php - Multi-byte strings and look-around weird bug

Question

Why following code behaves differently for different multi-bye strings?

echo preg_replace('@(?=\pL)@u', '*', 'م');  // prints: '*م'     ✓ 
echo preg_replace('@(?=\pL)@u', '*', 'ض');  // prints: '*ض'     ✓ 
echo preg_replace('@(?=\pL)@u', '*', 'غ');  // prints: '*�*�'   ✗ 
echo preg_replace('@(?=\pL)@u', '*', 'ص');  // prints: '*�*�'   ✗ 

See: http://3v4l.org/fvab1

Answer 1

您还需要包含修饰符字母 ( Lm)。请参阅以下遍历整个阿拉伯 unicode 块的脚本：

<?php
function uchar_2($dec)
{
    $utf = chr(192 + (($dec - ($dec % 64)) / 64)); 
    $utf .= chr(128 + ($dec % 64)); 


    return $utf;
}

$issues = 0;
$count = 0;
for ($dec = 1536; $dec <= 1791; $dec++) {
    $char = uchar_2($dec);
    if (preg_replace('@^(?=\pLm)$@u', '*', $char) !== $char) {
        printf("Issue with %s (%s)\n", $dec, $char);
        $issues++;
    }
    $count++;
}

printf("Found %d issues in %d rows\n", $issues, $count);

如果没有out Lm，大约一半的字符会失败。