ICE-6991 - Add support for unicode escape sequences

2 files changed, 184 insertions, 1 deletions

diff --git a/php/test/Ice/operations/Client.php b/php/test/Ice/operations/Client.php
index 40583fb9e05..da449507589 100644
--- a/php/test/Ice/operations/Client.php
+++ b/php/test/Ice/operations/Client.php
@@ -41,6 +41,123 @@ function twoways($communicator, $p)
     $myDerivedClass = $NS ? "Test\\MyDerivedClass" : "Test_MyDerivedClass";
     $myClass = $NS ? "Test\\MyClass" : "Test_MyClass";
     $objectPrxHelper = $NS ? "Ice\\ObjectPrxHelper" : "Ice_ObjectPrxHelper";
+    
+    
+    $s0 = $NS ? constant("Test\\s0") : constant("Test_s0");
+    $s1 = $NS ? constant("Test\\s1") : constant("Test_s1");
+    $s2 = $NS ? constant("Test\\s2") : constant("Test_s2");
+    $s3 = $NS ? constant("Test\\s3") : constant("Test_s3");
+    $s4 = $NS ? constant("Test\\s4") : constant("Test_s4");
+    $s5 = $NS ? constant("Test\\s5") : constant("Test_s5");
+    $s6 = $NS ? constant("Test\\s6") : constant("Test_s6");
+    $s7 = $NS ? constant("Test\\s7") : constant("Test_s7");
+    $s8 = $NS ? constant("Test\\s8") : constant("Test_s8");
+    $s9 = $NS ? constant("Test\\s9") : constant("Test_s9");
+    $s10 = $NS ? constant("Test\\s10") : constant("Test_s10");
+    
+    $sw0 = $NS ? constant("Test\\sw0") : constant("Test_sw0");
+    $sw1 = $NS ? constant("Test\\sw1") : constant("Test_sw1");
+    $sw2 = $NS ? constant("Test\\sw2") : constant("Test_sw2");
+    $sw3 = $NS ? constant("Test\\sw3") : constant("Test_sw3");
+    $sw4 = $NS ? constant("Test\\sw4") : constant("Test_sw4");
+    $sw5 = $NS ? constant("Test\\sw5") : constant("Test_sw5");
+    $sw6 = $NS ? constant("Test\\sw6") : constant("Test_sw6");
+    $sw7 = $NS ? constant("Test\\sw7") : constant("Test_sw7");
+    $sw8 = $NS ? constant("Test\\sw8") : constant("Test_sw8");
+    $sw9 = $NS ? constant("Test\\sw9") : constant("Test_sw9");
+    $sw10 = $NS ? constant("Test\\sw10") : constant("Test_sw10");
+
+    $ss0 = $NS ? constant("Test\\ss0") : constant("Test_ss0");
+    $ss1 = $NS ? constant("Test\\ss1") : constant("Test_ss1");
+    $ss2 = $NS ? constant("Test\\ss2") : constant("Test_ss2");
+    $ss3 = $NS ? constant("Test\\ss3") : constant("Test_ss3");
+    $ss4 = $NS ? constant("Test\\ss4") : constant("Test_ss4");
+    $ss5 = $NS ? constant("Test\\ss5") : constant("Test_ss5");
+
+    $su0 = $NS ? constant("Test\\su0") : constant("Test_su0");
+    $su1 = $NS ? constant("Test\\su1") : constant("Test_su1");
+    $su2 = $NS ? constant("Test\\su2") : constant("Test_su2");
+
+    {
+        $literals = $p->opStringLiterals();
+
+        test($s0 == "\\" && 
+             $s0 == $sw0 &&
+             $s0 == $literals[0] &&
+             $s0 == $literals[11]);
+
+        test($s1 == "A" &&
+             $s1 == $sw1 &&
+             $s1 == $literals[1] &&
+             $s1 == $literals[12]);
+
+        test($s2 == "Ice" &&
+             $s2 == $sw2 &&
+             $s2 == $literals[2] &&
+             $s2 == $literals[13]);
+        
+        test($s3 == "A21" && 
+             $s3 == $sw3 &&
+             $s3 == $literals[3] &&
+             $s3 == $literals[14]);
+
+        test($s4 == "\\u0041 \\U00000041" &&
+             $s4 == $sw4 &&
+             $s4 == $literals[4] &&
+             $s4 == $literals[15]);
+
+        test($s5 == "\xc3\xbf" &&
+             $s5 == $sw5 &&
+             $s5 == $literals[5] &&
+             $s5 == $literals[16]);
+        
+        test($s6 == "\xcf\xbf" &&
+             $s6 == $sw6 &&
+             $s6 == $literals[6] &&
+             $s6 == $literals[17]);
+        
+        test($s7 == "\xd7\xb0" &&
+             $s7 == $sw7 &&
+             $s7 == $literals[7] &&
+             $s7 == $literals[18]);
+        
+        test($s8 == "\xf0\x90\x80\x80" &&
+             $s8 == $sw8 &&
+             $s8 == $literals[8] &&
+             $s8 == $literals[19]);
+        
+        test($s9 == "\xf0\x9f\x8d\x8c" &&
+             $s9 == $sw9 &&
+             $s9 == $literals[9] &&
+             $s9 == $literals[20]);
+
+        test($s10 == "\xe0\xb6\xa7" &&
+             $s10 == $sw10 &&
+             $s10 == $literals[10] &&
+             $s10 == $literals[21]);
+        
+        test($ss0 == "'\"?\\\007\010\f\n\r\t\v" &&
+             $ss0 == $ss1 &&
+             $ss1 == $ss2 &&
+             $ss0 == $literals[22] &&
+             $ss0 == $literals[23] &&
+             $ss0 == $literals[24]);
+        
+        test($ss3 == "\\\\U\\u\\" &&
+             $ss3 == $literals[25]);
+        
+        test($ss4 == "\\A\\" &&
+             $ss4 == $literals[26]);
+        
+        test($ss5 == "\\u0041\\" &&
+             $ss5 == $literals[27]);
+        
+        test($su0 == $su1 &&
+             $su0 == $su2 &&
+             $su0 == $literals[28] &&
+             $su0 == $literals[29] &&
+             $su0 == $literals[30]);
+    }
 
     {
         $p->ice_ping();
diff --git a/php/test/Ice/operations/Test.ice b/php/test/Ice/operations/Test.ice
index 4d314fc8e0b..7c37371e0b9 100644
--- a/php/test/Ice/operations/Test.ice
+++ b/php/test/Ice/operations/Test.ice
@@ -224,6 +224,8 @@ class MyClass
     ByteBoolD opByteBoolD1(ByteBoolD opByteBoolD1);
     StringS opStringS2(StringS stringS);
     ByteBoolD opByteBoolD2(ByteBoolD byteBoolD);
+
+    StringS opStringLiterals();
 };
 
 // Test data member with same name as class
@@ -249,5 +251,69 @@ class MyDerivedClass extends MyClass
     MyClass1 opMyClass1(MyClass1 c);
 };
 
-};
+//
+// 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";
+
+};