allowed characters in a string (stripping it)

[Lo'oris]

I'd like to have a set of "allowed characters", and strip a string
from everything besides those.

I've tried and tried but so far every time I enter strings containing
unicode, it goes mad and output makes no sense.

I'm sure I'm missing something but no idea what.

$ACCENTED_ALL_L OW="Ã*èìòùà ¡Ã©Ã*óúâêî ôûäëïöüà ¿Ã£ÃµÃ±Ã§Ã¦Ã¸Ã¥ ăÄ�Ä•Ä“Ä*Ä«Å�Å �Å*ūəß";
$ACCENTED_ALL_B IG="ÀÈÌÒÙà �ÉÃ�ÓÚÂÊÎ ÔÛÄËÃ�ÖÜŠ¸ÃƒÃ•Ã‘ÇÆØÅ ĂĀĔĒĬĪŎŠŒÅ¬ÅªÆ�ß";
$ACCENTED_ALL=$ ACCENTED_ALL_LO W.$ACCENTED_ALL _BIG;
$ALPHABET_LOW=" qwertyuiopasdfg hjklzxcvbnm";
$ALPHABET_BIG=" QWERTYUIOPASDFG HJKLZXCVBNM";
$ALPHABET_ALL=$ ALPHABET_LOW.$A LPHABET_BIG;
$SYMBOLS_NAME=" .'- ";

first time I tried using something like this:
$name=preg_repl ace("/([a-zA-Z]|-|[$al])|./",'$1',$nam e);
(bear in mind I do *NOT* know regexp, a friend wrote this line)

now I tried instead using str_split:
function clear_name_comp lex ($name, $ok_chars) {
$ass=str_split( $ok_chars);
$al=array();
foreach ($ass as $a) {
$al[$a]=TRUE;
}

$s=str_split($n ame);
$ret="";
foreach ($s as $c) {
if (!$al[$c]) continue;
$ret.=$c;
}

return $ret;
}

still nothing.
unicode, and it goes mad and output makes no sense.
I belive that's because in both cases it treats unicode characters
splitting into single bytes, but still, I'm clueless about what am I
supposed to do.

[Steve]

"Tim Roberts" <ti**@probo.com wrote in message
news:i2******** *************** *********@4ax.c om...

Jerry Stuckle <js*******@attg lobal.netwrote:

>>Tim Roberts wrote:

>>"Lo'oris" <lo****@gmail.c omwrote:

I'd like to have a set of "allowed characters", and strip a string
from everything besides those.
I've tried and tried but so far every time I enter strings containing
unicode, it goes mad and output makes no sense.

How are you entering "strings containing unicode"? Browsers don't send
Unicode.

Excuse me? They sure can, depending on the language being used.

Yes, I know better. That was not the sentiment I intended to convey.

>>So the rest of your post is immaterial. Steve's suggestion is a lot
closer.

Damn you, Stuckle. How can you see anything at all from up there on your
high horse?

with jerry, it's a matter of people in glass houses. except when you start
throwing rocks at his, he will claim you have no rock and that, in fact,
you've not broken any windows. :)

Despite my faux pas, my suggestion was also correct, your invective
notwithstanding .

good word, invective...he likes doing that apparently. at least we encounter
it often in his posts.

cheers.

[Steve]

"Steve" <no****@example .comwrote in message
news:HI******** *****@newsfe07. lga...

>
"Tim Roberts" <ti**@probo.com wrote in message
news:dp******** *************** *********@4ax.c om...

>"Steve" <no****@example .comwrote:

>>>
$name=preg_ replace("/([a-zA-Z]|-|[$al])|./",'$1',$nam e);

it's not expensive at all. and a dot is any single character...not a
greedy
wild card. the only reason he wouldn't want a dot is because it could be
an
'illegal' character that he's trying to get rid of anyway. as it is, he
just
didn't escape the dot so that it is the character (period) and not the
directive (any single character).

That statement as written will replace each character with itself, one by
one, repeatedly, for each character in $name.

It is an expensive no-op.

that is true, however each character is analyzed *as a single character*.
there is no marker being set and a pattern being sought beyond that marker
to see if there is another pattern match. markers are set, the replacement
is made to those characters marked, the process is done. one of the least
expensive operations one could ask of preg.

may be a good idea to write a pattern you think would be less expense that
does similar things...see if you can time-test compare the two. you can
also measure memory consumption too. i don't think you'll find any
significant consumption of resources running the above, esp.
comparitively.

sorry tim, i needed to make it clear - as i'd mentioned in one of my first
responses to you - that i think the dot in his preg is just mistakenly not
excaped. i don't think he means "any single character", rather, "a period".
anyway, my comments above are made under that assumption. otherwise you are
more right than before, however more in the line of "that's dumb to put
or'ed patterns when one of those will basically make the other
conditions/patterns moot". still, in this case, the expense is nominal since
all conditions/patterns work over a single character.

just thought i'd clarify.

cheers