voku /
portable-utf8
| @@ 2260-2307 (lines=48) @@ | ||
| 2257 | * <strong>2</strong> for UTF-16BE. |
|
| 2258 | * </p> |
|
| 2259 | */ |
|
| 2260 | public static function is_utf16(string $str) |
|
| 2261 | { |
|
| 2262 | $str = self::remove_bom($str); |
|
| 2263 | ||
| 2264 | if (self::is_binary($str) === true) { |
|
| 2265 | ||
| 2266 | $maybeUTF16LE = 0; |
|
| 2267 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 2268 | if ($test) { |
|
| 2269 | $test2 = \mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 2270 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 2271 | if ($test3 === $test) { |
|
| 2272 | $strChars = self::count_chars($str, true); |
|
| 2273 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 2274 | if (\in_array($test3char, $strChars, true) === true) { |
|
| 2275 | $maybeUTF16LE++; |
|
| 2276 | } |
|
| 2277 | } |
|
| 2278 | } |
|
| 2279 | } |
|
| 2280 | ||
| 2281 | $maybeUTF16BE = 0; |
|
| 2282 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 2283 | if ($test) { |
|
| 2284 | $test2 = \mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 2285 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 2286 | if ($test3 === $test) { |
|
| 2287 | $strChars = self::count_chars($str, true); |
|
| 2288 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 2289 | if (\in_array($test3char, $strChars, true) === true) { |
|
| 2290 | $maybeUTF16BE++; |
|
| 2291 | } |
|
| 2292 | } |
|
| 2293 | } |
|
| 2294 | } |
|
| 2295 | ||
| 2296 | if ($maybeUTF16BE !== $maybeUTF16LE) { |
|
| 2297 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 2298 | return 1; |
|
| 2299 | } |
|
| 2300 | ||
| 2301 | return 2; |
|
| 2302 | } |
|
| 2303 | ||
| 2304 | } |
|
| 2305 | ||
| 2306 | return false; |
|
| 2307 | } |
|
| 2308 | ||
| 2309 | /** |
|
| 2310 | * Check if the string is UTF-32. |
|
| @@ 2320-2367 (lines=48) @@ | ||
| 2317 | * <strong>2</strong> for UTF-32BE. |
|
| 2318 | * </p> |
|
| 2319 | */ |
|
| 2320 | public static function is_utf32(string $str) |
|
| 2321 | { |
|
| 2322 | $str = self::remove_bom($str); |
|
| 2323 | ||
| 2324 | if (self::is_binary($str) === true) { |
|
| 2325 | ||
| 2326 | $maybeUTF32LE = 0; |
|
| 2327 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 2328 | if ($test) { |
|
| 2329 | $test2 = \mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 2330 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 2331 | if ($test3 === $test) { |
|
| 2332 | $strChars = self::count_chars($str, true); |
|
| 2333 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 2334 | if (\in_array($test3char, $strChars, true) === true) { |
|
| 2335 | $maybeUTF32LE++; |
|
| 2336 | } |
|
| 2337 | } |
|
| 2338 | } |
|
| 2339 | } |
|
| 2340 | ||
| 2341 | $maybeUTF32BE = 0; |
|
| 2342 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 2343 | if ($test) { |
|
| 2344 | $test2 = \mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 2345 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 2346 | if ($test3 === $test) { |
|
| 2347 | $strChars = self::count_chars($str, true); |
|
| 2348 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 2349 | if (\in_array($test3char, $strChars, true) === true) { |
|
| 2350 | $maybeUTF32BE++; |
|
| 2351 | } |
|
| 2352 | } |
|
| 2353 | } |
|
| 2354 | } |
|
| 2355 | ||
| 2356 | if ($maybeUTF32BE !== $maybeUTF32LE) { |
|
| 2357 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 2358 | return 1; |
|
| 2359 | } |
|
| 2360 | ||
| 2361 | return 2; |
|
| 2362 | } |
|
| 2363 | ||
| 2364 | } |
|
| 2365 | ||
| 2366 | return false; |
|
| 2367 | } |
|
| 2368 | ||
| 2369 | /** |
|
| 2370 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
