I just had to deal with a shitload of serialized php data that had become corrupted with Windows carriage returns.

Errors like this were produced:

Notice: unserialize(): Error at offset 1060 of 41775.

If you’re getting that on a call to unserialize, it’s very likely that your string lengths do not match your serialized data. That looks something like this:

s:3:”fuck”;

In my example, fuck is four letters long, yet unserialize is told to read in 3 characters. That will produce an error that looks like this:

Error at offset 8 of 11 bytes

You can change that above 3 to a 4 pretty easily, but it’s much harder if your file is 300K long.

Here’s a function that will re-measure all your serialized strings and let you unserialize them. You’ll still have fucked up characters, but at least you’ll get your data back.

function fixFuckedData($data){

  $splits = preg_split("/s:([0-9]*):/”, $data);
  preg_match_all(”/s:([0-9]*):/”, $data, $lengths);
  $lengths = $lengths[1];

  for($i = 0; $i < sizeof($splits)-1; $i++){
    $text = $splits[$i+1];
    $pos = strpos($text, '";');
    $text = substr($text, 1, $pos-1);
    $text_len = strlen($text);
    if($lengths[$i] != $text_len)
      $lengths[$i] = $text_len;
      //echo "{$lengths[$i]} -> $text_len\n”;
  }
  //var_dump($lengths);
  //var_dump(unserialize($data));

  $return = $splits[0];
  for($i = 0; $i < sizeof($splits)-1; $i++){
    $return .= "s:{$lengths[$i]}:";
    $return .= $splits[$i+1];
  }
  return $return;
}