voku /
portable-utf8
| @@ 2264-2311 (lines=48) @@ | ||
| 2261 | * <strong>2</strong> for UTF-16BE. |
|
| 2262 | * </p> |
|
| 2263 | */ |
|
| 2264 | public static function is_utf16(string $str) |
|
| 2265 | { |
|
| 2266 | $str = self::remove_bom($str); |
|
| 2267 | ||
| 2268 | if (self::is_binary($str) === true) { |
|
| 2269 | ||
| 2270 | $maybeUTF16LE = 0; |
|
| 2271 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 2272 | if ($test) { |
|
| 2273 | $test2 = \mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 2274 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 2275 | if ($test3 === $test) { |
|
| 2276 | $strChars = self::count_chars($str, true); |
|
| 2277 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 2278 | if (\in_array($test3char, $strChars, true) === true) { |
|
| 2279 | $maybeUTF16LE++; |
|
| 2280 | } |
|
| 2281 | } |
|
| 2282 | } |
|
| 2283 | } |
|
| 2284 | ||
| 2285 | $maybeUTF16BE = 0; |
|
| 2286 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 2287 | if ($test) { |
|
| 2288 | $test2 = \mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 2289 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 2290 | if ($test3 === $test) { |
|
| 2291 | $strChars = self::count_chars($str, true); |
|
| 2292 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 2293 | if (\in_array($test3char, $strChars, true) === true) { |
|
| 2294 | $maybeUTF16BE++; |
|
| 2295 | } |
|
| 2296 | } |
|
| 2297 | } |
|
| 2298 | } |
|
| 2299 | ||
| 2300 | if ($maybeUTF16BE !== $maybeUTF16LE) { |
|
| 2301 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 2302 | return 1; |
|
| 2303 | } |
|
| 2304 | ||
| 2305 | return 2; |
|
| 2306 | } |
|
| 2307 | ||
| 2308 | } |
|
| 2309 | ||
| 2310 | return false; |
|
| 2311 | } |
|
| 2312 | ||
| 2313 | /** |
|
| 2314 | * Check if the string is UTF-32. |
|
| @@ 2324-2371 (lines=48) @@ | ||
| 2321 | * <strong>2</strong> for UTF-32BE. |
|
| 2322 | * </p> |
|
| 2323 | */ |
|
| 2324 | public static function is_utf32(string $str) |
|
| 2325 | { |
|
| 2326 | $str = self::remove_bom($str); |
|
| 2327 | ||
| 2328 | if (self::is_binary($str) === true) { |
|
| 2329 | ||
| 2330 | $maybeUTF32LE = 0; |
|
| 2331 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 2332 | if ($test) { |
|
| 2333 | $test2 = \mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 2334 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 2335 | if ($test3 === $test) { |
|
| 2336 | $strChars = self::count_chars($str, true); |
|
| 2337 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 2338 | if (\in_array($test3char, $strChars, true) === true) { |
|
| 2339 | $maybeUTF32LE++; |
|
| 2340 | } |
|
| 2341 | } |
|
| 2342 | } |
|
| 2343 | } |
|
| 2344 | ||
| 2345 | $maybeUTF32BE = 0; |
|
| 2346 | $test = \mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 2347 | if ($test) { |
|
| 2348 | $test2 = \mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 2349 | $test3 = \mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 2350 | if ($test3 === $test) { |
|
| 2351 | $strChars = self::count_chars($str, true); |
|
| 2352 | foreach (self::count_chars($test3, true) as $test3char => $test3charEmpty) { |
|
| 2353 | if (\in_array($test3char, $strChars, true) === true) { |
|
| 2354 | $maybeUTF32BE++; |
|
| 2355 | } |
|
| 2356 | } |
|
| 2357 | } |
|
| 2358 | } |
|
| 2359 | ||
| 2360 | if ($maybeUTF32BE !== $maybeUTF32LE) { |
|
| 2361 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 2362 | return 1; |
|
| 2363 | } |
|
| 2364 | ||
| 2365 | return 2; |
|
| 2366 | } |
|
| 2367 | ||
| 2368 | } |
|
| 2369 | ||
| 2370 | return false; |
|
| 2371 | } |
|
| 2372 | ||
| 2373 | /** |
|
| 2374 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
