DTMBuilder is a glue layer which accepts input from the * CoroutineParser and issues calls to construct DTM nodes.
* *We're not really delighted with this; it's an extra layer of * call-and-return per node created. A better approach, since this is * so specific to DTM, might be to give it direct ("friend") access to * DTM's storage... or to fold it directly into the DTM class. We've * broken it out primarily to allow parallel development of the * builder and the DTM model. (Or, in fact, to make it a subclass * of CoroutineParser to avoid _that_ layer of call-and-return... though * I think tail-call optimization will help us in that case.
* *In its current form, this is basically a SAX-to-DTM call adapter. * SAX events coming in are restructured -- primarily by having their strings * and text content copied into appropriate storage and re-expressed as * index numbers. Note that this requires some negotiation between the DTM * and the builder to agree on which string pools this content will be placed * into.
* *We also restructure the call sequence somewhat, by merging * successive characters() calls into a single text-node creation, and * by breaking the attributes out of the SAX createEvent call and generating * create-attribute (and create-namespace-declaration) calls.
* *Status: MOSTLY DEAD CODE -- Merged into DTM implementations; this * class is being retained just for reference and can probably be discarded * some time soon -- Check with Joe.
* */ public class DTMBuilder implements ContentHandler, LexicalHandler { // // Data // /** Document being built. * *%TBD% The DTM API doesn't currently have construction calls, * so this is explicitly defined as a DTMDocumentImpl. That needs to * be fixed.
* */ private DTMDocumentImpl m_dtm; private CoroutineManager fCoroutineManager = null; private int fAppCoroutine = -1; private int fParserCoroutine = -1; private CoroutineParser co_parser; // Scott suggests sharing pools between DTMs. Note that this will require // threadsafety at the pool level. private static DTMStringPool commonLocalNames=new DTMSafeStringPool(); private static DTMStringPool commonNamespaceNames=new DTMSafeStringPool(); private static DTMStringPool commonPrefixes=new DTMSafeStringPool(); private DTMStringPool localNames; // For this DTM, may be common private DTMStringPool namespaceNames; // For this DTM, may be common private DTMStringPool prefixes; private FastStringBuffer content; // Unique per DTM int contentStart=0; // %TBD% The whole startup sequence has to be resolved -- // How much is passed builder-to-DTM and how much the other way? // Who decides whether we're sharing string pools or creating new ones? // When do we hook up to the coroutine parser system? // // Constructors // /* * @param dtm DTM object to be written into *%TBD% The DTM API doesn't currently have construction calls, * so this is explicitly defined as a DTMDocumentImpl. That needs to * be fixed.
* @param source SAX InputSource to read the XML document from. * @param parser SAX XMLReader to be used to parse source into dtm * */ public DTMBuilder(DTMDocumentImpl dtm, InputSource source, org.xml.sax.XMLReader parser) { m_dtm=dtm; // Start with persistant shared pools unless the DTM expresses // other preferences localNames=m_dtm.getLocalNameTable(); if(localNames==null) m_dtm.setLocalNameTable(localNames=commonLocalNames); namespaceNames=m_dtm.getNsNameTable(); if(namespaceNames==null) m_dtm.setNsNameTable(namespaceNames=commonNamespaceNames); prefixes=m_dtm.getPrefixNameTable(); if(prefixes==null) m_dtm.setPrefixNameTable(prefixes=commonPrefixes); // Unlike the other strings, which may be shared and thus should be // reset elsewhere (if at all), content starts empty each time we parse. content=m_dtm.getContentBuffer(); if(content==null) m_dtm.setContentBuffer(content=new FastStringBuffer()); else content.reset(); contentStart=0; // Establish incremental parsing hookups fCoroutineManager=new CoroutineManager(); fAppCoroutine = fCoroutineManager.co_joinCoroutineSet(-1); // %TBD% parser should be passed in so we can plug in the // Xalan version of other specific instances. co_parser=new CoroutineSAXParser(fCoroutineManager,fAppCoroutine,parser); fParserCoroutine=co_parser.getParserCoroutineID(); co_parser.setContentHandler(this); co_parser.setLexHandler(this); // Needed for comments, I think. // %TBD% MAJOR CONCERN: Are we sure we'll have reached the // startup point before we start the call? CHECK THIS! // Begin incremental parsing // Note that this doesn't return until the first chunk of parsing // has been completed and the parser coroutine yields. try { fCoroutineManager.co_resume(source,fAppCoroutine,fParserCoroutine); } catch(NoSuchMethodException e) { // Shouldn't happen unless we've miscoded our coroutine logic // "Shut down the garbage smashers on the detention level!" e.printStackTrace(System.err); fCoroutineManager.co_exit(fAppCoroutine); } } // // Public methods // // String accumulator support private void processAccumulatedText() { int len=content.length(); if(len!=contentStart) { // The FastStringBuffer has been previously agreed upon m_dtm.appendTextChild(contentStart,len-contentStart); contentStart=len; } } // // ContentHandler methods // Accept SAX events, reformat as DTM construction calls public void characters(char[] ch, int start, int length) throws org.xml.sax.SAXException { // Actually creating the text node is handled by // processAccumulatedText(); here we just accumulate the // characters into the buffer. content.append(ch,start,length); } public void endDocument() throws org.xml.sax.SAXException { // May need to tell the low-level builder code to pop up a level. // There _should't_ be any significant pending text at this point. m_dtm.appendEndDocument(); } public void endElement(java.lang.String namespaceURI, java.lang.String localName, java.lang.String qName) throws org.xml.sax.SAXException { processAccumulatedText(); // No args but we do need to tell the low-level builder code to // pop up a level. m_dtm.appendEndElement(); } public void endPrefixMapping(java.lang.String prefix) throws org.xml.sax.SAXException { // No-op } public void ignorableWhitespace(char[] ch, int start, int length) throws org.xml.sax.SAXException { // %TBD% I believe ignorable text isn't part of the DTM model...? } public void processingInstruction(java.lang.String target, java.lang.String data) throws org.xml.sax.SAXException { processAccumulatedText(); // %TBD% Which pools do target and data go into? } public void setDocumentLocator(Locator locator) { // No-op for DTM } public void skippedEntity(java.lang.String name) throws org.xml.sax.SAXException { processAccumulatedText(); //%TBD% } public void startDocument() throws org.xml.sax.SAXException { // No-op for DTM? //m_dtm.startDocument(); } public void startElement(java.lang.String namespaceURI, java.lang.String localName, java.lang.String qName, Attributes atts) throws org.xml.sax.SAXException { processAccumulatedText(); // %TBD% Split prefix off qname String prefix=null; int colon=qName.indexOf(':'); if(colon>0) prefix=qName.substring(0,colon); // %TBD% Where do we pool expandedName, or is it just the union, or... m_dtm.appendStartElement(namespaceNames.stringToIndex(namespaceURI), localNames.stringToIndex(localName), prefixes.stringToIndex(prefix)); /////// %TBD% // %TBD% I'm assuming that DTM will require resequencing of // NS decls before other attrs, hence two passes are taken. // %TBD% Is there an easier way to test for NSDecl? int nAtts=atts.getLength(); // %TBD% Countdown is more efficient if nobody cares about sequence. for(int i=nAtts-1;i>=0;--i) { qName=atts.getQName(i); if(qName.startsWith("xmlns:") || "xmlns".equals(qName)) { prefix=null; colon=qName.indexOf(':'); if(colon>0) { prefix=qName.substring(0,colon); } else { prefix=""; // Default prefix } m_dtm.appendNSDeclaration( prefixes.stringToIndex(prefix), namespaceNames.stringToIndex(atts.getValue(i)), atts.getType(i).equalsIgnoreCase("ID")); } } for(int i=nAtts-1;i>=0;--i) { qName=atts.getQName(i); if(qName.startsWith("xmlns:") || "xmlns".equals(qName)) { // %TBD% I hate having to extract the prefix into a new // string when we may never use it. Consider pooling whole // qNames, which are already strings? prefix=null; colon=qName.indexOf(':'); if(colon>0) { prefix=qName.substring(0,colon); localName=qName.substring(colon+1); } else { prefix=""; // Default prefix localName=qName; } content.append(atts.getValue(i)); // Single-string value int contentEnd=content.length(); if(!("xmlns".equals(prefix) || "xmlns".equals(qName))) m_dtm.appendAttribute(namespaceNames.stringToIndex(atts.getURI(i)), localNames.stringToIndex(localName), prefixes.stringToIndex(prefix), atts.getType(i).equalsIgnoreCase("ID"), contentStart, contentEnd-contentStart); contentStart=contentEnd; } } } public void startPrefixMapping(java.lang.String prefix, java.lang.String uri) throws org.xml.sax.SAXException { // No-op in DTM, handled during element/attr processing? } // // LexicalHandler support. Not all SAX2 parsers support these events // but we may want to pass them through when they exist... // public void comment(char[] ch, int start, int length) throws org.xml.sax.SAXException { processAccumulatedText(); content.append(ch,start,length); // Single-string value m_dtm.appendComment(contentStart,length); contentStart+=length; } public void endCDATA() throws org.xml.sax.SAXException { // No-op in DTM } public void endDTD() throws org.xml.sax.SAXException { // No-op in DTM } public void endEntity(java.lang.String name) throws org.xml.sax.SAXException { // No-op in DTM } public void startCDATA() throws org.xml.sax.SAXException { // No-op in DTM } public void startDTD(java.lang.String name, java.lang.String publicId, java.lang.String systemId) throws org.xml.sax.SAXException { // No-op in DTM } public void startEntity(java.lang.String name) throws org.xml.sax.SAXException { // No-op in DTM } // // coroutine support // public int getAppCoroutine() { return fAppCoroutine; } /** * getMoreNodes() tells the coroutine parser that we need more nodes. * * Parameter may be: * * null terminate the parser coroutine. * * Boolean.TRUE indication to continue parsing the current document. * Resumes normal SAX parsing. * * Boolean.FALSE indication to discontinue parsing and reset. * Throws UserRequestedStopException * to return control to the run() loop. * * We will be resumed with * co_resume(Boolean.TRUE, ...) on success with more remaining to parse. * co_resume(Boolean.FALSE, ...) on success after endDocument. */ private boolean co_yield(boolean getmore, boolean shutdown) { Object arg=null; if(!shutdown) arg= getmore ? Boolean.TRUE : Boolean.FALSE; try { arg = fCoroutineManager.co_resume(arg,fAppCoroutine,fParserCoroutine); if (arg instanceof Boolean) { return ((Boolean)arg).booleanValue(); } else // Unexpected! { System.err.println( "Active CoroutineSAXParser: unexpected resume parameter, " +arg.getClass ()+" with value=\""+arg+'"'); System.err.println("\tStopping parser rather than risk deadlock"); throw new RuntimeException("Coroutine parameter error ("+arg+')'); } } catch(java.lang.NoSuchMethodException e) { // Shouldn't happen unless we've miscoded our coroutine logic // "Shut down the garbage smashers on the detention level!" e.printStackTrace(System.err); fCoroutineManager.co_exit(fAppCoroutine); } // Only reached if NoSuchMethodException was thrown (no coroutine) return(false); } } // class DTMBuilder