s3-content-type-fixer.php

<?php

$opts = getopt ('', array(
  'access-key:',
  'secret-key:',
  'bucket:',
  'prefix:',
  'marker:'
));

if( !isset($opts['access-key'], $opts['secret-key'], $opts['bucket']) ) {
  echo "--access-key --secret-key and --bucket are required\n";
  exit(1);
}

require 'vendor/autoload.php';


function system_mime_type_extensions() {
    # Returns the system MIME type mapping of MIME types to extensions, as defined in /etc/mime.types (considering the first
    # extension listed to be canonical).
    $out = array();
    $file = fopen('/etc/mime.types', 'r');
    while(($line = fgets($file)) !== false) {
        $line = trim(preg_replace('/#.*/', '', $line));
        if(!$line)
            continue;
        $parts = preg_split('/\s+/', $line);
        if(count($parts) == 1)
            continue;
        $type = array_shift($parts);
        if(!isset($out[$type]))
            $out[$type] = array_shift($parts);
    }
    fclose($file);
    return $out;
}
$mimes = array_flip(system_mime_type_extensions());

$aws = \Aws\Common\Aws::factory(array(
  'key' => $opts['access-key'],
  'secret' => $opts['secret-key']
));
$s3 = $aws->get('s3');

$iterator = $s3->getIterator('ListObjects', array_filter(array(
    'Bucket' => $opts['bucket'],
    'Prefix' => $opts['prefix'],
    'Marker' => @$opts['marker'],
)));

foreach( $iterator as $item ) {
  if( false === ($pos = strrpos($item['Key'], '.')) ) {
    fwrite(STDERR, $item['Key'] . " - No file extension\n");
    continue;
  }
  $extension = strtolower(substr($item['Key'], $pos+1));
  if( !isset($mimes[$extension]) ) {
    fwrite(STDERR, $item['Key']  . " - No mime for extension\n");
    continue;
  }
  $mime = $mimes[$extension];
  // Head object
  $head = $s3->headObject(array(
    'Bucket' => $opts['bucket'],
    'Key' => $item['Key'],
  ));
  if( $head['ContentType'] === $mime ) {
    fwrite(STDERR, $item['Key'] 
            . ' - Object is already at expected content type ' . $mime . "\n");
    continue;
  }
  
  // Fix content type
  try {
    $result = $s3->copyObject(array_filter(array(
      'ACL' => 'public-read',
      'Bucket' => $opts['bucket'],
      'CacheControl' => $head['CacheControl'],
      'ContentDisposition' => $head['ContentDisposition'],
      'ContentEncoding' => $head['ContentEncoding'],
      'ContentLanguage' => $head['ContentLanguage'],
      'ContentType' => $mime,
      'CopySource' => $opts['bucket'] . '/' . $item['Key'],
      'Expires' => $head['Expires'],
      'Key' => $item['Key'],
      'StorageClass' => $item['StorageClass'],
      'ServerSideEncryption' => $head['ServerSideEncryption'],
    )));
    fwrite(STDERR, $item['Key'] . " - Content Type fixed to " . $mime . "\n");
  } catch( \Exception $e ) {
    echo $e;
  }

  
  }