complex RegExp - help needed

[saloon12yrd]

Hi there,

while I consider myself of good knowledge concerning RegExp, I'm in need of a RegExp that matches any content between two matching tokens, where the content in between can contain those tokens as well.

An easy example would be matching
[FONT=courier new]<td>[content]</td>[/FONT]
with nested tables (and thus with <td></td> as well) in mind.

Another would be
[FONT=courier new]if (...) {[content] }[/FONT]
again, with nested if - Statements in mind.

Detailed example:
[FONT=courier new]
if () {

&nbsp;&nbsp;echo 'something';

&nbsp;&nbsp;if () {
&nbsp;&nbsp;&nbsp;&nbsp;echo 'nested statement, matched as well';
&nbsp;&nbsp;} else {
&nbsp;&nbsp;&nbsp;&nbsp;echo 'else branch';
&nbsp;&nbsp;}

&nbsp;&nbsp;echo 'another thing';
&nbsp;&nbsp;// ...
}

if (...) {
&nbsp;&nbsp;echo 'content continues outside the match';
}
[/FONT]
where the RegExp matches everything highlighted in red.

The logical approach would seem to match all content until there are more closing tokens than opening tokens within the match and then stop.

I'm pretty sure this is doable with RegExp (BackReferences seem a good approach) but so far I haven't figured out how, so any help would be greatly appreciated.

Thanks in advance,
Sal-

[crimaniak]

Yes, you can use recursive patterns in preg_match (?R) but in general this is not a task for regexp. You need full parser to do this task good.

[saloon12yrd]

No no, there's no need to recurse anything.

What I need is to start at a precisely defined point in the source and match anything from that point on to the responding end.

There's no need to recurse all nested if-Statements (to use that example). All that is needed is to count all nested occurances to find the responding end.

I already have a parser for this class of problems that uses the tokenizer functionalities of PHP (~40 lines of code, linear runtime) but that approach simply is too slow (~0.1sec to return the content of all function() {} of a 50kb PHP class definition file).

Regards,
Sal-

[saloon12yrd]

@:

Damn I misunderstood your posting. Of course recursion is the answer I was looking for.

It performs around 500 times better than the parser that uses the tokenizer. It is even 15 times faster than PHP needs to read in the file.

Thanks for the hint!
Sal-