file1 /Users/boy/.Trash/, duplicate of $file2" . PHP_EOL;
            //rename($file1, '/Users/boy/.Trash/' . $hashes[$md5]);
        }
        else if (strstr($file2, '_')) {
            echo "mv $file2 /Users/boy/.Trash/, duplicate of $file1" . PHP_EOL;
            //rename($file2, '/Users/boy/.Trash/' . $file);
        }
        continue;
    }

    $hashes[$md5] = $file;

}

echo "DONE HASHING" . PHP_EOL;



//foreach ($files as $file) {
//    $p = [];
//    if (preg_match('/(?P<Y>\d{4})(?P<m>\d{2})(?P<d>\d{2})_(?P<H>\d{2})(?P<i>\d{2})(?P<s>\d{2})\.(?P<ext>\w+)/', $file, $p) < 1) {
//        continue;
//    }
//
//    $wrongFilePattern = "/IMG-{$p['Y']}{$p['m']}{$p['d']}-WA\\d+.{$p['ext']}/";
//
//    $wrongFile = false;
//    foreach ($files as $file1) {
//        if (preg_match($wrongFilePattern, $file1)) {
//            $wrongFile = $file1;
//        }
//    }
//    if (!$wrongFile) {
//        continue;
//    }
//
//    $rightFilePath = $path . DIRECTORY_SEPARATOR . $file;
//    $wrongFilePath = $path . DIRECTORY_SEPARATOR . $wrongFile;
//
//    if (sha1_file($rightFilePath) !== sha1_file($wrongFilePath)) {
//        continue;
//    }
//
//    echo "mv $wrongFilePath /Users/boy/.Trash/" . PHP_EOL;
//    //rename($wrongFilePath, '/Users/boy/.Trash/' . $wrongFile);
//}