Implement native scalar age

Author: Azoy

stdlib/public/SwiftShims/UnicodeData.h

@@ -87,6 +87,9 @@ SWIFT_RUNTIME_STDLIB_INTERNAL
 __swift_intptr_t _swift_stdlib_getScalarName(__swift_uint32_t scalar,
                                              __swift_uint8_t *buffer);
 
+SWIFT_RUNTIME_STDLIB_INTERNAL
+__swift_uint16_t _swift_stdlib_getAge(__swift_uint32_t scalar);
+
 #ifdef __cplusplus
 } // extern "C"
 #endif

stdlib/public/SwiftShims/UnicodeShims.h

@@ -271,10 +271,6 @@ typedef __swift_uint16_t __swift_stdlib_UChar;
 typedef __swift_uint8_t
     __swift_stdlib_UVersionInfo[__SWIFT_STDLIB_U_MAX_VERSION_LENGTH];
 
-SWIFT_RUNTIME_STDLIB_API
-void __swift_stdlib_u_charAge(
-    __swift_stdlib_UChar32, __swift_stdlib_UVersionInfo _Nonnull);
-
 SWIFT_RUNTIME_STDLIB_API
 __swift_int32_t
     __swift_stdlib_u_getIntPropertyValue(__swift_stdlib_UChar32,

stdlib/public/core/UnicodeScalarProperties.swift

@@ -872,15 +872,16 @@ extension Unicode.Scalar.Properties {
   /// This property corresponds to the "Age" property in the
   /// [Unicode Standard](http://www.unicode.org/versions/latest/).
   public var age: Unicode.Version? {
-    var versionInfo: __swift_stdlib_UVersionInfo = (0, 0, 0, 0)
-    withUnsafeMutablePointer(to: &versionInfo) { tuplePtr in
-      tuplePtr.withMemoryRebound(to: UInt8.self, capacity: 4) {
-        versionInfoPtr in
-        __swift_stdlib_u_charAge(icuValue, versionInfoPtr)
-      }
+    let age = _swift_stdlib_getAge(_scalar.value)
+
+    if age == .max {
+      return nil
     }
-    guard versionInfo.0 != 0 else { return nil }
-    return (major: Int(versionInfo.0), minor: Int(versionInfo.1))
+
+    let major = age & 0xFF
+    let minor = (age & 0xFF00) >> 8
+
+    return (major: Int(major), minor: Int(minor))
   }
 }
 

stdlib/public/stubs/UnicodeScalarProps.cpp

@@ -257,3 +257,41 @@ __swift_intptr_t _swift_stdlib_getScalarName(__swift_uint32_t scalar,
   // The return value is the number of initialized bytes.
   return c;
 }
+
+SWIFT_RUNTIME_STDLIB_INTERNAL
+__swift_uint16_t _swift_stdlib_getAge(__swift_uint32_t scalar) {
+  auto lowerBoundIndex = 0;
+  auto endIndex = 1659;
+  auto upperBoundIndex = endIndex - 1;
+
+  while (upperBoundIndex >= lowerBoundIndex) {
+    auto idx = lowerBoundIndex + (upperBoundIndex - lowerBoundIndex) / 2;
+
+    auto entry = _swift_stdlib_ages[idx];
+
+    auto lowerBoundScalar = (entry << 43) >> 43;
+    auto rangeCount = entry >> 32;
+    auto upperBoundScalar = lowerBoundScalar + rangeCount;
+
+    auto ageIdx = (__swift_uint8_t)((entry << 32) >> 32 >> 21);
+
+    if (scalar >= lowerBoundScalar && scalar <= upperBoundScalar) {
+      return _swift_stdlib_ages_data[ageIdx];
+    }
+
+    if (scalar > upperBoundScalar) {
+      lowerBoundIndex = idx + 1;
+      continue;
+    }
+
+    if (scalar < lowerBoundScalar) {
+      upperBoundIndex = idx - 1;
+      continue;
+    }
+  }
+
+  // If we made it out here, then our scalar was not found in the composition
+  // array.
+  // Return the max here to indicate that we couldn't find one.
+  return std::numeric_limits<__swift_uint16_t>::max();
+}

stdlib/public/stubs/UnicodeScalarProps.h

@@ -36,7 +36,6 @@ typedef __swift_stdlib_UProperty UProperty;
 typedef uint8_t UVersionInfo[U_MAX_VERSION_LENGTH];
 
 // Comparison and character property APIs
-void u_charAge(UChar32, UVersionInfo);
 int32_t u_getIntPropertyValue(UChar32, UProperty);
 }
 
@@ -58,21 +57,6 @@ int32_t u_getIntPropertyValue(UChar32, UProperty);
 
 #endif
 
-namespace {
-template <typename T, typename U> T *ptr_cast(U *p) {
-  return static_cast<T *>(static_cast<void *>(p));
-}
-template <typename T, typename U> const T *ptr_cast(const U *p) {
-  return static_cast<const T *>(static_cast<const void *>(p));
-}
-}
-
-void
-__swift_stdlib_u_charAge(__swift_stdlib_UChar32 c,
-                                __swift_stdlib_UVersionInfo versionInfo) {
-  return u_charAge(c, versionInfo);
-}
-
 __swift_int32_t
 __swift_stdlib_u_getIntPropertyValue(__swift_stdlib_UChar32 c,
                                             __swift_stdlib_UProperty p) {

stdlib/public/stubs/UnicodeShims.cpp

@@ -0,0 +1,97 @@
+//===----------------------------------------------------------------------===//
+//
+// This source file is part of the Swift.org open source project
+//
+// Copyright (c) 2021 Apple Inc. and the Swift project authors
+// Licensed under Apache License v2.0 with Runtime Library Exception
+//
+// See https://swift.org/LICENSE.txt for license information
+// See https://swift.org/CONTRIBUTORS.txt for the list of Swift project authors
+//
+//===----------------------------------------------------------------------===//
+
+import GenUtils
+
+func getAge(
+  from data: String,
+  into result: inout [(ClosedRange<UInt32>, [UInt8])]
+) {
+  for line in data.split(separator: "\n") {
+    // Skip comments
+    guard !line.hasPrefix("#") else {
+      continue
+    }
+    
+    let info = line.split(separator: "#")
+    let components = info[0].split(separator: ";")
+    
+    let scalars: ClosedRange<UInt32>
+    
+    let filteredScalars = components[0].filter { !$0.isWhitespace }
+    
+    // If we have . appear, it means we have a legitimate range. Otherwise,
+    // it's a singular scalar.
+    if filteredScalars.contains(".") {
+      let range = filteredScalars.split(separator: ".")
+      
+      scalars = UInt32(range[0], radix: 16)! ... UInt32(range[1], radix: 16)!
+    } else {
+      let scalar = UInt32(filteredScalars, radix: 16)!
+      
+      scalars = scalar ... scalar
+    }
+    
+    let version = components[1].filter { !$0.isWhitespace }
+    let parts = version.split(separator: ".")
+    
+    let major = UInt8(parts[0])!
+    let minor = UInt8(parts[1])!
+    
+    result.append((scalars, [major, minor]))
+  }
+}
+
+func emitAge(
+  data: [(ClosedRange<UInt32>, [UInt8])],
+  into result: inout String
+) {
+  var uniqueAges: Set<[UInt8]> = []
+  
+  for (_, age) in data {
+    uniqueAges.insert(age)
+  }
+  
+  let ages = uniqueAges.map {
+    UInt16($0[0]) | (UInt16($0[1]) << 8)
+  }
+  
+  emitCollection(ages, name: "_swift_stdlib_ages_data", into: &result)
+  
+  emitCollection(
+    data,
+    name: "_swift_stdlib_ages",
+    type: "__swift_uint64_t",
+    into: &result
+  ) {
+    var value: UInt64 = UInt64($0.0.lowerBound)
+    
+    let age = UInt16($0.1[0]) | (UInt16($0.1[1]) << 8)
+    let ageIndex = ages.firstIndex(of: age)!
+    
+    value |= UInt64(ageIndex) << 21
+    
+    value |= UInt64($0.0.count - 1) << 32
+    
+    return "0x\(String(value, radix: 16, uppercase: true))"
+  }
+}
+
+func generateAgeProp(into result: inout String) {
+  let derivedAge = readFile("Data/DerivedAge.txt")
+  
+  var ageData: [(ClosedRange<UInt32>, [UInt8])] = []
+  
+  getAge(from: derivedAge, into: &ageData)
+  
+  emitAge(data: flatten(ageData), into: &result)
+}

utils/gen-unicode-data/Data/DerivedAge.txt

@@ -20,6 +20,7 @@ func generateScalarProps() {
   generateNameAliasProp(into: &result)
   generateMappingProps(into: &result)
   generateNameProp(into: &result)
+  generateAgeProp(into: &result)
 
   write(result, to: "../../stdlib/public/stubs/UnicodeScalarProps.cpp")
 }