Ggtcf

I am trying to improve legacy PHP code that deals with cleaning-up BBCode from a string and am currently facing trouble with lists.

The current solution for lists does the following:

...
$search[] = sprintf('~[%s](.*)[/%s]~smUi', 'list', 'list');
$search[] = sprintf('~[%s=(.*)](.*)[/%s]~smUi', 'list', 'list');
$search[] = sprintf('~[%s]~i', '*');
$replace[] = '$1';
$replace[] = '$2';
$replace[] = '';
...
return preg_replace($search, $replace, $string);

This works fine when the string is something like

[list]
 [*]Item 1
 [*]Item 2
 [*]Item 3
[/list]

But it will also strip [*] if it's not inside lists and also fails with things like:

[list]
 [*][list]
 [*]Item 1.1
 [*]Item 1.2
 [*]Item 1.3
 [/list]
 [*]Item 2
 [*]Item 3
[/list]

Is it even possible using RegExp only to strip the [list] or [list=1] + [/list] tags along with [*] if they are within lists?

You may use

$search[] = sprintf('~[(%s)(?:=[^]]*)?]((?:(?![1b).)*?)[/1]s*~si', 'list');
$search[] = sprintf('~[%s]~i', '\*');
$replace[] = '$2';
$replace[] = '';
$count = 0;
do 
 $string = preg_replace($search, $replace, $string, -1, $count);

while ($count > 0);
return $string;

See the PHP demo.

I merged the first two regexps since they basically match the same (the =.*? part inside the open tag is just optional, I suggest using (?:=[^]]*)? to match = and then 0+ chars other than ] 1 or 0 times.

The ((?:(?![1b).)*?) pattern is a tempered greedy token that ensures that the innermost list tag is matched, 1 here matches the tag name that is captured with (%s).

The $count variable will hold the number of replacements done with preg_replace, and if nothing is replaced, the while block exits.