Added option for pretty printing whitespace in XML.

7 files changed, 79 insertions, 2 deletions

diff --git a/ChangeLog b/ChangeLog
index a82f430..05500be 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -4,6 +4,7 @@ Version 0.9.6
  - Add <link> tag for hyperlink fields
  - Add 'align' attribute to <para> tags for paragraph alignment
  - Add 'indent' attriute to <para> tags for paragraph indent
+ - Now has option for pretty printing output XML
 
 Version 0.9.5
  - Allow conversions on stdin and stdout
diff --git a/src/rtfx.1 b/src/rtfx.1
index 64f8547..3d234c6 100644
--- a/src/rtfx.1
+++ b/src/rtfx.1
@@ -42,7 +42,7 @@
 .Nd converts RTF to a generic XML format
 .Sh SYNOPSIS
 .Nm
-.Op Fl p
+.Op Fl pw
 .Op Ar inrtf
 .Op Ar outxml
 .Sh DESCRIPTION
@@ -62,6 +62,8 @@ The options are as follows:
 Puts
 .Nm 
 into 'presentation' output mode.
+.It Fl w
+Output XML is written with indents for easy reading.
 .It Ar inrtf
 The RTF file to convert. If not specified reads from stdin.
 .It Ar outxml
diff --git a/src/rtfx.cpp b/src/rtfx.cpp
index 8239b49..e305959 100644
--- a/src/rtfx.cpp
+++ b/src/rtfx.cpp
@@ -46,7 +46,7 @@
 
 int usage()
 {
-	fprintf(stderr, "usage: rtfx [-p] [inrtf] [outxml]\n");
+	fprintf(stderr, "usage: rtfx [-pw] [inrtf] [outxml]\n");
     exit(2);
 }
 
@@ -71,6 +71,10 @@ int main(int argc, char* argv[])
                 options.extras = true;
                 break;
 
+			case 'w':
+				options.pretty = true;
+				break;
+
             case '-':
                 arg = NULL;
                 break;
diff --git a/src/xmlcomposer.cpp b/src/xmlcomposer.cpp
index f10ce0c..6b1b725 100644
--- a/src/xmlcomposer.cpp
+++ b/src/xmlcomposer.cpp
@@ -131,6 +131,10 @@ void XmlComposer::endDocument()
     ASSERT(top != NULL);
     top.setAttribute(kNSPrefix, kNSRtfx);
 
+	// Pass 3: Pretty printing
+	if(m_options.pretty)
+		fix.prettyXml(m_document);
+
 	return;
 }
 
diff --git a/src/xmlcomposer.h b/src/xmlcomposer.h
index 26839d7..c9c321c 100644
--- a/src/xmlcomposer.h
+++ b/src/xmlcomposer.h
@@ -47,6 +47,7 @@ struct XmlComposerOptions
 		{ memset(this, 0, sizeof(*this)); }
 
 	bool extras;
+	bool pretty;
 };
 
 /*
diff --git a/src/xmlfixups.cpp b/src/xmlfixups.cpp
index 458798a..2c05e3a 100644
--- a/src/xmlfixups.cpp
+++ b/src/xmlfixups.cpp
@@ -65,6 +65,9 @@ static const char* kConsolidateEnd[] =
 static const char* kConsolidateStart[] =
     { kElStylesheet, kElInfo, NULL };
 
+static const char* kNoPretty[] =
+	{ kElPara, NULL };
+
 void loadStringSet(StringSet& set, const char** strings)
 {
     while(*strings)
@@ -74,6 +77,7 @@ void loadStringSet(StringSet& set, const char** strings)
 XmlFixups::XmlFixups()
 {
     loadStringSet(m_duplicates, kNoDuplicates);
+	loadStringSet(m_nopretty, kNoPretty);
     loadStringSet(m_removes, kRemoveTags);
     loadStringSet(m_removeEmpty, kRemoveEmpty);
     loadStringSet(m_requireAttrs, kRequireAttrs);
@@ -791,3 +795,56 @@ void XmlFixups::fixBlock(const DOM::Document& doc, DOM::Element& block)
         parent.replaceChild(el, block);
     }
 }
+
+void XmlFixups::prettyXml(DOM::Document& doc)
+{
+	DOM::Element el = doc.getDocumentElement();
+	internalPrettyXml(doc, el, 0);
+}
+
+void XmlFixups::internalPrettyXml(const DOM::Document& doc, DOM::Element& el,
+							      int level)
+{
+	DOM::Node child;
+
+	// Empty elements remain as before
+	if(!el.hasChildNodes())
+		return;
+
+	// Make sure we're allowed to pretty print this
+	if(m_nopretty.find(el.getNodeName()) != m_nopretty.end())
+		return;
+
+	// First see if we have any text nodes
+	for(child = el.getFirstChild(); child != NULL; child = child.getNextSibling())
+	{
+		// Any non-empty text nodes means we don't pretty print this
+		if(child.getNodeType() == DOM::Node::TEXT_NODE &&
+		   !child.getNodeValue().empty())
+			return;
+	}
+
+	// The main indent string
+	string indent("\r\n");
+	indent.append((level + 1) * 4, ' ');
+
+	for(child = el.getFirstChild(); child != NULL; child = child.getNextSibling())
+	{
+		if(child.getNodeType() == DOM::Node::TEXT_NODE)
+			continue;
+
+		DOM::Text text = doc.createTextNode(indent);
+		el.insertBefore(text, child);
+
+		if(child.getNodeType() == DOM::Node::ELEMENT_NODE)
+			internalPrettyXml(doc, (DOM::Element&)child, level + 1);
+	}
+
+	// The last indent
+	indent.assign("\r\n");
+	indent.append(level * 4, ' ');
+
+	DOM::Text text = doc.createTextNode(indent);
+	el.appendChild(text);
+}
+
diff --git a/src/xmlfixups.h b/src/xmlfixups.h
index f51d4a8..85d7c7e 100644
--- a/src/xmlfixups.h
+++ b/src/xmlfixups.h
@@ -133,7 +133,14 @@ public:
     // Replace blocks with 'fix' elements like paragraphs
     void fixBlock(const DOM::Document& doc, DOM::Element& block);
 
+	// Make XML ready for pretty printing
+	void prettyXml(DOM::Document& doc);
 
+// Internal Helpers
+protected:
+	void internalPrettyXml(const DOM::Document& doc, DOM::Element& el, int level);
+
+// Data
 protected:
 
     enum
@@ -145,6 +152,7 @@ protected:
 
     // Our tables cached for efficiency
     StringSet m_duplicates;
+	StringSet m_nopretty;
     StringSet m_removes;
     StringSet m_removeEmpty;
     StringSet m_consolidateStart;