[xmlsec] How to avoid the node passed to xmlSecEncCtxXmlEncrypt to be released

[Frank Gross]

I attached the patch that allows the replaced nodes during encryption 
and decryption to be returned in the 'xmlSecEncCtxPtr' structure. 
Actually, I added two entries in the 'xmlSecEncCtrPtr' structure :
  - nodeReplacementMode,  an enum of type 
'xmlEncCtxNodeReplacementMode', to define if the nodes will be released 
or returned in the second entry,
  - a pointer to a xmlNodePtr called 'replacedNodeList' that contains 
the list of the nodes that have been replaced.

One additional point, as it's up to the user to release the returned 
nodes, I'm not sure if the code I added in the 'xmlSecEncCtxReset' 
function to release the node list is necessary, because if someone 
releases the node but forget to set the pointer to NULL it will crash. 
But if he doesn't release the nodes, and this code is not there we got a 
memory leak.

If you could take a look

Thanks,
Frank

Aleksey Sanin a écrit :
> Sure, I love patches :) BTW, you can't find any docs about
> flags/flags2 in xmlSecEncCtx because they are not used at the moment :)
> Just reserved for the future :)
>
> Aleksey
>
> Frank Gross wrote:
>> Yes I agree, but it would be more efficient if I wouldn't have to do 
>> that. And I have the same problem when I try to encrypt only the 
>> content of an element, where all sub-nodes are removed. Actually, I 
>> try to write an API where you give the node to be encrypted, but the 
>> node passed to my function is still alive, and simply destroying the 
>> nodes inside the library is impossible in my case, because some other 
>> references can point on these nodes.
>>
>> Therefore may I suggest a request ? Could you add an option in the 
>> 'xmlSecEncCtxPtr' structure to specify how the nodes to be replaced 
>> should be handled, and return them in a list in the structure if for 
>> instance the option was set to not releasing them so that the user 
>> can choose to release them by hand or not. And the same kind of 
>> feature when decrypting a node, because in some cases you want to 
>> keep trace of the encrypted node.
>>
>> My 2 cents contribution that could really help in my case, and allow 
>> the programmer to choose instead of the library.
>>
>> If you agree I could make a patch, using flags or flags2 in the 
>> 'xmlSecEncCtxPtr' structure, and add a new entry of type xmlNodePtr 
>> with the list of nodes that were replaced but not released according 
>> to flags or flags2 value. BTW what is the difference between flags 
>> and flags2 ? Are they used because I didn't find any information in 
>> the documentation ?
>>
>> Best Regards,
>>
>> Frank
>>
>> Aleksey Sanin a écrit :
>>> Well, you can always copy the node yourself before encrypting it.
>>>
>>> Aleksey
>>>
>>> Frank Gross wrote:
>>>> Hi,
>>>>
>>>>  I noticed that function 'xmlSecEncCtxXmlEncrypt' releases the node 
>>>> that was encrypted when replaced by the 'EncryptedData' node.
>>>> Does it exist a way to not release that node, and let the user 
>>>> choose whether he wants to destroy it or not ?
>>>>
>>>> Thanks,
>>>>
>>>> Frank
>>>> _______________________________________________
>>>> xmlsec mailing list
>>>> xmlsec at aleksey.com
>>>> http://www.aleksey.com/mailman/listinfo/xmlsec
>>>
>> _______________________________________________
>> xmlsec mailing list
>> xmlsec at aleksey.com
>> http://www.aleksey.com/mailman/listinfo/xmlsec
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: patch.diff
Type: text/x-diff
Size: 12892 bytes
Desc: not available
Url : http://www.aleksey.com/pipermail/xmlsec/attachments/20080313/11d95918/patch-0002.bin