Some fixes for scalar names

Author: Azoy

stdlib/public/core/UnicodeScalarProperties.swift

@@ -1207,12 +1207,12 @@ extension Unicode.Scalar.Properties {
          (0x2B740 ... 0x2B81D),
          (0x2B820 ... 0x2CEA1),
          (0x2CEB0 ... 0x2EBE0),
-         (0x2F800 ... 0x2FA1D),
          (0x30000 ... 0x3134A):
       return "CJK UNIFIED IDEOGRAPH-\(scalarName)"
 
     case (0xF900 ... 0xFA6D),
-         (0xFA70 ... 0xFAD9):
+         (0xFA70 ... 0xFAD9),
+         (0x2F800 ... 0x2FA1D):
       return "CJK COMPATIBILITY IDEOGRAPH-\(scalarName)"
 
     case (0x17000 ... 0x187F7),

stdlib/public/stubs/Unicode/Apple/ScalarPropsData.h

@@ -20063,6 +20063,8 @@ static const __swift_uint8_t _swift_stdlib_names[215884] = {
   0xBD,
 };
 
+#define NAMES_SCALARS_MAX_INDEX 39039
+
 static const __swift_uint32_t _swift_stdlib_names_scalars[39040] = {
   0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0,
   0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x3, 0x7, 0xB, 0xD, 0x11,

stdlib/public/stubs/Unicode/Common/ScalarPropsData.h

@@ -20035,6 +20035,8 @@ static const __swift_uint8_t _swift_stdlib_names[215884] = {
   0xE1, 0xF, 0xBD,
 };
 
+#define NAMES_SCALARS_MAX_INDEX 39039
+
 static const __swift_uint32_t _swift_stdlib_names_scalars[39040] = {
   0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0,
   0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x3, 0x7, 0xB, 0xD, 0x11,

stdlib/public/stubs/Unicode/UnicodeScalarProps.cpp

@@ -233,18 +233,30 @@ __swift_intptr_t _swift_stdlib_getScalarName(__swift_uint32_t scalar,
   auto scalarIndex = (setOffset << 7) + (scalar & ((1 << 7) - 1));
   auto scalarOffset = _swift_stdlib_names_scalars[scalarIndex];
 
-  if (scalarOffset == 0) {
+  // U+20 is the first scalar that Unicode defines a name for, so their offset
+  // will the only valid 0.
+  if (scalarOffset == 0 && scalar != 0x20) {
     return 0;
   }
 
   __swift_uint32_t nextScalarOffset = 0;
-  int i = 1;
 
-  // Look for the next scalar who has a name and their position in the names
-  // array. This tells us exactly how many bytes our name takes up.
-  while (nextScalarOffset == 0) {
-    nextScalarOffset = _swift_stdlib_names_scalars[scalarIndex + i];
-    i += 1;
+  if (scalarIndex != NAMES_SCALARS_MAX_INDEX) {
+    int i = 1;
+
+    // Look for the next scalar who has a name and their position in the names
+    // array. This tells us exactly how many bytes our name takes up.
+    while (nextScalarOffset == 0) {
+      nextScalarOffset = _swift_stdlib_names_scalars[scalarIndex + i];
+      i += 1;
+    }
+  } else {
+    // This is the last element in the array which represents the last scalar
+    // name that Unicode defines (excluding variation selectors). This is
+    // U+E007F at the moment whose name is 'CANCEL TAG'. We add 4 because
+    // 'CANCEL' is found outside the byte area for words, but tag does.
+    // The resulting word indices look like this: [..., 0xFF, _, _, 0xXX];
+    nextScalarOffset = scalarOffset + 4;
   }
 
   auto nameSize = nextScalarOffset - scalarOffset;

utils/gen-unicode-data/Sources/GenScalarProps/Names.swift

@@ -159,6 +159,12 @@ func emitScalarNames(
     }
   }
 
+  result += """
+  #define NAMES_SCALARS_MAX_INDEX \(scalars.count - 1)
+  
+  
+  """
+  
   emitCollection(nameBytes, name: "_swift_stdlib_names", into: &result)
 
   return scalarNameIndices