Problem with mcrypt function*

Question

I am using these two functions below to encrypt and decrypt a string. I left the example as clean as possible.

_{Function for encryption}

$key = '123';
$iv  = md5( md5( 'key' ) );
mcrypt_encrypt(MCRYPT_RIJNDAEL_256, md5($key), $string, MCRYPT_MODE_CBC, $iv);

_{Function to decrypt}

$key = '123';
$iv  = md5( md5( 'key' ) );
mcrypt_decrypt(MCRYPT_RIJNDAEL_256, md5($key), $string, MCRYPT_MODE_CBC, $iv);

I don’t know the function at all, and I’m having trouble using json Encode/Decode. I can use Encode without problem, but Decode does not run the string decrypted.

I used as an example a simple array array( 123 ) and the decrypted output is exactly equal to json_encode( array( 123 ) ), the difference is the size, but neither before nor at the end.

_Dump

decrypt( $argument )        : string(32) "[123]"
json_encode( array( 123 ) ) : string(5) "[123]"

---

I found a question which reports the same problem as mine and the proposed solution was the use of rtrim( $decrypted , "\0" ). It worked for the author and me, but the explanation is superficial:

Says that the ENCRYPT/DECRYPT function adds garbage to get the correct size, and must remove nulls at the end of the string.

I also created null values at the end of the string but did not affect the size in the dump. I can’t understand how the string size doesn’t match.

---

1. The use of \0 in the trim covers which characters?
2. I would like to know more about the options Cipher and mode, I did not find in the DOC deeper references of each case.
3. It is possible to combine Cipher and mode so that the string does not create null values to satisfy the size.

Answer 1

For those who come to find this topic in the future, what Bacco said in that comment, basically, is to remove all the "garbage" added by the ciphers so that they have the same size which, in code, would be this:

$string = json_encode( array( 123 ) );

$key = '123';
$iv  = md5( md5( 'key' ) );

$encrypted = mcrypt_encrypt(

    MCRYPT_RIJNDAEL_256, md5( $key ),

    $string, MCRYPT_MODE_CBC, $iv
);

$decrypted = rtrim(

    mcrypt_decrypt( MCRYPT_RIJNDAEL_256, md5($key), $encrypted, MCRYPT_MODE_CBC, $iv),

    "\x00..\x1F"
);

Just compare with a var_dump() and see that without that rtrim() the encrypted string is quite different from the unencrypted string, but both have the same length:

string '’!Ù Èžtvûþ×Ij>+|oÙ(Œà¢qŒ)·Î"“Æ¡h' (length=32)
string '[123]���������������������������' (length=32)

But (and now comes my contribution), it is not enough to remove only the byte null ( 0 or 0x00) as each language can include its own trash.

That’s why my pseudo-code has a wider character range \x00.. x1F, that is, the first 31 characters of the ASCII Table.

By the way, responding to this survey of yours, if you only consider the removal of byte null ( 0) can peacefully use rtrim() without arguments as this character is already in the default deletion list of the function.