/**
* this function creates the appropriate regexp string of an token array
* you should not call this function directly, @see GeSHi::optimize_regexp_list().
*
* @param &$tokens array of tokens
* @param $recursed bool to know wether we recursed or not
* @return string
* @author Milian Wolff <*****@*****.**>
* @since 1.0.8
*/
function _optimize_regexp_list_tokens_to_string(&$tokens, $recursed = false)
{
$list = '';
foreach ($tokens as $token => $sub_tokens) {
$list .= $token;
$close_entry = isset($sub_tokens['']);
unset($sub_tokens['']);
if (!empty($sub_tokens)) {
$list .= '(?:' . GeSHi::_optimize_regexp_list_tokens_to_string($sub_tokens, true) . ')';
if ($close_entry) {
// make sub_tokens optional
$list .= '?';
}
}
$list .= '|';
}
if (!$recursed) {
// do some optimizations
// common trailing strings
// BUGGY!
//$list = preg_replace_callback('#(?<=^|\:|\|)\w+?(\w+)(?:\|.+\1)+(?=\|)#', create_function(
// '$matches', 'return "(?:" . preg_replace("#" . preg_quote($matches[1], "#") . "(?=\||$)#", "", $matches[0]) . ")" . $matches[1];'), $list);
// (?:p)? => p?
$list = preg_replace('#\\(\\?\\:(.)\\)\\?#', '\\1?', $list);
// (?:a|b|c|d|...)? => [abcd...]?
// TODO: a|bb|c => [ac]|bb
static $callback_2;
if (!isset($callback_2)) {
$callback_2 = create_function('$matches', 'return "[" . str_replace("|", "", $matches[1]) . "]";');
}
$list = preg_replace_callback('#\\(\\?\\:((?:.\\|)+.)\\)#', $callback_2, $list);
}
// return $list without trailing pipe
return substr($list, 0, -1);
}
