Replacing UTF-8 characters in PHP 5.3

Question

Why doesn't this test case work?

<?php
// cards with cyrillic inidices and suits in UTF-8 encoding
$a = array('7♠', 'Д♠', 'К♠', '8♦', 'В♦', 'Д♦', '10♣', '10♥', 'В♥', 'Т♥');
foreach ($a as $card) {
        $suit = substr($card, -1);

        $card = preg_replace('/(\d+)♥/', '<span class="red">$1&hearts;</span>', $card);
        $card = preg_replace('/(\d+)♦/', '<span class="red">$1&diams;</span>', $card);
        $card = preg_replace('/(\d+)♠/', '<span class="black">$1&spades;</span>', $card);
        $card = preg_replace('/(\d+)♣/', '<span class="black">$1&clubs;</span>', $card);

        printf("suit: %s, html: %s\n", $suit, $card);
}
?>

Output:

suit: ▒, html: <span class="black">7&spades;</span>
suit: ▒, html: Д♠
suit: ▒, html: К♠
suit: ▒, html: <span class="red">8&diams;</span>
suit: ▒, html: В♦
suit: ▒, html: Д♦
suit: ▒, html: <span class="black">10&clubs;</span>
suit: ▒, html: <span class="red">10&hearts;</span>
suit: ▒, html: В♥
suit: ▒, html: Т♥

Ie I'm struggling with 2 problems in my PHP-script:

1. Why isn't the last UTF-8 character extracted correctly?
2. Why only first suit is being replaced by preg_replace ?

Using PHP 5.3.3, PostgreSQL 8.4.12 holding UTF-8 JSON (with Russian text and card suits) on CentOS 6.2.

If 1. is a bug in PHP 5.3.3, then is there a good workaround? (I don't want to upgrade the stock package).

UPDATE:

<?php
$a = array('7♠', 'Д♠', 'К♠', '8♦', 'В♦', 'Д♦', '10♣', '10♥', 'В♥', 'Т♥');
foreach ($a as $card) {
        $suit = mb_substr($card, -1, 1, 'UTF-8');

        $card = preg_replace('/(\d+)♥/u', '<span class="red">$1&hearts;</span>', $card);
        $card = preg_replace('/(\d+)♦/u', '<span class="red">$1&diams;</span>', $card);
        $card = preg_replace('/(\d+)♠/u', '<span class="black">$1&spades;</span>', $card);
        $card = preg_replace('/(\d+)♣/u', '<span class="black">$1&clubs;</span>', $card);

        printf("suit: %s, html: %s\n", $suit, $card);
}
?>

The new output:

suit: ♠, html: <span class="black">7&spades;</span>
suit: ♠, html: Д♠
suit: ♠, html: К♠
suit: ♦, html: <span class="red">8&diams;</span>
suit: ♦, html: В♦
suit: ♦, html: Д♦
suit: ♣, html: <span class="black">10&clubs;</span>
suit: ♥, html: <span class="red">10&hearts;</span>
suit: ♥, html: В♥

Answer 1

substr is one of the naïve PHP core functions which assumes 1 byte = 1 character. substr(..., -1) extracts the last byte from the string. "♠" is longer than one byte though. You should use mb_substr($card, -1, 1, 'UTF-8') instead.

You need to add the u (PCRE_UTF8) modifier to the regular expression to make it deal with UTF-8 encoded expressions and strings correctly:

preg_replace('/(\d+)♥/u', ...