Fix bug when parsing named character references that contain numbers (#4

)

Sources/HTMLEntities/String+HTMLEntities.swift

@@ -97,9 +97,12 @@ public extension String {
         let unicodes = self.unicodeScalars
 
         // result buffer
-        var str: String = ""
+        var str: String? = nil
 
         // entity buffer
+        // use optional string since there are issues on Linux when checking
+        // again empty string, i.e., "\u{200C}" == "" is true; "\u{200C}" is
+        // the named character ‌
         var entity: String = ""
 
         // current parse state
@@ -139,7 +142,7 @@ public extension String {
                     // entity can only be a number type
                     state = .Number
                 }
-                else if unicode.isAlpha {
+                else if unicode.isAlphaNumeric {
                     // entity can only be named character reference type
                     state = .Named
 
@@ -241,10 +244,12 @@ public extension String {
                         let unicodeScalar = UnicodeScalar(code) {
                         // reached end of entity
                         // move unbuffered unicodes over to the result buffer
-                        str.append(String(unicodes[leftIndex..<ampersandIndex]))
+                        str = str == nil ? "" : str
+
+                        str?.append(String(unicodes[leftIndex..<ampersandIndex]))
 
                         // append unescaped character to result buffer
-                        str.append(Character(unicodeScalar))
+                        str?.append(Character(unicodeScalar))
 
                         // move left index since we have buffered everything
                         // up to and including the current entity
@@ -260,16 +265,14 @@ public extension String {
             // move currentIndex to the position of the next unicode to be consumed
             currentIndex = nextIndex
         }
-
-        if str == "" {
-            // no unescapable entity found
-            // return string as it is
-            return self
+
+        if var str = str {
+            // append rest of string to result buffer
+            str.append(String(unicodes[leftIndex..<unicodes.endIndex]))
+
+            return str
         }
-
-        // append rest of string to result buffer
-        str.append(String(unicodes[leftIndex..<unicodes.endIndex]))
-
-        return str
+
+        return self
     }
 }

Sources/HTMLEntities/Utilities.swift

@@ -27,9 +27,9 @@ func invert<K, V: Hashable>(_ dict: [K: V]) -> [V: K] {
 }
 
 extension UInt32 {
-    var isAlpha: Bool {
-        // ASCII values of [A-Z] and [a-z]
-        return 65...90 ~= self || 97...122 ~= self
+    var isAlphaNumeric: Bool {
+        // ASCII values of [0-9], [A-Z0, [and [a-z]
+        return self.isNumeral || 65...90 ~= self || 97...122 ~= self
     }
 
     var isAmpersand: Bool {
@@ -86,7 +86,7 @@ extension UInt32 {
         case .Hex:
             return self.isHexNumeral
         case .Named:
-            return self.isAlpha
+            return self.isAlphaNumeric
         default:
             return false
         }

Tests/HTMLEntitiesTests/HTMLEntitiesTest.swift

@@ -33,6 +33,18 @@ let str3Unescaped = "Jako efektivnější se nám jeví pořádání tzv. Road S
 let str3Escaped = "Jako efektivnější se nám jeví pořádání tzv. Road Show prostřednictvím našich autorizovaných dealerů v Čechách a na Moravě, které proběhnou v průběhu září a října."
 
 class HTMLEntitiesTests: XCTestCase {
+    func testNamedCharacterReferences() {
+        XCTAssertEqual(html4NamedCharactersDecodeMap.count, html4NamedCharactersEncodeMap.count)
+
+        for (reference, unicode) in html4NamedCharactersDecodeMap {
+            let unescaped = String(UnicodeScalar(unicode)!)
+            let escaped = reference
+
+            XCTAssertEqual(unescaped.htmlEscape(), escaped)
+            XCTAssertEqual(escaped.htmlUnescape(), unescaped)
+        }
+    }
+
     func testEncode() {
         XCTAssertEqual(str1Unescaped.htmlEscape(), str1Escaped)
         XCTAssertEqual(str2Unescaped.htmlEscape(), str2Escaped)
@@ -105,6 +117,7 @@ class HTMLEntitiesTests: XCTestCase {
 
     static var allTests : [(String, (HTMLEntitiesTests) -> () throws -> Void)] {
         return [
+            ("testNamedCharacterReferences", testNamedCharacterReferences),
             ("testEncode", testEncode),
             ("testDecode", testDecode),
             ("testInvertibility", testInvertibility),