Pre-parse character classes to reduce indexOf overhead

lib/sax.js

@@ -234,14 +234,37 @@ var whitespace = "\r\n\t "
   , nameBody = nameStart+number+"-."
   , quote = "'\""
   , entity = number+letter+"#"
+  , attribEnd = whitespace + ">"
   , CDATA = "[CDATA["
   , DOCTYPE = "DOCTYPE"
   , XML_NAMESPACE = "http://www.w3.org/XML/1998/namespace"
   , XMLNS_NAMESPACE = "http://www.w3.org/2000/xmlns/"
   , rootNS = { xml: XML_NAMESPACE, xmlns: XMLNS_NAMESPACE }
 
-function is (charclass, c) { return charclass.indexOf(c) !== -1 }
-function not (charclass, c) { return !is(charclass, c) }
+// turn all the string character sets into character class objects.
+whitespace = charClass(whitespace)
+number = charClass(number)
+letter = charClass(letter)
+nameStart = charClass(nameStart)
+nameBody = charClass(nameBody)
+quote = charClass(quote)
+entity = charClass(entity)
+attribEnd = charClass(attribEnd)
+
+function charClass (str) {
+  return str.split("").reduce(function (s, c) {
+    s[c] = true
+    return s
+  }, {})
+}
+
+function is (charclass, c) {
+  return charclass[c]
+}
+
+function not (charclass, c) {
+  return !charclass[c]
+}
 
 var S = 0
 sax.STATE =
@@ -905,7 +928,7 @@ function write (chunk) {
       continue
 
       case S.ATTRIB_VALUE_UNQUOTED:
-        if (not(whitespace+">",c)) {
+        if (not(attribEnd,c)) {
           if (c === "&") parser.state = S.ATTRIB_VALUE_ENTITY_U
           else parser.attribValue += c
           continue