Class XHTMLBalancingHandler

java.lang.Object

org.xml.sax.helpers.DefaultHandler

org.apache.tika.sax.ContentHandlerDecorator

org.apache.tika.sax.XHTMLBalancingHandler

All Implemented Interfaces:

ContentHandler, DTDHandler, EntityResolver, ErrorHandler

public class XHTMLBalancingHandler
extends ContentHandlerDecorator

SAX decorator that tracks open elements so a parser can recover well-formed XHTML when an exception interrupts the SAX stream mid-element.

The decorator is a thin passthrough on the happy path: it pushes and pops an internal stack on startElement/endElement and otherwise forwards every event to the wrapped handler unchanged. It deliberately does NOT mask bad event sequences (mismatched or excess endElement, duplicate attributes, etc.) -- those remain visible to StrictXHTMLValidator so parser bugs still surface as test failures.

The unhappy path -- a per-part SAX parser throwing mid-element after emitting one or more start tags -- is handled via drainOpenElements(), which emits a matching endElement (with the original uri/localName/qName) for every element still on the stack, in reverse open order. The wrapped handler is left in a well-formed state with no dangling elements from the failed sub-parse.

Typical use wraps the handler that receives events from an inner SAX parser, inside the catch arm that swallows the inner parser's exception:

XHTMLBalancingHandler balancer = new XHTMLBalancingHandler(contentHandler);
try {
    XMLReaderUtils.parseSAX(stream, new EmbeddedContentHandler(balancer), context);
} catch (SAXException e) {
    balancer.drainOpenElements();
    // ... log and continue ...
}

This handler does not touch startDocument/endDocument; the caller still owns the document lifecycle.

Constructor Summary

Constructors

Constructor	Description
XHTMLBalancingHandler(ContentHandler handler)

Method Summary

Modifier and Type	Method	Description
void	drainOpenElements()	Emits a matching endElement for every element still on the open stack, in reverse open order.
void	endElement(String uri, String localName, String qName)
int	openElementCount()	Number of elements currently open through this handler.
void	startElement(String uri, String localName, String qName, Attributes attrs)

Methods inherited from class ContentHandlerDecorator

characters, endDocument, endPrefixMapping, error, fatalError, handleException, ignorableWhitespace, processingInstruction, setContentHandler, setDocumentLocator, skippedEntity, startDocument, startPrefixMapping, toString, warning

Methods inherited from class DefaultHandler

notationDecl, resolveEntity, unparsedEntityDecl

Methods inherited from class Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface ContentHandler

declaration

Constructor Details

XHTMLBalancingHandler

public XHTMLBalancingHandler(ContentHandler handler)

Method Details

startElement

public void startElement(String uri,
 String localName,
 String qName,
 Attributes attrs)
                  throws SAXException

Specified by:

startElement in interface ContentHandler

Overrides:

startElement in class ContentHandlerDecorator

Throws:

SAXException

endElement

public void endElement(String uri,
 String localName,
 String qName)
                throws SAXException

Specified by:

endElement in interface ContentHandler

Overrides:

endElement in class ContentHandlerDecorator

Throws:

SAXException

drainOpenElements

public void drainOpenElements()
                       throws SAXException

Emits a matching endElement for every element still on the open stack, in reverse open order. After this call the stack is empty.

Intended for the catch arm of a caller that swallowed a SAXException from an inner SAX parser: the inner parser may have left one or more elements open mid-stream, and downstream serialization needs matching closers before any further events.

Does NOT emit endDocument -- document lifecycle stays with the caller.

Throws:

SAXException

openElementCount

public int openElementCount()

Number of elements currently open through this handler. Exposed for tests and for callers that want to know whether drainOpenElements() would emit anything.