- Preserve mode

- Document options
- Font fixes

8 files changed, 143 insertions, 60 deletions

diff --git a/src/domhelpers.cpp b/src/domhelpers.cpp
index 038e2df..4836d17 100644
--- a/src/domhelpers.cpp
+++ b/src/domhelpers.cpp
@@ -191,20 +191,6 @@ DOM::Element DOMHelpers::findChildElement(const DOM::Node& parent, const string&
     return DOM::Element();
 }
 
-DOM::Element DOMHelpers::findDescendantElement(const DOM::Node& parent, const string& name)
-{
-    DOM::Node child = parent.getFirstChild();
-    while(child != NULL)
-    {
-        if(isElement(child, name))
-            return (DOM::Element&)child;
-
-		child = child.getNextSibling();
-    }
-
-    return DOM::Element();
-}
-
 bool DOMHelpers::hasAncestor(const DOM::Node& ancestor, const DOM::Node& node)
 {
     DOM::Node n = node;
diff --git a/src/rtfformatting.h b/src/rtfformatting.h
index f2d0581..aa5ac58 100644
--- a/src/rtfformatting.h
+++ b/src/rtfformatting.h
@@ -155,7 +155,7 @@ public:
     void textSetFont(int font)
         { m_font = font; }
     void textSetFontSize(int fsize)
-        { m_fsize = fsize; }
+        { m_fsize = fsize == 24 ? -1 : fsize; } // default font size is always 24
 	void paraSetStyle(int style)
 		{ m_style = style; }
 	void paraSetList(int list)
diff --git a/src/rtfx.cpp b/src/rtfx.cpp
index cbecdb9..541432e 100644
--- a/src/rtfx.cpp
+++ b/src/rtfx.cpp
@@ -46,23 +46,56 @@
 
 int usage()
 {
-	fprintf(stderr, "usage: rtfx <inrtf> <outxml>\n");
-	return 2;
+	fprintf(stderr, "usage: rtfx [-p] <inrtf> <outxml>\n");
+    exit(2);
 }
 
 int main(int argc, char* argv[])
 {
-	if(argc < 3)
-		return usage();
+    XmlComposerOptions options;
+
+    while(argc > 1)
+    {
+        argc--;
+        argv++;
+
+        char* arg = argv[0];
+        if(*arg != '-')
+            break;
+
+        while(arg && *(++arg))
+        {
+            switch(*arg)
+            {
+            case 'p':
+                options.extras = true;
+                break;
+
+            case '-':
+                arg = NULL;
+                break;
+
+            case '?':
+            default:
+                usage();
+            }
+        }
+
+        if(arg == NULL)
+            break;
+    }
+
+	if(argc < 2)
+		usage();
 
 	try
 	{
         // The input file
-		FILE* file = fopen(argv[1], "rb");
+		FILE* file = fopen(argv[0], "rb");
 
 		if(!file)
 		{
-			fprintf(stderr, "rtfx: couldn't open rtf file: %s: %s\n", argv[1], strerror(errno));
+			fprintf(stderr, "rtfx: couldn't open rtf file: %s: %s\n", argv[0], strerror(errno));
 			return 1;
 		}
 
@@ -70,7 +103,6 @@ int main(int argc, char* argv[])
         RtfParser rtf;
 
         // Interprets tags and blocks from RTFParser
-        XmlComposerOptions options;
 		XmlComposer composer(options);
 		rtf.setHandler(&composer);
 
@@ -88,10 +120,10 @@ int main(int argc, char* argv[])
 		DOM::Document doc = composer.getDocument();
 		string xml = doc.serialize();
 
-		FILE* out = fopen(argv[2], "wb");
+		FILE* out = fopen(argv[1], "wb");
 		if(!out)
 		{
-			fprintf(stderr, "rtfx: couldn't open file: %s: %s\n", argv[2], strerror(errno));
+			fprintf(stderr, "rtfx: couldn't open file: %s: %s\n", argv[1], strerror(errno));
 			return 1;
 		}
 
diff --git a/src/tags.h b/src/tags.h
index cd7b211..98787d9 100644
--- a/src/tags.h
+++ b/src/tags.h
@@ -39,17 +39,24 @@
 #ifndef __TAGS_H__
 #define __TAGS_H__
 
+/*
+ * IMPORTANT: When adding tags, attributes or values here make sure to update
+ * the XSL Schema for the output formats.
+ */
+
 static const char* kNSPrefix = "xmlns";
+static const wchar_t* kNSRtfx = L"http://memberwebs.com/ns/rtfx/";
 
 // Tags used internally
 static const char* kElDest = "i_dest";
 static const char* kElBlock = "i_block";
-static const char* kAtFix = "i_fix";
-static const char* kAtCell = "i_cell";
 static const char* kElListtable = "i_listtable";
 static const char* kElListdef = "i_listdef";
 static const char* kElFontTable = "i_fonttable";
 static const char* kElFontDef = "i_fontdef";
+static const char* kAtFix = "i_fix";
+static const char* kAtCell = "i_cell";
+static const char* kAtList = "i_list";
 
 // All the main tags
 static const char* kElPara = "para";
@@ -70,7 +77,6 @@ static const char* kElHide = "hide";
 static const char* kElI = "i";
 static const char* kElStrike = "strike";
 static const char* kElU = "u";
-static const char* kElColor = "color";
 static const char* kElSuper = "super";
 static const char* kElSub = "sub";
 static const char* kElCell = "cell";
@@ -79,9 +85,9 @@ static const char* kElTable = "table";
 static const char* kElFootNote = "footnote";
 static const char* kElRef = "ref";
 static const char* kElFont = "font";
+static const char* kElOptions = "options";
 
 // Attributes
-static const char* kAtList = "list";
 static const char* kAtName = "name";
 static const char* kAtBold = "bold";
 static const char* kAtHidden = "hide";
@@ -93,7 +99,6 @@ static const char* kAtType = "type";
 static const char* kAtOrdered = "ordered";
 static const char* kAtStart = "start";
 static const char* kAtId = "id";
-static const char* kAtIndex = "id";
 static const char* kAtTo = "to";
 static const char* kAtSize = "size";
 
@@ -109,5 +114,7 @@ static const wchar_t* kValFootNote = L"footnote";
 static const wchar_t* kValList = L"list";
 static const wchar_t* kValPara = L"para";
 static const wchar_t* kValTable = L"table";
+static const wchar_t* kValTrue = L"true";
+static const wchar_t* kValZero = L"0";
 
 #endif // __TAGS_H__
diff --git a/src/xmlcomposer.cpp b/src/xmlcomposer.cpp
index 398cb55..a90c25a 100644
--- a/src/xmlcomposer.cpp
+++ b/src/xmlcomposer.cpp
@@ -111,13 +111,18 @@ void XmlComposer::startDocument(RtfParser* reader)
 	ASSERT(m_document != NULL);
 
     // Hook up the top level element
-	m_curLevel->setElement(m_document.getDocumentElement(), true);
+    DOM::Element top = m_document.getDocumentElement();
+	m_curLevel->setElement(top, true);
 
 	// Set the attributes on the top level
 	setAnalyser(AnalyserPtr(new Root));
 	setDestination(DestinationPtr(new Content));
 	getTextFormatting().resetPara();
 	getTextFormatting().resetText();
+
+    // Create the options element
+    m_docOptions = m_document.createElement(kElOptions);
+    top.appendChild(m_docOptions);
 }
 
 void XmlComposer::endDocument()
@@ -151,6 +156,11 @@ void XmlComposer::endDocument()
 
     // Pass 3: Final cleanup
     // XmlFixups::combineDuplicates(m_document);
+
+    DOM::Element top = m_document.getDocumentElement();
+    ASSERT(top != NULL);
+    top.setAttribute(kNSPrefix, kNSRtfx);
+
 	return;
 }
 
@@ -323,6 +333,12 @@ void XmlComposer::incrementAutoCount(int type)
     m_autocount[type]++;
 }
 
+void XmlComposer::addDocumentOption(DOM::Element& option)
+{
+    ASSERT(m_docOptions != NULL);
+    m_docOptions.appendChild(option);
+}
+
 
 /* ----------------------------------------------------------------------------------
  *  CONVENIENCE MACROS USED BELOW
@@ -694,23 +710,17 @@ ON_GROUPEND(Style)
 
     // Dig out all the formatting attributes
 	if(props.textIsBold())
-		AN_ATTRIBUTE(kAtBold, L"1");
+		AN_ATTRIBUTE(kAtBold, kValTrue);
 	if(props.textIsHidden())
-		AN_ATTRIBUTE(kAtHidden, L"1");
+		AN_ATTRIBUTE(kAtHidden, kValTrue);
 	if(props.textIsItalic())
-		AN_ATTRIBUTE(kAtItalic, L"1");
+		AN_ATTRIBUTE(kAtItalic, kValTrue);
 	if(props.textIsStrike())
-		AN_ATTRIBUTE(kAtStrike, L"1");
+		AN_ATTRIBUTE(kAtStrike, kValTrue);
 	if(props.textIsUnderline())
-		AN_ATTRIBUTE(kAtUnderline, L"1");
+		AN_ATTRIBUTE(kAtUnderline, kValTrue);
 
-    if(DO_EXTRAS())
-    {
-        // TODO: Do fonts here
-
-	    if(props.textColor() != -1)
-		    AN_ATTRIBUTE(kAtColor, props.textColor());
-    }
+    // TODO: Do fonts and colors here
 }
 
 
@@ -728,6 +738,21 @@ ON_GROUPSTART(FontTable)
     AN_ANALYSER(Font);
 }
 
+ON_DONE(FontTable)
+{
+    DOM::Element deffont = m_composer->createElement(kElFont);
+
+    // Default font is always the first in the list
+    deffont->setAttribute(kAtId, kValZero);
+
+    // Default size is always 12 pt
+    deffont->setAttribute(kAtSize, L"12");
+
+    // TODO: Is this correct?
+    deffont->setAttribute(kAtColor, kValZero);
+
+    m_composer->addDocumentOption(deffont);
+}
 
 // Font Analyser --------------------------------------------------------------------
 
@@ -1072,6 +1097,9 @@ ON_CHARDATA(Content)
             if(fontsize != -1)
                 AN_ATTRIBUTE(kAtSize, fontsize);
 
+            if(format.textColor() != -1)
+                AN_ATTRIBUTE(kAtColor, format.textColor());
+
             elements++;
         }
     }
@@ -1108,16 +1136,6 @@ ON_CHARDATA(Content)
 		elements++;
 	}
 
-    if(DO_EXTRAS())
-    {
-    	if(format.textColor() != -1)
-        {
-            AN_ELEMENT(kElColor);
-    		AN_ATTRIBUTE(kAtIndex, format.textColor());
-    		elements++;
-	    }
-    }
-
     if(format.textSuScript() == RtfFormatting::SUPERSCRIPT)
     {
         AN_ELEMENT(kElSuper);
diff --git a/src/xmlcomposer.h b/src/xmlcomposer.h
index 83ea4cd..41aad0e 100644
--- a/src/xmlcomposer.h
+++ b/src/xmlcomposer.h
@@ -117,6 +117,8 @@ public:
     int getAutoCount(int type);
     void incrementAutoCount(int type);
 
+    // Add a document option to the option block
+    void addDocumentOption(DOM::Element& option);
 
 	// Get the current formatting options
 	RtfFormatting& getTextFormatting();
@@ -139,7 +141,7 @@ protected:
 	DOM::Document m_document;              // The current document
 	XmlComposerOptions m_options;          // Configurable options for parsing
     int m_autocount[AUTOCOUNT_MAX];        // Auto counters for the document
-
+    DOM::Element m_docOptions;             // For storing document options
 
 // Sub classes
 protected:
@@ -250,6 +252,7 @@ protected:
     ANALYSER(FontTable)
         INITIALIZE
         GROUPSTART
+        DONE
     END_ANALYSER
 
     // Handle a Font in the Table
diff --git a/src/xmlfixups.cpp b/src/xmlfixups.cpp
index 10b35f1..45fa59a 100644
--- a/src/xmlfixups.cpp
+++ b/src/xmlfixups.cpp
@@ -42,16 +42,22 @@
 #include "tags.h"
 
 static const char* kNoDuplicates[] =
-    { kElB, kElU, kElI, kElColor, kElHide, kElSuper, kElSub, NULL };
+    { kElB, kElU, kElI, kElFont, kElHide, kElSuper, kElSub, NULL };
+
+static const char* kRequireAttrs[] =
+    { kElFont, NULL };
 
 static const char* kRemoveTags[] =
- { kElDest, kElListdef, kElListtable, kElFontTable, NULL };
+    { kElDest, kElListdef, kElListtable, kElFontTable, kElFontDef, NULL };
+
+static const char* kRemoveEmpty[] =
+    { kElOptions, kElList, NULL };
 
 static const char* kBlockTags[] =
-  { kElTable, kElPara, NULL };
+    { kElTable, kElPara, NULL };
 
 static const char* kHideList[] =
-   { kAtId, kAtList, NULL };
+    { kAtId, kAtList, NULL };
 
 static const char* kConsolidateEnd[] =
     { kElFootNote, NULL };
@@ -69,6 +75,8 @@ XmlFixups::XmlFixups()
 {
     loadStringSet(m_duplicates, kNoDuplicates);
     loadStringSet(m_removes, kRemoveTags);
+    loadStringSet(m_removes, kRemoveEmpty);
+    loadStringSet(m_requireAttrs, kRequireAttrs);
     loadStringSet(m_consolidateStart, kConsolidateStart);
     loadStringSet(m_consolidateEnd, kConsolidateEnd);
 }
@@ -502,6 +510,8 @@ void XmlFixups::runPassTwo(const DOM::Document& doc)
                 DOM::Element font = fonts.get(el.getAttribute(kAtId));
                 if(font != NULL)
                     el.setAttribute(kAtName, font.getAttribute(kAtName));
+
+                font.removeAttribute(kAtId);
             }
         }
 
@@ -532,8 +542,10 @@ void XmlFixups::runPassTwo(const DOM::Document& doc)
              */
         }
 
-        // Tags that just plain get removed
-        if(m_removes.find(name) != m_removes.end())
+        // Tags that get removed but contents preserved. Also here are
+        // tags that get removed if they have no attributes
+        if(m_removes.find(name) != m_removes.end() ||
+           (m_requireAttrs.find(name) != m_requireAttrs.end() && el.hasAttributes()))
         {
             DOM::Node parent = el->getParentNode();
 
@@ -556,6 +568,26 @@ void XmlFixups::runPassTwo(const DOM::Document& doc)
             }
         }
 
+        // Tags that get removed when no child nodes exist
+        if(m_removeEmpty.find(name) != m_removeEmpty.end() && el.hasChildNodes())
+        {
+            DOM::Node parent = el->getParentNode();
+
+            if(parent != NULL)
+            {
+                /*
+                 * After the element is removed, the current element is no longer
+                 * valid for iterating over the document. In addition we insert
+                 * all the child nodes of the current element before it. We need
+                 * to be sure to iterate over these elements, and to do so we
+                 * decrement the iterator.
+                 */
+                --it;
+
+                parent.removeChild(el);
+                continue;   /* Current element doesn't need any more processing */
+            }
+        }
 
         // Tags that need to get consolidated to start
         if(m_consolidateStart.find(name) != m_consolidateStart.end())
@@ -577,8 +609,11 @@ void XmlFixups::runPassTwo(const DOM::Document& doc)
                 {
                     DOM::Node next = el.getNextSibling();
 
+                    if(next == NULL || next.getNodeType() != DOM::Node::ELEMENT_NODE)
+                        break;
+
                     // If it's the same type of element ...
-                    if(!DOMHelpers::isElement(next, name))
+                    if(!DOMHelpers::isEqualElement((DOM::Element&)next, el))
                         break;
 
                     // NOTE: Notice we do nothing with attributes. Currently
diff --git a/src/xmlfixups.h b/src/xmlfixups.h
index 01c2d67..99fe876 100644
--- a/src/xmlfixups.h
+++ b/src/xmlfixups.h
@@ -146,8 +146,10 @@ protected:
     // Our tables cached for efficiency
     StringSet m_duplicates;
     StringSet m_removes;
+    StringSet m_removeEmpty;
     StringSet m_consolidateStart;
     StringSet m_consolidateEnd;
+    StringSet m_requireAttrs;
 };
 
 #endif // __XMLFIXUPS_H__