diff options
Diffstat (limited to 'script/encoder')
| -rw-r--r-- | script/encoder/utf16.lua | 12 |
1 files changed, 10 insertions, 2 deletions
diff --git a/script/encoder/utf16.lua b/script/encoder/utf16.lua index 9e71de68..744da174 100644 --- a/script/encoder/utf16.lua +++ b/script/encoder/utf16.lua @@ -86,9 +86,17 @@ local function utf8next(s, n) return n+1, utf8byte(s, n) elseif strmatch(s, "^[\xC2-\xDF][\x80-\xBF]", n) then return n+2, utf8byte(s, n) - elseif strmatch(s, "^[\xE0-\xEF][\x80-\xBF][\x80-\xBF]", n) then + elseif strmatch(s, "^[\xE0][\xA0-\xBF][\x80-\xBF]", n) then return n+3, utf8byte(s, n) - elseif strmatch(s, "^[\xF0-\xF4][\x80-\xBF][\x80-\xBF][\x80-\xBF]", n) then + elseif strmatch(s, "^[\xE1-\xEC][\x80-\xBF][\x80-\xBF]", n) then + return n+3, utf8byte(s, n) + elseif strmatch(s, "^[\xED][\x80-\x9F][\x80-\xBF]", n) then + return n+3, utf8byte(s, n) + elseif strmatch(s, "^[\xF0][\x90-\xBF][\x80-\xBF][\x80-\xBF]", n) then + return n+4, utf8byte(s, n) + elseif strmatch(s, "^[\xF1-\xF3][\x80-\xBF][\x80-\xBF][\x80-\xBF]", n) then + return n+4, utf8byte(s, n) + elseif strmatch(s, "^[\xF4][\x80-\x8F][\x80-\xBF][\x80-\xBF]", n) then return n+4, utf8byte(s, n) else return n+1 --invaild |