Fix umlauts in cpp reader

.gitignore

@@ -54,3 +54,5 @@ cmake-build*
 *.swp
 
 .cache
+
+src/version.hpp

CMakeLists.txt

@@ -31,6 +31,8 @@ set(libcimpp_MINOR_VERSION 2)
 set(libcimpp_PATCH_VERSION 0)
 set(libcimpp_VERSION ${libcimpp_MAJOR_VERSION}.${libcimpp_MINOR_VERSION}.${libcimpp_PATCH_VERSION})
 
+configure_file(${CMAKE_CURRENT_SOURCE_DIR}/src/version.hpp.in ${CMAKE_CURRENT_SOURCE_DIR}/src/version.hpp)
+
 # Options
 option(BUILD_SHARED_LIBS "Build shared library" OFF)
 option(CIMPP_BUILD_EXAMPLES "Build examples" OFF)

examples/src/main.cpp

@@ -1,26 +1,24 @@
 #include <iostream>
 #include <string>
 #include "CIMModel.hpp"
+#include "CimConstants.hpp"
 #include "BaseClass.hpp"
 #include "IdentifiedObject.hpp"
+#include "version.hpp"
 
-std::string formatName(std::string name) {
-	if (name.length() > 12) {
-		name.resize(10, ' ');
-		name += "..  ";
-	}
-	else {
-		name.resize(14, ' ');
-	}
-	return name;
-}
+// forward declerations
+static std::string format_name(std::string name);
+static std::size_t utf8_code_point_len(const std::string& txt);
+static std::size_t utf8_code_point_len(const char* ptr);
+static void utf8_code_point_resize(std::string& txt, std::size_t new_len, char fill);
 
 int main(int argc, char** argv)
 {
 	if(argc < 2)
 	{
 		std::cout << "Usage: example <cim file> [additional cim files]" << std::endl;
 		std::cout << "Provide at least one cim file" << std::endl;
+		std::cout << "Version: libcimpp " << RELEASE_VERSION << " (" << CimVersion << ")" << std::endl;
 		return 1;
 	}
 
@@ -52,7 +50,7 @@ int main(int argc, char** argv)
 			if (IdObj->name.initialized)
 			{
 				static unsigned int i = 0;
-				std::string outputName = formatName(IdObj->name);
+				std::string outputName = format_name(IdObj->name);
 				std::cout << outputName;
 				i++;
 				if (i % 5 == 0) {
@@ -64,3 +62,103 @@ int main(int argc, char** argv)
 	std::cout << std::endl;
 	return 0;
 }
+
+static std::string format_name(std::string name)
+{
+	if (utf8_code_point_len(name) > 12)
+	{
+		utf8_code_point_resize(name, 10, ' ');
+		name += "..  ";
+	}
+	else
+	{
+		utf8_code_point_resize(name, 14, ' ');
+	}
+	return name;
+}
+
+static std::size_t utf8_code_point_len(const std::string& txt)
+{
+	return utf8_code_point_len(txt.c_str());
+}
+
+static std::size_t utf8_code_point_len(const char* ptr)
+{
+	size_t len = 0;
+	for (; *ptr != 0; ++ptr)
+	{
+		if ((*ptr & 0b11000000) != 0b10000000)
+		{
+			++len;
+		}
+	}
+	return len;
+}
+
+static void utf8_code_point_resize(std::string& txt, std::size_t new_len, char fill)
+{
+	const char* ptr = txt.c_str();
+	size_t len = 0;
+	for (; *ptr != 0 && len <= new_len; ++ptr)
+	{
+		if ((*ptr & 0b11000000) != 0b10000000)
+		{
+			++len;
+		}
+	}
+	if (len <= new_len)
+	{
+		new_len += txt.length() - len;
+	}
+	else
+	{
+		new_len = ptr - 1 - txt.c_str();
+	}
+	txt.resize(new_len, fill);
+}
+
+static bool utf8_check_valid(const char* ptr, std::size_t* p_len = nullptr)
+{
+	bool valid = true;
+	size_t len = 0;
+	for (; *ptr != 0; ++ptr)
+	{
+		if ((*ptr & 0b11000000) != 0b10000000)
+		{
+			++len;
+			size_t inner_code_point_len = 0;
+			if ((*ptr & 0b11100000) == 0b11000000)  // 110xxxxx
+			{
+				inner_code_point_len = 1;
+			}
+			if ((*ptr & 0b11110000) == 0b11100000)  // 1110xxxx
+			{
+				inner_code_point_len = 2;
+			}
+			if ((*ptr & 0b11111000) == 0b11110000)  // 11110xxx
+			{
+				inner_code_point_len = 3;
+			}
+			for (size_t idx = 0; idx < inner_code_point_len; ++idx, ++ptr)
+			{
+				if ((*ptr & 0b11000000) != 0b10000000)
+				{
+					valid = false;
+				}
+				if (*ptr == 0)
+				{
+					break;
+				}
+			}
+		}
+		else
+		{
+			valid = false;
+		}
+	}
+	if (p_len != nullptr)
+	{
+		*p_len = len;
+	}
+	return valid;
+}

examples/src/readwrite.cpp

@@ -3,13 +3,16 @@
 
 #include "CIMModel.hpp"
 #include "CIMWriter.hpp"
+#include "CimConstants.hpp"
+#include "version.hpp"
 
 int main(int argc, char** argv)
 {
 	if (argc < 3)
 	{
 		std::cout << "Usage: readwrite <cim file> [additional cim files] <outputfile>" << std::endl;
 		std::cout << "Provide at least one cim file and outputfile" << std::endl;
+		std::cout << "Version: libcimpp " << RELEASE_VERSION << " (" << CimVersion << ")" << std::endl;
 		return 1;
 	}
 	std::string outputfile = argv[argc-1];

src/CIMContentHandler.cpp

@@ -88,6 +88,7 @@ void CIMContentHandler::startElement(const std::string &namespaceURI, const std:
 
 	// Remember last opened tag
 	tagStack.push(qName);
+	value.clear();
 
 	// If there is no RDF ID (an XML attribute!) then we don't have a new CIM
 	// object or RDF relation therefore the XML element will contain a value
@@ -148,11 +149,16 @@ void CIMContentHandler::startElement(const std::string &namespaceURI, const std:
 void CIMContentHandler::endElement(const std::string &namespaceURI, const std::string &localName, const std::string &qName)
 {
 	// Only process tags in cim namespace
-	if(qName.find("cim:") == std::string::npos)
+	if (qName.find("cim:") == std::string::npos)
 	{
 		return;
 	}
 
+	if (!tagStack.empty() && !objectStack.empty() && !value.empty())
+	{
+		assign(objectStack.top(), tagStack.top(), value);
+	}
+
 	// Pop Stacks
 	if (tagStack.size() == 0) {
 		std::cerr << "WARNING: Nearly tried to pop empty tag stack for tag: " << qName << std::endl;
@@ -174,27 +180,7 @@ void CIMContentHandler::endElement(const std::string &namespaceURI, const std::s
 
 void CIMContentHandler::characters(const std::string &characters)
 {
-	// Only process tags in "cim" namespace
-	if(tagStack.empty())
-	{
-		return;
-	}
-	if(objectStack.empty())
-	{
-		throw CriticalError("CIMContentHandler: Critical Error: objectStack empty");
-	}
-
-#ifndef DEBUG
-	assign(objectStack.top(), tagStack.top(), characters);
-#else
-	// Check if the characters only contain whitespace
-	if(is_only_whitespace(characters))
-	{
-		return;
-	}
-	if(!assign(objectStack.top(), tagStack.top(), characters))
-		std::cout << "CIMContentHandler: Note: Cannot assign '" << characters << "' to " << tagStack.top() << std::endl;
-#endif
+	value.append(characters);
 }
 
 void CIMContentHandler::ignorableWhitespace(const std::string &ch)
@@ -244,7 +230,7 @@ std::string CIMContentHandler::get_rdf_enum(const AttributesT &attributes)
 	{
 		if(attributes.getQName(i) == "rdf:resource")
 		{
-			std::regex expr("^http[s]*://[a-zA-Z0-9./_]*CIM-schema-cim[0-9]+#([a-zA-z0-9]*).([a-zA-z0-9]*)");
+			std::regex expr("^http[s]*://.*#([a-zA-z0-9]*).([a-zA-z0-9]*)");
 			std::smatch m;
 			std::string str = attributes.getValue(i);
 			if(std::regex_match(str, m, expr))

src/CIMContentHandler.hpp

@@ -50,6 +50,7 @@ class CIMContentHandler : public Arabica::SAX::ContentHandler<std::string>
 	std::stack<BaseClass*> objectStack;
 	std::stack<std::string> tagStack;
 	std::list<Task> taskQueue;
+	std::string value;
 };
 
 

src/CIMWriter.cpp

@@ -97,7 +97,7 @@ bool CIMWriter::writeCim(std::ostream& rdf, const std::vector<BaseClass*>& objLi
           std::stringstream stream;
           if (func(obj, stream))
           {
-            rdfObj << "    <" << attr << ">" << stream.str() << "</" << attr << ">" << std::endl;
+            rdfObj << "    <" << attr << ">" << xmlEscape(stream.str()) << "</" << attr << ">" << std::endl;
             ++attributesCount;
           }
         }
@@ -232,3 +232,26 @@ CGMESProfile CIMWriter::getAttributeProfile(const BaseClass* obj, const std::str
   }
   return UnknownProfile;
 }
+
+std::string CIMWriter::xmlEscape(const std::string& txt)
+{
+  if (txt.find_first_of("&<>'\"") == std::string::npos)
+  {
+    return txt;
+  }
+  std::string result;
+  result.reserve(txt.size() + 10);
+  for (size_t pos = 0; pos != txt.size(); ++pos)
+  {
+    switch (txt[pos])
+    {
+      case '&':   result.append("&amp;");       break;
+      case '<':   result.append("&lt;");        break;
+      case '>':   result.append("&gt;");        break;
+      case '\'':  result.append("&apos;");      break;
+      case '"':   result.append("&quot;");      break;
+      default:    result.append(&txt[pos], 1);  break;
+    }
+  }
+  return result;
+}

src/CIMWriter.hpp

@@ -104,6 +104,9 @@ class CIMWriter
    */
   static CGMESProfile getAttributeProfile(const BaseClass* obj, const std::string& attr,
                                           const CGMESProfile& classProfile);
+
+private:
+  static std::string xmlEscape(const std::string& txt);
 };
 
 #endif // CIMWRITER_HPP

src/version.hpp.in

@@ -0,0 +1 @@
+#define RELEASE_VERSION "${libcimpp_VERSION}"