stdlib/String: change == to perform string comparison after NFD

normalization

There is still some obscure bug with != on NSString, probably caused by
an ill-thought overload somewhere.

Part of rdar://17498444

Swift SVN r20495

stdlib/core/Character.swift

@@ -121,15 +121,8 @@ extension String {
 }
 
 public func ==(lhs: Character, rhs: Character) -> Bool {
-  switch (lhs, rhs) {
+  // FIXME(performance): constructing two temporary strings is extremely
-  case (.LargeRepresentation(let lhsValue), .LargeRepresentation(let rhsValue)):
+  // wasteful and inefficient.
-    return lhsValue._value == rhsValue._value
+  return String(lhs) == String(rhs)
-
-  case (.SmallRepresentation(let lhsValue), .SmallRepresentation(let rhsValue)):
-    return Character._smallValue(lhsValue) == Character._smallValue(rhsValue)
-
-  default:
-    return false
-  }
 }
 

stdlib/core/Misc.swift

@@ -67,6 +67,11 @@ public func _stdlib_getTypeName<T>(value: T) -> String {
   return result
 }
 
+/// Returns the human-readable type name for the given value.
+public func _stdlib_getDemangledTypeName<T>(value: T) -> String {
+  return _stdlib_demangleName(_stdlib_getTypeName(value))
+}
+
 @asmname("swift_stdlib_demangleName")
 func _stdlib_demangleNameImpl(
     mangledName: UnsafePointer<UInt8>,

stdlib/core/Reflection.swift

@@ -392,7 +392,7 @@ struct _StructMirror: MirrorType {
   }
 
   var summary: String {
-    return _stdlib_demangleName(_stdlib_getTypeName(value))
+    return _stdlib_getDemangledTypeName(value)
   }
   var quickLookObject: QuickLookObject? { return nil }
   var disposition: MirrorDisposition { return .Struct }
@@ -421,7 +421,7 @@ struct _ClassMirror: MirrorType {
     return _getClassChild(i, data)
   }
   var summary: String {
-    return _stdlib_demangleName(_stdlib_getTypeName(value))
+    return _stdlib_getDemangledTypeName(value)
   }
   var quickLookObject: QuickLookObject? {
     return _getClassQuickLookObject(data)
@@ -446,7 +446,7 @@ struct _ClassSuperMirror: MirrorType {
     return _getClassChild(i, data)
   }
   var summary: String {
-    return _stdlib_demangleName(_stdlib_getTypeName(value))
+    return _stdlib_getDemangledTypeName(value)
   }
   var quickLookObject: QuickLookObject? { return nil }
   var disposition: MirrorDisposition { return .Class }

stdlib/core/String.swift

@@ -173,19 +173,26 @@ extension String {
   }
 }
 
+/// Compare two strings, normalizing them to NFD first.
+///
+/// The behavior is equivalent to `NSString.compare()` with default options.
+///
+/// :returns:
+///   * -1 if `lhs < rhs`,
+///   * 0 if `lhs == rhs`,
+///   * 1 if `lhs > rhs`.
+@asmname("swift_stdlib_compareNSStringNormalizingToNFD")
+func _stdlib_compareNSStringNormalizingToNFD(lhs: AnyObject, rhs: AnyObject)
+    -> Int
+
 extension String: Equatable {
 }
 
 public func ==(lhs: String, rhs: String) -> Bool {
-  // FIXME: Compares UnicodeScalars, but should eventually do proper
+  // Note: this operation should be consistent with equality comparison of
-  // Unicode string comparison. This is above the level of the
+  // Character.
-  // standard equal algorithm because even the largest units
+  return _stdlib_compareNSStringNormalizingToNFD(
-  // (Characters/a.k.a. grapheme clusters) don't have a 1-for-1
+    lhs._bridgeToObjectiveCImpl(), rhs._bridgeToObjectiveCImpl()) == 0
-  // correspondence.  For example, "SS" == "ß" should be true.
-  //
-  // NOTE: if this algorithm is changed, consider updating equality comparison
-  // of Character.
-  return Swift.equal(lhs.unicodeScalars, rhs.unicodeScalars)
 }
 
 public func <(lhs: String, rhs: String) -> Bool {

stdlib/runtime/NSSwiftXXXBase.mm.gyb

@@ -93,6 +93,16 @@ extern "C" bool swift_stdlib_NSObject_isEqual(NSObject *lhs, NSObject *rhs) {
   return Result;
 }
 
+extern "C" int swift_stdlib_compareNSStringNormalizingToNFD(NSString *lhs,
+                                                            NSString *rhs) {
+  // 'kCFCompareNonliteral' actually means "normalize to NFD".
+  int Result = CFStringCompare((__bridge CFStringRef)lhs,
+                               (__bridge CFStringRef)rhs, kCFCompareNonliteral);
+  swift_unknownRelease(lhs);
+  swift_unknownRelease(rhs);
+  return Result;
+}
+
 // ${'Local Variables'}:
 // eval: (read-only-mode 1)
 // End:

stdlib/unittest/StdlibUnittest.swift.gyb

@@ -98,8 +98,8 @@ public func expectEqual<T>(
     _anyExpectFailed = true
     println("check failed at \(file), line \(line)")
     _printStackTrace(stackTrace)
-    println("expected: \"\(expected)\" (of type \(_stdlib_getTypeName(expected)))")
+    println("expected: \"\(expected)\" (of type \(_stdlib_getDemangledTypeName(expected)))")
-    println("actual: \"\(actual)\" (of type \(_stdlib_getTypeName(expected)))")
+    println("actual: \"\(actual)\" (of type \(_stdlib_getDemangledTypeName(expected)))")
     if collectMoreInfo { println(collectMoreInfo!()) }
     println()
   }
@@ -107,12 +107,14 @@ public func expectEqual<T>(
 
 public func expectNotEqual<T : Equatable>(
     expected: T, actual: T,
+    stackTrace: SourceLocStack? = nil,
     file: String = __FILE__, line: UWord = __LINE__
 ) {
   if expected == actual {
     _anyExpectFailed = true
     println("check failed at \(file), line \(line)")
-    println("unexpected value: \"\(actual)\" (of type \(_stdlib_getTypeName(actual)))")
+    _printStackTrace(stackTrace)
+    println("unexpected value: \"\(actual)\" (of type \(_stdlib_getDemangledTypeName(actual)))")
     println()
   }
 }
@@ -126,8 +128,8 @@ public func expectOptionalEqual<T : Equatable>(
   if !actual || expected != actual! {
     _anyExpectFailed = true
     println("check failed at \(file), line \(line)")
-    println("expected: \"\(expected)\" (of type \(_stdlib_getTypeName(expected)))")
+    println("expected: \"\(expected)\" (of type \(_stdlib_getDemangledTypeName(expected)))")
-    println("actual: \"\(actual)\" (of type \(_stdlib_getTypeName(actual)))")
+    println("actual: \"\(actual)\" (of type \(_stdlib_getDemangledTypeName(actual)))")
     println()
   }
 }
@@ -164,7 +166,7 @@ func _expectNotEqual${Generic}(
   if expected == actual {
     _anyExpectFailed = true
     println("check failed at \(file), line \(line)")
-    println("unexpected value: \"\(actual)\" (of type \(_stdlib_getTypeName(actual)))")
+    println("unexpected value: \"\(actual)\" (of type \(_stdlib_getDemangledTypeName(actual)))")
     println()
   }
 }
@@ -480,7 +482,7 @@ public func expectEqualSequence<
     println("check failed at \(file), line \(line)")
     _printStackTrace(stackTrace)
     println("expected elements: \"\(expected)\"")
-    println("actual: \"\(actual)\" (of type \(_stdlib_getTypeName(actual)))")
+    println("actual: \"\(actual)\" (of type \(_stdlib_getDemangledTypeName(actual)))")
     if collectMoreInfo { println(collectMoreInfo!()) }
     println()
   }

test/stdlib/NSStringAPI.swift

@@ -1138,12 +1138,103 @@ NSStringAPIs.test("writeToURL(_:atomically:encoding:error:)") {
   // FIXME
 }
 
-NSStringAPIs.test("OperatorEquals") {
+func checkEqualityImpl(
-  // FIXME
+  expected: Bool, lhs: String, rhs: String,
+  stackTrace: SourceLocStack
+) {
+  // String / String
+  expectEqual(expected, lhs == rhs, stackTrace: stackTrace)
+  expectEqual(!expected, lhs != rhs, stackTrace: stackTrace)
 
-  // NSString == NSString
+  // NSString / NSString
-  // String == NSString
+  let lhsNSString = lhs as NSString
-  // NSString == String
+  let rhsNSString = rhs as NSString
+  expectEqual(expected, lhsNSString == rhsNSString, stackTrace: stackTrace)
+  // FIXME: this fails now.
+  //expectEqual(!expected, lhsNSString != rhsNSString, stackTrace: stackTrace)
+
+  // String / NSString
+  expectEqual(expected, lhs == rhsNSString, stackTrace: stackTrace)
+  expectEqual(!expected, lhs != rhsNSString, stackTrace: stackTrace)
+
+  // NSString / String
+  expectEqual(expected, lhs == rhsNSString, stackTrace: stackTrace)
+  expectEqual(!expected, lhs != rhsNSString, stackTrace: stackTrace)
+}
+
+func checkEquality(
+  expected: Bool, lhs: String, rhs: String, stackTrace: SourceLocStack
+) {
+  checkEqualityImpl(expected, lhs, rhs, stackTrace.withCurrentLoc())
+  checkEqualityImpl(expected, rhs, lhs, stackTrace.withCurrentLoc())
+}
+
+struct EqualityTest {
+  let expected: Bool
+  let lhs: String
+  let rhs: String
+  let loc: SourceLoc
+
+  init(_ expected: Bool, _ lhs: String, _ rhs: String,
+       file: String = __FILE__, line: UWord = __LINE__) {
+    self.expected = expected
+    self.lhs = lhs
+    self.rhs = rhs
+    self.loc = SourceLoc(file, line, comment: "test data")
+  }
+}
+
+let equalityTests = [
+  EqualityTest(true, "", ""),
+  EqualityTest(false, "a", ""),
+
+  // U+0301 COMBINING ACUTE ACCENT
+  // U+00E1 LATIN SMALL LETTER A WITH ACUTE
+  EqualityTest(true, "a\u{301}", "\u{e1}"),
+  EqualityTest(false, "a\u{301}", "a"),
+  EqualityTest(false, "\u{e1}", "a"),
+
+  // U+304B HIRAGANA LETTER KA
+  // U+304C HIRAGANA LETTER GA
+  // U+3099 COMBINING KATAKANA-HIRAGANA VOICED SOUND MARK
+  EqualityTest(true, "\u{304b}", "\u{304b}"),
+  EqualityTest(true, "\u{304c}", "\u{304c}"),
+  EqualityTest(false, "\u{304b}", "\u{304c}"),
+  EqualityTest(false, "\u{304b}", "\u{304c}\u{3099}"),
+  EqualityTest(true, "\u{304c}", "\u{304b}\u{3099}"),
+  EqualityTest(false, "\u{304c}", "\u{304c}\u{3099}"),
+
+  // U+212B ANGSTROM SIGN
+  // U+030A COMBINING RING ABOVE
+  // U+00C5 LATIN CAPITAL LETTER A WITH RING ABOVE
+  EqualityTest(true, "\u{212b}", "A\u{30a}"),
+  EqualityTest(true, "\u{212b}", "\u{c5}"),
+  EqualityTest(true, "A\u{30a}", "\u{c5}"),
+  EqualityTest(false, "A\u{30a}", "a"),
+
+  // U+2126 OHM SIGN
+  // U+03A9 GREEK CAPITAL LETTER OMEGA
+  EqualityTest(true, "\u{2126}", "\u{03a9}"),
+
+  // U+1E69 LATIN SMALL LETTER S WITH DOT BELOW AND DOT ABOVE
+  // U+0323 COMBINING DOT BELOW
+  // U+0307 COMBINING DOT ABOVE
+  // U+1E63 LATIN SMALL LETTER S WITH DOT BELOW
+  EqualityTest(true, "\u{1e69}", "s\u{323}\u{307}"),
+  EqualityTest(true, "\u{1e69}", "s\u{307}\u{323}"),
+  EqualityTest(true, "\u{1e69}", "\u{1e63}\u{307}"),
+  EqualityTest(true, "\u{1e63}\u{307}", "s\u{323}\u{307}"),
+  EqualityTest(true, "\u{1e63}\u{307}", "s\u{307}\u{323}"),
+
+  // U+FB01 LATIN SMALL LIGATURE FI
+  EqualityTest(true, "\u{fb01}", "\u{fb01}"),
+  EqualityTest(false, "\u{fb01}", "fi"),
+]
+
+NSStringAPIs.test("OperatorEquals") {
+  for test in equalityTests {
+    checkEquality(test.expected, test.lhs, test.rhs, test.loc.withCurrentLoc())
+  }
 }
 
 // FIXME: these properties should be implemented in the core library.