Re: AW: Problem in Decryption

[Berin Lautenbach]

Dittman,

I think you are correct.

And to further re-inforce this - the Merlin interop tests (one of the 
basic sets of tests you have to pass to be considered "interoperable") 
have white space between the parent and the EncryptedData.  I found that 
if you remove the white space, you change the original document.

Finally - I don't believe (but it's been a long time since I really 
looked) there is anything in the spec that states that we *have* to 
encrypt all the child nodes of an element for a content encryption. 
I.e. we could provide an API that lets a user encrypt everything after 
the white space, and that would be legal.

So the implementation is very careful to only replace the EncryptedData, 
in the understanding that to do anything else would change the original 
document - and thus potentially invalidate signatures over that document 
(along wiht other possible problems).

Cheers,
        Berin
Dittmann Werner wrote:

Vishal, Berin

during some encrypt/decrypt tests with signature:

This is mainly a problem for "content" mode encryption.
AFAIK, during encryption XMLCipher creates a document fragment
with the content, serializes and encrypts it. Decryption
is reverse. After decryption XMLCipher first removes
(or shall remove) all child-nodes of the element e.g. 
CreditCard, then puts in the nodes of the document 
fragment (effectivly restores the original content).

Thus, IMHO, there is no problem with Signature/Encrypt.
During our test we already did tests signature/encryption
in content mode.
Regards,
Werner

-----Ursprüngliche Nachricht-----
Von: Vishal Mahajan [mailto:[EMAIL PROTECTED] 
Gesendet: Mittwoch, 11. Februar 2004 08:35
An: [EMAIL PROTECTED]
Betreff: Re: Problem in Decryption

Hi Berin,

I agree with your reading of the spec. But, the spec does not prevent 
the encrypter from putting whitespaces between the tags of 
EncryptedData 
and its parent node. Also there are example encrypted 
documents like the 
following in the spec:

     <EncryptedData xmlns='http://www.w3.org/2001/04/xmlenc#'
      Type='http://www.w3.org/2001/04/xmlenc#Content'>
       <CipherData>
         <CipherValue>A23B45C56</CipherValue>
       </CipherData>
     </EncryptedData>
Other thing to note is that if the document was signed prior to 
encryption, the signature verification would fail. I think this is an 
important interop issue. What do you think?

Thanks,
Vishal
Berin Lautenbach wrote:


Vishal,

I ran accross this when I was doing the checks for the Merlin Interop
examples - which have white space prior to the EncryptedData 
node, even

when that node is of type "element content".  According to 
the spec (from

section 4.2) :
#####################################
"The decryptor SHOULD support the ability to replace the 
EncryptedData

element with the decrypted 'element' or element 'content' 
represented by

the UTF-8 encoded characters. The decryptor is NOT REQUIRED 
to perform

validation on the result of this replacement operation."
######################################
The key words for me were in the first sentance - "...replace the
EncryptedData element with the decrypted 'element' or element
'content'...".
There is no mention of the parent of the EncryptedData, or any of the
surrounding nodes, so my reading is that the libraries (both 
the C++ and

Java libraries have this behaviour) are doing the correct thing.
Happy to be corrected if someone can find something else in the spec!
Cheers,
   Berin




Hi All,

I think there is a problem in the decryption process of the 
XMLCipher

class. To put the problem across, here's my question:

If there is some whitespace between the tags of 
EncryptedData and its

parent (See the example below) and the Type attribute of the
EncryptedData corresponds to "element content", then the current
implementation simply replaces the EncryptedData with the decrypted
DocumentFragment. Shouldn't it first remove the existing contents
(whitespace, etc) of the parent node of EncryptedData and then do an
appendChild(DocumentFragment) on the parent?
<CreditCard Limit='5,000' Currency='USD'>
  <EncryptedData xmlns='http://www.w3.org/2001/04/xmlenc#'
   Type='http://www.w3.org/2001/04/xmlenc#Content'>
    <CipherData>
      <CipherValue>A23B45C56</CipherValue>
    </CipherData>
  </EncryptedData>
</CreditCard>
If this is accepted as a bug, I even have a patch for it. 
Should I also

raise a bugzilla?

Thanks,
Vishal