Use Unicode.Scalar instead of Character

All the chars we care about are a single scalar, so this avoids spending
time on the grapheme breaking algorithm.

Sources/HTMLStreamer/AttributedStringConverter.swift

@@ -21,7 +21,7 @@ public struct AttributedStringConverter<Callbacks: AttributedStringCallbacks> {
     private let configuration: AttributedStringConverterConfiguration
     private var fontCache: [FontTrait: PlatformFont] = [:]
     
-    private var tokenizer: Tokenizer<String.Iterator>!
+    private var tokenizer: Tokenizer<String.UnicodeScalarView.Iterator>!
     private var str: NSMutableAttributedString!
     
     private var actionStack: [ElementAction] = []
@@ -38,7 +38,7 @@ public struct AttributedStringConverter<Callbacks: AttributedStringCallbacks> {
     }
     
     public mutating func convert(html: String) -> NSAttributedString {
-        tokenizer = Tokenizer(chars: html.makeIterator())
+        tokenizer = Tokenizer(chars: html.unicodeScalars.makeIterator())
         str = NSMutableAttributedString()
         
         actionStack = []
@@ -48,7 +48,7 @@ public struct AttributedStringConverter<Callbacks: AttributedStringCallbacks> {
         while let token = tokenizer.next() {
             switch token {
             case .character(let c):
-                currentRun.append(c)
+                currentRun.unicodeScalars.append(c)
             case .comment:
                 // ignored
                 continue

Sources/HTMLStreamer/Tokenizer.swift

@@ -7,11 +7,11 @@
 
 import Foundation
 
-struct Tokenizer<Chars: IteratorProtocol<Character>>: IteratorProtocol {
+struct Tokenizer<Chars: IteratorProtocol<Unicode.Scalar>>: IteratorProtocol {
     typealias Element = Token
     
     private var chars: Chars
-    private var reconsumeStack: [Character] = []
+    private var reconsumeStack: [Unicode.Scalar] = []
     private var state = State.data
     private var returnState: State?
     private var temporaryBuffer: String?
@@ -34,7 +34,7 @@ struct Tokenizer<Chars: IteratorProtocol<Character>>: IteratorProtocol {
                 state = returnState
                 return next()
             } else {
-                return .character(temporaryBuffer!.removeFirst())
+                return .character(temporaryBuffer!.unicodeScalars.removeFirst())
             }
         case .endOfFile:
             return nil
@@ -143,13 +143,13 @@ struct Tokenizer<Chars: IteratorProtocol<Character>>: IteratorProtocol {
         }
     }
     
-    private mutating func reconsume(_ c: Character?) {
+    private mutating func reconsume(_ c: Unicode.Scalar?) {
         if let c {
             reconsumeStack.append(c)
         }
     }
     
-    private mutating func nextChar() -> Character? {
+    private mutating func nextChar() -> Unicode.Scalar? {
         if !reconsumeStack.isEmpty {
             return reconsumeStack.removeLast()
         } else {
@@ -157,7 +157,7 @@ struct Tokenizer<Chars: IteratorProtocol<Character>>: IteratorProtocol {
         }
     }
     
-    private mutating func peekChar() -> Character? {
+    private mutating func peekChar() -> Unicode.Scalar? {
         if let nextToReconsume = reconsumeStack.last {
             return nextToReconsume
         } else {
@@ -172,7 +172,7 @@ struct Tokenizer<Chars: IteratorProtocol<Character>>: IteratorProtocol {
     // TODO: extract this all out into a standalone type and test it separately
     private mutating func peek(count: Int) -> String {
         precondition(count >= 0)
-        var buf = ""
+        var buf = String.UnicodeScalarView()
         for _ in 0..<count {
             if let c = nextChar() {
                 buf.append(c)
@@ -181,7 +181,7 @@ struct Tokenizer<Chars: IteratorProtocol<Character>>: IteratorProtocol {
             }
         }
         reconsumeStack.append(contentsOf: buf.reversed())
-        return buf
+        return String(buf)
     }
     
     private mutating func consume(count: Int) {
@@ -211,7 +211,7 @@ struct Tokenizer<Chars: IteratorProtocol<Character>>: IteratorProtocol {
 }
 
 enum Token: Equatable {
-    case character(Character)
+    case character(Unicode.Scalar)
     case comment(String)
     case startTag(String, selfClosing: Bool, attributes: [Attribute])
     case endTag(String)
@@ -371,9 +371,9 @@ private extension Tokenizer {
         loop: while let c = nextChar() {
             switch c {
             case "a"..."z", "A"..."Z", "0"..."9":
-                temporaryBuffer!.append(c)
+                temporaryBuffer!.unicodeScalars.append(c)
             case ";":
-                temporaryBuffer!.append(c)
+                temporaryBuffer!.unicodeScalars.append(c)
                 break loop
             default:
                 reconsume(c)
@@ -396,7 +396,7 @@ private extension Tokenizer {
                 }
             }
             if referent != nil {
-                for c in buf[index...].reversed() {
+                for c in buf[index...].unicodeScalars.reversed() {
                     reconsume(c)
                 }
                 temporaryBuffer!.removeSubrange(index...)
@@ -556,7 +556,7 @@ private extension Tokenizer {
         let c = nextChar()
         switch c {
         case .some("0"..."9"):
-            characterReferenceCode = (characterReferenceCode! * 10) + UInt32(c!.wholeNumberValue!)
+            characterReferenceCode = (characterReferenceCode! * 10) + UInt32(c!.hexDigitValue!)
             return tokenizeDecimalCharacterReference()
         case ";":
             state = .numericCharacterReferenceEnd
@@ -574,7 +574,7 @@ private extension Tokenizer {
         switch c {
         case .some("0"..."9"), .some("a"..."z"), .some("A"..."Z"):
             if case .attributeValue(_) = returnState {
-                currentStartTag!.attributes.uncheckedLast.value.append(c!)
+                currentStartTag!.attributes.uncheckedLast.value.unicodeScalars.append(c!)
                 return tokenizeAmbiguousAmpersand()
             } else {
                 return .character(c!)
@@ -669,10 +669,10 @@ private extension Tokenizer {
                     c = c.asciiLowercase
                 }
                 if currentStartTag != nil {
-                    currentStartTag!.0.append(c)
+                    currentStartTag!.0.unicodeScalars.append(c)
                     continue
                 } else if currentEndTag != nil {
-                    currentEndTag!.append(c)
+                    currentEndTag!.unicodeScalars.append(c)
                     continue
                 } else {
                     fatalError("bad current token")
@@ -752,7 +752,7 @@ private extension Tokenizer {
                 }
                 // if c in ["\"", "'", "<"], parse error: unexpected-character-in-attribute-name
                 if currentStartTag != nil {
-                    currentStartTag!.attributes.uncheckedLast.name.append(c)
+                    currentStartTag!.attributes.uncheckedLast.name.unicodeScalars.append(c)
                     continue
                 } else if currentEndTag != nil {
                     continue
@@ -837,7 +837,7 @@ private extension Tokenizer {
                 case .some(let c):
                     // if c in ["\"", "'", "<", "=", "`"], parse error: unexpected-character-in-unquoted-attribute-value
                     if currentStartTag != nil {
-                        currentStartTag!.attributes.uncheckedLast.value.append(c)
+                        currentStartTag!.attributes.uncheckedLast.value.unicodeScalars.append(c)
                         continue
                     } else if currentEndTag != nil {
                         continue
@@ -868,7 +868,7 @@ private extension Tokenizer {
                         c = "\u{FFFD}"
                     }
                     if currentStartTag != nil {
-                        currentStartTag!.attributes.uncheckedLast.value.append(c)
+                        currentStartTag!.attributes.uncheckedLast.value.unicodeScalars.append(c)
                         continue
                     } else if currentEndTag != nil {
                         continue
@@ -914,7 +914,7 @@ private extension Tokenizer {
                 // parse error: unexpected-null-character
                 c = "\u{FFFD}"
             }
-            currentComment!.append(c)
+            currentComment!.unicodeScalars.append(c)
             return tokenizeBogusComment()
         }
     }
@@ -998,7 +998,7 @@ private extension Tokenizer {
                 // parse error: unexpected-null-character
                 c = "\u{FFFD}"
             }
-            currentComment!.append(c)
+            currentComment!.unicodeScalars.append(c)
             return tokenizeComment()
         }
     }
@@ -1188,7 +1188,7 @@ private extension Tokenizer {
             } else if ("A"..."Z").contains(c) {
                 c = c.asciiLowercase
             }
-            currentDoctype!.0.append(c)
+            currentDoctype!.0.unicodeScalars.append(c)
             return tokenizeDoctypeName()
         }
     }
@@ -1309,7 +1309,7 @@ private extension Tokenizer {
                 // parse error: unexpected-null-character
                 c = "\u{FFFD}"
             }
-            currentDoctype!.publicIdentifier!.append(c)
+            currentDoctype!.publicIdentifier!.unicodeScalars.append(c)
             return tokenizeDoctypePublicIdentifier(quotes: quotes)
         }
     }
@@ -1450,7 +1450,7 @@ private extension Tokenizer {
                 // parse error: unexpected-null-character
                 c = "\u{FFFD}"
             }
-            currentDoctype!.systemIdentifier!.append(c)
+            currentDoctype!.systemIdentifier!.unicodeScalars.append(c)
             return tokenizeDoctypeSystemIdentifier(quotes: quotes)
         }
     }
@@ -1495,13 +1495,6 @@ private extension Tokenizer {
     }
 }
 
-private extension Character {
-    var asciiLowercase: Character {
-        assert(("A"..."Z").contains(self))
-        return Character(Unicode.Scalar(asciiValue! + 0x20))
-    }
-}
-
 private extension Array {
     // Optimization: allows in-place modification of the last element of the array.
     var uncheckedLast: Element {
@@ -1513,3 +1506,36 @@ private extension Array {
         }
     }
 }
+
+private extension Unicode.Scalar {
+    var asciiLowercase: Unicode.Scalar {
+        assert(("A"..."Z").contains(self))
+        return Unicode.Scalar(value + 0x20)!
+    }
+    
+    var hexDigitValue: Int? {
+        switch self {
+        case "0": 0
+        case "1": 1
+        case "2": 2
+        case "3": 3
+        case "4": 4
+        case "5": 5
+        case "6": 6
+        case "7": 7
+        case "8": 8
+        case "9": 9
+        case "A": 0xA
+        case "B": 0xB
+        case "C": 0xC
+        case "D": 0xD
+        case "E": 0xE
+        case "F": 0xF
+        default: nil
+        }
+    }
+    
+    var isNumber: Bool {
+        ("0"..."9").contains(self)
+    }
+}

Tests/HTMLStreamerTests/AttributedStringConverterTests.swift

@@ -207,4 +207,11 @@ final class AttributedStringConverterTests: XCTestCase {
         ]))
     }
     
+    func testMultiScalar() {
+        XCTAssertEqual(convert("🇺🇸"), NSAttributedString(string: "🇺🇸", attributes: [
+            .font: font,
+            .paragraphStyle: NSParagraphStyle.default,
+        ]))
+    }
+    
 }

Tests/HTMLStreamerTests/TokenizerTests.swift

@@ -11,7 +11,7 @@ import XCTest
 final class TokenizerTests: XCTestCase {
     
     private func tokenize(_ s: String) -> [Token] {
-        let iterator = Tokenizer(chars: s.makeIterator())
+        let iterator = Tokenizer(chars: s.unicodeScalars.makeIterator())
 //        let iterator = PrintIterator(inner: Tokenizer(chars: s.makeIterator()))
         return Array(AnySequence({ iterator }))
     }
@@ -23,8 +23,8 @@ final class TokenizerTests: XCTestCase {
         XCTAssertEqual(tokenize("&notin"), [.character("¬"), .character("i"), .character("n")])
         // unknown-named-character-reference:
         XCTAssertEqual(tokenize("&notit;"), [.character("¬"), .character("i"), .character("t"), .character(";")])
-        XCTAssertEqual(tokenize("&asdf"), "&asdf".map { .character($0) })
-        XCTAssertEqual(tokenize("&a"), "&a".map { .character($0) })
+        XCTAssertEqual(tokenize("&asdf"), "&asdf".unicodeScalars.map { .character($0) })
+        XCTAssertEqual(tokenize("&a"), "&a".unicodeScalars.map { .character($0) })
         
         // attribute special case
         XCTAssertEqual(tokenize("<a a='&nota' />"), [.startTag("a", selfClosing: true, attributes: [Attribute(name: "a", value: "&nota")])])
@@ -70,6 +70,10 @@ final class TokenizerTests: XCTestCase {
         XCTAssertEqual(tokenize(#"<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">"#), [.doctype("html", forceQuirks: false, publicIdentifier: "-//W3C//DTD HTML 4.01//EN", systemIdentifier: "http://www.w3.org/TR/html4/strict.dtd")])
     }
     
+    func testMultiScalar() {
+        XCTAssertEqual(tokenize("🇺🇸"), [.character("\u{1F1FA}"), .character("\u{1F1F8}")])
+    }
+    
 }
 
 private struct PrintIterator<Inner: IteratorProtocol>: IteratorProtocol {