1 files changed, 131 insertions, 5 deletions
diff --git a/Swiften/JID/JID.cpp b/Swiften/JID/JID.cpp
index e4611b3..9fa0574 100644
--- a/Swiften/JID/JID.cpp
+++ b/Swiften/JID/JID.cpp
@@ -7,17 +7,22 @@
 #define SWIFTEN_CACHE_JID_PREP
 
 #include <vector>
+#include <list>
 #include <iostream>
 
 #include <string>
 #ifdef SWIFTEN_CACHE_JID_PREP
 #include <boost/unordered_map.hpp>
 #endif
+#include <boost/assign/list_of.hpp>
+#include <boost/algorithm/string/find_format.hpp>
+#include <boost/algorithm/string/finder.hpp>
+#include <sstream>
 #include <stringprep.h>
 
 #include <Swiften/Base/String.h>
-#include "Swiften/JID/JID.h"
-#include "Swiften/IDN/StringPrep.h"
+#include <Swiften/JID/JID.h>
+#include <Swiften/IDN/StringPrep.h>
 
 #ifdef SWIFTEN_CACHE_JID_PREP
 typedef boost::unordered_map<std::string, std::string> PrepCache;
@@ -27,7 +32,85 @@ static PrepCache domainPrepCache;
 static PrepCache resourcePrepCache;
 #endif
 
-namespace Swift {
+static const std::list<char> escapedChars = boost::assign::list_of(' ')('"')('&')('\'')('/')('<')('>')('@')(':');
+
+static std::string getEscaped(char c) {
+	std::ostringstream s;
+	s << '\\' << std::hex << static_cast<int>(c);
+	return s.str();
+}
+
+static bool getEscapeSequenceValue(const std::string& sequence, unsigned char& value) {
+	std::stringstream s;
+	unsigned int v;
+	s << std::hex << sequence;
+	s >> v;
+	value = static_cast<unsigned char>(v);
+	return (!s.fail() && !s.bad() && (value == 0x5C || std::find(escapedChars.begin(), escapedChars.end(), value) != escapedChars.end()));
+}
+
+// Disabling this code for now, since GCC4.5+boost1.42 (on ubuntu) seems to
+// result in a bug. Replacing it with naive code.
+#if 0
+struct UnescapedCharacterFinder {
+	template<typename Iterator>	boost::iterator_range<Iterator> operator()(Iterator begin, Iterator end) {
+		for (; begin != end; ++begin) {
+			if (std::find(escapedChars.begin(), escapedChars.end(), *begin) != escapedChars.end()) {
+				return boost::iterator_range<Iterator>(begin, begin + 1);
+			}
+			else if (*begin == '\\') {
+				// Check if we have an escaped dissalowed character sequence
+				Iterator innerBegin = begin + 1;
+				if (innerBegin != end && innerBegin + 1 != end) {
+					Iterator innerEnd = innerBegin + 2;
+					unsigned char value;
+					if (getEscapeSequenceValue(std::string(innerBegin, innerEnd), value)) {
+						return boost::iterator_range<Iterator>(begin, begin + 1);
+					}
+				}
+			}
+		}
+		return boost::iterator_range<Iterator>(end, end);
+	}
+};
+
+struct UnescapedCharacterFormatter {
+	template<typename FindResult>	std::string operator()(const FindResult& match) const {
+		std::ostringstream s;
+		s << '\\' << std::hex << static_cast<int>(*match.begin());
+		return s.str();
+	}
+};
+
+struct EscapedCharacterFinder {
+	template<typename Iterator>	boost::iterator_range<Iterator> operator()(Iterator begin, Iterator end) {
+		for (; begin != end; ++begin) {
+			if (*begin == '\\') {
+				Iterator innerEnd = begin + 1;
+				for (size_t i = 0; i < 2 && innerEnd != end; ++i, ++innerEnd) {
+				}
+				unsigned char value;
+				if (getEscapeSequenceValue(std::string(begin + 1, innerEnd), value)) {
+					return boost::iterator_range<Iterator>(begin, innerEnd);
+				}
+			}
+		}
+		return boost::iterator_range<Iterator>(end, end);
+	}
+};
+
+struct EscapedCharacterFormatter {
+	template<typename FindResult>	std::string operator()(const FindResult& match) const {
+		unsigned char value;
+		if (getEscapeSequenceValue(std::string(match.begin() + 1, match.end()), value)) {
+			return std::string(reinterpret_cast<const char*>(&value), 1);
+		}
+		return boost::copy_range<std::string>(match);
+	}
+};
+#endif
+
+using namespace Swift;
 
 JID::JID(const char* jid) {
 	initializeFromString(std::string(jid));
@@ -126,5 +209,48 @@ int JID::compare(const Swift::JID& o, CompareType compareType) const {
 	return 0;
 }
 
-} // namespace Swift
-
+std::string JID::getEscapedNode(const std::string& node) {
+	std::string result;
+	for (std::string::const_iterator i = node.begin(); i != node.end(); ++i) {
+		if (std::find(escapedChars.begin(), escapedChars.end(), *i) != escapedChars.end()) {
+			result += getEscaped(*i);
+			continue;
+		}
+		else if (*i == '\\') {
+			// Check if we have an escaped dissalowed character sequence
+			std::string::const_iterator innerBegin = i + 1;
+			if (innerBegin != node.end() && innerBegin + 1 != node.end()) {
+				std::string::const_iterator innerEnd = innerBegin + 2;
+				unsigned char value;
+				if (getEscapeSequenceValue(std::string(innerBegin, innerEnd), value)) {
+					result += getEscaped(*i);
+					continue;
+				}
+			}
+		}
+		result += *i;
+	}
+	return result;
+	//return boost::find_format_all_copy(node, UnescapedCharacterFinder(), UnescapedCharacterFormatter());
+}
+					
+std::string JID::getUnescapedNode() const {
+	std::string result;
+	for (std::string::const_iterator j = node_.begin(); j != node_.end();) {
+		if (*j == '\\') {
+			std::string::const_iterator innerEnd = j + 1;
+			for (size_t i = 0; i < 2 && innerEnd != node_.end(); ++i, ++innerEnd) {
+			}
+			unsigned char value;
+			if (getEscapeSequenceValue(std::string(j + 1, innerEnd), value)) {
+				result += std::string(reinterpret_cast<const char*>(&value), 1);
+				j = innerEnd;
+				continue;
+			}
+		}
+		result += *j;
+		++j;
+	}
+	return result;
+	//return boost::find_format_all_copy(node_, EscapedCharacterFinder(), EscapedCharacterFormatter());
+}