voku /
portable-utf8
| @@ 3284-3330 (lines=47) @@ | ||
| 3281 | * |
|
| 3282 | * @return int|false false if is't not UTF16, 1 for UTF-16LE, 2 for UTF-16BE. |
|
| 3283 | */ |
|
| 3284 | public static function is_utf16($str) |
|
| 3285 | { |
|
| 3286 | if (self::is_binary($str)) { |
|
| 3287 | self::checkForSupport(); |
|
| 3288 | ||
| 3289 | $maybeUTF16LE = 0; |
|
| 3290 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 3291 | if ($test !== false && strlen($test) > 1) { |
|
| 3292 | $test2 = mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 3293 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 3294 | if ($test3 == $test) { |
|
| 3295 | $strChars = self::count_chars($str); |
|
| 3296 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3297 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3298 | $maybeUTF16LE++; |
|
| 3299 | } |
|
| 3300 | } |
|
| 3301 | } |
|
| 3302 | } |
|
| 3303 | ||
| 3304 | $maybeUTF16BE = 0; |
|
| 3305 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 3306 | if ($test !== false && strlen($test) > 1) { |
|
| 3307 | $test2 = mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 3308 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 3309 | if ($test3 == $test) { |
|
| 3310 | $strChars = self::count_chars($str); |
|
| 3311 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3312 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3313 | $maybeUTF16BE++; |
|
| 3314 | } |
|
| 3315 | } |
|
| 3316 | } |
|
| 3317 | } |
|
| 3318 | ||
| 3319 | if ($maybeUTF16BE != $maybeUTF16LE) { |
|
| 3320 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 3321 | return 1; |
|
| 3322 | } else { |
|
| 3323 | return 2; |
|
| 3324 | } |
|
| 3325 | } |
|
| 3326 | ||
| 3327 | } |
|
| 3328 | ||
| 3329 | return false; |
|
| 3330 | } |
|
| 3331 | ||
| 3332 | /** |
|
| 3333 | * Check if the string is UTF-32. |
|
| @@ 3339-3385 (lines=47) @@ | ||
| 3336 | * |
|
| 3337 | * @return int|false false if is't not UTF16, 1 for UTF-32LE, 2 for UTF-32BE. |
|
| 3338 | */ |
|
| 3339 | public static function is_utf32($str) |
|
| 3340 | { |
|
| 3341 | if (self::is_binary($str)) { |
|
| 3342 | self::checkForSupport(); |
|
| 3343 | ||
| 3344 | $maybeUTF32LE = 0; |
|
| 3345 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 3346 | if ($test !== false && strlen($test) > 1) { |
|
| 3347 | $test2 = mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 3348 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 3349 | if ($test3 == $test) { |
|
| 3350 | $strChars = self::count_chars($str); |
|
| 3351 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3352 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3353 | $maybeUTF32LE++; |
|
| 3354 | } |
|
| 3355 | } |
|
| 3356 | } |
|
| 3357 | } |
|
| 3358 | ||
| 3359 | $maybeUTF32BE = 0; |
|
| 3360 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 3361 | if ($test !== false && strlen($test) > 1) { |
|
| 3362 | $test2 = mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 3363 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 3364 | if ($test3 == $test) { |
|
| 3365 | $strChars = self::count_chars($str); |
|
| 3366 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 3367 | if (in_array($test3char, $strChars, true) === true) { |
|
| 3368 | $maybeUTF32BE++; |
|
| 3369 | } |
|
| 3370 | } |
|
| 3371 | } |
|
| 3372 | } |
|
| 3373 | ||
| 3374 | if ($maybeUTF32BE != $maybeUTF32LE) { |
|
| 3375 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 3376 | return 1; |
|
| 3377 | } else { |
|
| 3378 | return 2; |
|
| 3379 | } |
|
| 3380 | } |
|
| 3381 | ||
| 3382 | } |
|
| 3383 | ||
| 3384 | return false; |
|
| 3385 | } |
|
| 3386 | ||
| 3387 | /** |
|
| 3388 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
