Merge pull request #20438 from milseman/uniterator

[String] Custom Iterators for String Views

Author: milseman

stdlib/public/core/StringCharacterView.swift

@@ -197,16 +197,8 @@ extension String: BidirectionalCollection {
 
       let i = _guts.scalarAlign(i)
       let distance = _characterStride(startingAt: i)
-
-      if _fastPath(_guts.isFastUTF8) {
-        let start = i.encodedOffset
-        let end = start + distance
-        return _guts.withFastUTF8(range: start..<end) { utf8 in
-          return Character(unchecked: String._uncheckedFromUTF8(utf8))
-        }
-      }
-
-      return _foreignSubscript(position: i, distance: distance)
+      return _guts.errorCorrectedCharacter(
+        startingAt: i.encodedOffset, endingAt: i.encodedOffset &+ distance)
     }
   }
 
@@ -228,40 +220,40 @@ extension String: BidirectionalCollection {
   }
 }
 
-// Foreign string support
 extension String {
-  @usableFromInline @inline(never)
-  @_effects(releasenone)
-  internal func _foreignSubscript(position: Index, distance: Int) -> Character {
-#if _runtime(_ObjC)
-    _sanityCheck(_guts.isForeign)
+  @_fixed_layout
+  public struct Iterator: IteratorProtocol {
+    @usableFromInline
+    internal var _guts: _StringGuts
 
-    // Both a fast-path for single-code-unit graphemes and validation:
-    //   ICU treats isolated surrogates as isolated graphemes
-    if distance == 1 {
-      return Character(
-        String(_guts.foreignErrorCorrectedScalar(startingAt: position).0))
-    }
+    @usableFromInline
+    internal var _position: Int = 0
 
-    let start = position.encodedOffset
-    let end = start + distance
-    let count = end - start
+    @usableFromInline
+    internal var _end: Int
 
-    // TODO(String performance): Stack buffer if small enough
-
-    var cus = Array<UInt16>(repeating: 0, count: count)
-    cus.withUnsafeMutableBufferPointer {
-      _cocoaStringCopyCharacters(
-        from: _guts._object.cocoaObject,
-        range: start..<end,
-        into: $0.baseAddress._unsafelyUnwrappedUnchecked)
+    @inlinable
+    internal init(_ guts: _StringGuts) {
+      self._guts = guts
+      self._end = guts.count
     }
-    return cus.withUnsafeBufferPointer {
-      return Character(String._uncheckedFromUTF16($0))
+
+    @inlinable
+    public mutating func next() -> Character? {
+      guard _fastPath(_position < _end) else { return nil }
+
+      let len = _guts._opaqueCharacterStride(startingAt: _position)
+      let nextPosition = _position &+ len
+      let result = _guts.errorCorrectedCharacter(
+        startingAt: _position, endingAt: nextPosition)
+      _position = nextPosition
+      return result
     }
-#else
-    fatalError("No foreign strings on Linux in this version of Swift")
-#endif
+  }
+
+  @inlinable
+  public __consuming func makeIterator() -> Iterator {
+    return Iterator(_guts)
   }
 }
 

stdlib/public/core/StringUnicodeScalarView.swift

@@ -161,13 +161,42 @@ extension String.UnicodeScalarView: BidirectionalCollection {
     @inline(__always) get {
       String(_guts)._boundsCheck(position)
       let i = _guts.scalarAlign(position)
-      if _fastPath(_guts.isFastUTF8) {
-        return _guts.fastUTF8Scalar(startingAt: i.encodedOffset)
-      }
+      return _guts.errorCorrectedScalar(startingAt: i.encodedOffset).0
+    }
+  }
+}
+
+extension String.UnicodeScalarView {
+  @_fixed_layout
+  public struct Iterator: IteratorProtocol {
+    @usableFromInline
+    internal var _guts: _StringGuts
+
+    @usableFromInline
+    internal var _position: Int = 0
+
+    @usableFromInline
+    internal var _end: Int
+
+    @inlinable
+    internal init(_ guts: _StringGuts) {
+      self._guts = guts
+      self._end = guts.count
+    }
 
-      return _foreignSubscript(aligned: i)
+    @inlinable
+    public mutating func next() -> Unicode.Scalar? {
+      guard _fastPath(_position < _end) else { return nil }
+
+      let (result, len) = _guts.errorCorrectedScalar(startingAt: _position)
+      _position &+= len
+      return result
     }
   }
+  @inlinable
+  public __consuming func makeIterator() -> Iterator {
+    return Iterator(_guts)
+  }
 }
 
 extension String.UnicodeScalarView: CustomStringConvertible {
@@ -403,14 +432,4 @@ extension String.UnicodeScalarView {
 
     return i.encoded(offsetBy: -len)
   }
-
-  @usableFromInline @inline(never)
-  @_effects(releasenone)
-  internal func _foreignSubscript(aligned i: Index) -> Unicode.Scalar {
-    _sanityCheck(_guts.isForeign)
-    _sanityCheck(_guts.isOnUnicodeScalarBoundary(i),
-      "should of been aligned prior")
-
-    return _guts.foreignErrorCorrectedScalar(startingAt: i).0
-  }
 }

stdlib/public/core/UnicodeHelpers.swift

@@ -262,6 +262,7 @@ extension _StringGuts {
 #endif
   }
 
+  @usableFromInline
   @_effects(releasenone)
   internal func foreignErrorCorrectedScalar(
     startingAt idx: String.Index
@@ -374,4 +375,65 @@ extension _StringGuts {
       "Error-correction shouldn't give trailing surrogate at position zero")
     return String.Index(encodedOffset: idx.encodedOffset &- 1)
   }
+
+  @usableFromInline @inline(never)
+  @_effects(releasenone)
+  internal func foreignErrorCorrectedGrapheme(
+    startingAt start: Int, endingAt end: Int
+  ) -> Character {
+#if _runtime(_ObjC)
+    _sanityCheck(self.isForeign)
+
+    // Both a fast-path for single-code-unit graphemes and validation:
+    //   ICU treats isolated surrogates as isolated graphemes
+    let count = end &- start
+    if start &- end == 1 {
+      return Character(String(self.foreignErrorCorrectedScalar(
+        startingAt: String.Index(encodedOffset: start)
+      ).0))
+    }
+
+    // TODO(String performance): Stack buffer if small enough
+    var cus = Array<UInt16>(repeating: 0, count: count)
+    cus.withUnsafeMutableBufferPointer {
+      _cocoaStringCopyCharacters(
+        from: self._object.cocoaObject,
+        range: start..<end,
+        into: $0.baseAddress._unsafelyUnwrappedUnchecked)
+    }
+    return cus.withUnsafeBufferPointer {
+      return Character(String._uncheckedFromUTF16($0))
+    }
+#else
+    fatalError("No foreign strings on Linux in this version of Swift")
+#endif
+  }
+}
+
+// Higher level aggregate operations. These should only be called when the
+// result is the sole operation done by a caller, otherwise it's always more
+// efficient to use `withFastUTF8` in the caller.
+extension _StringGuts {
+  @inlinable @inline(__always)
+  internal func errorCorrectedScalar(
+    startingAt i: Int
+  ) -> (Unicode.Scalar, scalarLength: Int) {
+    if _fastPath(isFastUTF8) {
+      return withFastUTF8 { _decodeScalar($0, startingAt: i) }
+    }
+    return foreignErrorCorrectedScalar(
+      startingAt: String.Index(encodedOffset: i))
+  }
+  @inlinable @inline(__always)
+  internal func errorCorrectedCharacter(
+    startingAt start: Int, endingAt end: Int
+  ) -> Character {
+    if _fastPath(isFastUTF8) {
+      return withFastUTF8(range: start..<end) { utf8 in
+        return Character(unchecked: String._uncheckedFromUTF8(utf8))
+      }
+    }
+
+    return foreignErrorCorrectedGrapheme(startingAt: start, endingAt: end)
+  }
 }

test/SILOptimizer/licm_exclusivity.swift

@@ -35,12 +35,14 @@ func run_ReversedArray(_ N: Int) {
 // TEST2: Hoist and Sink pairs attempt
 // TEST2: Hoisted
 
-// TESTSIL-LABEL: sil @$s16licm_exclusivity20count_unicodeScalarsyySS17UnicodeScalarViewVF : $@convention(thin) (@guaranteed String.UnicodeScalarView) -> () {
-// TESTSIL: bb0(%0 : $String.UnicodeScalarView)
-// TESTSIL-NEXT: %1 = global_addr @$s16licm_exclusivity5countSivp : $*Int
-// TESTSIL: begin_access [modify] [dynamic] [no_nested_conflict] %1 : $*Int
-// TESTSIL: end_access
-// TESTSIL: return
+// FIXME: <rdar://problem/45931225> Re-enable the below
+//
+// xTESTSIL-LABEL: sil @$s16licm_exclusivity20count_unicodeScalarsyySS17UnicodeScalarViewVF : $@convention(thin) (@guaranteed String.UnicodeScalarView) -> () {
+// xTESTSIL: bb0(%0 : $String.UnicodeScalarView)
+// xTESTSIL-NEXT: %1 = global_addr @$s16licm_exclusivity5countSivp : $*Int
+// xTESTSIL: begin_access [modify] [dynamic] [no_nested_conflict] %1 : $*Int
+// xTESTSIL: end_access
+// xTESTSIL: return
 var count: Int = 0
 public func count_unicodeScalars(_ s: String.UnicodeScalarView) {
   for _ in s {

test/api-digester/Outputs/stability-stdlib-abi.swift.expected

@@ -141,6 +141,12 @@ Class _SharedStringStorage has removed conformance to _NSStringCore
 Class _StringStorage has removed conformance to _NSStringCore
 Protocol _NSStringCore has been removed
 
+Func String.UnicodeScalarView._foreignSubscript(aligned:) has been removed
+Struct String.UnicodeScalarView has type witness type for Collection.Iterator changing from IndexingIterator<String.UnicodeScalarView> to String.UnicodeScalarView.Iterator
+Struct String.UnicodeScalarView has type witness type for Sequence.Iterator changing from IndexingIterator<String.UnicodeScalarView> to String.UnicodeScalarView.Iterator
+Func String._foreignSubscript(position:distance:) has been removed
+Struct String has type witness type for Collection.Iterator changing from IndexingIterator<String> to String.Iterator
+Struct String has type witness type for Sequence.Iterator changing from IndexingIterator<String> to String.Iterator
 Func Unicode.UTF32._decode(_:) has been removed
 Func _UnicodeParser._decode(_:repairingIllFormedSequences:into:) has been removed
 Func _UnicodeParser._parse(_:repairingIllFormedSequences:into:) has been removed

validation-test/stdlib/String.swift

@@ -126,7 +126,7 @@ StringTests.test("AssociatedTypes-UTF16View") {
   typealias View = String.UTF16View
   expectCollectionAssociatedTypes(
     collectionType: View.self,
-    iteratorType: IndexingIterator<View>.self,
+    iteratorType: View.Iterator.self,
     subSequenceType: Substring.UTF16View.self,
     indexType: View.Index.self,
     indicesType: View.Indices.self)
@@ -145,7 +145,7 @@ StringTests.test("AssociatedTypes-UnicodeScalarView") {
 StringTests.test("AssociatedTypes-CharacterView") {
   expectCollectionAssociatedTypes(
     collectionType: String.self,
-    iteratorType: IndexingIterator<String>.self,
+    iteratorType: String.Iterator.self,
     subSequenceType: Substring.self,
     indexType: String.Index.self,
     indicesType: DefaultIndices<String>.self)