Aleksey:
  Done. I didn't think the overload was particularly troublesome since
the XPath syntax was clearly different from a node-name but the approach
you suggested is cleaner and without ambiguity. I'm attaching diffs
against 0.0.10 as before plus an additional diff (-b) which is smaller
and removes some (possibly gratuitous) white-space editing I'd done
while implementing the code. Your choice as to which to check-in.
Ferrell

-----Original Message-----
From: Aleksey Sanin [mailto:[EMAIL PROTECTED]] 
Sent: Thursday, December 19, 2002 4:50 AM
To: Moultrie, Ferrell (ISSAtlanta)
Cc: [EMAIL PROTECTED]
Subject: Re: [xmlsec] enhancement to xmlsec


Hi, Ferrell!

I like this idea but might I propose to add a new argument 
"--node-xpath" instead of
changing the meaning of the old one? I just don't like this tricky "if 
argument starts with slash..."
part because I 'll be the first one who will forget about this :)
Will you mind to make this change, please? And I'll be happy to commit 
it to CVS :)

Thanks,
Aleksey


Moultrie,  (ISSAtlanta) wrote:

>Aleksey:
>  My xml documents that I'm signing/verifying contain repeated nodes of
>the same name so any beyond the first node are not addressable by the
>--node-name construct supported by xmlsec. Additionally, I'm not using
a
>dtd so the --node-id doesn't help me out either. Finally, the
>sign/verify code accept the --node-id/--node-name arguments but ignore
>them, always signing/verifying the first Signature node found. 
>  To solve the above issues, I'm attaching suggested changes (relative
>to 0.0.10) that allow the --node-name argument to contain an absolute
>XPath expression to locate the section of the document to be
>signed/verified/etc. If the argument starts with a '/' character, it is
>processed as an XPath expression, otherwise it is just a ns:node-name
as
>before. This shouldn't break anything but allows much greater
>flexibility in using xmlsec with complex xml structures. 
>  For example, "--node-name //Module[2]" would select the 2nd Module
>element in the document skeletonized below:
><?xml version="1.0" encoding="UTF-8"?>
><Keys>
>  <Modules>
>    <Module> first module element including <sig:Signature/> </Module>
>    <Module> second module element including <sig:Signature/> </Module>
>    <Module> third module element including <sig:Signature/> </Module>
>  </Modules>
></Keys>
>  Using either "--node-name //Module[1]" or "--node-name Module" would
>select the first occurance (note node-name format works just as it did
>before).
>  While I could have obviously written a new utility to handle manual
>signing of my test documents, it seems much better to add this
>functionality to the xmlsec application where it could easily be reused
>by others. I'm attaching the diff's -- let me know if you need another
>format, or, a full copy of the xmlsec.c file before/after my changes.
>  I have rather thoroughly tested the digital signature
>creation/verification but haven't tested encryption/decryption -- but
>since it now uses a common routine, what could go wrong, right? (:>)
>Thanks!
>  Ferrell
>
>
>  
>
>-----------------------------------------------------------------------
-
>
>306a307
>  
>
>>char *nodeXPath = NULL;
>>    
>>
>388,395c389,400
><              nodeName = strrchr(argv[++pos], ':');
><              if(nodeName != NULL) {
><                  *(nodeName++) = '\0';
><                  nodeNs = argv[pos];
><              } else {
><                  nodeName = argv[pos];
><                  nodeNs = NULL;
><              }
>---
>  
>
>>                      if (argv[++pos][0] == '/') { /* absolute XPath
specification? */ 
>>                              nodeXPath = argv[pos];
>>                      } else {
>>                              nodeName = strrchr(argv[pos], ':');
>>                              if(nodeName != NULL) {
>>                                  *(nodeName++) = '\0';
>>                                  nodeNs = argv[pos];
>>                              } else {
>>                                  nodeName = argv[pos];
>>                                  nodeNs = NULL;
>>                              }
>>                      }
>>    
>>
>1245a1251,1281
>  
>
>>/* Function to find base node for operation as specified by user using
either
>> * DTD Id, XPath abs expression, NodeName, or default to root node */
>>xmlNodePtr findBaseNode(xmlDocPtr doc) {
>>      xmlNodePtr cur = NULL;
>>      if(nodeId != NULL) {
>>          xmlAttrPtr attr;
>>          attr = xmlGetID(doc, BAD_CAST nodeId);
>>          cur = (attr != NULL) ? attr->parent : NULL;
>>      } else if(nodeXPath != NULL) {
>>              xmlNodeSetPtr pNodeList = NULL;
>>              xmlXPathContextPtr pCtx = NULL;
>>              xmlXPathObjectPtr pXPathObj = NULL;
>>              pCtx = xmlXPathNewContext(doc);
>>              pXPathObj = xmlXPathEval(BAD_CAST nodeXPath, pCtx);
>>              if (pXPathObj != NULL) {
>>                      pNodeList = pXPathObj->nodesetval;
>>              }
>>              if (pNodeList != NULL && pNodeList->nodeNr > 0) { 
>>                      cur = pNodeList->nodeTab[0];
>>              }
>>              xmlXPathFreeContext(pCtx);
>>              xmlXPathFreeObject(pXPathObj);
>>      } else if(nodeName != NULL) {
>>              cur = xmlSecFindNode(xmlDocGetRootElement(doc), BAD_CAST
nodeName, BAD_CAST nodeNs);
>>      } else {
>>          cur = xmlDocGetRootElement(doc);
>>      }
>>      return cur; /* can be NULL .. caller must check */
>>} /* of findBaseNode() */
>>
>>    
>>
>1248a1285
>  
>
>>      xmlNodePtr cur;
>>    
>>
>1255,1256c1292,1301
><     signNode = xmlSecFindNode(xmlDocGetRootElement(doc), 
><                            BAD_CAST "Signature", xmlSecDSigNs);
>---
>  
>
>>      /**
>>       * What do we want to sign?
>>       */    
>>      cur = findBaseNode(doc);
>>      if (cur == NULL) { 
>>              fprintf(stderr,"Error: failed to find specified
node\n");
>>              return(-1);
>>      }
>>      signNode = xmlSecFindNode(cur, BAD_CAST "Signature",
xmlSecDSigNs);
>>    
>>
>1258,1259c1303,1304
><         fprintf(stderr,"Error: failed to find Signature node\n");
><      return(-1);
>---
>  
>
>>              fprintf(stderr,"Error: failed to find Signature
node\n");
>>              return(-1);
>>    
>>
>1310a1356
>  
>
>>      xmlNodePtr cur;
>>    
>>
>1313,1315c1359,1368
><          
><     signNode = xmlSecFindNode(xmlDocGetRootElement(doc), 
><                            BAD_CAST "Signature", xmlSecDSigNs);
>---
>  
>
>>      /**
>>       * What do we want to verify?
>>       */    
>>      cur = findBaseNode(doc);
>>      if (cur == NULL) { 
>>              fprintf(stderr,"Error: failed to find specified
node\n");
>>              return(-1);
>>      }
>>      signNode = xmlSecFindNode(cur, BAD_CAST "Signature",
xmlSecDSigNs);
>>    
>>
>1317,1318c1370,1371
><         fprintf(stderr,"Error: failed to find Signature node\n");
><      return(-1);
>---
>  
>
>>              fprintf(stderr,"Error: failed to find Signature
node\n");
>>              return(-1);
>>    
>>
>1414,1424c1467,1468
><      if(nodeId != NULL) {
><          xmlAttrPtr attr;
><          
><          attr = xmlGetID(doc, BAD_CAST nodeId);
><          cur = (attr != NULL) ? attr->parent : NULL;
><      } else if(nodeName != NULL) {
><          cur = xmlSecFindNode(xmlDocGetRootElement(doc), BAD_CAST
nodeName, BAD_CAST nodeNs);
><      } else {
><          cur = xmlDocGetRootElement(doc);
><      }
><      
>---
>  
>
>>      cur = findBaseNode(doc);
>>
>>    
>>
>1498c1542,1547
><     cur = xmlSecFindNode(xmlDocGetRootElement(doc), BAD_CAST
"EncryptedData", xmlSecEncNs);
>---
>  
>
>>      cur = findBaseNode(doc);
>>      if (cur == NULL) { 
>>              fprintf(stderr,"Error: failed to find specified
node\n");
>>              return(-1);
>>      }
>>    cur = xmlSecFindNode(cur, BAD_CAST "EncryptedData", xmlSecEncNs);
>>    
>>
>
>


_______________________________________________
xmlsec mailing list
[EMAIL PROTECTED]
http://www.aleksey.com/mailman/listinfo/xmlsec

Reply via email to