list files by type

files-by-type.php

#!/usr/bin/php
<?php
/**
 * files-by-type.php - get metadata from files readable with exiftool
 * and presents options for renaming
 * creates cache files, metadata.ser
 *
 * https://www.sno.phy.queensu.ca/~phil/exiftool/#filename
 * see https://www.sno.phy.queensu.ca/~phil/exiftool/exiftool_pod.html
 *
 * relies on command-line tools, tested on MacOS.
 *
 * @license GPLv3 (http://www.gnu.org/licenses/gpl-3.0.html)
 */

define('DEBUG',1);
define('VERBOSE',1);

date_default_timezone_set('UTC');
ini_set('default_charset', 'utf-8');
ini_set('mbstring.encoding_translation', 'On');
ini_set('mbstring.func_overload', 6);
ini_set('auto_detect_line_endings', TRUE);

// check CLI commands are available
$folder = '.';
$commands = getCommands();
if (empty($commands)) exit;

$cmd = 'find '.$folder.' -type f -print';
$files = cmd_execute($cmd);
$types = [];
foreach ($files as $k => $path) {
    $ext = strtolower(pathinfo($path, PATHINFO_EXTENSION));
    if (!array_key_exists($ext, $types)) {
        $types[$ext] = 1;
    } else {
        $types[$ext]++;
    }
}

$total = array_sum($types);
ksort($types);
foreach ($types as $type => $count) {
    printf("\n%s:\t%".strlen($total)."s", $type, $count);
}
printf("\n\nTotal:\t%s\n", $total);

exit;

//-----------------------------------------------------------------------------
// functions used above

/**
 * Execute a command and return streams as an array of
 * stdin, stdout, stderr
 *
 * @param  string $cmd command to execute
 * @return array|false array $streams | boolean false if failure
 * @see    https://secure.php.net/manual/en/function.proc-open.php
 */
function shell_execute($cmd)
{
    $process = proc_open(
        $cmd,
        [
        ['pipe', 'r'],
        ['pipe', 'w'],
        ['pipe', 'w']
        ], $pipes
    );
    if (is_resource($process)) {
        $streams = [];
        foreach ($pipes as $p => $v) {
            $streams[] = stream_get_contents($pipes[$p]);
        }
        proc_close($process);
        return [
            'stdin'  => $streams[0],
            'stdout' => $streams[1],
            'stderr' => $streams[2]
        ];
    }
    return false;
}
/**
 * Execute a command and return output of stdout or throw exception of stderr
 *
 * @param  string $cmd command to execute
 * @param  boolean $split split returned results? default on newline
 * @param  string $exp regular expression to preg_split to split on
 * @return mixed string $stdout | Exception if failure
 * @see    shell_execute($cmd)
 */
function cmd_execute($cmd, $split = true, $exp = "/\n/")
{
    $result = shell_execute($cmd);
    if (!empty($result['stderr'])) {
        throw new Exception($result['stderr']);
    }
    $data = $result['stdout'];
    if (empty($split) || empty($exp) || empty($data)) {
        return $data;
    }
    return preg_split($exp, $data);
}

// check required commands installed and get path
function getCommands()
{
  static $commands = []; // cli command paths
  if (!empty($commands)) {
    return $commands;
  }

  $requirements = [
    'find' => 'system find commmand'
  ];

  $errors = [];
  foreach ($requirements as $tool => $description) {
    $cmd = cmd_execute("which $tool");
    if (empty($cmd)) {
      $errors[] = "Error: Missing requirement: $tool - " . $description;
    } else {
      $commands[$tool] = $cmd[0];
    }
  }

  if (!empty($errors)) {
    debug(join("\n", $errors) . "\n");
  }

  return $commands;
}