Question

I have a string which is build up like this:

[propertyname]=[value]

Both the propertyname and the value can be encapsulated by either single or double quotes.

So i could receive a string which looks like this:

"height"='max'

or:

'height'='max'

As long as both the propertyname and value are encapsulated by the same type of quotes.

What i need to do is remove the quotes. But only around the propertyname and value! Because the following could very well be a valid string too:

"blaat"="Some 'random' blaat"

The end result should be:

blaat=Some 'random' blaat

I have the following regex which works. But it only works when i’m either checking for double quotes or single quotes. When i try to combine them with the | operator, then it doesn’t work anymore.

<?php
$string = '"height"=\'something "else" in here\'';

//echo preg_replace ( '#"(.*?)"#', '$1', $string );
//echo preg_replace ( '#\'(.*?)\'#', '$1', $string );
echo preg_replace ( '#("(.*?)"|\'(.*?)\')#', '$1', $string );
?>

So i could simply do two preg_replace calls, but that’s a nasty work around considering regex should be able to handle this in one call…

Any idea what the problem is?

Answer 1

Your regex should be matching correctly, but you have a problem: In your “combined regex”, $1 refers to the entire match (because the first set of parentheses encloses the entire match), so you’re replacing the match with itself, including the quotes.

("(.*?)"|\'(.*?)\')
^ ^        ^
|-+--- $1  |---- $3
  |--- $2

Now, you could simply drop the outer parentheses:

"(.*?)"|\'(.*?)\'
 ^        ^
 |--- $1  |---- $2

But then you have a different problem: You either need to replace the match with $1 or $2, depending on which half of the regex did match. Since you can’t know that in advance, that won’t be easy. You could possibly try and replace with $1$2, but I don’t know whether PHP would allow a backreference to a group that didn’t participate in the match.

Better play it safe and use a regex that can handle both cases at once, including escaped quotes within the quoted strings:

$result = preg_replace(
    '/(        # Match and capture (group 1):
     ["\']     # an opening quote character
    )          # (End of group 1).
    (          # Now match and capture (group 2):
     (?:       #  Either...
      \\\\.    #   an escaped character
     |         #  or...
      (?!\1)   #   (as long as it is not the closing quote)
      .        #   any other character.
     )*        #  Repeat as needed.
    )          # (End of group 2)
    \1         # Now match the closing quote./x', 
    '\2', $subject);