[
https://issues.apache.org/jira/browse/XERCESC-1978?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Scott Cantor closed XERCESC-1978.
---------------------------------
Applied to 3.1 branch, r1662888
This is a modified fix that is ABI safe. The original patch modified only an
internal struct that isn't supposed to be public, but to eliminate any
possibility of a change, this is a less efficient fix that manually checks
string lengths instead of storing them in-struct.
> DOMDocumentImpl:: getPooledNString(const XMLCh *in, XMLSize_t n) returns
> incorrect string
> -----------------------------------------------------------------------------------------
>
> Key: XERCESC-1978
> URL: https://issues.apache.org/jira/browse/XERCESC-1978
> Project: Xerces-C++
> Issue Type: Bug
> Components: DOM
> Affects Versions: 3.1.1
> Environment: Windows x32, but should be everywhere
> Reporter: Rainer Prosi
> Assignee: Alberto Massari
> Priority: Blocker
> Fix For: 3.1.2, 3.2.0
>
>
> the methods returns the original string rather than the substring(n) if the
> substring and original string have the same hash code.
> This leeds to corrupt xml files due to prefixes being replaced by the entire
> string:
> Original version:
> while (*pspe != 0)
> {
> if (XMLString::equals((*pspe)->fString, in))
> return (*pspe)->fString;
> ...
> Bugfix suggestion:
> //RP 120416 we need to check for length of the returned <=n; else any hash
> matching string longer than n will be accepted --- bad snafu!
> if (XMLString::equalsN((*pspe)->fString, in, n) && XMLString::stringLen(
> (*pspe)->fString)<=n)
> Example prefix + attribute name that will fail:
> HDM:
> HDM:OffsetBack
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
