Unicode should now work properly. See #37

4 files changed, 10 insertions, 10 deletions

diff --git a/src/rtfparser.cpp b/src/rtfparser.cpp
index 31e5971..7e70fe5 100644
--- a/src/rtfparser.cpp
+++ b/src/rtfparser.cpp
@@ -363,6 +363,10 @@ bool RtfParser::parseControlWord()
 		else if(ch >= '0' && ch <= '9')
 			param.append(1, (char)ch);
 
+		// Hyphens are part of the parameter of a control word
+		else if(ch == '-' && !controlword.empty())
+			param.append(1, (char)ch);
+
 		// Now handle escapes and other special types of
 		// control words. These are all only valid at beginning
 		// of the "control word"
@@ -475,16 +479,11 @@ bool RtfParser::parseControlWord()
 		else if(m_parseUnicode && flags & RtfHandler::kHasParam &&
 			    controlword == "u" )
 		{
-			sendData(numPar);
+			// RTF plays hokey and uses negative values in unicode
+			sendData((unsigned short)((short)numPar));
 			m_uniEat = m_uniEatStack.top();
 		}
 
-		// Unicode destination
-		else if(m_parseUnicode && controlword == "ud")
-		{
-
-		}
-
 		// Skip value for unicode characters
 		else if(m_parseUnicode && controlword == "uc")
 		{
diff --git a/src/rtfparser.h b/src/rtfparser.h
index fe2d415..9f0289e 100644
--- a/src/rtfparser.h
+++ b/src/rtfparser.h
@@ -71,7 +71,8 @@ public:
 		{ return m_depth; }
 	void setHexParse(bool parse)
 		{ m_parseHex = parse; }
-	void setUnicode(bool unicode);
+	void setUnicode(bool unicode)
+		{ m_parseUnicode = unicode; }
 
 protected:
 	RtfHandler* m_handler;         // The current handler
@@ -79,7 +80,6 @@ protected:
 	bool m_parseHex;               // Whether to parse hex chars or not
 	string m_parseErrors;          // A list of all the RTF parsing errors
 
-    // TODO: Look at exactly what this is doing
     // Unicode char handling
 	bool m_parseUnicode;
 	typedef std::stack<int> StackInt;
diff --git a/src/rtfx.cpp b/src/rtfx.cpp
index 5428148..7f35f7c 100644
--- a/src/rtfx.cpp
+++ b/src/rtfx.cpp
@@ -145,6 +145,7 @@ int main(int argc, char* argv[])
 
         // Reads RTF tags and blocks
         RtfParser rtf;
+		rtf.setUnicode(true);
 
         // Interprets tags and blocks from RTFParser
 		XmlComposer composer(options);
diff --git a/src/xmlcomposer.cpp b/src/xmlcomposer.cpp
index f94ce29..93ae306 100644
--- a/src/xmlcomposer.cpp
+++ b/src/xmlcomposer.cpp
@@ -1129,7 +1129,7 @@ ON_CONTROLWORD(Root)
 	}
 	else if(cw == "field")
 		AN_ANALYSER(Field);
-	else if(flags & kAsterisk)
+	else if(flags & kAsterisk && cw != "ud")
 		AN_ANALYSER(Skip);
 	else if(processTextContent(cw, flags, param))
 		DUMMY;