added non-pretty output of XML serialization, changed DemoOutput accordingly and changed DemoOutputTest to have some kind of automatic inspection instead of visual inspection.

6 files changed, 79 insertions, 32 deletions

diff --git a/src/core/XML.cpp b/src/core/XML.cpp
index 9cfb5dc..b7d1511 100644
--- a/src/core/XML.cpp
+++ b/src/core/XML.cpp
@@ -4,18 +4,23 @@
 namespace ousia {
 namespace xml {
 
-void Node::serialize(std::ostream &out, const std::string &doctype)
+void Node::serialize(std::ostream &out, const std::string &doctype, bool pretty)
 {
 	if (doctype != "") {
-		out << doctype << "\n";
+		out << doctype;
+		if (pretty) {
+			out << '\n';
+		}
 	}
-	doSerialize(out, 0);
+	doSerialize(out, 0, pretty);
 }
 
-void Element::doSerialize(std::ostream &out, unsigned int tabdepth)
+void Element::doSerialize(std::ostream &out, unsigned int tabdepth, bool pretty)
 {
-	for (unsigned int t = 0; t < tabdepth; t++) {
-		out << '\t';
+	if (pretty) {
+		for (unsigned int t = 0; t < tabdepth; t++) {
+			out << '\t';
+		}
 	}
 	out << '<' << name;
 	for (auto &a : attributes) {
@@ -23,25 +28,41 @@ void Element::doSerialize(std::ostream &out, unsigned int tabdepth)
 	}
 	// if we have no children, we close the tag immediately.
 	if (children.size() == 0) {
-		out << "/>\n";
+		out << "/>";
+		if (pretty) {
+			out << '\n';
+		}
 	} else {
-		out << ">\n";
+		out << ">";
+		if (pretty) {
+			out << '\n';
+		}
 		for (auto &n : children) {
-			n->doSerialize(out, tabdepth + 1);
+			n->doSerialize(out, tabdepth + 1, pretty);
 		}
-		for (unsigned int t = 0; t < tabdepth; t++) {
-			out << '\t';
+		if (pretty) {
+			for (unsigned int t = 0; t < tabdepth; t++) {
+				out << '\t';
+			}
+		}
+		out << "</" << name << ">";
+		if (pretty) {
+			out << '\n';
 		}
-		out << "</" << name << ">\n";
 	}
 }
 
-void Text::doSerialize(std::ostream &out, unsigned int tabdepth)
+void Text::doSerialize(std::ostream &out, unsigned int tabdepth, bool pretty)
 {
-	for (unsigned int t = 0; t < tabdepth; t++) {
-		out << '\t';
+	if (pretty) {
+		for (unsigned int t = 0; t < tabdepth; t++) {
+			out << '\t';
+		}
+	}
+	out << text;
+	if (pretty) {
+		out << '\n';
 	}
-	out << text << '\n';
 }
 }
 
diff --git a/src/core/XML.hpp b/src/core/XML.hpp
index e55ecba..b05d4c6 100644
--- a/src/core/XML.hpp
+++ b/src/core/XML.hpp
@@ -73,9 +73,12 @@ public:
 	 * @param out     is the output stream the serialized data shall be
 	 *                written to.
 	 * @param doctype enables you to add a prefix specifying the doctype.
+	 * @param pretty is a flag that manipulates whether newlines and tabs are
+	 *               used.
 	 */
 	void serialize(std::ostream &out,
-	               const std::string &doctype = "<?xml version=\"1.0\"?>");
+	               const std::string &doctype = "<?xml version=\"1.0\"?>",
+	               bool pretty = true);
 	/**
 	 * This method just writes the XML representation of this node to the
 	 * output stream.
@@ -83,8 +86,11 @@ public:
 	 * @param out      the output stream the serialized data shall be written
 	 *                 to.
 	 * @param tabdepth the current tabdepth for prettier output.
+	 * @param pretty is a flag that manipulates whether newlines and tabs are
+	 *               used.
 	 */
-	virtual void doSerialize(std::ostream &out, unsigned int tabdepth) = 0;
+	virtual void doSerialize(std::ostream &out, unsigned int tabdepth,
+	                         bool pretty) = 0;
 
 	/**
 	 * @return the parent XML element of this node.
@@ -130,7 +136,8 @@ public:
 	 * * The end tag of this element.
 	 *
 	 */
-	void doSerialize(std::ostream &out, unsigned int tabdepth) override;
+	void doSerialize(std::ostream &out, unsigned int tabdepth,
+	                 bool pretty) override;
 
 	const ManagedVector<Node> &getChildren() const { return children; }
 
@@ -155,7 +162,8 @@ public:
 	 * This just writes the text to the output.
 	 *
 	 */
-	void doSerialize(std::ostream &out, unsigned int tabdepth) override;
+	void doSerialize(std::ostream &out, unsigned int tabdepth,
+	                 bool pretty) override;
 };
 }
 
diff --git a/src/plugins/html/DemoOutput.cpp b/src/plugins/html/DemoOutput.cpp
index 69f2756..4cadf12 100644
--- a/src/plugins/html/DemoOutput.cpp
+++ b/src/plugins/html/DemoOutput.cpp
@@ -29,12 +29,12 @@ namespace ousia {
 namespace html {
 
 void DemoHTMLTransformer::writeHTML(Handle<model::Document> doc,
-                                    std::ostream &out)
+                                    std::ostream &out, bool pretty)
 {
 	Manager &mgr = doc->getManager();
 	// Create an XML object tree for the document first.
 	Rooted<xml::Element> html{new xml::Element{
-	    mgr, {nullptr}, "html", {{"xlmns", "http://www.w3.org/1999/xhtml"}}}};
+	    mgr, {nullptr}, "html"}};
 	// add the head Element
 	Rooted<xml::Element> head{new xml::Element{mgr, html, "head"}};
 	html->addChild(head);
@@ -78,7 +78,7 @@ void DemoHTMLTransformer::writeHTML(Handle<model::Document> doc,
 	body->addChild(book);
 
 	// After the content has been transformed, we serialize it.
-	html->serialize(out, "<!DOCTYPE html>");
+	html->serialize(out, "<!DOCTYPE html>", pretty);
 }
 
 /**
diff --git a/src/plugins/html/DemoOutput.hpp b/src/plugins/html/DemoOutput.hpp
index 07d758b..b755aac 100644
--- a/src/plugins/html/DemoOutput.hpp
+++ b/src/plugins/html/DemoOutput.hpp
@@ -83,11 +83,14 @@ public:
 	 * Therefore this is not an adequate model of our algorithms but only a
 	 * Demo.
 	 *
-	 * @param doc is a Document using concepts of the book, headings, emphasis
-	 *            and lists domains but no other.
-	 * @param out is the output stream the data shall be written to.
+	 * @param doc    is a Document using concepts of the book, headings,
+	 *               emphasis and lists domains but no other.
+	 * @param out    is the output stream the data shall be written to.
+	 * @param pretty is a flag that manipulates whether newlines and tabs are
+	 *               used.
 	 */
-	void writeHTML(Handle<model::Document> doc, std::ostream &out);
+	void writeHTML(Handle<model::Document> doc, std::ostream &out,
+	               bool pretty = true);
 };
 }
 }
diff --git a/test/core/model/TestAdvanced.hpp b/test/core/model/TestAdvanced.hpp
index a1510e7..769a0df 100644
--- a/test/core/model/TestAdvanced.hpp
+++ b/test/core/model/TestAdvanced.hpp
@@ -254,7 +254,7 @@ static Rooted<Document> constructAdvancedDocument(Manager &mgr, Logger &logger,
 		{
 			if (!addAnnotation(logger, doc, p,
 			                   "Aufklärung ist der Ausgang des Menschen aus "
-			                   "seiner selbstverschuldeten Unmündigkeit",
+			                   "seiner selbstverschuldeten Unmündigkeit!",
 			                   "strong")) {
 				return {nullptr};
 			}
diff --git a/test/plugins/html/DemoOutputTest.cpp b/test/plugins/html/DemoOutputTest.cpp
index 5a34112..5b2758d 100644
--- a/test/plugins/html/DemoOutputTest.cpp
+++ b/test/plugins/html/DemoOutputTest.cpp
@@ -19,6 +19,7 @@
 #include <gtest/gtest.h>
 
 #include <iostream>
+#include <sstream>
 
 #include <plugins/html/DemoOutput.hpp>
 
@@ -56,9 +57,17 @@ TEST(DemoHTMLTransformer, writeHTML)
 	mgr.exportGraphviz("bookDocument.dot");
 #endif
 
-	// TODO: change this. Don't use printouts
+	// we can only do a rough check here.
 	DemoHTMLTransformer transformer;
-	transformer.writeHTML(doc, std::cout);
+	std::stringstream out;
+	transformer.writeHTML(doc, out);
+	const std::string res = out.str();
+	ASSERT_FALSE(res == "");
+	ASSERT_TRUE(res.find("Was ist Aufklärung?") != std::string::npos);
+	ASSERT_TRUE(res.find(
+	                "Aufklärung ist der Ausgang des Menschen aus seiner "
+	                "selbstverschuldeten Unmündigkeit!") != std::string::npos);
+	ASSERT_TRUE(res.find("Sapere aude!") != std::string::npos);
 }
 
 TEST(DemoHTMLTransformer, AnnotationProcessing)
@@ -100,9 +109,15 @@ TEST(DemoHTMLTransformer, AnnotationProcessing)
 	buildAnnotationEntity(doc, logger, {"emphasized"}, em_start, em_end);
 	buildAnnotationEntity(doc, logger, {"strong"}, strong_start, strong_end);
 
-	// TODO: change this. Don't use printouts
+	// Check serialization.
 	DemoHTMLTransformer transformer;
-	transformer.writeHTML(doc, std::cout);
+	std::stringstream out;
+	transformer.writeHTML(doc, out, false);
+	const std::string res = out.str();
+	// In HTML the overlapping structure must be serialized as follows:
+	ASSERT_TRUE(
+	    res.find("<em>bla<strong>blub</strong></em><strong>bla</strong>") !=
+	    std::string::npos);
 }
 }
 }