Implement native scalar name aliases

Author: Azoy

stdlib/public/SwiftShims/UnicodeData.h

@@ -73,6 +73,9 @@ __swift_uint8_t _swift_stdlib_getNumericType(__swift_uint32_t scalar);
 SWIFT_RUNTIME_STDLIB_INTERNAL
 double _swift_stdlib_getNumericValue(__swift_uint32_t scalar);
 
+SWIFT_RUNTIME_STDLIB_INTERNAL
+const char *_swift_stdlib_getNameAlias(__swift_uint32_t);
+
 #ifdef __cplusplus
 } // extern "C"
 #endif

stdlib/public/core/UnicodeScalarProperties.swift

@@ -1238,7 +1238,13 @@ extension Unicode.Scalar.Properties {
   /// This property corresponds to the "Name_Alias" property in the
   /// [Unicode Standard](http://www.unicode.org/versions/latest/).
   public var nameAlias: String? {
-    return _scalarName(__swift_stdlib_U_CHAR_NAME_ALIAS)
+    let nameAliasPtr = _swift_stdlib_getNameAlias(_scalar.value)
+
+    guard nameAliasPtr != nil else {
+      return nil
+    }
+
+    return String(cString: nameAliasPtr!)
   }
 }
 

stdlib/public/stubs/UnicodeScalarProps.cpp

@@ -117,3 +117,16 @@ double _swift_stdlib_getNumericValue(__swift_uint32_t scalar) {
   auto valueIdx = _swift_stdlib_numeric_values_indices[scalarIdx];
   return _swift_stdlib_numeric_values[valueIdx];
 }
+
+SWIFT_RUNTIME_STDLIB_INTERNAL
+const char *_swift_stdlib_getNameAlias(__swift_uint32_t scalar) {
+  auto dataIdx = _swift_stdlib_getScalarBitArrayIdx(scalar,
+                                                    _swift_stdlib_nameAlias,
+                                                  _swift_stdlib_nameAlias_ranks);
+
+  if (dataIdx == std::numeric_limits<__swift_intptr_t>::max()) {
+    return nullptr;
+  }
+
+  return _swift_stdlib_nameAlias_data[dataIdx];
+}

stdlib/public/stubs/UnicodeScalarProps.h

@@ -909,3 +909,37 @@ static const __swift_uint8_t _swift_stdlib_numeric_values_indices[1872] = {
   0x42, 0xB, 0x31, 0x42, 0x1C, 0x42, 0x65, 0x1A, 0x65, 0x42, 0x42, 0x2, 0x65, 0x87, 0x87, 0x42,
   0x50, 0x87, 0xE, 0x38, 0xB, 0x1E, 0x87, 0x6, 0xE,
 };
+
+static const char * const _swift_stdlib_nameAlias_data[29] = {
+  "LATIN CAPITAL LETTER GHA", "LATIN SMALL LETTER GHA", "SYRIAC SUBLINEAR COLON SKEWED LEFT",
+  "KANNADA LETTER LLLA", "LAO LETTER FO FON", "LAO LETTER FO FAY", "LAO LETTER RO", "LAO LETTER LO",
+  "TIBETAN MARK BKA- SHOG GI MGO RGYAN", "HANGUL JONGSEONG YESIEUNG-KIYEOK",
+  "HANGUL JONGSEONG YESIEUNG-SSANGKIYEOK", "HANGUL JONGSEONG SSANGYESIEUNG",
+  "HANGUL JONGSEONG YESIEUNG-KHIEUKH", "WEIERSTRASS ELLIPTIC FUNCTION", "MICR ON US SYMBOL",
+  "MICR DASH SYMBOL", "LEFTWARDS TRIANGLE-HEADED ARROW WITH DOUBLE VERTICAL STROKE",
+  "RIGHTWARDS TRIANGLE-HEADED ARROW WITH DOUBLE VERTICAL STROKE", "YI SYLLABLE ITERATION MARK",
+  "MYANMAR LETTER KHAMTI LLA", "PRESENTATION FORM FOR VERTICAL RIGHT WHITE LENTICULAR BRACKET",
+  "CUNEIFORM SIGN NU11 TENU", "CUNEIFORM SIGN NU11 OVER NU11 BUR OVER BUR",
+  "MEDEFAIDRIN CAPITAL LETTER H", "MEDEFAIDRIN CAPITAL LETTER NG", "MEDEFAIDRIN SMALL LETTER H",
+  "MEDEFAIDRIN SMALL LETTER NG", "HENTAIGANA LETTER E-1",
+  "BYZANTINE MUSICAL SYMBOL FTHORA SKLIRON CHROMA VASIS",
+};
+
+static const __swift_uint16_t _swift_stdlib_nameAlias_ranks[87] = {
+  0x0, 0x9, 0x9, 0xB, 0xC, 0xD, 0xE, 0x0, 0x0, 0x0, 0x2, 0x2, 0x0, 0x0, 0x0, 0x1, 0x1, 0x0, 0x1,
+  0x1, 0x1, 0x1, 0x0, 0x0, 0x0, 0x0, 0x4, 0x0, 0x0, 0x0, 0x0, 0x1, 0x0, 0x0, 0x0, 0x0, 0x4, 0x0,
+  0x1, 0x1, 0x1, 0x1, 0x0, 0x2, 0x2, 0x2, 0x2, 0x0, 0x0, 0x0, 0x0, 0x2, 0x0, 0x0, 0x0, 0x1, 0x1,
+  0x0, 0x0, 0x1, 0x1, 0x1, 0x0, 0x1, 0x1, 0x1, 0x1, 0x0, 0x0, 0x0, 0x0, 0x2, 0x0, 0x0, 0x0, 0x0,
+  0x4, 0x0, 0x0, 0x0, 0x1, 0x1, 0x0, 0x0, 0x1, 0x1, 0x1,
+};
+
+static const __swift_uint64_t _swift_stdlib_nameAlias[88] = {
+  0x7, 0x10480017042, 0x0, 0x100400000, 0x800000000000, 0x20000, 0x1000000, 0x20000000400000, 0x0,
+  0x0, 0xC0000, 0x0, 0x0, 0x0, 0x0, 0x20000000000, 0x0, 0x20000, 0x40000000, 0x0, 0x0, 0x0, 0x30000,
+  0x0, 0x0, 0x0, 0x28A000, 0x40000, 0x0, 0x0, 0x0, 0x1000000000000, 0x80000, 0x0, 0x0, 0x0,
+  0xF00000000000, 0x90000, 0x10000000000, 0x0, 0x0, 0x0, 0xD0000, 0x30000000000, 0x0, 0x0, 0x0,
+  0xE0000, 0x0, 0x0, 0x0, 0x1400000000000000, 0x100000, 0x0, 0x0, 0x20000000000000, 0x0, 0x120000,
+  0x0, 0x400000000000, 0x0, 0x0, 0x130000, 0x10000000000, 0x0, 0x0, 0x0, 0x140000, 0x0, 0x0, 0x0,
+  0x30, 0x150000, 0x0, 0x0, 0x0, 0xC0000000C00000, 0x170000, 0x0, 0x0, 0x200000000, 0x0, 0x1B0000,
+  0x0, 0x20000000000000, 0x0, 0x0, 0x1C0000,
+};