Why is my regex removing spaces?

Question

$str = "& &svnips   Â ∴ ≈ osidnviosd & sopinsdo";   
$regex = "/&[^\w;]/";
echo preg_replace($regex, "&", $str);

I'm trying to replace all un-encoded ampersands with encoded ones.
The problem is it's removing the space between & and sopinsdo.

Any idea why?

Answer 1

You search for 2 Characters ("&" and a character that is NOT (; or \w))and replace it with &

You should replace with & (add a space to the end of replace string)

Answer 2

So you don't want the space between & and sopinsdo removed. Just add one

echo preg_replace($regex, "& ", $str);

Answer 3

Why use regex? Why not use htmlspecialchars()?

echo htmlspecialchars($str, ENT_NOQUOTES, 'UTF-8', false);

Note the fourth parameter. It tells it not to double encode anything. So basically, this will turn all < into <, all > into > and all & that are not part of an existing entity into &

But, if you must use regex, you could do:

$regex = '/&([^\w;])/';
echo preg_replace($regex, '&\1', $str);

Basically, it saves the non-word character and then adds it back...

Answer 4

This regex does what you're looking for.

preg_replace('/&(?!\w+;)/', '&', $text);

So for a few simple test cases you can get properly escaped HTML:

'& sopinsdo'          -> '& sopinsdo'
'& sopinsdo'      -> '& sopinsdo'
'sopinsdo & foo; bar' -> 'sopinsdo & foo; bar'
'sopinsdo &foo bar'   -> 'sopinsdo &foo bar'