Don't break IRC control characters in some encodings

Fix #577

include/znc/Client.h

@@ -82,9 +82,9 @@ protected:
 	CClient* m_pClient;
 };
 
-class CClient : public CZNCSock {
+class CClient : public CIRCSocket {
 public:
-	CClient() : CZNCSock() {
+	CClient() : CIRCSocket() {
 		m_pUser = NULL;
 		m_pNetwork = NULL;
 		m_bGotPass = false;

include/znc/IRCSock.h

@@ -30,7 +30,8 @@ class CIRCNetwork;
 class CClient;
 // !Forward Declarations
 
-class CIRCSock : public CZNCSock {
+// TODO: This class needs new name
+class CIRCSock : public CIRCSocket {
 public:
 	CIRCSock(CIRCNetwork* pNetwork);
 	virtual ~CIRCSock();

include/znc/Socket.h

@@ -220,4 +220,37 @@ protected:
 	CModule*  m_pModule; //!< pointer to the module that this sock instance belongs to
 };
 
+/**
+ * @class CIRCSocket
+ * @brief Base IRC socket for client<->ZNC, and ZNC<->server
+ */
+class CIRCSocket : public CZNCSock {
+public:
+#ifdef HAVE_ICU
+	/**
+	 * @brief Allow IRC control characters to appear even if protocol encoding explicitly disallows them.
+	 *
+	 * E.g. ISO-2022-JP disallows 0x0F, which in IRC means "reset format",
+	 * so by default it gets replaced with U+FFFD ("replacement character").
+	 * https://code.google.com/p/chromium/issues/detail?id=277062#c3
+	 *
+	 * In case if protocol encoding uses these code points for something else, the encoding takes preference,
+	 * and they are not IRC control characters anymore.
+	 */
+	void IcuExtToUCallback(
+		UConverterToUnicodeArgs* toArgs,
+		const char* codeUnits,
+		int32_t length,
+		UConverterCallbackReason reason,
+		UErrorCode* err) override;
+	void IcuExtFromUCallback(
+		UConverterFromUnicodeArgs* fromArgs,
+		const UChar* codeUnits,
+		int32_t length,
+		UChar32 codePoint,
+		UConverterCallbackReason reason,
+		UErrorCode* err) override;
+#endif
+};
+
 #endif /* SOCKET_H */

src/IRCSock.cpp

@@ -54,7 +54,7 @@ bool CIRCSock::IsFloodProtected(double fRate) {
 	return fRate > FLOOD_MINIMAL_RATE;
 }
 
-CIRCSock::CIRCSock(CIRCNetwork* pNetwork) : CZNCSock() {
+CIRCSock::CIRCSock(CIRCNetwork* pNetwork) : CIRCSocket() {
 	m_pNetwork = pNetwork;
 	m_bAuthed = false;
 	m_bNamesx = false;

src/Socket.cpp

@@ -22,6 +22,10 @@
 #include <znc/znc.h>
 #include <signal.h>
 
+#ifdef HAVE_ICU
+#include <unicode/ucnv_cb.h>
+#endif
+
 CZNCSock::CZNCSock(int timeout) : Csock(timeout) {
 #ifdef HAVE_LIBSSL
 	DisableSSLCompression();
@@ -487,3 +491,54 @@ bool CSocket::Listen(unsigned short uPort, bool bSSL, unsigned int uTimeout) {
 
 CModule* CSocket::GetModule() const { return m_pModule; }
 /////////////////// !CSocket ///////////////////
+
+#ifdef HAVE_ICU
+void CIRCSocket::IcuExtToUCallback(
+		UConverterToUnicodeArgs* toArgs,
+		const char* codeUnits,
+		int32_t length,
+		UConverterCallbackReason reason,
+		UErrorCode* err) {
+	// From http://www.mirc.com/colors.html
+	// The Control+O key combination in mIRC inserts ascii character 15,
+	// which turns off all previous attributes, including color, bold, underline, and italics.
+	//
+	// \x02 bold
+	// \x03 mIRC-compatible color
+	// \x04 RRGGBB color
+	// \x0F normal/reset (turn off bold, colors, etc.)
+	// \x12 reverse (weechat)
+	// \x16 reverse (mirc, kvirc)
+	// \x1D italic
+	// \x1F underline
+	// Also see http://www.visualirc.net/tech-attrs.php
+	//
+	// Keep in sync with CUser::AddTimestamp and CIRCSocket::IcuExtFromUCallback
+	static const std::set<char> scAllowedChars = {'\x02', '\x03', '\x04', '\x0F', '\x12', '\x16', '\x1D', '\x1F'};
+	if (reason == UCNV_ILLEGAL && length == 1 && scAllowedChars.count(*codeUnits)) {
+		*err = U_ZERO_ERROR;
+		UChar c = *codeUnits;
+		ucnv_cbToUWriteUChars(toArgs, &c, 1, 0, err);
+		return;
+	}
+	Csock::IcuExtToUCallback(toArgs, codeUnits, length, reason, err);
+}
+
+void CIRCSocket::IcuExtFromUCallback(
+		UConverterFromUnicodeArgs* fromArgs,
+		const UChar* codeUnits,
+		int32_t length,
+		UChar32 codePoint,
+		UConverterCallbackReason reason,
+		UErrorCode* err) {
+	// See comment in CIRCSocket::IcuExtToUCallback
+	static const std::set<UChar32> scAllowedChars = {0x02, 0x03, 0x04, 0x0F, 0x12, 0x16, 0x1D, 0x1F};
+	if (reason == UCNV_ILLEGAL && scAllowedChars.count(codePoint)) {
+		*err = U_ZERO_ERROR;
+		char c = codePoint;
+		ucnv_cbFromUWriteBytes(fromArgs, &c, 1, 0, err);
+		return;
+	}
+	Csock::IcuExtFromUCallback(fromArgs, codeUnits, length, codePoint, reason, err);
+}
+#endif

src/User.cpp

@@ -577,6 +577,8 @@ CString CUser::AddTimestamp(time_t tm, const CString& sStr) const {
 			// \x1D italic
 			// \x1F underline
 			// Also see http://www.visualirc.net/tech-attrs.php
+			//
+			// Keep in sync with CIRCSocket::IcuExt__UCallback
 			if (CString::npos != sRet.find_first_of("\x02\x03\x04\x0F\x12\x16\x1D\x1F")) {
 				sRet += "\x0F";
 			}