Question

I have this simple code to count punctuation in a string. ie “there’s 2 commas, 3 semi-colons…” etc. But when it sees an em-dash (—) it doesn’t work. Note that it is not a hyphen (-), I don’t care about those.

Is there something special about em-dash that makes it weird in a PHP string and/or as an array key? Maybe a weird unicode problem?

$punc_counts = array(
    "," => 0,
    ";" => 0,
    "—" => 0, //exists, really!
    "'" => 0,
    "\"" => 0,
    "(" => 0,
    ")" => 0,
);

// $str is a long string of text

//remove all non-punctuation chars from $str (works correctly, keeping em-dashes)
$puncs = "";
foreach($punc_counts as $key => $value)
    $puncs .= $key;
$str = preg_replace("/[^{$puncs}]/", "", $str);

//$str now equals something like:
//$str == ",;'—\"—()();;,";

foreach(str_split($str) as $char)
{       
    //if it's a puncutation char we care about, count it
    if(isset($punc_counts[$char]))
        $punc_counts[$char]++;
    else
        print($char);
}

print("<br/>");
print_r($punc_counts);
print("<br/>");

The code above prints:

——
Array ( [,] => 2 [;] => 3 [—] => 0 ['] => 1 ["] => 1 [(] => 2 [)] => 2 ) 

Answer 1

It’s probably not multibyte compatible. There is a useful comment on the PHP doc page for str_split that suggests the following:

function str_split_unicode($str, $l = 0) {
    if ($l > 0) {
        $ret = array();
        $len = mb_strlen($str, "UTF-8");
        for ($i = 0; $i < $len; $i += $l) {
            $ret[] = mb_substr($str, $i, $l, "UTF-8");
        }
        return $ret;
    }
    return preg_split("//u", $str, -1, PREG_SPLIT_NO_EMPTY);
}