DOM with Lua DataModel and dropped V8

4 files changed, 172 insertions, 158 deletions

diff --git a/src/uscxml/interpreter/BasicContentExecutor.cpp b/src/uscxml/interpreter/BasicContentExecutor.cpp
index 5344957..9ea2b6a 100644
--- a/src/uscxml/interpreter/BasicContentExecutor.cpp
+++ b/src/uscxml/interpreter/BasicContentExecutor.cpp
@@ -593,155 +593,89 @@ void BasicContentExecutor::processParams(std::multimap<std::string, Data>& param
 		paramMap.insert(make_pair(name, d));
 	}
 }
-
+    
 Data BasicContentExecutor::elementAsData(XERCESC_NS::DOMElement* element, bool asExpression) {
-	if (HAS_ATTR(element, kXMLCharExpr)) {
-//        return _callbacks->evalAsData(ATTR(element, "expr"));
-#if 0
-		if (LOCALNAME(element) == "content") {
-			// test 528
-			return _callbacks->evalAsData(ATTR(element, kXMLCharExpr));
-		} else {
-			// test 326
-			return Data(ATTR(element, kXMLCharExpr), Data::INTERPRETED);
-		}
-#endif
-		if (asExpression) // test 453
-			return Data(ATTR(element, kXMLCharExpr), Data::INTERPRETED);
-		return _callbacks->evalAsData(ATTR(element, kXMLCharExpr));
-	}
-
-	if (HAS_ATTR(element, kXMLCharSource)) {
-		// remote content from URL
-
-		// test 446, test 552, test 558
-		std::string src = ATTR(element, kXMLCharSource);
-		URL url(ATTR(element, kXMLCharSource));
-		if (!url.isAbsolute()) {
-			url = URL::resolve(url, _callbacks->getBaseURL());
-		}
-
-		std::string content = url.getInContent();
-
-		// make an attempt to parse as XML
-		try {
-#if 0
-			XERCESC_NS::XercesDOMParser parser;
-			parser.setValidationScheme(XERCESC_NS::XercesDOMParser::Val_Never);
-			parser.setDoNamespaces(true);
-			parser.useScanner(XERCESC_NS::XMLUni::fgWFXMLScanner);
-
-			XERCESC_NS::HandlerBase errHandler;
-			parser.setErrorHandler(&errHandler);
-
-			std::string tmp = url;
-			XERCESC_NS::MemBufInputSource is((XMLByte*)content.c_str(), content.size(), X("fake"));
-
-			parser.parse(is);
-
-			Data d;
-			XERCESC_NS::DOMDocument* doc = parser.adoptDocument();
-			d.adoptedDoc = std::shared_ptr<XERCESC_NS::DOMDocument>(doc);
-			d.node = doc->getDocumentElement();
-			return d;
-#else
-			std::unique_ptr<XERCESC_NS::XercesDOMParser> parser(new XERCESC_NS::XercesDOMParser());
-			parser->setValidationScheme(XERCESC_NS::XercesDOMParser::Val_Always);
-			parser->setDoNamespaces(true);
-			parser->useScanner(XERCESC_NS::XMLUni::fgWFXMLScanner);
-
-			std::unique_ptr<XERCESC_NS::ErrorHandler> errHandler(new XERCESC_NS::HandlerBase());
-			parser->setErrorHandler(errHandler.get());
-
-			try {
-				std::string tmp = url;
-				parser->parse(tmp.c_str());
-
-				XERCESC_NS::DOMNode* newNode = element->getOwnerDocument()->importNode(parser->getDocument()->getDocumentElement(), true);
-
-				// remove any old child elements
-				while(element->getFirstElementChild() != NULL) {
-					element->removeChild(element->getFirstElementChild());
-				}
-				// we need to save the DOM somewhere .. Data::adoptedDoc was not good enough
-				element->appendChild(newNode);
-
-				Data d;
-//                d.adoptedDoc = std::shared_ptr<XERCESC_NS::DOMDocument>(parser->adoptDocument());
-				d.node = newNode;
-				return d;
-			}
-
-			catch (const XERCESC_NS::SAXParseException& toCatch) {
-				ERROR_PLATFORM_THROW(X(toCatch.getMessage()).str());
-			} catch (const XERCESC_NS::RuntimeException& toCatch) {
-				ERROR_PLATFORM_THROW(X(toCatch.getMessage()).str());
-			} catch (const XERCESC_NS::XMLException& toCatch) {
-				ERROR_PLATFORM_THROW(X(toCatch.getMessage()).str());
-			} catch (const XERCESC_NS::DOMException& toCatch) {
-				ERROR_PLATFORM_THROW(X(toCatch.getMessage()).str());
-			}
-
-#endif
-
-		} catch (...) {
-			// just ignore and return as an interpreted string below
-		}
-		try {
-			Data d = _callbacks->getAsData(content);
-			if (!d.empty())
-				return d;
-		} catch(...) {}
-
-		return Data(spaceNormalize(content), Data::VERBATIM);
-
-	} else {
-		// local content in document
-
-		std::list<DOMNode*> elementChildren = DOMUtils::filterChildType(DOMNode::ELEMENT_NODE, element);
-		if (elementChildren.size() > 0) {
-			// always return parent element, even with a single child node
-			return Data(static_cast<DOMNode*>(element));
-		}
-
-		std::list<DOMNode*> textChildren = DOMUtils::filterChildType(DOMNode::TEXT_NODE, element);
-		if (textChildren.size() > 0) {
-			std::stringstream contentSS;
-			for (auto textIter = textChildren.begin(); textIter != textChildren.end(); textIter++) {
-				contentSS << X((*textIter)->getNodeValue());
-			}
-
-			// this must be handled in getAsData
-			// test294, test562
-			if (LOCALNAME(element) == "content") {
-				// need first try getAsData because how to pass 179 ?
-				try {
-					// test153, we need to throw for test150 in promela
-					Data d = _callbacks->getAsData(contentSS.str());
-					if (!d.empty())
-						return d;
-				} catch (ErrorEvent &) {
-					return Data(spaceNormalize(contentSS.str()), Data::VERBATIM);
-				}
-			}
-
-			if (asExpression) // not actually used, but likely expected
-				return Data(contentSS.str(), Data::INTERPRETED);
-
-			// test153, we need to throw for test150 in promela, but we need to space normalize for test558
-			try {
-				Data d = _callbacks->getAsData(contentSS.str());
-				if (!d.empty())
-					return d;
-			} catch ( ... ) {}
-
-			// test558
-			return Data(spaceNormalize(contentSS.str()), Data::VERBATIM);
-		}
-	}
-
-//	LOG(USCXML_WARN) << "Element " << DOMUtils::xPathForNode(element) << " did not yield any data";
-	return Data();
+    // element with expr
+    if (HAS_ATTR(element, kXMLCharExpr)) {
+        std::string expr = ATTR(element, kXMLCharExpr);
+        if (_callbacks->isLegalDataValue(expr)) {
+            return Data(expr, Data::INTERPRETED);
+        } else {
+            ERROR_EXECUTION_THROW2("Expression '" + expr + "' is not a legal data value", element);
+        }
+    }
+    
+    // element with external src - this ought to behave just as with child nodes below
+    if (HAS_ATTR(element, kXMLCharSource)) {
+
+        // remove any old child elements
+        while(element->getFirstElementChild() != NULL) {
+            element->removeChild(element->getFirstElementChild());
+        }
+
+        std::string src = ATTR(element, kXMLCharSource);
+        URL url(ATTR(element, kXMLCharSource));
+        if (!url.isAbsolute()) {
+            url = URL::resolve(url, _callbacks->getBaseURL());
+        }
+        std::string content = url.getInContent();
+
+        // append as XML?
+        try {
+            std::unique_ptr<XERCESC_NS::XercesDOMParser> parser(new XERCESC_NS::XercesDOMParser());
+            parser->setValidationScheme(XERCESC_NS::XercesDOMParser::Val_Always);
+            parser->setDoNamespaces(true);
+            parser->useScanner(XERCESC_NS::XMLUni::fgWFXMLScanner);
+            
+            std::unique_ptr<XERCESC_NS::ErrorHandler> errHandler(new XERCESC_NS::HandlerBase());
+            parser->setErrorHandler(errHandler.get());
+
+            XERCESC_NS::MemBufInputSource is((XMLByte*)content.c_str(), content.size(), X("fake"));
+            is.setPublicId(X(url));
+            
+            parser->parse(is);
+            XERCESC_NS::DOMNode* newNode = element->getOwnerDocument()->importNode(parser->getDocument()->getDocumentElement(), true);
+            
+            // we need to save the DOM somewhere .. Data::adoptedDoc was not good enough
+            element->appendChild(newNode);
+            goto SOURCE_APPEND_DONE;
+        } catch (...) {
+        }
+
+        // append as text (are we leaking?)
+        XERCESC_NS::DOMText* textNode = element->getOwnerDocument()->createTextNode(X(content));
+        element->appendChild(textNode);
+    }
+SOURCE_APPEND_DONE:
+
+    if (element->hasChildNodes()) {
+        // XML elements e.g. for content with invoke
+        std::list<DOMNode*> elementChildren = DOMUtils::filterChildType(DOMNode::ELEMENT_NODE, element);
+        if (elementChildren.size() > 0) {
+            // always return parent element, even with a single child node
+            return Data(static_cast<DOMNode*>(element));
+        }
+
+        // expression in text element
+        std::list<DOMNode*> textChildren = DOMUtils::filterChildType(DOMNode::TEXT_NODE, element);
+        if (textChildren.size() > 0) {
+            std::stringstream contentSS;
+            for (auto textIter = textChildren.begin(); textIter != textChildren.end(); textIter++) {
+                contentSS << X((*textIter)->getNodeValue());
+            }
+
+            try {
+                Data d = _callbacks->getAsData(contentSS.str());
+                if (!d.empty())
+                    return d;
+            } catch(...) {}
+
+            // anything else is considered verbatim - space normalize?
+            return Data(spaceNormalize(contentSS.str()), Data::VERBATIM);
+        }
+    }
+    
+    return Data();
 }
 
 }
diff --git a/src/uscxml/interpreter/ContentExecutorImpl.h b/src/uscxml/interpreter/ContentExecutorImpl.h
index c85d9a0..0c080ca 100644
--- a/src/uscxml/interpreter/ContentExecutorImpl.h
+++ b/src/uscxml/interpreter/ContentExecutorImpl.h
@@ -62,6 +62,7 @@ public:
 	virtual Data evalAsData(const std::string& expr) = 0;
 	virtual void eval(const std::string& expr) = 0;
 	virtual Data getAsData(const std::string& expr) = 0;
+    virtual bool isLegalDataValue(const std::string& expr) = 0;
 	virtual void assign(const std::string& location, const Data& data, const std::map<std::string, std::string>& attrs) = 0;
 
 
diff --git a/src/uscxml/interpreter/FastMicroStep.cpp b/src/uscxml/interpreter/FastMicroStep.cpp
index ab6b5f9..1258279 100644
--- a/src/uscxml/interpreter/FastMicroStep.cpp
+++ b/src/uscxml/interpreter/FastMicroStep.cpp
@@ -509,6 +509,9 @@ COMPLETION_STABLISHED:
 	_transitions.resize(tmp.size());
 	_exitSets.resize(tmp.size());
 
+//    std::list<Transition*> transSpontaneous;
+//    std::list<Transition*> transEventful;
+    
 	for (i = 0; i < _transitions.size(); i++) {
 		_transitions[i] = new Transition(i);
 		_transitions[i]->element = tmp.front();
@@ -579,7 +582,10 @@ TARGET_SET_ESTABLISHED:
 
 		if (!HAS_ATTR(_transitions[i]->element, kXMLCharEvent)) {
 			_transitions[i]->type |= USCXML_TRANS_SPONTANEOUS;
-		}
+//            transSpontaneous.push_back(_transitions[i]);
+        } else {
+//            transEventful.push_back(_transitions[i]);
+        }
 
 		if (iequals(TAGNAME_CAST(_transitions[i]->element->getParentNode()), _xmlPrefix.str() + "history")) {
 			_transitions[i]->type |= USCXML_TRANS_HISTORY;
@@ -606,14 +612,88 @@ TARGET_SET_ESTABLISHED:
 		{
 			_exitSets[i] = getExitSet(_transitions[i]);
 		}
-
 	}
 
 	/**
 	 * This bound by cache locality!
 	 * Before you change anything, do benchmark!
 	 */
-
+    
+#if 0
+#define exit1 _exitSets[t1->postFixOrder]
+#define exit2 _exitSets[t2->postFixOrder]
+
+    for (size_t set = 0; set < 2; set++) {
+        std::list<Transition*>::const_iterator transBegin;
+        std::list<Transition*>::const_iterator transEnd;
+        if (set == 0) {
+            transBegin = transEventful.begin();
+            transEnd = transEventful.end();
+        } else {
+            transBegin = transSpontaneous.begin();
+            transEnd = transSpontaneous.end();
+        }
+
+        for (auto t1Iter = transBegin; t1Iter != transEnd; t1Iter++) {
+            Transition* t1 = *t1Iter;
+            auto anc1 = _states[t1->source]->ancestors;
+            const uint32_t source1 = t1->source;
+
+            for (auto t2Iter = t1Iter; t2Iter != transEnd; t2Iter++) {
+                Transition* t2 = *t2Iter;
+                
+                if (exit1.first == 0 && exit2.first == 0) {
+                    goto COMPATIBLE_TRANS;
+                }
+                
+                if (exit1.first <= exit2.first && exit1.second >= exit2.first) {
+                    goto CONFLICTING_TRANS;
+                }
+                
+                if (exit2.first <= exit1.first && exit2.second >= exit1.first) {
+                    goto CONFLICTING_TRANS;
+                }
+                
+            COMPATIBLE_TRANS:
+                if (t2->source == source1) {
+                    goto CONFLICTING_TRANS;
+                }
+                
+                if (anc1[t2->source]) {
+                    goto CONFLICTING_TRANS;
+                }
+                
+                if (_states[t2->source]->ancestors[source1]) {
+                    goto CONFLICTING_TRANS;
+                }
+                
+                t1->conflicts[t2->postFixOrder] = false;
+                //            _transitions[j]->conflicts[i] = false;
+                continue;
+                
+            CONFLICTING_TRANS:
+                t1->conflicts[t2->postFixOrder] = true;
+                //            _transitions[j]->conflicts[i] = true;
+                
+                continue;
+
+            }
+        }
+        
+        for (auto t1Iter = transBegin; t1Iter != transEnd; t1Iter++) {
+            Transition* t1 = *t1Iter;
+            // conflicts matrix is symmetric
+            for (auto t2Iter = transBegin; t2Iter != t1Iter; t2Iter++) {
+                Transition* t2 = *t2Iter;
+
+                t1->conflicts[t2->postFixOrder] = t2->conflicts[t1->postFixOrder];
+            }
+        }
+
+    }
+    
+    
+#else
 #define anc1 _states[_transitions[i]->source]->ancestors
 #define exit1 _exitSets[i]
 #define exit2 _exitSets[j]
@@ -659,12 +739,6 @@ CONFLICTING_TRANS:
 			continue;
 
 		}
-#if 0 // moved down for cache locality
-		// conflicts matrix is symmetric
-		for (j = 0; j < i; j++) {
-			_transitions[i]->conflicts[j] = _transitions[j]->conflicts[i];
-		}
-#endif
 	}
 
 	for (i = 0; i < _transitions.size(); i++) {
@@ -673,7 +747,8 @@ CONFLICTING_TRANS:
 			_transitions[i]->conflicts[j] = _transitions[j]->conflicts[i];
 		}
 	}
-
+#endif
+    
 	// initialize bitarrays for step()
 	_exitSet = boost::dynamic_bitset<BITSET_BLOCKTYPE>(_states.size(), false);
 	_entrySet = boost::dynamic_bitset<BITSET_BLOCKTYPE>(_states.size(), false);
diff --git a/src/uscxml/interpreter/InterpreterImpl.h b/src/uscxml/interpreter/InterpreterImpl.h
index 2f5fb07..1f87fd2 100644
--- a/src/uscxml/interpreter/InterpreterImpl.h
+++ b/src/uscxml/interpreter/InterpreterImpl.h
@@ -196,6 +196,10 @@ public:
 		return _dataModel.getAsData(expr);
 	}
 
+    virtual bool isLegalDataValue(const std::string& expr) {
+        return _dataModel.isLegalDataValue(expr);
+    }
+
 	virtual void assign(const std::string& location, const Data& data, const std::map<std::string, std::string>& attrs);
 
 	virtual std::string getInvokeId() {