From 255362d5f815ca0e04e18518bb3629d241af5117 Mon Sep 17 00:00:00 2001
From: Benjamin Paassen <bpaassen@techfak.uni-bielefeld.de>
Date: Sun, 22 Feb 2015 11:48:19 +0100
Subject: corrected XML serialization for primitive text content regarding
 whitespace and newline handling and extended XMLTest to check that.

---
 src/core/XML.cpp      | 114 ++++++++++++++++++++++++++++++++++++++------------
 test/core/XMLTest.cpp |  22 ++++------
 2 files changed, 94 insertions(+), 42 deletions(-)

diff --git a/src/core/XML.cpp b/src/core/XML.cpp
index e25d18a..404af92 100644
--- a/src/core/XML.cpp
+++ b/src/core/XML.cpp
@@ -66,13 +66,82 @@ static std::string escapePredefinedEntities(const std::string &input)
 
 void Element::doSerialize(std::ostream &out, unsigned int tabdepth, bool pretty)
 {
+	bool hasText = false;
 	if (pretty) {
+		// print tabs at the beginning, if we are in pretty mode.
 		for (unsigned int t = 0; t < tabdepth; t++) {
 			out << '\t';
 		}
+		/*
+		 * if we are in pretty mode we also need to check if we have a text
+		 * node as child.
+		 * If so this changes our further output processing because of the way
+		 * XML treats primitive data: The structure
+		 *
+		 * \code{.xml}
+		 * <Element name="A">
+		 *   <Text>content</Text>
+		 *   <Text>content2</Text>
+		 * </Element>
+		 * \endcode
+		 *
+		 * has to be serialized as
+		 *
+		 * \code{.xml}
+		 * <A>contentcontent2</A>
+		 * \endcode
+		 *
+		 * because otherwise we introduce whitespaces and newlines where no
+		 * such things had been before.
+		 *
+		 * On the other hand the structure
+		 *
+		 * \code{.xml}
+		 * <Element name="A">
+		 *   <Element name="B">
+		 *     <Text>content</Text>
+		 *   </Element>
+		 * </Element>
+		 * \endcode
+		 *
+		 * Can be serialized as
+		 *
+		 * \code{.xml}
+		 * <A>
+		 *   <B>content</B>
+		 * </A>
+		 * \endcode
+		 *
+		 * As last example consider the case
+		 *
+		 * \code{.xml}
+		 * <Element name="A">
+		 *   <Element name="B">
+		 *     <Text>content</Text>
+		 *   </Element>
+		 *   <Text>content2</Text>
+		 * </Element>
+		 * \endcode
+		 *
+		 * Here the A-Element again has primitive text content, such that we
+		 * are not allowed to prettify. It has to be serialized like this:
+		 *
+		 * \code{.xml}
+		 * <A><B>content</B>content2</A>
+		 * \endcode
+		 *
+		 *
+		 */
+		for (auto n : children) {
+			if (n->isa(&RttiTypes::XMLText)) {
+				hasText = true;
+				break;
+			}
+		}
 	}
+
 	out << '<';
-	if(!nspace.empty()){
+	if (!nspace.empty()) {
 		out << nspace << ":";
 	}
 	out << name;
@@ -88,19 +157,19 @@ void Element::doSerialize(std::ostream &out, unsigned int tabdepth, bool pretty)
 		}
 	} else {
 		out << ">";
-		if (pretty) {
+		if (pretty && !hasText) {
 			out << std::endl;
 		}
-		for (auto &n : children) {
-			n->doSerialize(out, tabdepth + 1, pretty);
+		for (auto n : children) {
+			n->doSerialize(out, tabdepth + 1, pretty && !hasText);
 		}
-		if (pretty) {
+		if (pretty && !hasText) {
 			for (unsigned int t = 0; t < tabdepth; t++) {
 				out << '\t';
 			}
 		}
 		out << "</";
-		if(!nspace.empty()){
+		if (!nspace.empty()) {
 			out << nspace << ":";
 		}
 		out << name << ">";
@@ -112,29 +181,20 @@ void Element::doSerialize(std::ostream &out, unsigned int tabdepth, bool pretty)
 
 void Text::doSerialize(std::ostream &out, unsigned int tabdepth, bool pretty)
 {
-	if (pretty) {
-		for (unsigned int t = 0; t < tabdepth; t++) {
-			out << '\t';
-		}
-	}
 	out << escapePredefinedEntities(text);
-	if (pretty) {
-		out << std::endl;
-	}
 }
 }
 
-namespace RttiTypes
-{
-	const Rtti XMLNode = RttiBuilder<xml::Node>("XMLNode");
-	const Rtti XMLElement =
-	    RttiBuilder<xml::Element>("XMLElement")
-	        .parent(&XMLNode)
-	        .composedOf(&XMLNode)
-	        .property("name", {&RttiTypes::String,
-	                           {[](const xml::Element *obj) {
-		                           return Variant::fromString(obj->getName());
-		                       }}});
-	const Rtti XMLText = RttiBuilder<xml::Text>("XMLText").parent(&XMLNode);
-}
+namespace RttiTypes {
+const Rtti XMLNode = RttiBuilder<xml::Node>("XMLNode");
+const Rtti XMLElement =
+    RttiBuilder<xml::Element>("XMLElement")
+        .parent(&XMLNode)
+        .composedOf(&XMLNode)
+        .property("name", {&RttiTypes::String,
+                           {[](const xml::Element *obj) {
+	                           return Variant::fromString(obj->getName());
+	                       }}});
+const Rtti XMLText = RttiBuilder<xml::Text>("XMLText").parent(&XMLNode);
 }
+}
\ No newline at end of file
diff --git a/test/core/XMLTest.cpp b/test/core/XMLTest.cpp
index b11a24c..c264570 100644
--- a/test/core/XMLTest.cpp
+++ b/test/core/XMLTest.cpp
@@ -42,29 +42,21 @@ TEST(XMLNode, testSerialize)
 	body->addChild(div);
 	Rooted<Element> p{new Element{mgr, div, "p"}};
 	div->addChild(p);
-	p->addChild(new Text(mgr, p, "my text"));
-	Rooted<Element> p2{new Element{mgr, div, "p"}};
-	div->addChild(p2);
-	p2->addChild(new Text(mgr, p2, "my text"));
+	p->addChild(new Text(mgr, p, "A"));
+	div->addChild(new Text(mgr, div, "B"));
+	Rooted<Element> myTag{new Element{mgr, div, "myTag", {}, "myNameSpace"}};
+	div->addChild(myTag);
+	myTag->addChild(new Text(mgr, myTag, "C"));
 
 	// Now this is what we expect to see:
 	std::string expected{
 	    "<?xml version=\"1.0\"?>\n"
 	    "<html>\n"
 	    "\t<head>\n"
-	    "\t\t<title>\n"
-	    "\t\t\tmy title\n"
-	    "\t\t</title>\n"
+	    "\t\t<title>my title</title>\n"
 	    "\t</head>\n"
 	    "\t<body>\n"
-	    "\t\t<div class=\"content\" id=\"1\">\n"
-	    "\t\t\t<p>\n"
-	    "\t\t\t\tmy text\n"
-	    "\t\t\t</p>\n"
-	    "\t\t\t<p>\n"
-	    "\t\t\t\tmy text\n"
-	    "\t\t\t</p>\n"
-	    "\t\t</div>\n"
+	    "\t\t<div class=\"content\" id=\"1\"><p>A</p>B<myNameSpace:myTag>C</myNameSpace:myTag></div>\n"
 	    "\t</body>\n"
 	    "</html>\n"};
 
-- 
cgit v1.2.3