| 1 | /* |
| 2 | ** $Id: lutf8lib.c $ |
| 3 | ** Standard library for UTF-8 manipulation |
| 4 | ** See Copyright Notice in lua.h |
| 5 | */ |
| 6 | |
| 7 | #define lutf8lib_c |
| 8 | #define LUA_LIB |
| 9 | |
| 10 | #include "lprefix.h" |
| 11 | |
| 12 | |
| 13 | #include <assert.h> |
| 14 | #include <limits.h> |
| 15 | #include <stdlib.h> |
| 16 | #include <string.h> |
| 17 | |
| 18 | #include "lua.h" |
| 19 | |
| 20 | #include "lauxlib.h" |
| 21 | #include "lualib.h" |
| 22 | |
| 23 | |
| 24 | #define MAXUNICODE 0x10FFFFu |
| 25 | |
| 26 | #define MAXUTF 0x7FFFFFFFu |
| 27 | |
| 28 | /* |
| 29 | ** Integer type for decoded UTF-8 values; MAXUTF needs 31 bits. |
| 30 | */ |
| 31 | #if (UINT_MAX >> 30) >= 1 |
| 32 | typedef unsigned int utfint; |
| 33 | #else |
| 34 | typedef unsigned long utfint; |
| 35 | #endif |
| 36 | |
| 37 | |
| 38 | #define iscont(p) ((*(p) & 0xC0) == 0x80) |
| 39 | |
| 40 | |
| 41 | /* from strlib */ |
| 42 | /* translate a relative string position: negative means back from end */ |
| 43 | static lua_Integer u_posrelat (lua_Integer pos, size_t len) { |
| 44 | if (pos >= 0) return pos; |
| 45 | else if (0u - (size_t)pos > len) return 0; |
| 46 | else return (lua_Integer)len + pos + 1; |
| 47 | } |
| 48 | |
| 49 | |
| 50 | /* |
| 51 | ** Decode one UTF-8 sequence, returning NULL if byte sequence is |
| 52 | ** invalid. The array 'limits' stores the minimum value for each |
| 53 | ** sequence length, to check for overlong representations. Its first |
| 54 | ** entry forces an error for non-ascii bytes with no continuation |
| 55 | ** bytes (count == 0). |
| 56 | */ |
| 57 | static const char *utf8_decode (const char *s, utfint *val, int strict) { |
| 58 | static const utfint limits[] = |
| 59 | {~(utfint)0, 0x80, 0x800, 0x10000u, 0x200000u, 0x4000000u}; |
| 60 | unsigned int c = (unsigned char)s[0]; |
| 61 | utfint res = 0; /* final result */ |
| 62 | if (c < 0x80) /* ascii? */ |
| 63 | res = c; |
| 64 | else { |
| 65 | int count = 0; /* to count number of continuation bytes */ |
| 66 | for (; c & 0x40; c <<= 1) { /* while it needs continuation bytes... */ |
| 67 | unsigned int cc = (unsigned char)s[++count]; /* read next byte */ |
| 68 | if ((cc & 0xC0) != 0x80) /* not a continuation byte? */ |
| 69 | return NULL; /* invalid byte sequence */ |
| 70 | res = (res << 6) | (cc & 0x3F); /* add lower 6 bits from cont. byte */ |
| 71 | } |
| 72 | res |= ((utfint)(c & 0x7F) << (count * 5)); /* add first byte */ |
| 73 | if (count > 5 || res > MAXUTF || res < limits[count]) |
| 74 | return NULL; /* invalid byte sequence */ |
| 75 | s += count; /* skip continuation bytes read */ |
| 76 | } |
| 77 | if (strict) { |
| 78 | /* check for invalid code points; too large or surrogates */ |
| 79 | if (res > MAXUNICODE || (0xD800u <= res && res <= 0xDFFFu)) |
| 80 | return NULL; |
| 81 | } |
| 82 | if (val) *val = res; |
| 83 | return s + 1; /* +1 to include first byte */ |
| 84 | } |
| 85 | |
| 86 | |
| 87 | /* |
| 88 | ** utf8len(s [, i [, j [, lax]]]) --> number of characters that |
| 89 | ** start in the range [i,j], or nil + current position if 's' is not |
| 90 | ** well formed in that interval |
| 91 | */ |
| 92 | static int utflen (lua_State *L) { |
| 93 | lua_Integer n = 0; /* counter for the number of characters */ |
| 94 | size_t len; /* string length in bytes */ |
| 95 | const char *s = luaL_checklstring(L, 1, &len); |
| 96 | lua_Integer posi = u_posrelat(luaL_optinteger(L, 2, 1), len); |
| 97 | lua_Integer posj = u_posrelat(luaL_optinteger(L, 3, -1), len); |
| 98 | int lax = lua_toboolean(L, 4); |
| 99 | luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 2, |
| 100 | "initial position out of bounds" ); |
| 101 | luaL_argcheck(L, --posj < (lua_Integer)len, 3, |
| 102 | "final position out of bounds" ); |
| 103 | while (posi <= posj) { |
| 104 | const char *s1 = utf8_decode(s + posi, NULL, !lax); |
| 105 | if (s1 == NULL) { /* conversion error? */ |
| 106 | luaL_pushfail(L); /* return fail ... */ |
| 107 | lua_pushinteger(L, posi + 1); /* ... and current position */ |
| 108 | return 2; |
| 109 | } |
| 110 | posi = s1 - s; |
| 111 | n++; |
| 112 | } |
| 113 | lua_pushinteger(L, n); |
| 114 | return 1; |
| 115 | } |
| 116 | |
| 117 | |
| 118 | /* |
| 119 | ** codepoint(s, [i, [j [, lax]]]) -> returns codepoints for all |
| 120 | ** characters that start in the range [i,j] |
| 121 | */ |
| 122 | static int codepoint (lua_State *L) { |
| 123 | size_t len; |
| 124 | const char *s = luaL_checklstring(L, 1, &len); |
| 125 | lua_Integer posi = u_posrelat(luaL_optinteger(L, 2, 1), len); |
| 126 | lua_Integer pose = u_posrelat(luaL_optinteger(L, 3, posi), len); |
| 127 | int lax = lua_toboolean(L, 4); |
| 128 | int n; |
| 129 | const char *se; |
| 130 | luaL_argcheck(L, posi >= 1, 2, "out of bounds" ); |
| 131 | luaL_argcheck(L, pose <= (lua_Integer)len, 3, "out of bounds" ); |
| 132 | if (posi > pose) return 0; /* empty interval; return no values */ |
| 133 | if (pose - posi >= INT_MAX) /* (lua_Integer -> int) overflow? */ |
| 134 | return luaL_error(L, "string slice too long" ); |
| 135 | n = (int)(pose - posi) + 1; /* upper bound for number of returns */ |
| 136 | luaL_checkstack(L, n, "string slice too long" ); |
| 137 | n = 0; /* count the number of returns */ |
| 138 | se = s + pose; /* string end */ |
| 139 | for (s += posi - 1; s < se;) { |
| 140 | utfint code; |
| 141 | s = utf8_decode(s, &code, !lax); |
| 142 | if (s == NULL) |
| 143 | return luaL_error(L, "invalid UTF-8 code" ); |
| 144 | lua_pushinteger(L, code); |
| 145 | n++; |
| 146 | } |
| 147 | return n; |
| 148 | } |
| 149 | |
| 150 | |
| 151 | static void pushutfchar (lua_State *L, int arg) { |
| 152 | lua_Unsigned code = (lua_Unsigned)luaL_checkinteger(L, arg); |
| 153 | luaL_argcheck(L, code <= MAXUTF, arg, "value out of range" ); |
| 154 | lua_pushfstring(L, "%U" , (long)code); |
| 155 | } |
| 156 | |
| 157 | |
| 158 | /* |
| 159 | ** utfchar(n1, n2, ...) -> char(n1)..char(n2)... |
| 160 | */ |
| 161 | static int utfchar (lua_State *L) { |
| 162 | int n = lua_gettop(L); /* number of arguments */ |
| 163 | if (n == 1) /* optimize common case of single char */ |
| 164 | pushutfchar(L, 1); |
| 165 | else { |
| 166 | int i; |
| 167 | luaL_Buffer b; |
| 168 | luaL_buffinit(L, &b); |
| 169 | for (i = 1; i <= n; i++) { |
| 170 | pushutfchar(L, i); |
| 171 | luaL_addvalue(&b); |
| 172 | } |
| 173 | luaL_pushresult(&b); |
| 174 | } |
| 175 | return 1; |
| 176 | } |
| 177 | |
| 178 | |
| 179 | /* |
| 180 | ** offset(s, n, [i]) -> index where n-th character counting from |
| 181 | ** position 'i' starts; 0 means character at 'i'. |
| 182 | */ |
| 183 | static int byteoffset (lua_State *L) { |
| 184 | size_t len; |
| 185 | const char *s = luaL_checklstring(L, 1, &len); |
| 186 | lua_Integer n = luaL_checkinteger(L, 2); |
| 187 | lua_Integer posi = (n >= 0) ? 1 : len + 1; |
| 188 | posi = u_posrelat(luaL_optinteger(L, 3, posi), len); |
| 189 | luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 3, |
| 190 | "position out of bounds" ); |
| 191 | if (n == 0) { |
| 192 | /* find beginning of current byte sequence */ |
| 193 | while (posi > 0 && iscont(s + posi)) posi--; |
| 194 | } |
| 195 | else { |
| 196 | if (iscont(s + posi)) |
| 197 | return luaL_error(L, "initial position is a continuation byte" ); |
| 198 | if (n < 0) { |
| 199 | while (n < 0 && posi > 0) { /* move back */ |
| 200 | do { /* find beginning of previous character */ |
| 201 | posi--; |
| 202 | } while (posi > 0 && iscont(s + posi)); |
| 203 | n++; |
| 204 | } |
| 205 | } |
| 206 | else { |
| 207 | n--; /* do not move for 1st character */ |
| 208 | while (n > 0 && posi < (lua_Integer)len) { |
| 209 | do { /* find beginning of next character */ |
| 210 | posi++; |
| 211 | } while (iscont(s + posi)); /* (cannot pass final '\0') */ |
| 212 | n--; |
| 213 | } |
| 214 | } |
| 215 | } |
| 216 | if (n == 0) /* did it find given character? */ |
| 217 | lua_pushinteger(L, posi + 1); |
| 218 | else /* no such character */ |
| 219 | luaL_pushfail(L); |
| 220 | return 1; |
| 221 | } |
| 222 | |
| 223 | |
| 224 | static int iter_aux (lua_State *L, int strict) { |
| 225 | size_t len; |
| 226 | const char *s = luaL_checklstring(L, 1, &len); |
| 227 | lua_Integer n = lua_tointeger(L, 2) - 1; |
| 228 | if (n < 0) /* first iteration? */ |
| 229 | n = 0; /* start from here */ |
| 230 | else if (n < (lua_Integer)len) { |
| 231 | n++; /* skip current byte */ |
| 232 | while (iscont(s + n)) n++; /* and its continuations */ |
| 233 | } |
| 234 | if (n >= (lua_Integer)len) |
| 235 | return 0; /* no more codepoints */ |
| 236 | else { |
| 237 | utfint code; |
| 238 | const char *next = utf8_decode(s + n, &code, strict); |
| 239 | if (next == NULL) |
| 240 | return luaL_error(L, "invalid UTF-8 code" ); |
| 241 | lua_pushinteger(L, n + 1); |
| 242 | lua_pushinteger(L, code); |
| 243 | return 2; |
| 244 | } |
| 245 | } |
| 246 | |
| 247 | |
| 248 | static int iter_auxstrict (lua_State *L) { |
| 249 | return iter_aux(L, 1); |
| 250 | } |
| 251 | |
| 252 | static int iter_auxlax (lua_State *L) { |
| 253 | return iter_aux(L, 0); |
| 254 | } |
| 255 | |
| 256 | |
| 257 | static int iter_codes (lua_State *L) { |
| 258 | int lax = lua_toboolean(L, 2); |
| 259 | luaL_checkstring(L, 1); |
| 260 | lua_pushcfunction(L, lax ? iter_auxlax : iter_auxstrict); |
| 261 | lua_pushvalue(L, 1); |
| 262 | lua_pushinteger(L, 0); |
| 263 | return 3; |
| 264 | } |
| 265 | |
| 266 | |
| 267 | /* pattern to match a single UTF-8 character */ |
| 268 | #define UTF8PATT "[\0-\x7F\xC2-\xFD][\x80-\xBF]*" |
| 269 | |
| 270 | |
| 271 | static const luaL_Reg funcs[] = { |
| 272 | {"offset" , byteoffset}, |
| 273 | {"codepoint" , codepoint}, |
| 274 | {"char" , utfchar}, |
| 275 | {"len" , utflen}, |
| 276 | {"codes" , iter_codes}, |
| 277 | /* placeholders */ |
| 278 | {"charpattern" , NULL}, |
| 279 | {NULL, NULL} |
| 280 | }; |
| 281 | |
| 282 | |
| 283 | LUAMOD_API int luaopen_utf8 (lua_State *L) { |
| 284 | luaL_newlib(L, funcs); |
| 285 | lua_pushlstring(L, UTF8PATT, sizeof(UTF8PATT)/sizeof(char) - 1); |
| 286 | lua_setfield(L, -2, "charpattern" ); |
| 287 | return 1; |
| 288 | } |
| 289 | |
| 290 | |