rdar://106965817 (FoundationEssentials: Internal character set)

Add a Swift-native character set that mirrors CF/NSCharacterSet to support FoundationEssentials. We will use this type to back `CharacterSet`, which is currently NS-bridged, when we get to re-core it when the time comes.

Currently it's only used for String capitalization.

Sources/FoundationEssentials/BuiltInUnicodeScalarSet.swift

@@ -0,0 +1,92 @@
+//===----------------------------------------------------------------------===//
+//
+// This source file is part of the Swift Collections open source project
+//
+// Copyright (c) 2022 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
+//
+//===----------------------------------------------------------------------===//
+
+@_implementationOnly import _CShims
+
+// Native implementation of CFCharacterSet.
+// Represents sets of unicode scalars of those whose bitmap data we own.
+// whitespace, whitespaceAndNewline, and newline are not included since they're not stored with bitmaps
+// This only contains a subset of predefined CFCharacterSet that are in use for now.
+internal struct BuiltInUnicodeScalarSet {
+    enum SetType {
+        case lowercaseLetter
+        case uppercaseLetter
+        case canonicalDecomposable
+
+        // Below are internal
+        case caseIgnorable
+        case graphemeExtend
+    }
+
+    var charset: SetType
+    init(type: SetType) {
+        charset = type
+    }
+
+    // Equivalent to  __CFUniCharMapExternalSetToInternalIndex(__CFUniCharMapCompatibilitySetID())
+    private var _bitmapTableIndex: Int {
+        switch charset {
+        case .lowercaseLetter:
+            return 2
+        case .uppercaseLetter:
+            return 3
+        case .canonicalDecomposable:
+            return 5
+        case .caseIgnorable:
+            return 20
+        case .graphemeExtend:
+            return 21
+        }
+    }
+
+    // CFUniCharIsMemberOf
+    func contains(_ scalar: Unicode.Scalar) -> Bool {
+        let planeNo = Int((scalar.value >> 16) & 0xFF)
+        let bitmp = _bitmapPtrForPlane(planeNo)
+        return _isMemberOfBitmap(scalar, bitmp)
+    }
+
+    // CFUniCharGetBitmapPtrForPlane
+    func _bitmapPtrForPlane(_ plane: Int) -> UnsafePointer<UInt8>? {
+        let tableIndex = _bitmapTableIndex
+        guard tableIndex < __CFUniCharNumberOfBitmaps else {
+            return nil
+        }
+
+        let data = withUnsafePointer(to: __CFUniCharBitmapDataArray) { ptr in
+            ptr.withMemoryRebound(to: __CFUniCharBitmapData.self, capacity: Int(__CFUniCharNumberOfBitmaps)) { bitmapDataPtr in
+                bitmapDataPtr.advanced(by: tableIndex).pointee
+            }
+        }
+        return plane < data._numPlanes ? data._planes[plane] : nil
+    }
+
+    let bitShiftForByte = UInt16(3)
+    let bitShiftForMask = UInt16(7)
+
+    // CFUniCharIsMemberOfBitmap
+    func _isMemberOfBitmap(_ scalar: Unicode.Scalar, _ bitmap: UnsafePointer<UInt8>?) -> Bool {
+        guard let bitmap else { return false }
+        let theChar = UInt16(truncatingIfNeeded: scalar.value) // intentionally truncated
+
+        let position = bitmap[Int(theChar >> bitShiftForByte)]
+        let mask = theChar & bitShiftForMask
+        let new = (Int(position) & Int(UInt32(1) << mask)) != 0
+        return new
+    }
+
+    static let uppercaseLetter = Self.init(type: .uppercaseLetter)
+    static let lowercaseLetter = Self.init(type: .lowercaseLetter)
+    static let caseIgnorable = Self.init(type: .caseIgnorable)
+    static let graphemeExtend = Self.init(type: .graphemeExtend)
+    static let canonicalDecomposable = Self.init(type: .canonicalDecomposable)
+}
+

Sources/_CShims/include/CFUniCharBitmapData.h

@@ -0,0 +1,17 @@
+/*
+	CFUniCharBitmapData.h
+	Copyright (c) 1999-2021, Apple Inc. and the Swift project authors. All rights reserved.
+	This file is generated. Don't touch this file directly.
+*/
+
+#ifndef _cfunichar_bitmap_data_h
+#define _cfunichar_bitmap_data_h
+
+#include "_CStdlib.h"
+
+typedef struct {
+    uint32_t _numPlanes;
+    const uint8_t **_planes;
+} __CFUniCharBitmapData;
+
+#endif /* _cfunichar_bitmap_data_h */

Sources/_CShims/include/_CStdlib.h

@@ -130,5 +130,9 @@
 #include <uchar.h>
 #endif
 
+#if __has_include(<stdint.h>)
+#include <stdint.h>
+#endif
+
 #endif // FOUNDATION_CSTDLIB
 

Tests/FoundationEssentialsTests/BuiltInUnicodeScalarSetTest.swift

@@ -0,0 +1,61 @@
+//===----------------------------------------------------------------------===//
+//
+// This source file is part of the Swift.org open source project
+//
+// Copyright (c) 2014 - 2017 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
+//
+//===----------------------------------------------------------------------===//
+
+#if canImport(TestSupport)
+import TestSupport
+#endif
+
+#if FOUNDATION_FRAMEWORK
+@testable import Foundation
+#else
+@testable import FoundationEssentials
+#endif // FOUNDATION_FRAMEWORK
+
+final class BuiltInUnicodeScalarSetTest: XCTestCase {
+
+    func testMembership() {
+        func setContainsScalar(_ set: BuiltInUnicodeScalarSet, _ scalar: Unicode.Scalar, _ expect: Bool, file: StaticString = #file, line: UInt = #line) {
+            let actual = set.contains(scalar)
+            XCTAssertEqual(actual, expect, file: file, line: line)
+        }
+
+        setContainsScalar(.lowercaseLetter, "a", true)
+        setContainsScalar(.lowercaseLetter, "ô", true)
+        setContainsScalar(.lowercaseLetter, "\u{01FB}", true)
+        setContainsScalar(.lowercaseLetter, "\u{1FF7}", true)
+        setContainsScalar(.lowercaseLetter, "\u{1D467}", true)
+        setContainsScalar(.lowercaseLetter, "A", false)
+
+        setContainsScalar(.uppercaseLetter, "A", true)
+        setContainsScalar(.uppercaseLetter, "À", true)
+        setContainsScalar(.uppercaseLetter, "\u{01CF}", true)
+        setContainsScalar(.uppercaseLetter, "\u{1E5C}", true)
+        setContainsScalar(.uppercaseLetter, "\u{1D4A9}", true)
+        setContainsScalar(.uppercaseLetter, "a", false)
+
+        setContainsScalar(.caseIgnorable, "'", true)
+        setContainsScalar(.caseIgnorable, "ʻ", true)
+        setContainsScalar(.caseIgnorable, "\u{00B4}", true) // ACUTE ACCENT
+        setContainsScalar(.caseIgnorable, "\u{10792}", true) // MODIFIER LETTER SMALL CAPITAL G
+        setContainsScalar(.caseIgnorable, "\u{E0020}", true)
+        setContainsScalar(.caseIgnorable, "0", false)
+
+        setContainsScalar(.graphemeExtend, "\u{0300}", true)
+        setContainsScalar(.graphemeExtend, "\u{0610}", true)
+        setContainsScalar(.graphemeExtend, "\u{302A}", true) // IDEOGRAPHIC LEVEL TONE MARK
+        setContainsScalar(.graphemeExtend, "\u{1D17B}", true) // MUSICAL SYMBOL COMBINING ACCENT
+        setContainsScalar(.graphemeExtend, "\u{E0020}", true) // TAG SPACE
+        setContainsScalar(.graphemeExtend, "A", false)
+        setContainsScalar(.graphemeExtend, "~", false)
+    }
+
+}