preg_match and nested BBCodes

[Jack_McSlay]

I'm trying to make a BBCode parser but i'm having a real hard time with nested bbcodes. I have this code:

<pre><?php
$text = 'start[quote]sub-start[quote]middle 1[/quote][quote]middle 2[/quote]sub-end[/quote]end';

function format_split ($text){
	$string = (string) $text;
	$out = array('');
	$key = 0;

$unique_long = 'b|i|u|s|sup|sub|left|right|center|justify|quote';

while (strpos ($text,'[')!==false){
  	$vtype = false;
	if (preg_match('@\\[('.$unique_long.')\](.*?)\[/\1\]@',$text,$matches)){
	  	$vtype = 3;
		$value = null;
		$cont = format_split ($matches[2]);
	}

	if ($vtype){
		$text = explode ($matches[0],$text,2);

		$out[$key] .= $text[0];
		$out[] = array (
			'tag' => $matches[1],
			'vtype' => $vtype,
			'value' => $value,
			'cont' => $cont
		);
		$out[] = '';
		$key += 2;
		$text = $text[1];
	}
	else {
		$text = explode ('[',$text,2);
		$out[$key] .= $text[0].'[';
		$text = $text[1];
	}
}
$out[$key] .= $text;

return $out;
}

var_dump (format_split ($text));

?></pre>

as you may realize, the expected result would be something like this:

array(3) {
  [0]=>
  string(5) "start"
  [1]=>
  array(4) {
    ["tag"]=>
    string(5) "quote"
    ["vtype"]=>
    int(3)
    ["value"]=>
    NULL
    ["cont"]=>
    array(5) {
      [0]=>
      string(9) "sub-start"
      [1]=>
      array(4) {
        ["tag"]=>
        string(1) "quote"
        ["vtype"]=>
        int(3)
        ["value"]=>
        NULL
        ["cont"]=>
        array(1) {
          [0]=>
          string(8) "middle 1"
        }
      }
      [2]=>
      string(0) ""
      [3]=>
      array(4) {
        ["tag"]=>
        string(1) "quote"
        ["vtype"]=>
        int(3)
        ["value"]=>
        NULL
        ["cont"]=>
        array(1) {
          [0]=>
          string(8) "middle 2"
        }
      }
      [4]=>
      string(7) "sub-end"
    }
  }
  [2]=>
  string(3) "end"
}

but if you test, you'll see the first quote tag ends where should be the end of the second quote tag that's inside it:

array(5) {
  [0]=>
  string(5) "start"
  [1]=>
  array(4) {
    ["tag"]=>
    string(5) "quote"
    ["vtype"]=>
    int(3)
    ["value"]=>
    NULL
    ["cont"]=>
    array(1) {
      [0]=>
      string(24) "sub-start[quote]middle 1"
    }
  }
  [2]=>
  string(0) ""
  [3]=>
  array(4) {
    ["tag"]=>
    string(5) "quote"
    ["vtype"]=>
    int(3)
    ["value"]=>
    NULL
    ["cont"]=>
    array(1) {
      [0]=>
      string(8) "middle 2"
    }
  }
  [4]=>
  string(18) "sub-end[/quote]end"
}

[Weedpacket]

Thing is of course that you need to match stuff after the opening "[bbtag]" only as long as "[/bbtag]" doesn't appear at all (except at the very end). Doing that repeatedly would deal with the innermost pair of [bbtag]...[/bbtag] pairs, then the next innermost, and so on.

!\[bbtag\]((?:(?!\[/bbtag\]).)*)\[/bbtag\]!

Ugly; but basically

!\[bbtag\] # The start tag
(
    (?:
        (?!\[/bbtag\])  # Don't start matching this here
        .                      # Any character
     )* # zero or more times
) # the contents of the tag
\[/bbtag\] # The end tag
!x

it says to match zero or more characters, as long as they aren't the start of the string "[/bbtag]". (I've also used the (?🙂 assertions so that the parentheses used don't count towards your $matches).

[Jack_McSlay]

no effect, but it seems to be having some progress with

@\[bbtag\]((?:(?!\[/bbtag\]).|.*\[bbtag\].*\[/bbtag\].*)*)\[/bbtag\]@

however, it only works properly when there's an even amount of children tags in it, whenever there's an odd amount, it will keep skipping tags for whatever reason. further ideas?

[Weedpacket]

Correcting for my typo (having both made one and omitted the standard "cut-and-paste code never works" disclaimer), and adding example use:

$string = "This has [bbtag] some bbtags[/bbtag] in it; [bbtag]some of them [bbtag]nested[/bbtag] for some reason[/bbtag] and some [bbtag]not[/bbtag].";


$regex="#\[bbtag\]((?:(?!\[/bbtag\]).)*)\[/bbtag\]#";

while(preg_match($regex, $string, $m))
{
	$string preg_replace($regex, "<foo>".$m[1]."</foo>" $string, 1);
}
echo $string;

Of course you could just replace [bbtag] with the appropriate opening text, and
[/bbtag] with the appropriate closing text; but this obviously will fail if the
tags are mismatched.