Regular expressions blues

Barak

I'm not exactly knowledgable when it comes to regular expressions and that hasn't helped me this evening as I've got quite a complex one to write (or perhaps it isn't, but I know that little).

An example of what I'm trying to do is replace text smileys in a string, but not if they're between [ code ][ /code ] tags (just a couple of tags I made up for formatting ease, but without the spaces as vB uses them too).

First of all, is it possible to do this with a single regular expression or will I have to stick with the current, convoluted method of splitting the string up into parts and applying simpler replaces on the relevant parts?

Any help is greatly appreciated on this. I can do most things in PHP, but regex ain't one of 'em 😉

Rohan_Hill

Yeah, I'd say you have to split the string up into parts 🙁

Weedpacket

It's probably more hassle than it's worth to try doing it with a single preg_replace, especially since you'll probably be wanting to do this for a whole bunch of smilies.

In fact, once you break it down, regexps are probably excessive.

$bits = explode('[ /code'],$text);
foreach($bits as $k=>$bit)
{	list($before_code,$after_code) = explode('[ code]',$bit);
	str_replace(':-)','<img etc.>'$before_code);
	$bits[$k]=$before_code.'[ code]'.$after_code;
}
$text = join('[ /code]',$bits);

With judicious whitespace slipped in so vBulletin doesn't try and process the pseudotags.

Totally untested, totally OTTOMH. Possibly not the best solution. Back in an hour. Distributed on an "as is" basis, with all express and implied warranties and conditions disclaimed, including, without limitation, any implied warranties and conditions of merchantability, satisfactory quality, fitness for a particular purpose, and non-infringement.

The ':-)' and '<img etc.>' can be replaced by arrays of smilies and their replacements.

Barak

Yeah, that's pretty much what I'm doing at the moment, I was just posting in the hope of there being a nice regular expression to make it easier 😉

Thanks for your help 🙂

Weedpacket

As my disclaimer warned, there may be a better solution. And it does use a single preg_replace.

Consider the result of the following code.

$strings = array(
'word word word this word word',
'word word word {code}this{/code} word word',
'word this word this word word',
'word this word {code}this{/code} word word',
'word {code}this{/code} word this word word',
'word {code}this{/code} word {code}this{/code} word word',
'word word word {code}this{/code} word this word',
'word this word {code}this{/code} word this word',
'word {code}this{/code} word this word this word',
'word {code}this{/code} word {code}this{/code} word this word',
'word word word {code}word this word{/code} word word',
'word this word this word word',
'word this word {code}word this word{/code} word word',
'word {code}word this word{/code} word this word word',
'word {code}word this word{/code} word {code}word this word{/code} word word',
'word word word {code}word this word{/code} word this word',
'word this word {code}word this word{/code} word this word',
'word {code}word this word{/code} word this word this word',
'word {code}word this word{/code} word {code}word this word{/code} word this word',
);

foreach($strings as $string)
	echo preg_replace(
		'/(this)(?=((?!\{\/code\}).)*(\{code\}|$))/s',
		'that',
	$string)."<br>";

(Note that for the sake of rendering, [ and ] have been replaced by { and }).

Replace "this" and "that" with the smiley and its replacement. If you're going to use arrays of smileys, then the entire regexp will need to be stored in each element of the array, not just the smiley itself.

Barak

AHA!
Thanks a lot 🙂
I may not understand it just yet (too early in the morning for that), but it works, so I thank you 🙂

Weedpacket

code[/code]
Self-explanatory 🙂
code[/code]
Positive forward assertion - to match, the text must be followed by something matching the indicated pattern. (That something isn't considered part of the match, though. This is good, because we don't want it getting caught up in the replacement, nor do we want additional smilies in the same section of text to be missed.)
code[/code]
Forward negative assertion; matched text must not be followed by anything matching this. (In this case, a closing code pseudotag.) So....
code*[/code]
...would mean zero or more instances of a character that is not the start of a closing code pseudotag. (No, [^[] isn't enough, becuase that would match the start of any other pseudotag as well. No, swapping the assertion and the . won't work either - that first matches a character and then forbids the pattern - if the character you first matched started the pattern you're trying to forbid, you're hosed. I found this out the hard way...)

code[/code]
Followed by an opening code pseudotag, or the end of the string. (This is where we want to stop looking - we don't want to continue on and mangle bits inside code blocks.)

/s

makes \n qualify as a character when it comes to matching ".".

It would be theoretically possible to do this with ereg, but I for one am not about to go about trying it. Those of you who are still wedded to POSIX-style regexps are invited to write an equivalent that uses a call to ereg_replace() instead of one to preg_replace().... :rolleyes: