Cant find problem

[PeteSpain]

Hi, I have the following bad word filter function but for some reason it is classing the word "test" as a bad word. In my database table I only have one word that is similar to this and that is "testa". My site is in 8 languages so often a bad word is similar to a god word in a different language but as this is not exactly the same, I can't work out why it is classing it as bad.

Any suggestions would be great.

Peter

<?php
function ReplaceBadWords( $str )
{
    $words = getbadwords();
    foreach ( $words as $badword => $goodword )
    if ( preg_match( $expr = '~(^|[^a-z0-9]?)(' . $badword . '?)($|[^a-z0-9])~iU', $str ) )
        $str = preg_replace( $expr, "\\1{$goodword}\\3", $str );

return $str;
}

function getbadwords()
{
    $sql = "select word from word_filter";
    $words = runsql( $sql );

foreach( $words AS $w ) {
    $bwords[$w["word"]] = '******';
}
return $bwords;
}

?>

And function runsql( $sql ) is

function runsql( $sql )
{
    $res = mysql_query( $sql ) or die( mysql_error() );

$out = array();

if ( mysql_num_rows( $res ) > 0 ) {
    while ( $a = mysql_fetch_assoc( $res ) ) {
        $out[] = $a;
    }
}

return $out;
}

[bradgrafelman]

In your regexp, you've got a '?' (question mark) after you insert $badword into the regexp; this means that the previous entity is optional. The "previous entity" in this case happens to be the last character in the word. Thus that section of your regexp looks like:
code[/code]which will match 'testa' or 'test' (since the 'a' at the end appears 0 or 1 times in the word 'test').

[PeteSpain]

Thanks bradgrafelman,

I removed the "?" from ...(' . $badword . '?)... and that looks like it's ok.

If this is correct I can mark this as resolved.

Thanks