Enabled XML Schema-based insertion helpers and validation-as-you-type.

src/xmlpromptgenerator.cpp

@@ -29,6 +29,20 @@
 #include "catalogresolver.h"
 #include "xmlschemaparser.h"
 
+// Xerces-C req'd for Schema parsing
+#define XERCES_TMPLSINC
+
+#include <xercesc/util/NameIdPool.hpp>
+#include <xercesc/util/PlatformUtils.hpp>
+#include <xercesc/framework/XMLValidator.hpp>
+#include <xercesc/parsers/SAXParser.hpp>
+#include <xercesc/parsers/XercesDOMParser.hpp>
+#include <xercesc/validators/schema/SchemaValidator.hpp>
+#include <xercesc/validators/common/ContentSpecNode.hpp>
+#include <xercesc/validators/schema/SchemaSymbols.hpp>
+
+using namespace xercesc;
+
 XmlPromptGenerator::XmlPromptGenerator (
     const std::string& catalogPath,
     const std::string& basePath,
@@ -39,8 +53,9 @@ XmlPromptGenerator::XmlPromptGenerator (
 	d->catalogPath = catalogPath;
 	d->auxPath = auxPath;
 	d->elementDeclRecurseLevel = 0;
-    d->rootElement = true;
+	d->isRootElement = true;
-    d->dtdFound = false;
+	d->grammarFound = false;
+	d->attributeValueCutoff = 12; // this prevents enums being stored in their thousands
 	XML_SetParamEntityParsing ( p, XML_PARAM_ENTITY_PARSING_UNLESS_STANDALONE );
 	XML_SetElementHandler ( p, starthandler, endhandler );
 	XML_SetDoctypeDeclHandler ( p, doctypedeclstarthandler, doctypedeclendhandler );
@@ -66,6 +81,19 @@ void XMLCALL XmlPromptGenerator::starthandler (
 	PromptGeneratorData *d;
 	d = ( PromptGeneratorData * ) data;
 
+	if (d->isRootElement)
+	{
+		d->rootElement = el;
+		handleSchema ( d, el, attr );
+		d->isRootElement = false;
+		if ( ! (d->elementMap.empty() )  )//if ( d->elementMap.size() == 1) // must be 1 for success
+		{
+			d->grammarFound = true;
+			XML_StopParser ( d->p, false );
+			return;
+		}
+	}
+
 	d->push ( el );
 
 	std::string parent, element;
@@ -97,25 +125,9 @@ void XMLCALL XmlPromptGenerator::starthandler (
 		attributeName = *attr;
 		attributeValue = * ( attr + 1 );
 
+		if (d->attributeMap[element][attributeName].size() < d->attributeValueCutoff)
 			d->attributeMap[element][attributeName].insert ( attributeValue );
-
-        /*
-            // TBD: may not be xsi: check for http://www.w3.org/2001/XMLSchema-instance
-            while (d->rootElement && strstr((const char *)attr, "xsi:noNamespaceSchemaLocation"))
-            {
-              std::string schemaPath = PathResolver::run(attributeValue, d->auxPath);
-              std::string buffer;
-              if (!ReadFile::run(schemaPath, buffer))
-                break;
-              XmlSchemaParser xsp(d, true);
-              if (!xsp.parse(buffer))
-                break;
-              XML_StopParser(d->p, false);
-              return;
 	}
-        */
-    }
-    d->rootElement = false;
 }
 
 void XMLCALL XmlPromptGenerator::endhandler ( void *data, const XML_Char *el )
@@ -125,9 +137,9 @@ void XMLCALL XmlPromptGenerator::endhandler ( void *data, const XML_Char *el )
 	d->pop();
 }
 
-bool XmlPromptGenerator::getDtdFound()
+bool XmlPromptGenerator::getGrammarFound()
 {
-    return d->dtdFound;
+	return d->grammarFound;
 }
 
 void XmlPromptGenerator::getAttributeMap (
@@ -174,8 +186,8 @@ void XMLCALL XmlPromptGenerator::doctypedeclendhandler ( void *data )
 	d = ( PromptGeneratorData * ) data;
 	if ( !d->elementMap.empty() )
 	{
-        d->dtdFound = true;
+		d->grammarFound = true;
-        XML_StopParser ( d->p, false );
+		XML_StopParser ( d->p, false ); // experimental
 	}
 }
 
@@ -287,12 +299,16 @@ int XMLCALL XmlPromptGenerator::externalentityrefhandler (
 		stdPublicId = publicId;
 
 	std::string stdSystemId = CatalogResolver::lookupPublicId ( stdPublicId, d->catalogPath );
-    if ( stdSystemId.empty() && systemId )
-        stdSystemId = systemId;
 	
-    Replace::run ( stdSystemId, "file:///", "", false );
+	if ( !stdSystemId.empty() )
+	{
+		Replace::run ( stdSystemId, "file://", "", false );
 		Replace::run ( stdSystemId, "%20", " ", false );
-
+	}
+	else
+	{
+		if (systemId )
+			stdSystemId = systemId;
 		if ( base )
 		{
 			std::string test = PathResolver::run ( stdSystemId, base );
@@ -301,6 +317,7 @@ int XMLCALL XmlPromptGenerator::externalentityrefhandler (
 				stdSystemId = test;
 			}
 		}
+	}
 		
 	if ( !stdSystemId.empty() )
 	{
@@ -308,7 +325,7 @@ int XMLCALL XmlPromptGenerator::externalentityrefhandler (
 	}
 
 	std::string encoding = XmlEncodingHandler::get ( buffer );
-    XML_Parser dtdParser = XML_ExternalEntityParserCreate ( d->p, context, encoding.c_str() );//"UTF-8");
+	XML_Parser dtdParser = XML_ExternalEntityParserCreate ( d->p, context, encoding.c_str() );
 	if ( !dtdParser )
 		return false;
 
@@ -345,3 +362,145 @@ void XMLCALL XmlPromptGenerator::entitydeclhandler (
 		d->entitySet.insert ( entityName );
 	}
 }
+
+void XmlPromptGenerator::handleSchema (
+    PromptGeneratorData *d,
+    const XML_Char *el,
+    const XML_Char **attr )
+{
+	// first check for XML Schema association
+	XML_Char **schemaAttr = ( XML_Char ** ) attr; // now redundant; could use attr
+	std::string path;
+	for ( ; d->isRootElement && *schemaAttr; schemaAttr += 2 )
+	{
+		// no namespace
+		if ( !strcmp ( ( const char * ) *schemaAttr, "xsi:noNamespaceSchemaLocation" ) )
+		{
+			path = ( const char * ) * ( schemaAttr + 1 );
+			break;
+		}
+		// with namespace -- check if this works
+		else if ( !strcmp ( ( const char * ) *schemaAttr, "xsi:schemaLocation" ) )
+		{
+			char *searchIterator;
+			for ( searchIterator = ( char * ) * ( schemaAttr + 1 ); *searchIterator && *searchIterator != ' ' && *searchIterator != '\t' && *searchIterator != '\n'; searchIterator++ )
+				;
+			if ( *searchIterator )
+			{
+				path = ( const char * ) ( searchIterator + 1 );
+				break;
+			}
+		}
+	}
+
+	if ( path.empty() )
+	{
+		return;
+	}
+	std::string schemaPath = PathResolver::run ( path, d->auxPath );
+
+	try
+	{
+		XMLPlatformUtils::Initialize();
+	}
+	catch ( const XMLException& toCatch )
+	{
+		XMLPlatformUtils::Terminate();
+		return;
+	}
+
+	XercesDOMParser *parser = new XercesDOMParser();
+	parser->setDoNamespaces ( true );
+	parser->setDoSchema ( true );
+	parser->setValidationSchemaFullChecking ( true );
+
+	Grammar *rootGrammar = parser->loadGrammar ( schemaPath.c_str(), Grammar::SchemaGrammarType );
+	if ( !rootGrammar )
+	{
+		delete parser;
+		return;
+	}
+
+	SchemaGrammar* grammar = ( SchemaGrammar* ) rootGrammar;
+	RefHash3KeysIdPoolEnumerator<SchemaElementDecl> elemEnum = grammar->getElemEnumerator();
+
+	if ( !elemEnum.hasMoreElements() )
+	{
+		delete grammar;
+		delete parser;
+		
+		return;
+	}
+
+	while ( elemEnum.hasMoreElements() )
+	{
+		const SchemaElementDecl& curElem = elemEnum.nextElement();
+
+		std::string element;
+		std::set<std::string> children;
+		
+		const QName *qnm = curElem.getElementName();
+		if ( qnm )
+		{
+			element = XMLString::transcode ( qnm->getRawName() ); // this includes any prefix:localname combinations
+		}
+		if ( element.empty() )
+			continue;
+
+		const XMLCh* fmtCntModel = curElem.getFormattedContentModel();
+		if ( fmtCntModel != NULL ) // tbd: this does not yet pick up prefix:localname combinations
+		{
+			size_t len;
+			char *s, *word;
+			s = ( char * ) XMLString::transcode ( fmtCntModel );
+	
+			while ( ( word = GetWord::run ( &s, &len ) ) != NULL )
+			{
+				std::string currentValue ( word, len );
+				if ( currentValue.size() )
+					children.insert ( currentValue );
+			}
+		}
+		if ( !children.empty() )
+			d->elementMap.insert ( make_pair ( element, children ) );
+		
+		// fetch attributes
+		if ( curElem.hasAttDefs() && ! ( curElem.getAttDefList().isEmpty() ) )
+		{
+			std::map<std::string, std::set<std::string> > attributeMap;
+
+			XMLAttDefList& attIter = curElem.getAttDefList();
+			for ( unsigned int i = 0; i < attIter.getAttDefCount(); i++ )
+			{
+				std::string attribute, attributeValue;
+				std::set<std::string> attributeValueSet;
+
+				XMLAttDef& attr = attIter.getAttDef ( i );
+				XMLAttDef::DefAttTypes ty = attr.getDefaultType();
+				if ( ty == XMLAttDef::Prohibited )
+					continue;
+				SchemaAttDef *pAttr = ( SchemaAttDef * ) &attr;
+
+				const QName *qnm = pAttr->getAttName();
+				if ( qnm )
+				{
+					attribute = XMLString::transcode ( qnm->getRawName() );
+				}
+				if ( attribute.empty() )
+					continue;
+
+				// Value
+				if ( pAttr->getValue() )
+				{
+					attributeValue = XMLString::transcode ( pAttr->getValue() );
+					attributeValueSet.insert ( attributeValue );
+					attributeMap.insert ( make_pair ( attribute, attributeValueSet ) );
+				}
+			}
+			if ( !attributeMap.empty() )
+				d->attributeMap.insert( make_pair ( element, attributeMap ) );
+		}
+	}
+}
+
+

src/xmlpromptgenerator.h

@@ -33,9 +33,10 @@ struct PromptGeneratorData : public ParserData
 	std::map<std::string, std::set<std::string> > elementMap;
 	std::map<std::string, std::set<std::string> > requiredAttributeMap;
 	std::set<std::string> entitySet;
-    std::string catalogPath, auxPath;
+	std::string catalogPath, auxPath, rootElement;
 	int elementDeclRecurseLevel;
-    bool rootElement, dtdFound;
+	bool isRootElement, grammarFound;
+	unsigned attributeValueCutoff;
 	XML_Parser p;
 };
 
@@ -56,7 +57,7 @@ public:
 		    std::map<std::string, std::set<std::string> > &elementMap );
 		void getEntitySet (
 		    std::set<std::string> &entitySet );
-    bool getDtdFound();
+		bool getGrammarFound();
 	private:
 		std::auto_ptr<PromptGeneratorData> d;
 		static void XMLCALL starthandler (
@@ -100,6 +101,10 @@ private:
 		    const XML_Char *systemId,
 		    const XML_Char *publicId,
 		    const XML_Char *notationName );
+		static void handleSchema (
+		    PromptGeneratorData *d,
+		    const XML_Char *el,
+		    const XML_Char **attr );
 };
 
 #endif