IDN testsuite and fixes

Net/include/Poco/Net/DNS.h

@@ -32,7 +32,17 @@ class Net_API DNS
 	/// This class provides an interface to the
 	/// domain name service.
 	///
-	/// An internal DNS cache is used to speed up name lookups.
+	/// Starting with POCO C++ Libraries release 1.9.0,
+	/// this class also supports Internationalized Domain Names (IDNs).
+	///
+	/// Regarding IDNs, the following rules apply:
+	///
+	///   * An IDN passed to hostByName() must be encoded manually, by calling
+	///     encodeIDN() (after testing with isIDN() first).
+	///   * An UTF-8 IDN passed to resolve() or resolveOne() is automatically encoded.
+	///   * IDNs returned in HostEntry objects are never decoded. They can be
+	///     decoded by calling decodeIDN() (after testing for an encoded IDN by
+	///     calling isEncodedIDN()).
 {
 public:
 	enum HintFlag
@@ -60,6 +70,9 @@ public:
 		/// for the host with the given name. HintFlag argument is only
 		/// used on platforms that have getaddrinfo().
 		///
+		/// Note that Internationalized Domain Names must be encoded
+		/// using Punycode (see encodeIDN()) before calling this method.
+		///
 		/// Throws a HostNotFoundException if a host with the given
 		/// name cannot be found.
 		///
@@ -93,8 +106,8 @@ public:
 		/// for the host with the given IP address or host name.
 		///
 		/// If address contains a UTF-8 encoded IDN (internationalized
-		/// domain name), the domain name will be encoded first according
-		/// to Punycode.
+		/// domain name), the domain name will be encoded first using
+		/// Punycode.
 		///
 		/// Throws a HostNotFoundException if a host with the given
 		/// name cannot be found.

Net/src/DNS.cpp

@@ -164,9 +164,18 @@ HostEntry DNS::resolve(const std::string& address)
 {
 	IPAddress ip;
 	if (IPAddress::tryParse(address, ip))
+	{
 		return hostByAddress(ip);
+	}
+	else if (isIDN(address))
+	{
+		std::string encoded = encodeIDN(address);
+		return hostByName(encoded);
+	}
 	else
+	{
 		return hostByName(address);
+	}
 }
 
 
@@ -235,11 +244,16 @@ std::string DNS::encodeIDN(const std::string& idn)
 	while (it != end)
 	{
 		std::string label;
+		bool mustEncode = false;
 		while (it != end && *it != '.')
 		{
+			if (static_cast<unsigned char>(*it) >= 0x80) mustEncode = true;
 			label += *it++;
 		}
-		encoded += encodeIDNLabel(label);
+		if (mustEncode)
+			encoded += encodeIDNLabel(label);
+		else
+			encoded += label;
 		if (it != end) encoded += *it++;
 	}
 	return encoded;
@@ -258,10 +272,7 @@ std::string DNS::decodeIDN(const std::string& encodedIDN)
 		{
 			label += *it++;
 		}
-		if (label.compare(0, 4, "xn--") == 0)
-			decoded += decodeIDNLabel(label);
-		else
-			decoded += label;
+		decoded += decodeIDNLabel(label);
 		if (it != end) decoded += *it++;
 	}
 	return decoded;
@@ -289,8 +300,8 @@ std::string DNS::encodeIDNLabel(const std::string& label)
 	char buffer[64];
 	std::size_t size = 64;
 	int rc = punycode_encode(uniLabel.size(), &uniLabel[0], &size, buffer);
-	if (rc > 0)
-		return encoded.append(buffer, size);
+	if (rc == punycode_success)
+		encoded.append(buffer, size);
 	else
 		throw DNSException("Failed to encode IDN label", label);
 	return encoded;
@@ -305,12 +316,12 @@ std::string DNS::decodeIDNLabel(const std::string& encodedIDN)
 		std::size_t size = 64;
 		punycode_uint buffer[64];
 		int rc = punycode_decode(encodedIDN.size() - 4, encodedIDN.data() + 4, &size, buffer);
-		if (rc > 0)
+		if (rc == punycode_success)
 		{
 			Poco::UTF32Encoding utf32;
 			Poco::UTF8Encoding utf8;
 			Poco::TextConverter converter(utf32, utf8);
-			converter.convert(buffer, size*rc, decoded);
+			converter.convert(buffer, size*sizeof(punycode_uint), decoded);
 		}
 		else throw DNSException("Failed to decode IDN label: ", encodedIDN);
 	}
@@ -394,7 +405,7 @@ void DNS::aierror(int code, const std::string& arg)
   2018-02-17 with SHA-1 a966a8017f6be579d74a50a226accc7607c40133
   labeled punycode-spec 1.0.3 (2006-Mar-23-Thu).
 
-  Modified for POCO C++ Libraries by Günter Obiltschnig.
+  Modified for POCO C++ Libraries by Guenter Obiltschnig.
 
   License on the original code:
 
@@ -526,8 +537,8 @@ int punycode_encode(size_t input_length_orig, const punycode_uint input[], size_
 			if (max_out - out < 2) return punycode_big_output;
 			output[out++] = (char) input[j];
 		}
-	/* else if (input[j] < n) return punycode_bad_input; */
-	/* (not needed for Punycode with unsigned code points) */
+		/* else if (input[j] < n) return punycode_bad_input; */
+		/* (not needed for Punycode with unsigned code points) */
 	}
 
 	h = b = (punycode_uint) out;

Net/testsuite/src/DNSTest.cpp

@@ -93,14 +93,89 @@ void DNSTest::testResolve()
 
 void DNSTest::testEncodeIDN()
 {
-	std::string idn("d\xc3\xb6m\xc3\xa4in.example"); // domain.example with umlauts
+	std::string idn("d\xc3\xb6m\xc3\xa4in.example"); // d"om"ain.example 
 	assert (DNS::isIDN(idn));
 	assert (DNS::encodeIDN(idn) == "xn--dmin-moa0i.example");
+
+	idn = ".d\xc3\xb6m\xc3\xa4in.example"; // .d"om"ain.example 
+	assert(DNS::isIDN(idn));
+	assert(DNS::encodeIDN(idn) == ".xn--dmin-moa0i.example");
+
+	idn = "d\xc3\xb6m\xc3\xa4in.example."; // .d"om"ain.example.
+	assert(DNS::isIDN(idn));
+	assert(DNS::encodeIDN(idn) == "xn--dmin-moa0i.example.");
+
+	idn = "d\xc3\xb6m\xc3\xa4in"; // d"om"ain
+	assert(DNS::isIDN(idn));
+	assert(DNS::encodeIDN(idn) == "xn--dmin-moa0i");
+
+	idn = "\xc3\xa4""aaa.example"; // "aaaa.example
+	assert (DNS::isIDN(idn));
+	assert (DNS::encodeIDN(idn) == "xn--aaa-pla.example");
+
+	idn = "a\xc3\xa4""aa.example"; // a"aaa.example
+	assert(DNS::isIDN(idn));
+	assert(DNS::encodeIDN(idn) == "xn--aaa-qla.example");
+
+	idn = "foo.\xc3\xa2""bcd\xc3\xa9""f.example"; // foo.^abcd'ef.example
+	assert(DNS::isIDN(idn));
+	assert(DNS::encodeIDN(idn) == "foo.xn--bcdf-9na9b.example");
+
+	idn = "\xe2\x98\x83.example"; // <snowman>.example
+	assert(DNS::isIDN(idn));
+	assert(DNS::encodeIDN(idn) == "xn--n3h.example");
+
+	idn = "\xe2\x98\x83."; // <snowman>.
+	assert(DNS::isIDN(idn));
+	assert(DNS::encodeIDN(idn) == "xn--n3h.");
+
+	idn = "\xe2\x98\x83"; // <snowman>
+	assert(DNS::isIDN(idn));
+	assert(DNS::encodeIDN(idn) == "xn--n3h");
+
+	std::string dn = "www.pocoproject.org";
+	assert (!DNS::isIDN(dn));
+	assert (DNS::encodeIDN(dn) == "www.pocoproject.org");
 }
 
 
 void DNSTest::testDecodeIDN()
 {
+	std::string enc("xn--dmin-moa0i.example");
+	assert (DNS::isEncodedIDN(enc));
+	assert (DNS::decodeIDN(enc) == "d\xc3\xb6m\xc3\xa4in.example"); // d"om"ain.example 
+
+	enc = ".xn--dmin-moa0i.example";
+	assert(DNS::isEncodedIDN(enc));
+	assert(DNS::decodeIDN(enc) == ".d\xc3\xb6m\xc3\xa4in.example"); // .d"om"ain.example 
+
+	enc = "xn--dmin-moa0i.example.";
+	assert(DNS::isEncodedIDN(enc));
+	assert(DNS::decodeIDN(enc) == "d\xc3\xb6m\xc3\xa4in.example."); // d"om"ain.example.
+
+	enc = "xn--dmin-moa0i";
+	assert(DNS::isEncodedIDN(enc));
+	assert(DNS::decodeIDN(enc) == "d\xc3\xb6m\xc3\xa4in"); // d"om"ain
+
+	enc = "foo.xn--bcdf-9na9b.example";
+	assert (DNS::isEncodedIDN(enc));
+	assert (DNS::decodeIDN(enc) == "foo.\xc3\xa2""bcd\xc3\xa9""f.example"); // foo.^abcd'ef.example
+
+	enc = "xn--n3h.example";
+	assert (DNS::isEncodedIDN(enc));
+	assert (DNS::decodeIDN(enc) == "\xe2\x98\x83.example"); // <snowman>.example
+
+	enc = "xn--n3h.";
+	assert(DNS::isEncodedIDN(enc));
+	assert(DNS::decodeIDN(enc) == "\xe2\x98\x83."); // <snowman>.
+
+	enc = "xn--n3h";
+	assert(DNS::isEncodedIDN(enc));
+	assert(DNS::decodeIDN(enc) == "\xe2\x98\x83"); // <snowman>
+
+	std::string dn = "www.pocoproject.org";
+	assert (!DNS::isEncodedIDN(dn));
+	assert (DNS::decodeIDN(dn) == "www.pocoproject.org");
 }