Merge pull request #40644 from Azoy/scalar-name-fixes

Azoy · web-flow · commit 6475e098c55b · 2021-12-20T20:38:11.000-08:00
[stdlib] Some fixes for scalar names
diff --git a/stdlib/public/core/UnicodeScalarProperties.swift b/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),
diff --git a/stdlib/public/stubs/Unicode/Apple/ScalarPropsData.h b/stdlib/public/stubs/Unicode/Apple/ScalarPropsData.h
@@ -7420,6 +7420,8 @@ static const __swift_uint32_t _swift_stdlib_word_indices[12866] = {
   0x13139,
 };
 
+#define NAMES_LAST_SCALAR_OFFSET 215884
+
 static const __swift_uint8_t _swift_stdlib_names[215884] = {
   0xFF, 0xA3, 0x2, 0xFF, 0x76, 0x2, 0x2B, 0xFF, 0x36, 0x2, 0x2B, 0x1D, 0x1, 0xFF, 0x1B, 0x5, 0x1,
   0xFF, 0x28, 0x6, 0x1, 0xFF, 0x74, 0x4, 0xFF, 0x45, 0x4, 0x27, 0xFF, 0xB7, 0x1, 0x25, 0xFF, 0xB7,
@@ -20063,6 +20065,8 @@ 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,
diff --git a/stdlib/public/stubs/Unicode/Common/ScalarPropsData.h b/stdlib/public/stubs/Unicode/Common/ScalarPropsData.h
@@ -7391,6 +7391,8 @@ static const __swift_uint32_t _swift_stdlib_word_indices[12866] = {
   0x130FC, 0x130FE, 0x13103, 0x13108, 0x1310D, 0x13111, 0x1311A, 0x13127, 0x1312B, 0x13131, 0x1313A,
 };
 
+#define NAMES_LAST_SCALAR_OFFSET 215884
+
 static const __swift_uint8_t _swift_stdlib_names[215884] = {
   0xFF, 0xA7, 0x2, 0xFF, 0x77, 0x2, 0x2B, 0xFF, 0x3A, 0x2, 0x2B, 0x1D, 0x1, 0xFF, 0xDD, 0x4, 0x1,
   0xFF, 0x7C, 0x6, 0x1, 0xFF, 0x9F, 0x4, 0xFF, 0x46, 0x4, 0x27, 0xFF, 0xB6, 0x1, 0x25, 0xFF, 0xB6,
@@ -20035,6 +20037,8 @@ 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,
diff --git a/stdlib/public/stubs/Unicode/UnicodeScalarProps.cpp b/stdlib/public/stubs/Unicode/UnicodeScalarProps.cpp
@@ -233,18 +233,27 @@ __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).
+    nextScalarOffset = NAMES_LAST_SCALAR_OFFSET;
   }
 
   auto nameSize = nextScalarOffset - scalarOffset;
diff --git a/utils/gen-unicode-data/Sources/GenScalarProps/Names.swift b/utils/gen-unicode-data/Sources/GenScalarProps/Names.swift
@@ -159,6 +159,12 @@ func emitScalarNames(
     }
   }
   
+  result += """
+  #define NAMES_LAST_SCALAR_OFFSET \(nameBytes.count)
+  
+  
+  """
+  
   emitCollection(nameBytes, name: "_swift_stdlib_names", into: &result)
   
   return scalarNameIndices
@@ -198,6 +204,12 @@ func emitScalars(
     }
   }
   
+  result += """
+  #define NAMES_SCALARS_MAX_INDEX \(scalars.count - 1)
+  
+  
+  """
+  
   emitCollection(scalars, name: "_swift_stdlib_names_scalars", into: &result)
   
   return scalarSetIndices

Original file line number	Diff line number	Diff line change
`@@ -159,6 +159,12 @@ func emitScalarNames(`
`159`	`159`	`}`
`160`	`160`	`}`
`161`	`161`
	`162`	`+ result += """`
	`163`	`+ #define NAMES_LAST_SCALAR_OFFSET \(nameBytes.count)`
	`164`	`+`
	`165`	`+`
	`166`	`+ """`
	`167`	`+`
`162`	`168`	`emitCollection(nameBytes, name: "_swift_stdlib_names", into: &result)`
`163`	`169`
`164`	`170`	`return scalarNameIndices`
`@@ -198,6 +204,12 @@ func emitScalars(`
`198`	`204`	`}`
`199`	`205`	`}`
`200`	`206`
	`207`	`+ result += """`
	`208`	`+ #define NAMES_SCALARS_MAX_INDEX \(scalars.count - 1)`
	`209`	`+`
	`210`	`+`
	`211`	`+ """`
	`212`	`+`
`201`	`213`	`emitCollection(scalars, name: "_swift_stdlib_names_scalars", into: &result)`
`202`	`214`
`203`	`215`	`return scalarSetIndices`