voku /
portable-utf8
| @@ 2864-2910 (lines=47) @@ | ||
| 2861 | * |
|
| 2862 | * @return int|false false if is't not UTF16, 1 for UTF-16LE, 2 for UTF-16BE. |
|
| 2863 | */ |
|
| 2864 | public static function is_utf16($str) |
|
| 2865 | { |
|
| 2866 | if (self::is_binary($str)) { |
|
| 2867 | self::checkForSupport(); |
|
| 2868 | ||
| 2869 | $maybeUTF16LE = 0; |
|
| 2870 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-16LE'); |
|
| 2871 | if ($test !== false && strlen($test) > 1) { |
|
| 2872 | $test2 = mb_convert_encoding($test, 'UTF-16LE', 'UTF-8'); |
|
| 2873 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-16LE'); |
|
| 2874 | if ($test3 == $test) { |
|
| 2875 | $strChars = self::count_chars($str); |
|
| 2876 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 2877 | if (in_array($test3char, $strChars, true) === true) { |
|
| 2878 | $maybeUTF16LE++; |
|
| 2879 | } |
|
| 2880 | } |
|
| 2881 | } |
|
| 2882 | } |
|
| 2883 | ||
| 2884 | $maybeUTF16BE = 0; |
|
| 2885 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-16BE'); |
|
| 2886 | if ($test !== false && strlen($test) > 1) { |
|
| 2887 | $test2 = mb_convert_encoding($test, 'UTF-16BE', 'UTF-8'); |
|
| 2888 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-16BE'); |
|
| 2889 | if ($test3 == $test) { |
|
| 2890 | $strChars = self::count_chars($str); |
|
| 2891 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 2892 | if (in_array($test3char, $strChars, true) === true) { |
|
| 2893 | $maybeUTF16BE++; |
|
| 2894 | } |
|
| 2895 | } |
|
| 2896 | } |
|
| 2897 | } |
|
| 2898 | ||
| 2899 | if ($maybeUTF16BE != $maybeUTF16LE) { |
|
| 2900 | if ($maybeUTF16LE > $maybeUTF16BE) { |
|
| 2901 | return 1; |
|
| 2902 | } else { |
|
| 2903 | return 2; |
|
| 2904 | } |
|
| 2905 | } |
|
| 2906 | ||
| 2907 | } |
|
| 2908 | ||
| 2909 | return false; |
|
| 2910 | } |
|
| 2911 | ||
| 2912 | /** |
|
| 2913 | * Check if the string is UTF-32. |
|
| @@ 2919-2965 (lines=47) @@ | ||
| 2916 | * |
|
| 2917 | * @return int|false false if is't not UTF16, 1 for UTF-32LE, 2 for UTF-32BE. |
|
| 2918 | */ |
|
| 2919 | public static function is_utf32($str) |
|
| 2920 | { |
|
| 2921 | if (self::is_binary($str)) { |
|
| 2922 | self::checkForSupport(); |
|
| 2923 | ||
| 2924 | $maybeUTF32LE = 0; |
|
| 2925 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-32LE'); |
|
| 2926 | if ($test !== false && strlen($test) > 1) { |
|
| 2927 | $test2 = mb_convert_encoding($test, 'UTF-32LE', 'UTF-8'); |
|
| 2928 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-32LE'); |
|
| 2929 | if ($test3 == $test) { |
|
| 2930 | $strChars = self::count_chars($str); |
|
| 2931 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 2932 | if (in_array($test3char, $strChars, true) === true) { |
|
| 2933 | $maybeUTF32LE++; |
|
| 2934 | } |
|
| 2935 | } |
|
| 2936 | } |
|
| 2937 | } |
|
| 2938 | ||
| 2939 | $maybeUTF32BE = 0; |
|
| 2940 | $test = mb_convert_encoding($str, 'UTF-8', 'UTF-32BE'); |
|
| 2941 | if ($test !== false && strlen($test) > 1) { |
|
| 2942 | $test2 = mb_convert_encoding($test, 'UTF-32BE', 'UTF-8'); |
|
| 2943 | $test3 = mb_convert_encoding($test2, 'UTF-8', 'UTF-32BE'); |
|
| 2944 | if ($test3 == $test) { |
|
| 2945 | $strChars = self::count_chars($str); |
|
| 2946 | foreach (self::count_chars($test3) as $test3char => $test3charEmpty) { |
|
| 2947 | if (in_array($test3char, $strChars, true) === true) { |
|
| 2948 | $maybeUTF32BE++; |
|
| 2949 | } |
|
| 2950 | } |
|
| 2951 | } |
|
| 2952 | } |
|
| 2953 | ||
| 2954 | if ($maybeUTF32BE != $maybeUTF32LE) { |
|
| 2955 | if ($maybeUTF32LE > $maybeUTF32BE) { |
|
| 2956 | return 1; |
|
| 2957 | } else { |
|
| 2958 | return 2; |
|
| 2959 | } |
|
| 2960 | } |
|
| 2961 | ||
| 2962 | } |
|
| 2963 | ||
| 2964 | return false; |
|
| 2965 | } |
|
| 2966 | ||
| 2967 | /** |
|
| 2968 | * Checks whether the passed string contains only byte sequences that appear valid UTF-8 characters. |
|
