[stdlib] String: Implement happy paths for index validation

Author: lorentey

stdlib/public/core/CMakeLists.txt

@@ -157,6 +157,7 @@ set(SWIFTLIB_ESSENTIAL
   StringProtocol.swift
   StringIndex.swift
   StringIndexConversions.swift
+  StringIndexValidation.swift
   StringInterpolation.swift
   StringLegacy.swift
   StringNormalization.swift

stdlib/public/core/GroupInfo.json

@@ -26,6 +26,7 @@
     "StringHashable.swift",
     "StringIndex.swift",
     "StringIndexConversions.swift",
+    "StringIndexValidation.swift",
     "StringInterpolation.swift",
     "StringLegacy.swift",
     "StringNormalization.swift",

stdlib/public/core/StringCharacterView.swift

@@ -54,7 +54,7 @@ extension String: BidirectionalCollection {
   ///   `endIndex`.
   /// - Returns: The index value immediately after `i`.
   public func index(after i: Index) -> Index {
-    let i = _guts.roundDownToNearestCharacter(_guts.validateScalarIndex(i))
+    let i = _guts.validateCharacterIndex(i)
     return _uncheckedIndex(after: i)
   }
 
@@ -83,8 +83,7 @@ extension String: BidirectionalCollection {
   ///   `startIndex`.
   /// - Returns: The index value immediately before `i`.
   public func index(before i: Index) -> Index {
-    let i = _guts.roundDownToNearestCharacter(
-      _guts.validateInclusiveScalarIndex(i))
+    let i = _guts.validateInclusiveCharacterIndex(i)
     // Note: Aligning an index may move it closer towards the `startIndex`, so
     // the `i > startIndex` check needs to come after rounding.
     _precondition(i > startIndex, "String index is out of bounds")
@@ -138,8 +137,7 @@ extension String: BidirectionalCollection {
 
     // TODO: known-ASCII and single-scalar-grapheme fast path, etc.
 
-    var i = _guts.roundDownToNearestCharacter(
-      _guts.validateInclusiveScalarIndex(i))
+    var i = _guts.validateInclusiveCharacterIndex(i)
 
     if distance >= 0 {
       for _ in stride(from: 0, to: distance, by: 1) {
@@ -214,8 +212,7 @@ extension String: BidirectionalCollection {
     let limit = _guts.ensureMatchingEncoding(limit)
     let start = _guts.ensureMatchingEncoding(i)
 
-    var i = _guts.roundDownToNearestCharacter(
-      _guts.validateInclusiveScalarIndex(i))
+    var i = _guts.validateInclusiveCharacterIndex(i)
 
     if distance >= 0 {
       for _ in stride(from: 0, to: distance, by: 1) {
@@ -253,10 +250,8 @@ extension String: BidirectionalCollection {
     // grapheme breaks -- swapping `start` and `end` may change the magnitude of
     // the result.
 
-    let start = _guts.roundDownToNearestCharacter(
-      _guts.validateInclusiveScalarIndex(start))
-    let end = _guts.roundDownToNearestCharacter(
-      _guts.validateInclusiveScalarIndex(end))
+    let start = _guts.validateInclusiveCharacterIndex(start)
+    let end = _guts.validateInclusiveCharacterIndex(end)
 
     // TODO: known-ASCII and single-scalar-grapheme fast path, etc.
 

stdlib/public/core/StringGuts.swift

@@ -400,186 +400,6 @@ extension _StringGuts {
   }
 }
 
-// Index validation
-extension _StringGuts {
-  /// Validate `i` and adjust its position toward the start, returning the
-  /// resulting index or trapping as appropriate. If this function returns, then
-  /// the returned value
-  ///
-  /// - has an encoding that matches this string,
-  /// - is within the bounds of this string, and
-  /// - is aligned on a scalar boundary.
-  @_alwaysEmitIntoClient
-  internal func validateScalarIndex(_ i: String.Index) -> String.Index {
-    let i = ensureMatchingEncoding(i)
-    _precondition(i._encodedOffset < count, "String index is out of bounds")
-    return scalarAlign(i)
-  }
-
-  /// Validate `i` and adjust its position toward the start, returning the
-  /// resulting index or trapping as appropriate. If this function returns, then
-  /// the returned value
-  ///
-  /// - has an encoding that matches this string,
-  /// - is within `start ..< end`, and
-  /// - is aligned on a scalar boundary.
-  @_alwaysEmitIntoClient
-  internal func validateScalarIndex(
-    _ i: String.Index,
-    in bounds: Range<String.Index>
-  ) -> String.Index {
-    _internalInvariant(bounds.upperBound <= endIndex)
-
-    let i = ensureMatchingEncoding(i)
-    _precondition(i >= bounds.lowerBound && i < bounds.upperBound,
-      "Substring index is out of bounds")
-    return scalarAlign(i)
-  }
-}
-
-extension _StringGuts {
-  /// Validate `i` and adjust its position toward the start, returning the
-  /// resulting index or trapping as appropriate. If this function returns, then
-  /// the returned value
-  ///
-  /// - has an encoding that matches this string,
-  /// - is within the bounds of this string (including the `endIndex`), and
-  /// - is aligned on a scalar boundary.
-  @_alwaysEmitIntoClient
-  internal func validateInclusiveScalarIndex(
-    _ i: String.Index
-  ) -> String.Index {
-    let i = ensureMatchingEncoding(i)
-    _precondition(i._encodedOffset <= count, "String index is out of bounds")
-    return scalarAlign(i)
-  }
-
-  /// Validate `i` and adjust its position toward the start, returning the
-  /// resulting index or trapping as appropriate. If this function returns, then
-  /// the returned value
-  ///
-  /// - has an encoding that matches this string,
-  /// - is within the bounds of this string (including the `endIndex`), and
-  /// - is aligned on a scalar boundary.
-  internal func validateInclusiveScalarIndex(
-    _ i: String.Index,
-    in bounds: Range<String.Index>
-  ) -> String.Index {
-    _internalInvariant(bounds.upperBound <= endIndex)
-
-    let i = ensureMatchingEncoding(i)
-    _precondition(i >= bounds.lowerBound && i <= bounds.upperBound,
-      "Substring index is out of bounds")
-    return scalarAlign(i)
-  }
-}
-
-extension _StringGuts {
-  @_alwaysEmitIntoClient
-  internal func validateSubscalarRange(
-    _ range: Range<String.Index>
-  ) -> Range<String.Index> {
-    let upper = ensureMatchingEncoding(range.upperBound)
-    let lower = ensureMatchingEncoding(range.lowerBound)
-
-    // Note: if only `lower` was miscoded, then the range invariant `lower <=
-    // upper` may no longer hold after the above conversions, so we need to
-    // re-check it here.
-    _precondition(upper._encodedOffset <= count && lower <= upper,
-      "String index range is out of bounds")
-
-    return Range(_uncheckedBounds: (lower, upper))
-  }
-
-  @_alwaysEmitIntoClient
-  internal func validateSubscalarRange(
-    _ range: Range<String.Index>,
-    in bounds: Range<String.Index>
-  ) -> Range<String.Index> {
-    _internalInvariant(bounds.upperBound <= endIndex)
-
-    let upper = ensureMatchingEncoding(range.upperBound)
-    let lower = ensureMatchingEncoding(range.lowerBound)
-
-    // Note: if only `lower` was miscoded, then the range invariant `lower <=
-    // upper` may no longer hold after the above conversions, so we need to
-    // re-check it here.
-    _precondition(
-      upper <= bounds.upperBound
-      && lower >= bounds.lowerBound
-      && lower <= upper,
-      "Substring index range is out of bounds")
-
-    return Range(_uncheckedBounds: (lower, upper))
-  }
-}
-
-extension _StringGuts {
-  /// Validate `range` and adjust the position of its bounds, returning the
-  /// resulting range or trapping as appropriate. If this function returns, then
-  /// the bounds of the returned value
-  ///
-  /// - have an encoding that matches this string,
-  /// - are within the bounds of this string, and
-  /// - are aligned on a scalar boundary.
-  internal func validateScalarRange(
-    _ range: Range<String.Index>
-  ) -> Range<String.Index> {
-    var upper = ensureMatchingEncoding(range.upperBound)
-    var lower = ensureMatchingEncoding(range.lowerBound)
-
-    // Note: if only `lower` was miscoded, then the range invariant `lower <=
-    // upper` may no longer hold after the above conversions, so we need to
-    // re-check it here.
-    _precondition(upper._encodedOffset <= count && lower <= upper,
-      "String index range is out of bounds")
-
-    upper = scalarAlign(upper)
-    lower = scalarAlign(lower)
-
-    // Older binaries may generate `startIndex` without the
-    // `_isCharacterAligned` flag. Compensate for that here so that substrings
-    // that start at the beginning will never get the sad path in
-    // `index(after:)`. Note that we don't need to do this for `upper` and we
-    // don't need to compare against the `endIndex` -- those aren't nearly as
-    // critical.
-    if lower._encodedOffset == 0 { lower = lower._characterAligned }
-
-    return Range(_uncheckedBounds: (lower, upper))
-  }
-
-  /// Validate `range` and adjust the position of its bounds, returning the
-  /// resulting range or trapping as appropriate. If this function returns, then
-  /// the bounds of the returned value
-  ///
-  /// - have an encoding that matches this string,
-  /// - are within `start ..< end`, and
-  /// - are aligned on a scalar boundary.
-  internal func validateScalarRange(
-    _ range: Range<String.Index>,
-    in bounds: Range<String.Index>
-  ) -> Range<String.Index> {
-    _internalInvariant(bounds.upperBound <= endIndex)
-
-    var upper = ensureMatchingEncoding(range.upperBound)
-    var lower = ensureMatchingEncoding(range.lowerBound)
-
-    // Note: if only `lower` was miscoded, then the range invariant `lower <=
-    // upper` may no longer hold after the above conversions, so we need to
-    // re-check it here.
-    _precondition(
-      upper <= bounds.upperBound
-      && lower >= bounds.lowerBound
-      && lower <= upper,
-      "Substring index range is out of bounds")
-
-    upper = scalarAlign(upper)
-    lower = scalarAlign(lower)
-
-    return Range(_uncheckedBounds: (lower, upper))
-  }
-}
-
 // Old SPI(corelibs-foundation)
 extension _StringGuts {
   @available(*, deprecated)

stdlib/public/core/StringIndex.swift

@@ -479,6 +479,18 @@ extension String.Index {
   }
 }
 
+extension String.Index {
+  @_alwaysEmitIntoClient @inline(__always) // Swift 5.7
+  internal var _isUTF8CharacterIndex: Bool {
+    _canBeUTF8 && _isCharacterAligned
+  }
+
+  @_alwaysEmitIntoClient @inline(__always) // Swift 5.7
+  internal var _isUTF8ScalarIndex: Bool {
+    _canBeUTF8 && _isScalarAligned
+  }
+}
+
 extension String.Index: Equatable {
   @inlinable @inline(__always)
   public static func == (lhs: String.Index, rhs: String.Index) -> Bool {