ICE-6991 - Add support for unicode escape sequences

5 files changed, 226 insertions, 0 deletions

diff --git a/python/test/Ice/operations/ServerAMD.py b/python/test/Ice/operations/ServerAMD.py
index 39e51b08c77..3561221335a 100755
--- a/python/test/Ice/operations/ServerAMD.py
+++ b/python/test/Ice/operations/ServerAMD.py
@@ -403,6 +403,12 @@ class MyDerivedClassI(Test.MyDerivedClass):
     def opMyStruct1_async(self, cb, value, current=None):
         return cb.ice_response(value)
     
+    def opStringLiterals_async(self, cb, current=None):
+        return cb.ice_response([
+                Test.s0, Test.s1, Test.s2, Test.s3, Test.s4, Test.s5, Test.s6, Test.s7, Test.s8, Test.s9, Test.s10,
+                Test.sw0, Test.sw1, Test.sw2, Test.sw3, Test.sw4, Test.sw5, Test.sw6, Test.sw7, Test.sw8, Test.sw9, Test.sw10,
+                Test.ss0, Test.ss1, Test.ss2, Test.ss3, Test.ss4, Test.ss5,
+                Test.su0, Test.su1, Test.su2])
 
 def run(args, communicator):
     communicator.getProperties().setProperty("TestAdapter.Endpoints", "default -p 12010:udp")
diff --git a/python/test/Ice/operations/Test.ice b/python/test/Ice/operations/Test.ice
index f2a46f54e9b..7f5ec510834 100644
--- a/python/test/Ice/operations/Test.ice
+++ b/python/test/Ice/operations/Test.ice
@@ -246,6 +246,8 @@ class MyClass
     ByteBoolD opByteBoolD1(ByteBoolD opByteBoolD1);
     StringS opStringS2(StringS stringS);
     ByteBoolD opByteBoolD2(ByteBoolD byteBoolD);
+    
+    StringS opStringLiterals();
 };
 
 struct MyStruct1
@@ -269,4 +271,69 @@ class MyDerivedClass extends MyClass
     MyStruct1 opMyStruct1(MyStruct1 opMyStruct1);
 };
 
+//
+// String literals
+//
+
+const string s0 = "\u005c";                           // backslash
+const string s1 = "\u0041";                           // A
+const string s2 = "\u0049\u0063\u0065";               // Ice
+const string s3 = "\u004121";                         // A21
+const string s4 = "\\u0041 \\U00000041";              // \\u0041 \\U00000041
+const string s5 = "\u00FF";                           // ÿ
+const string s6 = "\u03FF";                           // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL (U+03FF)
+const string s7 = "\u05F0";                           // HEBREW LIGATURE YIDDISH DOUBLE VAV (U+05F0)
+const string s8 = "\U00010000";                       // LINEAR B SYLLABLE B008 A (U+10000)
+const string s9 = "\U0001F34C";                       // BANANA (U+1F34C)
+const string s10 = "\u0DA7";                          // Sinhala Letter Alpapraana Ttayanna
+
+const string sw0 = "\U0000005c";                      // backslash
+const string sw1 = "\U00000041";                      // A
+const string sw2 = "\U00000049\U00000063\U00000065";  // Ice
+const string sw3 = "\U0000004121";                    // A21
+const string sw4 = "\\u0041 \\U00000041";             // \\u0041 \\U00000041
+const string sw5 = "\U000000FF";                      // ÿ
+const string sw6 = "\U000003FF";                      // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL (U+03FF)
+const string sw7 = "\U000005F0";                      // HEBREW LIGATURE YIDDISH DOUBLE VAV (U+05F0)
+const string sw8 = "\U00010000";                      // LINEAR B SYLLABLE B008 A (U+10000)
+const string sw9 = "\U0001F34C";                      // BANANA (U+1F34C)
+const string sw10 = "\U00000DA7";                     // Sinhala Letter Alpapraana Ttayanna
+
+/**
+\'	single quote	byte 0x27 in ASCII encoding
+\"	double quote	byte 0x22 in ASCII encoding
+\?	question mark	byte 0x3f in ASCII encoding
+\\	backslash	byte 0x5c in ASCII encoding
+\a	audible bell	byte 0x07 in ASCII encoding
+\b	backspace	byte 0x08 in ASCII encoding
+\f	form feed - new page	byte 0x0c in ASCII encoding
+\n	line feed - new line	byte 0x0a in ASCII encoding
+\r	carriage return	byte 0x0d in ASCII encoding
+\t	horizontal tab	byte 0x09 in ASCII encoding
+\v	vertical tab	byte 0x0b in ASCII encoding
+**/
+
+const string ss0 = "\'\"\?\\\a\b\f\n\r\t\v";
+const string ss1 = "\u0027\u0022\u003f\u005c\u0007\u0008\u000c\u000a\u000d\u0009\u000b";
+const string ss2 = "\U00000027\U00000022\U0000003f\U0000005c\U00000007\U00000008\U0000000c\U0000000a\U0000000d\U00000009\U0000000b";
+
+const string ss3 = "\\\\U\\u\\"; /* \\U\u\  */
+const string ss4 = "\\\u0041\\"; /* \A\     */
+const string ss5 = "\\u0041\\";  /* \u0041\ */
+
+//
+// ÿ - Unicode Character 'LATIN SMALL LETTER Y WITH DIAERESIS' (U+00FF)
+// Ā - Unicode Character 'LATIN CAPITAL LETTER A WITH MACRON' (U+0100)
+// ἀ - Unicode Character 'GREEK SMALL LETTER ALPHA WITH PSILI' (U+1F00)
+// 𐆔 - Unicode Character 'ROMAN DIMIDIA SEXTULA SIGN' (U+10194)
+// 𐅪 - Unicode Character 'GREEK ACROPHONIC THESPIAN ONE HUNDRED' (U+1016A)
+// 𐆘 - Unicode Character 'ROMAN SESTERTIUS SIGN' (U+10198)
+// 🍀 - Unicode Character 'FOUR LEAF CLOVER' (U+1F340)
+// 🍁 - Unicode Character 'MAPLE LEAF' (U+1F341)
+// 🍂 - Unicode Character 'FALLEN LEAF' (U+1F342)
+// 🍃 - Unicode Character 'LEAF FLUTTERING IN WIND' (U+1F343)
+const string su0 = "ÿĀἀ𐆔𐅪𐆘🍀🍁🍂🍃";
+const string su1 = "\u00FF\u0100\u1F00\U00010194\U0001016A\U00010198\U0001F340\U0001F341\U0001F342\U0001F343";
+const string su2 = "\U000000FF\U00000100\U00001F00\U00010194\U0001016A\U00010198\U0001F340\U0001F341\U0001F342\U0001F343";
+
 };
diff --git a/python/test/Ice/operations/TestAMD.ice b/python/test/Ice/operations/TestAMD.ice
index 022dca4bba4..ba567612d60 100644
--- a/python/test/Ice/operations/TestAMD.ice
+++ b/python/test/Ice/operations/TestAMD.ice
@@ -244,6 +244,8 @@ dictionary<MyEnum, MyEnumS> MyEnumMyEnumSD;
     ByteBoolD opByteBoolD1(ByteBoolD opByteBoolD1);
     StringS opStringS2(StringS stringS);
     ByteBoolD opByteBoolD2(ByteBoolD byteBoolD);
+    
+    StringS opStringLiterals();
 };
 
 struct MyStruct1
@@ -267,4 +269,69 @@ class MyClass1
     MyStruct1 opMyStruct1(MyStruct1 opMyStruct1);
 };
 
+//
+// String literals
+//
+
+const string s0 = "\u005c";                           // backslash
+const string s1 = "\u0041";                           // A
+const string s2 = "\u0049\u0063\u0065";               // Ice
+const string s3 = "\u004121";                         // A21
+const string s4 = "\\u0041 \\U00000041";              // \\u0041 \\U00000041
+const string s5 = "\u00FF";                           // ÿ
+const string s6 = "\u03FF";                           // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL (U+03FF)
+const string s7 = "\u05F0";                           // HEBREW LIGATURE YIDDISH DOUBLE VAV (U+05F0)
+const string s8 = "\U00010000";                       // LINEAR B SYLLABLE B008 A (U+10000)
+const string s9 = "\U0001F34C";                       // BANANA (U+1F34C)
+const string s10 = "\u0DA7";                          // Sinhala Letter Alpapraana Ttayanna
+
+const string sw0 = "\U0000005c";                      // backslash
+const string sw1 = "\U00000041";                      // A
+const string sw2 = "\U00000049\U00000063\U00000065";  // Ice
+const string sw3 = "\U0000004121";                    // A21
+const string sw4 = "\\u0041 \\U00000041";             // \\u0041 \\U00000041
+const string sw5 = "\U000000FF";                      // ÿ
+const string sw6 = "\U000003FF";                      // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL (U+03FF)
+const string sw7 = "\U000005F0";                      // HEBREW LIGATURE YIDDISH DOUBLE VAV (U+05F0)
+const string sw8 = "\U00010000";                      // LINEAR B SYLLABLE B008 A (U+10000)
+const string sw9 = "\U0001F34C";                      // BANANA (U+1F34C)
+const string sw10 = "\U00000DA7";                     // Sinhala Letter Alpapraana Ttayanna
+
+/**
+\'	single quote	byte 0x27 in ASCII encoding
+\"	double quote	byte 0x22 in ASCII encoding
+\?	question mark	byte 0x3f in ASCII encoding
+\\	backslash	byte 0x5c in ASCII encoding
+\a	audible bell	byte 0x07 in ASCII encoding
+\b	backspace	byte 0x08 in ASCII encoding
+\f	form feed - new page	byte 0x0c in ASCII encoding
+\n	line feed - new line	byte 0x0a in ASCII encoding
+\r	carriage return	byte 0x0d in ASCII encoding
+\t	horizontal tab	byte 0x09 in ASCII encoding
+\v	vertical tab	byte 0x0b in ASCII encoding
+**/
+
+const string ss0 = "\'\"\?\\\a\b\f\n\r\t\v";
+const string ss1 = "\u0027\u0022\u003f\u005c\u0007\u0008\u000c\u000a\u000d\u0009\u000b";
+const string ss2 = "\U00000027\U00000022\U0000003f\U0000005c\U00000007\U00000008\U0000000c\U0000000a\U0000000d\U00000009\U0000000b";
+
+const string ss3 = "\\\\U\\u\\"; /* \\U\u\  */
+const string ss4 = "\\\u0041\\"; /* \A\     */
+const string ss5 = "\\u0041\\";  /* \u0041\ */
+
+//
+// ÿ - Unicode Character 'LATIN SMALL LETTER Y WITH DIAERESIS' (U+00FF)
+// Ā - Unicode Character 'LATIN CAPITAL LETTER A WITH MACRON' (U+0100)
+// ἀ - Unicode Character 'GREEK SMALL LETTER ALPHA WITH PSILI' (U+1F00)
+// 𐆔 - Unicode Character 'ROMAN DIMIDIA SEXTULA SIGN' (U+10194)
+// 𐅪 - Unicode Character 'GREEK ACROPHONIC THESPIAN ONE HUNDRED' (U+1016A)
+// 𐆘 - Unicode Character 'ROMAN SESTERTIUS SIGN' (U+10198)
+// 🍀 - Unicode Character 'FOUR LEAF CLOVER' (U+1F340)
+// 🍁 - Unicode Character 'MAPLE LEAF' (U+1F341)
+// 🍂 - Unicode Character 'FALLEN LEAF' (U+1F342)
+// 🍃 - Unicode Character 'LEAF FLUTTERING IN WIND' (U+1F343)
+const string su0 = "ÿĀἀ𐆔𐅪𐆘🍀🍁🍂🍃";
+const string su1 = "\u00FF\u0100\u1F00\U00010194\U0001016A\U00010198\U0001F340\U0001F341\U0001F342\U0001F343";
+const string su2 = "\U000000FF\U00000100\U00001F00\U00010194\U0001016A\U00010198\U0001F340\U0001F341\U0001F342\U0001F343";
+
 };
diff --git a/python/test/Ice/operations/TestI.py b/python/test/Ice/operations/TestI.py
index 5035d00cb3a..6700fbfa26e 100644
--- a/python/test/Ice/operations/TestI.py
+++ b/python/test/Ice/operations/TestI.py
@@ -361,5 +361,11 @@ class MyDerivedClassI(Test.MyDerivedClass):
 
     def opMyStruct1(self, value, current=None):
         return value
+    
+    def opStringLiterals(self, current=None):
+        return [Test.s0, Test.s1, Test.s2, Test.s3, Test.s4, Test.s5, Test.s6, Test.s7, Test.s8, Test.s9, Test.s10,
+                Test.sw0, Test.sw1, Test.sw2, Test.sw3, Test.sw4, Test.sw5, Test.sw6, Test.sw7, Test.sw8, Test.sw9, Test.sw10,
+                Test.ss0, Test.ss1, Test.ss2, Test.ss3, Test.ss4, Test.ss5,
+                Test.su0, Test.su1, Test.su2]
 
     
\ No newline at end of file
diff --git a/python/test/Ice/operations/Twoways.py b/python/test/Ice/operations/Twoways.py
index ed0d9e3ae3e..04dcee69f1a 100644
--- a/python/test/Ice/operations/Twoways.py
+++ b/python/test/Ice/operations/Twoways.py
@@ -14,6 +14,86 @@ def test(b):
         raise RuntimeError('test assertion failed')
 
 def twoways(communicator, p):
+    
+    literals = p.opStringLiterals();
+        
+    test(Test.s0 == "\\" and
+         Test.s0 == Test.sw0 and
+         Test.s0 == literals[0] and
+         Test.s0 == literals[11]);
+
+    test(Test.s1 == "A" and
+         Test.s1 == Test.sw1 and
+         Test.s1 == literals[1] and
+         Test.s1 == literals[12]);
+
+    test(Test.s2 == "Ice" and
+         Test.s2 == Test.sw2 and
+         Test.s2 == literals[2] and
+         Test.s2 == literals[13]);
+    
+    test(Test.s3 == "A21" and 
+         Test.s3 == Test.sw3 and
+         Test.s3 == literals[3] and
+         Test.s3 == literals[14]);
+
+    test(Test.s4 == "\\u0041 \\U00000041" and
+         Test.s4 == Test.sw4 and
+         Test.s4 == literals[4] and
+         Test.s4 == literals[15]);
+
+    test(Test.s5 == "\xc3\xbf" and
+         Test.s5 == Test.sw5 and
+         Test.s5 == literals[5] and
+         Test.s5 == literals[16]);
+
+    test(Test.s6 == "\xcf\xbf" and
+         Test.s6 == Test.sw6 and
+         Test.s6 == literals[6] and
+         Test.s6 == literals[17]);
+
+    test(Test.s7 == "\xd7\xb0" and
+         Test.s7 == Test.sw7 and
+         Test.s7 == literals[7] and
+         Test.s7 == literals[18]);
+
+    test(Test.s8 == "\xf0\x90\x80\x80" and
+         Test.s8 == Test.sw8 and
+         Test.s8 == literals[8] and
+         Test.s8 == literals[19]);
+    
+    test(Test.s9 == "\xf0\x9f\x8d\x8c" and
+         Test.s9 == Test.sw9 and
+         Test.s9 == literals[9] and
+         Test.s9 == literals[20]);
+
+    test(Test.s10 == "\xe0\xb6\xa7" and
+         Test.s10 == Test.sw10 and
+         Test.s10 == literals[10] and
+         Test.s10 == literals[21]);
+
+    test(Test.ss0 == "\'\"\x3f\\\a\b\f\n\r\t\v" and
+         Test.ss0 == Test.ss1 and
+         Test.ss0 == Test.ss2 and
+         Test.ss0 == literals[22] and
+         Test.ss0 == literals[23] and
+         Test.ss0 == literals[24]);
+    
+    test(Test.ss3 == "\\\\U\\u\\" and
+         Test.ss3 == literals[25]);
+
+    test(Test.ss4 == "\\A\\" and
+         Test.ss4 == literals[26]);
+
+    test(Test.ss5 == "\\u0041\\" and
+         Test.ss5 == literals[27]);
+            
+    test(Test.su0 == Test.su1 and
+         Test.su0 == Test.su2 and
+         Test.su0 == literals[28] and
+         Test.su0 == literals[29] and
+         Test.su0 == literals[30]);
+
     #
     # ice_ping
     #