[RESOLVED] help with regexp on a multibyte string

[itaym02]

I have the following string:

$text="א אב אבי אביהו מדינה שול של";

In which I wish to add '&#1488;&#1488;&#1488;' to all <4 chars word, so the string will turn into:
"&#1488;&#1488;&#1488;&#1488; &#1488;&#1489;&#1488;&#1488;&#1488; &#1488;&#1489;&#1497;&#1488;&#1488;&#1488; &#1488;&#1489;&#1497;&#1492;&#1493; &#1502;&#1491;&#1497;&#1504;&#1492; &#1513;&#1493;&#1500;&#1488;&#1488;&#1488; &#1513;&#1500;&#1488;&#1488;&#1488;"

The code I am using is:

	$text="א אב אבי אביהו מדינה שול של";
	$pattern='/\s(.{1,6})\s/';
	$text=preg_replace($pattern,' $1אאא ',$text);
	echo $text;

Which results in:
&#1488; &#1488;&#1489;&#1488;&#1488;&#1488; &#1488;&#1489;&#1497; &#1488;&#1489;&#1497;&#1492;&#1493; &#1502;&#1491;&#1497;&#1504;&#1492; &#1513;&#1493;&#1500;&#1488;&#1488;&#1488; &#1513;&#1500;

Problems:
1. It seems word boundry is not recognized (hence my use of \s).
2. Why was the &#1488;&#1489;&#1497; not replaced?

[nrg_alpha]

I'm not familiar with the Israel alphabet (I'm assuming that's israeli - if not, my apologies, as I'm not knowledgable in middle eastern writing). But here's how I would do it:

$text = 'To be simply a test. Go!
A little bit goes a long way!';

$pattern = '#\b[a-z]{1,3}\K(?=\s)#i';
$text = preg_replace($pattern, '***', $text);
echo $text // Output: To*** be*** simply a*** test. Go! A*** little bit*** goes a*** long way!

Granted, this makes assumptions that 'words' of 3 characters or less preceding a punctuation mark isn't altered.. so 'Go!' remains as is. If you desire these kind of words to be altered, simply change the look ahead assertion (?=\s) to (?=\b ). So with regards to the alphabet itself, you can simply replace [a-z] with the range that contains your alphabet / allowable characters instead.

[itaym02]

TNX

[Weedpacket]

You might not have said that you were using Hebrew; preg_match() uses whatever locale setting is in place at the time it's called. This can be changed with [man]setlocale[/man]. Try putting

setlocale(LC_ALL, 'he');

before it. Maybe then it will know the difference between "wordlike" and "nonwordlike" characters in Hebrew, and therefore find the boundaries between them. Unfortunately there seems to be some variation on different systems about how locales are specified - you might need to use "heb" or "he_IL" or something along those lines.