voku /
portable-utf8
| @@ 3398-3444 (lines=47) @@ | ||
| 3395 | * |
|
| 3396 | * @return int|false false if is't not UTF16, 1 for UTF-16LE, 2 for UTF-16BE. |
|
| 3397 | */ |
|
| 3398 | public static function is_utf16($str) |
|
| 3399 | { |
|
| 3400 | if (self::is_binary($str)) { |
|
| 3401 | self::checkForSupport(); |
|
| 3402 | ||
| 3403 | $maybeUTF16LE = 0; |
|
| 3404 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 3405 | if ($test !== false && strlen($test) > 1) { |
|
| 3406 | $test2 = \mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 3407 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 3408 | if ($test3 === $test) { |
|
| 3409 | $strChars = self::count_chars($str); |
|
| 3410 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3411 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3412 | $maybeUTF16LE++; |
|
| 3413 | } |
|
| 3414 | } |
|
| 3415 | } |
|
| 3416 | } |
|
| 3417 | ||
| 3418 | $maybeUTF16BE = 0; |
|
| 3419 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 3420 | if ($test !== false && strlen($test) > 1) { |
|
| 3421 | $test2 = \mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 3422 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 3423 | if ($test3 === $test) { |
|
| 3424 | $strChars = self::count_chars($str); |
|
| 3425 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3426 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3427 | $maybeUTF16BE++; |
|
| 3428 | } |
|
| 3429 | } |
|
| 3430 | } |
|
| 3431 | } |
|
| 3432 | ||
| 3433 | if ($maybeUTF16BE !== $maybeUTF16LE) { |
|
| 3434 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 3435 | return 1; |
|
| 3436 | } else { |
|
| 3437 | return 2; |
|
| 3438 | } |
|
| 3439 | } |
|
| 3440 | ||
| 3441 | } |
|
| 3442 | ||
| 3443 | return false; |
|
| 3444 | } |
|
| 3445 | ||
| 3446 | /** |
|
| 3447 | * Check if the string is UTF-32. |
|
| @@ 3453-3499 (lines=47) @@ | ||
| 3450 | * |
|
| 3451 | * @return int|false false if is't not UTF16, 1 for UTF-32LE, 2 for UTF-32BE. |
|
| 3452 | */ |
|
| 3453 | public static function is_utf32($str) |
|
| 3454 | { |
|
| 3455 | if (self::is_binary($str)) { |
|
| 3456 | self::checkForSupport(); |
|
| 3457 | ||
| 3458 | $maybeUTF32LE = 0; |
|
| 3459 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 3460 | if ($test !== false && strlen($test) > 1) { |
|
| 3461 | $test2 = \mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 3462 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 3463 | if ($test3 === $test) { |
|
| 3464 | $strChars = self::count_chars($str); |
|
| 3465 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3466 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3467 | $maybeUTF32LE++; |
|
| 3468 | } |
|
| 3469 | } |
|
| 3470 | } |
|
| 3471 | } |
|
| 3472 | ||
| 3473 | $maybeUTF32BE = 0; |
|
| 3474 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 3475 | if ($test !== false && strlen($test) > 1) { |
|
| 3476 | $test2 = \mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 3477 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 3478 | if ($test3 === $test) { |
|
| 3479 | $strChars = self::count_chars($str); |
|
| 3480 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3481 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3482 | $maybeUTF32BE++; |
|
| 3483 | } |
|
| 3484 | } |
|
| 3485 | } |
|
| 3486 | } |
|
| 3487 | ||
| 3488 | if ($maybeUTF32BE !== $maybeUTF32LE) { |
|
| 3489 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 3490 | return 1; |
|
| 3491 | } else { |
|
| 3492 | return 2; |
|
| 3493 | } |
|
| 3494 | } |
|
| 3495 | ||
| 3496 | } |
|
| 3497 | ||
| 3498 | return false; |
|
| 3499 | } |
|
| 3500 | ||
| 3501 | /** |
|
| 3502 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
