ASCII fast paths for grapheme stride (#72064)

Author: Catfish-Man

stdlib/public/core/StringCharacterView.swift

@@ -249,7 +249,7 @@ extension String: BidirectionalCollection {
     }
     return i
   }
-
+  
   /// Returns the distance between two indices.
   ///
   /// - Parameters:
@@ -269,23 +269,25 @@ extension String: BidirectionalCollection {
     let start = _guts.validateInclusiveCharacterIndex_5_7(start)
     let end = _guts.validateInclusiveCharacterIndex_5_7(end)
 
-    // TODO: known-ASCII and single-scalar-grapheme fast path, etc.
-
     // Per SE-0180, `start` and `end` are allowed to fall in between Character
     // boundaries, in which case this function must still terminate without
     // trapping and return a result that makes sense.
-
-    var i = start
+    var i = start._encodedOffset
     var count = 0
-    if i < end {
-      while i < end { // Note `<` instead of `==`
-        count += 1
-        i = _uncheckedIndex(after: i)
+    if start < end {
+      while i < end._encodedOffset { // Note `<` instead of `==`
+        count &+= 1
+        /*
+         For the purposes of this loop, this should be equivalent to
+         _uncheckedIndex(after: i). We don't need to spend time setting up
+         actual Indexes when we only care about counting strides.
+         */
+        i &+= _guts._opaqueCharacterStride(startingAt: i)
       }
-    } else if i > end {
-      while i > end { // Note `<` instead of `==`
-        count -= 1
-        i = _uncheckedIndex(before: i)
+    } else if start > end {
+      while i > end._encodedOffset { // Note `<` instead of `==`
+        count &-= 1
+        i &-= _guts._opaqueCharacterStride(endingAt: i)
       }
     }
     return count

stdlib/public/core/StringGraphemeBreaking.swift

@@ -196,9 +196,30 @@ extension _StringGuts {
   /// inconsistent with `_opaqueCharacterStride(endingAt:)`. On the other hand,
   /// this behavior makes this suitable for use in substrings whose start index
   /// itself does not fall on a cluster boundary.
-  @usableFromInline @inline(never)
+  @usableFromInline @inline(__always)
   @_effects(releasenone)
   internal func _opaqueCharacterStride(startingAt i: Int) -> Int {
+    _internalInvariant(i < endIndex._encodedOffset)
+    if isFastUTF8 {
+      let fast = withFastUTF8 { utf8 in
+        if i &+ 1 == utf8.count { return true }
+        let pair = UnsafeRawPointer(
+          utf8.baseAddress.unsafelyUnwrapped
+        ).loadUnaligned(fromByteOffset: i, as: UInt16.self)
+        //& 0x8080 == 0 is "both not ASCII", != 0x0A0D is "not CRLF"
+        return pair & 0x8080 == 0 && pair != 0x0A0D
+      }
+      if _fastPath(fast) {
+        _internalInvariant(_opaqueComplexCharacterStride(startingAt: i) == 1)
+        return 1
+      }
+    }
+    
+    return _opaqueComplexCharacterStride(startingAt: i)
+  }
+
+  @_effects(releasenone) @inline(never)
+  internal func _opaqueComplexCharacterStride(startingAt i: Int) -> Int {
     if _slowPath(isForeign) {
       return _foreignOpaqueCharacterStride(startingAt: i)
     }
@@ -221,9 +242,32 @@ extension _StringGuts {
   ///
   /// Note: unlike `_opaqueCharacterStride(startingAt:)`, this method always
   /// finds a correct grapheme cluster boundary.
-  @usableFromInline @inline(never)
+
+  @usableFromInline @inline(__always)
   @_effects(releasenone)
   internal func _opaqueCharacterStride(endingAt i: Int) -> Int {
+    if i <= 1 {
+      return i
+    }
+    if isFastUTF8 {
+      let fast = withFastUTF8 { utf8 in
+        let pair = UnsafeRawPointer(
+          utf8.baseAddress.unsafelyUnwrapped
+        ).loadUnaligned(fromByteOffset: i &- 2, as: UInt16.self)
+        //& 0x8080 == 0 is "both not ASCII", != 0x0A0D is "not CRLF"
+        return pair & 0x8080 == 0 && pair != 0x0A0D
+      }
+      if _fastPath(fast) {
+        _internalInvariant(_opaqueComplexCharacterStride(endingAt: i) == 1)
+        return 1
+      }
+    }
+
+    return _opaqueComplexCharacterStride(endingAt: i)
+  }
+
+  @_effects(releasenone) @inline(never)
+  internal func _opaqueComplexCharacterStride(endingAt i: Int) -> Int {
     if _slowPath(isForeign) {
       return _foreignOpaqueCharacterStride(endingAt: i)
     }