Question

I have very large HTML that, if being parsed into DOM tree, would take much time, so this option despite being “proper” is not available. I need to remove all the inside-tag style declarations.

There is a regular expression that seem to work in most cases:

> re
/\sstyle\s*=(\"[^\">]*\"*|\'[^\'>]*\'*|[^\s>]*)/gi
> test
[ '<img src="some.jpg" style="width:auto" width="50" height="60">',
  '<img style=\'width:auto\'>',
  '<img style=\'width:auto>',
  '<img style=width:auto>',
  '<div style=\'\'>',
  '<div style=\'background-image:url(\'paper.gif\');\'',
  '<div style=\'background-image:url(\\\'paper.gif\\\');\'' ]
> test.forEach(function(t){console.log(t.replace(re,''))})
<img src="some.jpg" width="50" height="60">
<img>
<img>
<img>
<div>
<divpaper.gif');'
<divpaper.gif\');'

As you see, in case there are repeated quotes inside the value part, either with or without proper escaping, the regular expression doesn’t work. Any ideas how I can improve it?

Answer 1

Why would you want to write one big regular expression to do all of that at once?

Parsing it into a DOM tree might take too much time, but writing a hand-crafted parser will probably be better.

You can also mix the two: use a regular expression to isolate each and every tag (which is easy), then parse the attributes inside the tag, isolating (and removing) any style attribute you encounter.