voku /
portable-utf8
| @@ 2832-2878 (lines=47) @@ | ||
| 2829 | * |
|
| 2830 | * @return int|false false if is't not UTF16, 1 for UTF-16LE, 2 for UTF-16BE. |
|
| 2831 | */ |
|
| 2832 | public static function is_utf16($str) |
|
| 2833 | { |
|
| 2834 | if (self::is_binary($str)) { |
|
| 2835 | self::checkForSupport(); |
|
| 2836 | ||
| 2837 | $maybeUTF16LE = 0; |
|
| 2838 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 2839 | if ($test !== false && strlen($test) > 1) { |
|
| 2840 | $test2 = mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 2841 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 2842 | if ($test3 == $test) { |
|
| 2843 | $strChars = self::count_chars($str); |
|
| 2844 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 2845 | if (in_array($test3char, $strChars, true) === true) { |
|
| 2846 | $maybeUTF16LE++; |
|
| 2847 | } |
|
| 2848 | } |
|
| 2849 | } |
|
| 2850 | } |
|
| 2851 | ||
| 2852 | $maybeUTF16BE = 0; |
|
| 2853 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 2854 | if ($test !== false && strlen($test) > 1) { |
|
| 2855 | $test2 = mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 2856 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 2857 | if ($test3 == $test) { |
|
| 2858 | $strChars = self::count_chars($str); |
|
| 2859 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 2860 | if (in_array($test3char, $strChars, true) === true) { |
|
| 2861 | $maybeUTF16BE++; |
|
| 2862 | } |
|
| 2863 | } |
|
| 2864 | } |
|
| 2865 | } |
|
| 2866 | ||
| 2867 | if ($maybeUTF16BE != $maybeUTF16LE) { |
|
| 2868 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 2869 | return 1; |
|
| 2870 | } else { |
|
| 2871 | return 2; |
|
| 2872 | } |
|
| 2873 | } |
|
| 2874 | ||
| 2875 | } |
|
| 2876 | ||
| 2877 | return false; |
|
| 2878 | } |
|
| 2879 | ||
| 2880 | /** |
|
| 2881 | * Check if the string is UTF-32. |
|
| @@ 2887-2933 (lines=47) @@ | ||
| 2884 | * |
|
| 2885 | * @return int|false false if is't not UTF16, 1 for UTF-32LE, 2 for UTF-32BE. |
|
| 2886 | */ |
|
| 2887 | public static function is_utf32($str) |
|
| 2888 | { |
|
| 2889 | if (self::is_binary($str)) { |
|
| 2890 | self::checkForSupport(); |
|
| 2891 | ||
| 2892 | $maybeUTF32LE = 0; |
|
| 2893 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 2894 | if ($test !== false && strlen($test) > 1) { |
|
| 2895 | $test2 = mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 2896 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 2897 | if ($test3 == $test) { |
|
| 2898 | $strChars = self::count_chars($str); |
|
| 2899 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 2900 | if (in_array($test3char, $strChars, true) === true) { |
|
| 2901 | $maybeUTF32LE++; |
|
| 2902 | } |
|
| 2903 | } |
|
| 2904 | } |
|
| 2905 | } |
|
| 2906 | ||
| 2907 | $maybeUTF32BE = 0; |
|
| 2908 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 2909 | if ($test !== false && strlen($test) > 1) { |
|
| 2910 | $test2 = mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 2911 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 2912 | if ($test3 == $test) { |
|
| 2913 | $strChars = self::count_chars($str); |
|
| 2914 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 2915 | if (in_array($test3char, $strChars, true) === true) { |
|
| 2916 | $maybeUTF32BE++; |
|
| 2917 | } |
|
| 2918 | } |
|
| 2919 | } |
|
| 2920 | } |
|
| 2921 | ||
| 2922 | if ($maybeUTF32BE != $maybeUTF32LE) { |
|
| 2923 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 2924 | return 1; |
|
| 2925 | } else { |
|
| 2926 | return 2; |
|
| 2927 | } |
|
| 2928 | } |
|
| 2929 | ||
| 2930 | } |
|
| 2931 | ||
| 2932 | return false; |
|
| 2933 | } |
|
| 2934 | ||
| 2935 | /** |
|
| 2936 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
