PrettyPrinterPerformance PrettyPrinterPerformance PrettyPrinterPerfor…

…mance PrettyPrinterPerformance Optimized the PrettyPrinter for swiftlang#894

Worked to get the perfomance to be closer to where we were before the changes in swiftlang#883. This code should be only about 1.5% slower rather than 7% slower.

Using a lazy filter as `count(where:_)` isn't avaliable < Swift 6.0.

One forward loop and using the UTF8 view makes this faster than the original code pre-swiftlang#883.

Sources/SwiftFormat/PrettyPrint/PrettyPrintBuffer.swift

@@ -122,16 +122,17 @@ struct PrettyPrintBuffer {
     // In case of comments, we may get a multi-line string.
     // To account for that case, we need to correct the lineNumber count.
     // The new column is only the position within the last line.
-    let lines = text.split(separator: "\n")
-    lineNumber += lines.count - 1
-    if lines.count > 1 {
-      // in case we have inserted new lines, we need to reset the column
-      column = lines.last?.count ?? 0
-    } else {
-      // in case it is an end of line comment or a single line comment,
-      // we just add to the current column
-      column += lines.last?.count ?? 0
+    var lastLength = 0
+    // We are only interested in "\n" we can use the UTF8 view and skip the grapheme clustering.
+    for element in text.utf8 {
+      if element == 10 {
+        lineNumber += 1
+        lastLength = 0
+      } else {
+        lastLength += 1
+      }
     }
+    column += lastLength
   }
 
   /// Request that the given number of spaces be printed out before the next text token.