From 5ca5cdd3966bfa9dea94f196b060405f01335481 Mon Sep 17 00:00:00 2001
From: Erik Mogensen <erma@escenic.com>
Date: Wed, 26 Sep 2018 16:28:53 +0200
Subject: [PATCH] Improve performance of parsing

Looping through strings in Javascript land is a lot slower than having
the JS engine do it natively.  String's indexOf does this faster than
looping through each character.

This change deals with parsed data (i.e. the raw text between the
tags), attribute values (what's in the quotes) and inside XML
comments.  These three types of data account for a very large portion
of characters in any XML document, leaving behind mainly names of tags
and of attributes.

It might be worth it to rewrite the switch statement, or move the
optimisations into the switch statement itself.
---
 lib/parsers/ltx.js | 27 ++++++++++++++++++++++++++-
 1 file changed, 26 insertions(+), 1 deletion(-)

diff --git a/lib/parsers/ltx.js b/lib/parsers/ltx.js
index b770524d..70da1071 100644
--- a/lib/parsers/ltx.js
+++ b/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
           }