[XERCESC-1598] SGXMLScanner omits namespace prefixes when reporting start/end element QNames

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.7.0
• Fix Version/s: 2.8.0
• Component/s: Validating Parser (XML Schema)
• Environment: 64-bit; Solaris 10; Sun CC: Sun C++ 5.7 Patch 117830-07 2006/03/15
• Assignee: Unassigned
• Reporter: Michael Fuller
• Created: 26/May/06 13:20
• Updated: 12/Mar/18 04:05
• Resolved: 30/May/06 16:29

Description

When parsing a document the SGXMLScanner and (sometimes) the IGXMLScanner does not
correctly report QNames when calling startElement() and endElement(): specifically, it omits
the namespace prefix. (Other scanners behave differently=>correctly.)

For example, using the SAX2Print sample against the document:
<doc xsi:schemaLocation="doc doc.xsd ent ent.xsd"
xmlns="doc" xmlns:ent="ent"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<ent:company>HP</ent:company>
</doc>

(with appropriate "doc.xsd" and "ent.xsd" files in the CWD), "SAX2Print -s" reports:
<?xml version="1.0" encoding="LATIN1"?>
<doc xsi:schemaLocation="doc doc.xsd ent ent.xsd">
<ent:company>HP</ent:company>
</doc>

but "SAX2Print" reports:
<?xml version="1.0" encoding="LATIN1"?>
<doc xsi:schemaLocation="doc doc.xsd ent ent.xsd">
<company>HP</company>
</doc>

Note "<company>" vs. "<ent:company>".

Note also that SAX2Print by default will be using the IGXMLScanner;
if you force the use of the SGXMLScanner by modifying SAX2Print.cpp:

• • • SAX2Print.cpp Fri May 26 16:18:07 2006
    • SAX2Print.cpp.ORIG Fri May 26 16:16:13 2006
      ***************
    • 319,329 ****
      parser->setFeature(XMLUni::fgXercesSchemaFullChecking, schemaFullChecking);
      parser->setFeature(XMLUni::fgSAX2CoreNameSpacePrefixes, namespacePrefixes);

• parser->setProperty(
• xercesc::XMLUni::fgXercesScannerName,
• static_cast<void*>(const_cast<XMLCh*>(
• xercesc::XMLUni::fgSGXMLScanner)));
  -
  //
  // Create the handler object and install it as the document and error
  // handler for the parser. Then parse the file and catch any exceptions

then both "SAX2Print -s" and SAX2Print" fail, printing <company> rather than <ent:company>.

I will attach the test document and schema files.

We have a kludge for SAX2XMLReaderImpl.cpp that works around the above bug,
but it's not the right fix, as there is nothing really wrong with SAX2XMLReaderImpl.cpp,
just the underlying scanners. Nonetheless, I will attach that also.