voku /
portable-utf8
| @@ 3534-3580 (lines=47) @@ | ||
| 3531 | * |
|
| 3532 | * @return int|false false if is't not UTF16, 1 for UTF-16LE, 2 for UTF-16BE. |
|
| 3533 | */ |
|
| 3534 | public static function is_utf16($str) |
|
| 3535 | { |
|
| 3536 | if (self::is_binary($str)) { |
|
| 3537 | self::checkForSupport(); |
|
| 3538 | ||
| 3539 | $maybeUTF16LE = 0; |
|
| 3540 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 3541 | if ($test !== false && strlen($test) > 1) { |
|
| 3542 | $test2 = \mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 3543 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 3544 | if ($test3 === $test) { |
|
| 3545 | $strChars = self::count_chars($str); |
|
| 3546 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3547 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3548 | $maybeUTF16LE++; |
|
| 3549 | } |
|
| 3550 | } |
|
| 3551 | } |
|
| 3552 | } |
|
| 3553 | ||
| 3554 | $maybeUTF16BE = 0; |
|
| 3555 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 3556 | if ($test !== false && strlen($test) > 1) { |
|
| 3557 | $test2 = \mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 3558 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 3559 | if ($test3 === $test) { |
|
| 3560 | $strChars = self::count_chars($str); |
|
| 3561 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3562 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3563 | $maybeUTF16BE++; |
|
| 3564 | } |
|
| 3565 | } |
|
| 3566 | } |
|
| 3567 | } |
|
| 3568 | ||
| 3569 | if ($maybeUTF16BE !== $maybeUTF16LE) { |
|
| 3570 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 3571 | return 1; |
|
| 3572 | } else { |
|
| 3573 | return 2; |
|
| 3574 | } |
|
| 3575 | } |
|
| 3576 | ||
| 3577 | } |
|
| 3578 | ||
| 3579 | return false; |
|
| 3580 | } |
|
| 3581 | ||
| 3582 | /** |
|
| 3583 | * Check if the string is UTF-32. |
|
| @@ 3589-3635 (lines=47) @@ | ||
| 3586 | * |
|
| 3587 | * @return int|false false if is't not UTF16, 1 for UTF-32LE, 2 for UTF-32BE. |
|
| 3588 | */ |
|
| 3589 | public static function is_utf32($str) |
|
| 3590 | { |
|
| 3591 | if (self::is_binary($str)) { |
|
| 3592 | self::checkForSupport(); |
|
| 3593 | ||
| 3594 | $maybeUTF32LE = 0; |
|
| 3595 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 3596 | if ($test !== false && strlen($test) > 1) { |
|
| 3597 | $test2 = \mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 3598 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 3599 | if ($test3 === $test) { |
|
| 3600 | $strChars = self::count_chars($str); |
|
| 3601 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3602 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3603 | $maybeUTF32LE++; |
|
| 3604 | } |
|
| 3605 | } |
|
| 3606 | } |
|
| 3607 | } |
|
| 3608 | ||
| 3609 | $maybeUTF32BE = 0; |
|
| 3610 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 3611 | if ($test !== false && strlen($test) > 1) { |
|
| 3612 | $test2 = \mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 3613 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 3614 | if ($test3 === $test) { |
|
| 3615 | $strChars = self::count_chars($str); |
|
| 3616 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3617 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3618 | $maybeUTF32BE++; |
|
| 3619 | } |
|
| 3620 | } |
|
| 3621 | } |
|
| 3622 | } |
|
| 3623 | ||
| 3624 | if ($maybeUTF32BE !== $maybeUTF32LE) { |
|
| 3625 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 3626 | return 1; |
|
| 3627 | } else { |
|
| 3628 | return 2; |
|
| 3629 | } |
|
| 3630 | } |
|
| 3631 | ||
| 3632 | } |
|
| 3633 | ||
| 3634 | return false; |
|
| 3635 | } |
|
| 3636 | ||
| 3637 | /** |
|
| 3638 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
