[xmlsec] XMLSEC Reference URI question

Moultrie, Ferrell (ISSAtlanta) FMoultrie at iss.net
Wed Jul 24 11:38:25 PDT 2002


Aleksey:
  Thanks for the background. My biggest problem is that my document is a
collection of logically unrelated documents (think of a file folder full of
signed letters). They were all created/signed independently of each other at
different times/places. Because of this, the "section#" attribute is a
problem since I have no globally unique ID to reference when
creating/validating the signature. Since I presume that the Id="tag" is
itself actually computed into the signature hash, I can't assign the tag
when building the compound document. If I'm wrong on that and only the
children of Section# would be signed/verified then I could work around this
by assigning section numbers on a document-by-document basis when assembling
the master document. The only other solution I've thought of would be to
clone the subdocuments into a new/empty document and sign/verify them there.
That seems like a lot of extra work but then an empty URI would work. 
  Does the above description make sense? Any other ideas on how to solve
this problem without having to implement full URI/xpointer support?
Thanks!
  Ferrell

-----Original Message-----
From: Aleksey Sanin [mailto:aleksey at aleksey.com] 
Sent: Tuesday, July 23, 2002 6:48 PM
To: Moultrie, Ferrell (ISSAtlanta)
Cc: 'xmlsec at aleksey.com'; Dodd, Tim (ISS Atlanta)
Subject: Re: [xmlsec] XMLSEC Reference URI question


Hi, Ferrell!

The current XMLDSig does not require full XInclude support and limits the
possible reference URIs to the URIs you've listed plus full 
qualified URLs.
The xmlsec does not support full XIncude simply because there were no demand
for this yet :) The problem with "id" is that the according to the XML specs
attribute with
*any* name could have ID flag in the DTD/schema and by this used as "id".
For example, you can have something like this in your DTD:
   <!ATTLIST Section ThisIsAnID ID #IMPLIED>
and attribute "ThisIsAnID" MUST be treated as an "id" (for example, when 
you
are using "#xpointer(id('tag'))" to reference the element).
As you can see, the "id" attribute means nothing w/o DTD or shema :) I 
decided that
it's a bad idea to require application to provide DTD/schema in order to 
validate document.
So I've implemented a "hack" when application can tell xmlsec library 
which attributes
are "ids" in the particular document.
Regarding your question, I think that can do something like this:  

<Document>
  <Section1 id="section1">
    ... content ...
    <dsig:Signature>
	<dsig:SignedInfo>
	...
	<dsig:Reference URI="#xpointer(id('section1'))">
		<dsig:Transforms>
			<!-- exclude dsig:Siganture element from digest! -->
			<dsig:Transform	/>			
		</dsig:Transforms>			
		...
	</dsig:Reference>
	</dsig:SignedInfo>
	...
    </dsig:Signature>
  </Section1>
  <Section2 id="section2">
    ... content ...
    <dsig:Signature>
	<dsig:SignedInfo>
	...
	<dsig:Reference URI="#xpointer(id('section2'))">
		<dsig:Transforms>
			<!-- exclude dsig:Siganture element from digest! -->
			<dsig:Transform	/>			
		</dsig:Transforms>			
		...
	</dsig:Reference>
	</dsig:SignedInfo>
	...
    </dsig:Signature>
  </Section2>
  ...
</Document>

Please note, that you need to put additional XPath transform to exclude 
<dsig:Signature> element
itself from digesting (XPath transofrm is supported by xmlsec)!


With best regards,

Aleksey





Moultrie, Ferrell (ISSAtlanta) wrote:

>Aleksey:
>  Looking in xmlSecTransformStateParseUri() [transforms.c:1069] it 
>appears that your support of current-document URI references is limited 
>to:  o URI="" (empty URI, whole document signed/verified)  o 
>URI="#xpointer(/)"  o URI="#xpointer(id('tag'))"
>  Further, it looks like the id('tag') actually resolves to looking for the
>first element in the document with the attribute Id="tag". This is
commented
>as a hack for documents w/o schemas or DTDs. Can you explain what's behind
>this "hack" and where you are headed with regard to the complete URI
>specification?
>
>  Also, since the URI processing appears to be limited, I'm wondering 
>if you support the use of an <XPath> element child of the <Transform> 
>element fully, partially, or not at all.
>
>  The problem I'm trying to solve is that I have documents which 
>consist of multiple sections that each have an individual signature on 
>that section only. In other words,
>
><Document>
>  <Section1>
>    ... content ...
>    <Signature ... />
>  </Section1>
>  <Section2>
>    ... content ...
>    <Signature ... />
>  </Section2>
>  ...
></Document>
>
>  I need to have some way (presumably the Reference URI or the 
>Transform) to limit the signature (and verification) to just the 
>content of <Section1> when computing <Section1>'s signature block, etc. 
>What is the best way to support this case with the current XMLSEC 
>library?
>
>Thanks!
>  Ferrell
>
>=====================================
>Ferrell Moultrie (ferrell at iss.net)
>Software Engineer
>
>Internet Security Systems, Inc.
>6303 Barfield Road
>Atlanta, Georgia 30328
>Phone:  404-236-2600
>Direct: 404-236-2849
>Fax:    404-236-2632
>http://www.iss.net
>
>Internet Security Systems -- The Power to Protect 
>=====================================
>_______________________________________________
>xmlsec mailing list
>xmlsec at aleksey.com http://www.aleksey.com/mailman/listinfo/xmlsec
>  
>




More information about the xmlsec mailing list