Issue 2467 - Ingore inner quotes when parsing CSS selectors

include.preload.js

 /*
  * This file is part of Adblock Plus <https://adblockplus.org/>,
  * Copyright (C) 2006-2015 Eyeo GmbH
  *
  * Adblock Plus is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 3 as
  * published by the Free Software Foundation.
  *
  * Adblock Plus is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
@@ skipping 135 matching lines @@
 {
   var result = [];
   var prefix = "::content ";
 
   for (var i = 0; i < selectors.length; i++)
   {
     var selector = selectors[i];
     var start = 0;
     var sep = "";
 
     for (var j = 0; j < selector.length; j++)

[Thomas Greiner, 2015/05/07 17:42:58]

This loop is only necessary if there's a `,` in the string and since that's only
the case for less than 0.5% of element hiding filters in EasyListGermany it
seems to be well worth checking for it in advance.

Not really related to this issue though so we can tackle this in a separate
ticket.

[Sebastian Noack, 2015/05/07 18:59:40]

Not a bad idea, of course assuming that it makes a notable difference
performance-wise. After all this state machine doesn't perform too bad.

[Sebastian Noack, 2015/05/07 19:17:22]

So I tried following:

function convertSelectorsForShadowDOM(selectors)
{
  t = Date.now();
  ...
  for (var i = 0; i < selectors.length; i++)
  {
    var selector = selectors[i];
    if (selector.indexOf(",") == -1)
    {
      result.push(prefix + selector);
      continue;
    }
    ...
  }
  console.log(Date.now() - t);
  result;
}

However, I couldn't measure any difference. Both, with and without the shortcut
the function runs for 4-8ms on my notebook while loading nytimes.com. But feel
free to do a little more benchmarking. 

[Thomas Greiner, 2015/05/08 17:03:25]

I'd recommend using `window.performance.now()` for that instead of `Date.now()`
since it's much more accurate. Also don't forget to consider the accumulative
number since it might run in multiple frames and thereby slow down the page load
by more than just a few milliseconds.

I saw an up to 50% speed improvement on my computer on recode.net (5 frames)
compared to the approach without. Generally, there is a speed improvement for
about 90% of page loads but the absolute numbers for time saved depend on how
many frames the page has and seem to vary between 2ms to 6ms per frame (e.g.
10-30ms on recode.net) from what I've observed.

[Sebastian Noack, 2015/05/08 17:21:28]

I'm not sure whether I can confirm your benchmark. But if you think its an
improvement I wouldn't mind having a shortcut here, as its a trivial change.
Feel free to submit a patch.

[Sebastian Noack, 2015/05/15 11:35:21]

There you go: http://codereview.adblockplus.org/4759827771293696/

     {
       var chr = selector[j];
       if (chr == "\\")
         j++;
       else if (chr == sep)
         sep = "";
-      else if (chr == '"' || chr == "'")
-        sep = chr;
-      else if (chr == "," && sep == "")
+      else if (sep == "")
       {
-        result.push(prefix + selector.substring(start, j));
-        start = j + 1;
+        if (chr == '"' || chr == "'")
+          sep = chr;
+        else if (chr == ",")
+        {
+          result.push(prefix + selector.substring(start, j));
+          start = j + 1;
+        }
       }
     }
 
     result.push(prefix + selector.substring(start));
   }
 
   return result;
 }
 
 function init(document)
@@ skipping 108 matching lines @@
   }, true);
 
   return updateStylesheet;
 }
 
 if (document instanceof HTMLDocument)
 {
   checkSitekey();
   window.updateStylesheet = init(document);
 }