[Jaey]

I wrote a script in perl that effectively censors out certain words. I was hoping to translate this script into php, but even after reading the docs at php.net I got some errors. So, I was hoping someone could translate it for me (and/or simplify it if possible).

Code:
#!/usr/bin/perl -w
$strings=<>;
@tstrings=split(" ", $strings);
for($i=0; $i<=$#tstrings; $i++)
{
if($tstrings[$i] eq "for" || $tstrings[$i] eq "bar")
{
$tstrings[$i]="*****";
}
}
$nstrings=join(" ", @tstrings);
print "$nstrings\n";

-thanks

[Simon Mundy]

[php]

$censored = "foo|bar|foobar|americanidolsingers";

$text = preg_replace("!(?b)(".$censored.")(?b)!msi","* ***",$text);

[/php]

[Jaey]

That doesn't seem to... work. It doesn't replace anything...

EDIT: After reading the docs on preg_replace I got it fixed and working.. thanks for the help

[Jaey]

Actually, could you explain the !(?:b) stuff? I had to trash that to get it to work, but I'm not sure what it does. Thanks

[Simon Mundy]

My bad, I should have tested it first (but I thought I'd take the Microsoft attitude of getting everyone else to beta test for me )

The '!' were simply meant to be delimeters for the GREP pattern - I could have used any old character there.

The ?b is meant to match a sub-pattern without actually including it in the captured result. So for instance with the following:-

preg_replace('!((?:HELLO)THERE)!msi','\\1','HELLOT HERE')

the search string matches 'HELLOTHERE' but only 'THERE' is captured and used as the replacement subpattern....

Actually, come to think of it, I didn't even need it in the example I gave up top. Ooops - sorry! The reason I was using it was to use the '\b' (or word boundary character) as a means of checking that the full word was matched, but not to include that boundary when it was replaced. What I SHOULD have written was:-

preg_replace('!?<=\b(".$censored.")?=\b!msi","**** ",$text);

...which uses assertions. It means that it should find a word boundary BEFORE ( ?<\b ) then the string you want to find, then another word boundary AFTER ( ?=\b ) but not to capture those word boundaries in the match.

The grep manual is full of very cool stuff - I just learned how to use the recursion modifier to match balanced tags within an RTF file, but it's taken me a week or so to fully get my head around it.

I can see why the true web geeks go for PERL, as it really does some powerful stuff - it's just that my poor brain is better suited to the less cryptic PHP for day-to-day stuff.

[Simon Mundy]

[Double-post]