index.php

<?php

/**
 * FIXME: This is first working version of the script.
 * I do not like it - not optimal and broke DRY, needs to be refactored
 * 
 * Usage example:
 * 
 * $strFolderName = $argv[1] ? $argv[1] : '.';
 * var_dump(mainFunc($strFolderName, FORMAT_ARRAY));
 */


define('FORMAT_ARRAY', 0);
define('FORMAT_JSON', 1);

/**
 * @param string $argStrFolderName
 * @param int $argIntFormat
 * @return array|string
 * @throws Exception
 */
function mainFunc($argStrFolderName = '', $argIntFormat = FORMAT_ARRAY)
{
	$arrOut = array();
	$strFolderName = $argStrFolderName ? $argStrFolderName : '.';

	if (!is_dir($strFolderName))
		throw (new Exception("Path not found"));

	foreach (scandir($strFolderName) as $strEntity)
	{
		$strFullPath = $strFolderName . DIRECTORY_SEPARATOR . $strEntity;
		if (isNotDotFolder($strEntity) && is_dir($strFullPath))
		{
			$arrOut[] = [
				'name' => $strEntity,
				'size' => folderSize($strFullPath),
				'files' => filesCount($strFullPath),
				'dupes' => dupesCount($strFullPath)];
		}
	}

	usort($arrOut, "sizeCmp");
	return ($argIntFormat == FORMAT_JSON) ? json_encode($arrOut) : $arrOut;
}

/**
 * Recursive function to get folder size
 *
 * @param string $argStrPath
 * @return int
 */
function folderSize($argStrPath)
{
	$intTotalSize = 0;
	$arrFiles = scandir($argStrPath);
	$strCleanPath = rtrim($argStrPath, DIRECTORY_SEPARATOR) . DIRECTORY_SEPARATOR;

	foreach ($arrFiles as $strEntity)
	{
		if (isNotDotFolder($strEntity))
		{
			$strCurrentFile = $strCleanPath . $strEntity;
			$intTotalSize += is_dir($strCurrentFile) ? folderSize($strCurrentFile) : filesize($strCurrentFile);
		}
	}

	return $intTotalSize;
}

/**
 * Recursive function to get files count
 *
 * @param string $argStrPath
 * @return int
 */
function filesCount($argStrPath)
{
	$intTotalFiles = 0;
	$arrFiles = scandir($argStrPath);
	$strCleanPath = rtrim($argStrPath, DIRECTORY_SEPARATOR) . DIRECTORY_SEPARATOR;
	foreach ($arrFiles as $strEntity)
	{
		if (isNotDotFolder($strEntity))
		{
			$strCurrentFile = $strCleanPath . $strEntity;
			$intTotalFiles += is_dir($strCurrentFile) ? filesCount($strCurrentFile) : 1;
		}
	}
	return $intTotalFiles;
}

/**
 * Count number of identical files in folder
 *
 *
 * @param $argStrPath
 * @return int
 */
function dupesCount($argStrPath)
{
	$intTotalDupes = 0;
	$arrFiles = scandir($argStrPath);
	$strCleanPath = rtrim($argStrPath, DIRECTORY_SEPARATOR) . DIRECTORY_SEPARATOR;
	foreach ($arrFiles as $strEntity)
	{
		$strCurrentFile = $strCleanPath . $strEntity;
		if (isNotDotFolder($strEntity))
		{
			foreach ($arrFiles as $strEntityCompareTo)
			{
				if (isNotDotFolder($strEntityCompareTo)
					&& !is_dir($strCleanPath . $strEntityCompareTo)
					&& $strEntity != $strEntityCompareTo
				)
					$intTotalDupes += isIdentical($strCurrentFile, $strCleanPath . $strEntityCompareTo) ? 1 : 0;
			}
		}
	}
	return $intTotalDupes;
}

/**
 * @param $strEntity
 * @return bool
 */
function isNotDotFolder($strEntity)
{
	return ($strEntity <> "." && $strEntity <> "..");
}

/**
 * Return true if files are identical by content
 *
 * @param string $argStrFileName1
 * @param string $argStrFileName2
 * @return bool
 */
function isIdentical($argStrFileName1, $argStrFileName2)
{
	// Fixme: remove errors suppression
	return (@md5_file($argStrFileName1) == @md5_file($argStrFileName2));
}

/**
 * @param array $argArrA
 * @param array $argArrB
 * @return int
 */
function sizeCmp($argArrA, $argArrB)
{
	if ($argArrA['size'] == $argArrB['size'])
	{
		return 0;
	}
	return ($argArrA['size'] < $argArrB['size']) ? -1 : 1;
}