From ab6fbbdfed3b85fdd6f4ea2aafd272602699dde7 Mon Sep 17 00:00:00 2001
From: Edouard DUPIN <yui.heero@gmail.com>
Date: Mon, 13 Oct 2014 21:42:08 +0200
Subject: [PATCH] [DEV] add parsing of &"'<> spacial value in generic text.

---
 exml/Element.cpp | 16 +++++++++++-----
 exml/Element.h   |  2 +-
 exml/Text.cpp    | 47 ++++++++++++++++++++++++++++++++++++++++++++++-
 exml/Text.h      |  1 +
 4 files changed, 59 insertions(+), 7 deletions(-)

diff --git a/exml/Element.cpp b/exml/Element.cpp
index a925910..6e5a89d 100644
--- a/exml/Element.cpp
+++ b/exml/Element.cpp
@@ -137,11 +137,18 @@ void exml::Element::append(exml::Node* _node) {
 }
 
 std::string exml::Element::getText() {
-	// TODO : add more capabilities ...
 	std::string res;
-	for (size_t iii=0; iii<m_listSub.size(); iii++) {
-		if (NULL!=m_listSub[iii]) {
-			m_listSub[iii]->iGenerate(res, 0);
+	if (m_listSub.size() == 1) {
+		if (m_listSub[0]->getType() == typeText) {
+			res = m_listSub[0]->getValue();
+		} else {
+			m_listSub[0]->iGenerate(res, 0);
+		}
+	} else {
+		for (size_t iii=0; iii<m_listSub.size(); iii++) {
+			if (NULL!=m_listSub[iii]) {
+				m_listSub[iii]->iGenerate(res, 0);
+			}
 		}
 	}
 	return res;
@@ -309,7 +316,6 @@ bool exml::Element::subParse(const std::string& _data, int32_t& _pos, bool _case
 					CREATE_ERROR(_doc, _data, _pos, _filePos, std::string("End file with '<!") + _data[iii+white+2] + "' chars  == > invalide XML");
 					return false;
 				}
-				
 				continue;
 			}
 			if(_data[iii+white+1] == '/') {
diff --git a/exml/Element.h b/exml/Element.h
index d1b927f..348899f 100644
--- a/exml/Element.h
+++ b/exml/Element.h
@@ -77,7 +77,7 @@ namespace exml {
 			const Element* getNamed(const std::string& _name) const;
 			/**
 			 * @brief get the internal data of the element (if the element has some sub node thay are converted in xml string  == > like this it is not needed to use <![CDATA[...]]>
-			 * @return the curent data string.
+			 * @return the curent data string. if Only one text node, then we get the parssed data (no &amp; ...) if more than one node, then we transform &,",',<,> in xml normal text...
 			 */
 			std::string getText();
 		protected:
diff --git a/exml/Text.cpp b/exml/Text.cpp
index acd0a19..900978a 100644
--- a/exml/Text.cpp
+++ b/exml/Text.cpp
@@ -9,11 +9,50 @@
 #include <exml/Text.h>
 #include <exml/debug.h>
 #include <exml/Document.h>
+#include <regex>
 
 #undef __class__
 #define __class__ "Text"
 
 
+// transform the Text with :
+//     "&lt;"   == "<"
+//     "&gt;"   == ">"
+//     "&amp;"  == "&"
+//     "&apos;" == "'"
+//     "&quot;" == """
+static std::string replaceSpecialChar(const std::string& _inval) {
+	std::string out;
+	static std::regex regexLT("&lt;");
+	static std::regex regexGT("&gt;");
+	static std::regex regexAPOS("&apos;");
+	static std::regex regexQUOT("&quot;");
+	static std::regex regexAMP("&amp;");
+	
+	out = std::regex_replace(_inval, regexLT, "<");
+	out = std::regex_replace(out, regexGT, ">");
+	out = std::regex_replace(out, regexAPOS, "'");
+	out = std::regex_replace(out, regexQUOT, "\"");
+	out = std::regex_replace(out, regexAMP, "&");
+	//EXML_ERROR("plop "<< _inval << " => " << out);
+	return out;
+}
+static std::string replaceSpecialCharOut(const std::string& _inval) {
+	std::string out;
+	static std::regex regexLT("<");
+	static std::regex regexGT(">;");
+	static std::regex regexAMP("&");
+	static std::regex regexAPOS("'");
+	static std::regex regexQUOT("\"");
+	
+	out = std::regex_replace(_inval, regexAMP, "&amp;");
+	out = std::regex_replace(out, regexQUOT, "&quot;");
+	out = std::regex_replace(out, regexAPOS, "&apos;");
+	out = std::regex_replace(out, regexGT, "&gt;");
+	out = std::regex_replace(out, regexLT, "&lt;");
+	return out;
+}
+
 static bool isWhiteChar(char32_t _val) {
 	if(    _val == ' '
 	    || _val == '\t'
@@ -25,7 +64,7 @@ static bool isWhiteChar(char32_t _val) {
 }
 
 bool exml::Text::iGenerate(std::string& _data, int32_t _indent) const {
-	_data += m_value;
+	_data += replaceSpecialCharOut(m_value);
 	return true;
 }
 
@@ -65,6 +104,7 @@ bool exml::Text::iParse(const std::string& _data, int32_t& _pos, bool _caseSensi
 			m_value = std::string(_data, _pos, newEnd-(_pos));
 			EXML_VERBOSE(" find text '" << m_value << "'");
 			_pos = iii-1;
+			m_value = replaceSpecialChar(m_value);
 			return true;
 		}
 	}
@@ -73,6 +113,11 @@ bool exml::Text::iParse(const std::string& _data, int32_t& _pos, bool _caseSensi
 	return false;
 }
 
+bool exml::TextCDATA::iGenerate(std::string& _data, int32_t _indent) const {
+	_data += "<![CDATA[" + m_value +"]]>";
+	return true;
+}
+
 bool exml::TextCDATA::iParse(const std::string& _data, int32_t& _pos, bool _caseSensitive, exml::filePos& _filePos, exml::Document& _doc) {
 	EXML_VERBOSE("start parse : 'text::CDATA'");
 	m_pos = _filePos;
diff --git a/exml/Text.h b/exml/Text.h
index 1a88aba..3e89ae9 100644
--- a/exml/Text.h
+++ b/exml/Text.h
@@ -58,6 +58,7 @@ namespace exml {
 			virtual ~TextCDATA() { };
 		public: // herited function:
 			virtual bool iParse(const std::string& _data, int32_t& _pos, bool _caseSensitive, exml::filePos& _filePos, exml::Document& _doc);
+			virtual bool iGenerate(std::string& _data, int32_t _indent) const;
 	};
 };