Improve performance of parsing

benchmarks/parsers.js

@@ -64,7 +64,7 @@ var suite = new benchmark.Suite('XML parsers comparison')
 parsers.forEach(function (parser) {
   parser.parse('<r>')
   suite.add(parser.name, function () {
-    parser.parse('<foo bar="baz">quux</foo>')
+    parser.parse('<foo bar="urn:uuid:60a76c80-d399-11d9-b91C-0003939e0af6">urn:uuid:60a76c80-d399-11d9-b91C-0003939e0af6 urn:uuid:60a76c80-d399-11d9-b91C-0003939e0af6</foo>')
   })
 })
 

lib/parsers/ltx.js

@@ -25,6 +25,7 @@ var SaxLtx = module.exports = function SaxLtx () {
   var endTag
   var selfClosing
   var attrQuote
+  var attrQuoteChar
   var recordStart = 0
   var attrName
 
@@ -54,13 +55,36 @@ var SaxLtx = module.exports = function SaxLtx () {
 
     function endRecording () {
       if (typeof recordStart === 'number') {
-        var recorded = data.slice(recordStart, pos)
+        var recorded = data.substring(recordStart, pos)
         recordStart = undefined
         return recorded
       }
     }
 
     for (; pos < data.length; pos++) {
+      if (state === STATE_TEXT) {
+        // if we're looping through text, fast-forward using indexOf to
+        // the next '<' character
+        const lt = data.indexOf('<', pos)
+        if (lt !== -1 && pos !== lt) {
+          pos = lt
+        }
+      } else if (state === STATE_ATTR_VALUE) {
+        // if we're looping through an attribute, fast-forward using
+        // indexOf to the next end quote character
+        const quot = data.indexOf(attrQuoteChar, pos)
+        if (quot !== -1) {
+          pos = quot
+        }
+      } else if (state === STATE_IGNORE_COMMENT) {
+        // if we're looping through a comment, fast-forward using
+        // indexOf to the first end-comment character
+        const endcomment = data.indexOf('-->', pos)
+        if (endcomment !== -1) {
+          pos = endcomment + 2 // target the '>' character
+        }
+      }
+
       var c = data.charCodeAt(pos)
       switch (state) {
         case STATE_TEXT:
@@ -153,6 +177,7 @@ var SaxLtx = module.exports = function SaxLtx () {
         case STATE_ATTR_QUOT:
           if (c === 34 /* " */ || c === 39 /* ' */) {
             attrQuote = c
+            attrQuoteChar = c === 34 ? '"' : "'"
             state = STATE_ATTR_VALUE
             recordStart = pos + 1
           }