3 ** Standard library for UTF-8 manipulation
29 #define MSGInvalid	"invalid UTF-8 code"
32 ** Integer type for decoded UTF-8 values; MAXUTF needs 31 bits.
47 static lua_Integer u_posrelat (lua_Integer pos, size_t len) {  in u_posrelat()  argument
48   if (pos >= 0) return pos;  in u_posrelat()
49   else if (0u - (size_t)pos > len) return 0;  in u_posrelat()
50   else return (lua_Integer)len + pos + 1;  in u_posrelat()
55 ** Decode one UTF-8 sequence, returning NULL if byte sequence is
58 ** entry forces an error for non-ascii bytes with no continuation
92 ** utf8len(s [, i [, j [, lax]]]) --> number of characters that
101   lua_Integer posj = u_posrelat(luaL_optinteger(L, 3, -1), len);  in utflen()
103   luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 2,  in utflen()
105   luaL_argcheck(L, --posj < (lua_Integer)len, 3,  in utflen()
114     posi = s1 - s;  in utflen()
123 ** codepoint(s, [i, [j [, lax]]]) -> returns codepoints for all
133   const char *se;  in codepoint()  local
137   if (pose - posi >= INT_MAX)  /* (lua_Integer -> int) overflow? */  in codepoint()
139   n = (int)(pose -  posi) + 1;  /* upper bound for number of returns */  in codepoint()
142   se = s + pose;  /* string end */  in codepoint()
143   for (s += posi - 1; s < se;) {  in codepoint()
163 ** utfchar(n1, n2, ...)  -> char(n1)..char(n2)...
184 ** offset(s, n, [i])  -> index where n-th character counting from
193   luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 3,  in byteoffset()
197     while (posi > 0 && iscontp(s + posi)) posi--;  in byteoffset()
205            posi--;  in byteoffset()
211        n--;  /* do not move for 1st character */  in byteoffset()
216          n--;  in byteoffset()
269 /* pattern to match a single UTF-8 character */
270 #define UTF8PATT	"[\0-\x7F\xC2-\xFD][\x80-\xBF]*"
287   lua_pushlstring(L, UTF8PATT, sizeof(UTF8PATT)/sizeof(char) - 1);  in luaopen_utf8()
288   lua_setfield(L, -2, "charpattern");  in luaopen_utf8()