[XERCESC-2145] indefinite memory allocations using DOMNode::getTextContent()

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Not A Bug
• Affects Version/s: 3.1.2, 3.2.1
• Component/s: DOM
• Environment: Windows 10, Visual Studio 2017
• External issue URL: https://stackoverflow.com/questions/49677254/xerces-c-memory
• Assignee: Unassigned
• Reporter: Blair Fonville
• Created: 05/Apr/18 22:04
• Updated: 06/Apr/18 16:38
• Resolved: 05/Apr/18 23:59

Description

In a program which reads (parses) an xml file once into a DOMDocument, and then accesses the content of the nodes/elements often, through getTextContent(), the heap memory builds indefinitely.

For example: 

#include <xercesc/dom/DOM.hpp>

XERCES_CPP_NAMESPACE_USE

int main(int argc, char **argv)
{
   XMLPlatformUtils::Initialize();

   XMLCh tempStr[100];
   XMLString::transcode("LS", tempStr, 99);
   DOMImplementation *impl = DOMImplementationRegistry::getDOMImplementation(tempStr);
   DOMLSParser *parser = ((DOMImplementationLS*)impl) ->createLSParser( DOMImplementationLS::MODE_SYNCHRONOUS, 0);

   DOMDocument *doc = impl->createDocument(0, 0, 0);
   doc = parser->parseURI("config.xml");

   DOMElement *el = doc->getDocumentElement();

   // Heap blows up here
   while (1) {
      char *cstr = XMLString::transcode(el->getTextContent());
      XMLString::release(&cstr);
   }

   // and/or here
   while (1) {
      XMLCh *xstr = XMLString::replicate(el->getTextContent());
      char *cstr = XMLString::transcode(xstr);
      XMLString::release(&cstr);
      XMLString::release(&xstr);
   }
}

I realize that the memory is supposed to be freed when the Document is released, but in a program where the Document is populated at startup, and used forever, this is, for all practical purposes, a memory leak.