optimizations

scinfu · Aug 2, 2024 · a690c8a · a690c8a
1 parent 7df0685
commit a690c8a
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 15 deletions.
diff --git a/Sources/CharacterReader.swift b/Sources/CharacterReader.swift
@@ -59,14 +59,12 @@ public final class CharacterReader {
         return String(scalar)
     }
 
-    public func consumeToAny(_ chars: Set<UnicodeScalar>) -> String {
+    public func consumeToAny(_ chars: Set<Unicode.Scalar.UTF8View.Element>) -> String {
         let start = pos
 
-        let utf8CodeUnits = Set(chars.flatMap { $0.utf8 })
-
         while pos < input.endIndex {
             let utf8Byte = input[pos]
-            if utf8CodeUnits.contains(utf8Byte) {
+            if chars.contains(utf8Byte) {
                 break
             }
             input.formIndex(after: &pos)
@@ -276,19 +274,19 @@ public final class CharacterReader {
         }
     }
 
-    static let dataTerminators: Set<UnicodeScalar> = [.Ampersand, .LessThan, TokeniserStateVars.nullScalr]
+    static let dataTerminators = Set([.Ampersand, .LessThan, TokeniserStateVars.nullScalr].flatMap { $0.utf8 })
 
     public func consumeData() -> String {
         return consumeToAny(CharacterReader.dataTerminators)
     }
 
-    static let tagNameTerminators: Set<UnicodeScalar> = [.BackslashT, .BackslashN, .BackslashR, .BackslashF, .Space, .Slash, .GreaterThan, TokeniserStateVars.nullScalr]
+    static let tagNameTerminators = Set([.BackslashT, .BackslashN, .BackslashR, .BackslashF, .Space, .Slash, .GreaterThan, TokeniserStateVars.nullScalr].flatMap { $0.utf8 })
 
     public func consumeTagName() -> String {
         return consumeToAny(CharacterReader.tagNameTerminators)
     }
 
-    public func consumeToAnySorted(_ chars: Set<UnicodeScalar>) -> String {
+    public func consumeToAnySorted(_ chars: Set<Unicode.Scalar.UTF8View.Element>) -> String {
         return consumeToAny(chars)
     }
 }

diff --git a/Sources/Entities.swift b/Sources/Entities.swift
@@ -51,7 +51,7 @@ public class Entities {
             return left.value != right.value
         }
 
-        private static let codeDelims: Set<UnicodeScalar>  = Set([",", ";"])
+        private static let codeDelims = Set([",", ";"].flatMap { $0.utf8 })
 
         init(string: String, size: Int, id: Int) {
 

diff --git a/Sources/TokeniserState.swift b/Sources/TokeniserState.swift
@@ -15,14 +15,14 @@ protocol TokeniserStateProtocol {
 public class TokeniserStateVars {
 	public static let nullScalr: UnicodeScalar = "\u{0000}"
 
-    static let attributeSingleValueChars = Set(["'", UnicodeScalar.Ampersand, nullScalr])
-    static let attributeDoubleValueChars = Set(["\"", UnicodeScalar.Ampersand, nullScalr])
-    static let attributeNameChars = Set([UnicodeScalar.BackslashT, "\n", "\r", UnicodeScalar.BackslashF, " ", "/", "=", ">", nullScalr, "\"", "'", UnicodeScalar.LessThan])
-    static let attributeValueUnquoted = Set([UnicodeScalar.BackslashT, "\n", "\r", UnicodeScalar.BackslashF, " ", UnicodeScalar.Ampersand, ">", nullScalr, "\"", "'", UnicodeScalar.LessThan, "=", "`"])
+    static let attributeSingleValueChars = Set(["'", UnicodeScalar.Ampersand, nullScalr].flatMap { $0.utf8 })
+    static let attributeDoubleValueChars = Set(["\"", UnicodeScalar.Ampersand, nullScalr].flatMap { $0.utf8 })
+    static let attributeNameChars = Set([UnicodeScalar.BackslashT, "\n", "\r", UnicodeScalar.BackslashF, " ", "/", "=", ">", nullScalr, "\"", "'", UnicodeScalar.LessThan].flatMap { $0.utf8 })
+    static let attributeValueUnquoted = Set([UnicodeScalar.BackslashT, "\n", "\r", UnicodeScalar.BackslashF, " ", UnicodeScalar.Ampersand, ">", nullScalr, "\"", "'", UnicodeScalar.LessThan, "=", "`"].flatMap { $0.utf8 })
 
-    static let dataDefaultStopChars: Set<UnicodeScalar> = [UnicodeScalar.Ampersand, UnicodeScalar.LessThan, TokeniserStateVars.nullScalr]
-    static let commentDefaultStopChars: Set<UnicodeScalar> = ["-", TokeniserStateVars.nullScalr]
-    static let readDataDefaultStopChars: Set<UnicodeScalar> = [UnicodeScalar.LessThan, TokeniserStateVars.nullScalr]
+    static let dataDefaultStopChars = Set([UnicodeScalar.Ampersand, UnicodeScalar.LessThan, TokeniserStateVars.nullScalr].flatMap { $0.utf8 })
+    static let commentDefaultStopChars = Set(["-", TokeniserStateVars.nullScalr].flatMap { $0.utf8 })
+    static let readDataDefaultStopChars = Set([UnicodeScalar.LessThan, TokeniserStateVars.nullScalr].flatMap { $0.utf8 })
 
 
     static let replacementChar: UnicodeScalar = Tokeniser.replacementChar