voku /
portable-utf8
| @@ 3508-3556 (lines=49) @@ | ||
| 3505 | * |
|
| 3506 | * @return int|false false if is't not UTF-16, 1 for UTF-16LE, 2 for UTF-16BE. |
|
| 3507 | */ |
|
| 3508 | public static function is_utf16($str) |
|
| 3509 | { |
|
| 3510 | $str = self::remove_bom($str); |
|
| 3511 | ||
| 3512 | if (self::is_binary($str)) { |
|
| 3513 | self::checkForSupport(); |
|
| 3514 | ||
| 3515 | $maybeUTF16LE = 0; |
|
| 3516 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 3517 | if ($test) { |
|
| 3518 | $test2 = \mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 3519 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 3520 | if ($test3 === $test) { |
|
| 3521 | $strChars = self::count_chars($str, true); |
|
| 3522 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 3523 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3524 | $maybeUTF16LE++; |
|
| 3525 | } |
|
| 3526 | } |
|
| 3527 | } |
|
| 3528 | } |
|
| 3529 | ||
| 3530 | $maybeUTF16BE = 0; |
|
| 3531 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 3532 | if ($test) { |
|
| 3533 | $test2 = \mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 3534 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 3535 | if ($test3 === $test) { |
|
| 3536 | $strChars = self::count_chars($str, true); |
|
| 3537 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 3538 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3539 | $maybeUTF16BE++; |
|
| 3540 | } |
|
| 3541 | } |
|
| 3542 | } |
|
| 3543 | } |
|
| 3544 | ||
| 3545 | if ($maybeUTF16BE !== $maybeUTF16LE) { |
|
| 3546 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 3547 | return 1; |
|
| 3548 | } else { |
|
| 3549 | return 2; |
|
| 3550 | } |
|
| 3551 | } |
|
| 3552 | ||
| 3553 | } |
|
| 3554 | ||
| 3555 | return false; |
|
| 3556 | } |
|
| 3557 | ||
| 3558 | /** |
|
| 3559 | * Check if the string is UTF-32. |
|
| @@ 3565-3613 (lines=49) @@ | ||
| 3562 | * |
|
| 3563 | * @return int|false false if is't not UTF-16, 1 for UTF-32LE, 2 for UTF-32BE. |
|
| 3564 | */ |
|
| 3565 | public static function is_utf32($str) |
|
| 3566 | { |
|
| 3567 | $str = self::remove_bom($str); |
|
| 3568 | ||
| 3569 | if (self::is_binary($str)) { |
|
| 3570 | self::checkForSupport(); |
|
| 3571 | ||
| 3572 | $maybeUTF32LE = 0; |
|
| 3573 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 3574 | if ($test) { |
|
| 3575 | $test2 = \mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 3576 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 3577 | if ($test3 === $test) { |
|
| 3578 | $strChars = self::count_chars($str, true); |
|
| 3579 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 3580 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3581 | $maybeUTF32LE++; |
|
| 3582 | } |
|
| 3583 | } |
|
| 3584 | } |
|
| 3585 | } |
|
| 3586 | ||
| 3587 | $maybeUTF32BE = 0; |
|
| 3588 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 3589 | if ($test) { |
|
| 3590 | $test2 = \mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 3591 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 3592 | if ($test3 === $test) { |
|
| 3593 | $strChars = self::count_chars($str, true); |
|
| 3594 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 3595 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3596 | $maybeUTF32BE++; |
|
| 3597 | } |
|
| 3598 | } |
|
| 3599 | } |
|
| 3600 | } |
|
| 3601 | ||
| 3602 | if ($maybeUTF32BE !== $maybeUTF32LE) { |
|
| 3603 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 3604 | return 1; |
|
| 3605 | } else { |
|
| 3606 | return 2; |
|
| 3607 | } |
|
| 3608 | } |
|
| 3609 | ||
| 3610 | } |
|
| 3611 | ||
| 3612 | return false; |
|
| 3613 | } |
|
| 3614 | ||
| 3615 | /** |
|
| 3616 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
