| 1 | /* |
| 2 | ** $Id: lutf8lib.c,v 1.16.1.1 2017/04/19 17:29:57 roberto Exp $ |
| 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 | #define MAXUNICODE 0x10FFFF |
| 24 | |
| 25 | #define iscont(p) ((*(p) & 0xC0) == 0x80) |
| 26 | |
| 27 | |
| 28 | /* from strlib */ |
| 29 | /* translate a relative string position: negative means back from end */ |
| 30 | static lua_Integer u_posrelat (lua_Integer pos, size_t len) { |
| 31 | if (pos >= 0) return pos; |
| 32 | else if (0u - (size_t)pos > len) return 0; |
| 33 | else return (lua_Integer)len + pos + 1; |
| 34 | } |
| 35 | |
| 36 | |
| 37 | /* |
| 38 | ** Decode one UTF-8 sequence, returning NULL if byte sequence is invalid. |
| 39 | */ |
| 40 | static const char *utf8_decode (const char *o, int *val) { |
| 41 | static const unsigned int limits[] = {0xFF, 0x7F, 0x7FF, 0xFFFF}; |
| 42 | const unsigned char *s = (const unsigned char *)o; |
| 43 | unsigned int c = s[0]; |
| 44 | unsigned int res = 0; /* final result */ |
| 45 | if (c < 0x80) /* ascii? */ |
| 46 | res = c; |
| 47 | else { |
| 48 | int count = 0; /* to count number of continuation bytes */ |
| 49 | while (c & 0x40) { /* still have continuation bytes? */ |
| 50 | int cc = s[++count]; /* read next byte */ |
| 51 | if ((cc & 0xC0) != 0x80) /* not a continuation byte? */ |
| 52 | return NULL; /* invalid byte sequence */ |
| 53 | res = (res << 6) | (cc & 0x3F); /* add lower 6 bits from cont. byte */ |
| 54 | c <<= 1; /* to test next bit */ |
| 55 | } |
| 56 | res |= ((c & 0x7F) << (count * 5)); /* add first byte */ |
| 57 | if (count > 3 || res > MAXUNICODE || res <= limits[count]) |
| 58 | return NULL; /* invalid byte sequence */ |
| 59 | s += count; /* skip continuation bytes read */ |
| 60 | } |
| 61 | if (val) *val = res; |
| 62 | return (const char *)s + 1; /* +1 to include first byte */ |
| 63 | } |
| 64 | |
| 65 | |
| 66 | /* |
| 67 | ** utf8len(s [, i [, j]]) --> number of characters that start in the |
| 68 | ** range [i,j], or nil + current position if 's' is not well formed in |
| 69 | ** that interval |
| 70 | */ |
| 71 | static int utflen (lua_State *L) { |
| 72 | int n = 0; |
| 73 | size_t len; |
| 74 | const char *s = luaL_checklstring(L, 1, &len); |
| 75 | lua_Integer posi = u_posrelat(luaL_optinteger(L, 2, 1), len); |
| 76 | lua_Integer posj = u_posrelat(luaL_optinteger(L, 3, -1), len); |
| 77 | luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 2, |
| 78 | "initial position out of string" ); |
| 79 | luaL_argcheck(L, --posj < (lua_Integer)len, 3, |
| 80 | "final position out of string" ); |
| 81 | while (posi <= posj) { |
| 82 | const char *s1 = utf8_decode(s + posi, NULL); |
| 83 | if (s1 == NULL) { /* conversion error? */ |
| 84 | lua_pushnil(L); /* return nil ... */ |
| 85 | lua_pushinteger(L, posi + 1); /* ... and current position */ |
| 86 | return 2; |
| 87 | } |
| 88 | posi = s1 - s; |
| 89 | n++; |
| 90 | } |
| 91 | lua_pushinteger(L, n); |
| 92 | return 1; |
| 93 | } |
| 94 | |
| 95 | |
| 96 | /* |
| 97 | ** codepoint(s, [i, [j]]) -> returns codepoints for all characters |
| 98 | ** that start in the range [i,j] |
| 99 | */ |
| 100 | static int codepoint (lua_State *L) { |
| 101 | size_t len; |
| 102 | const char *s = luaL_checklstring(L, 1, &len); |
| 103 | lua_Integer posi = u_posrelat(luaL_optinteger(L, 2, 1), len); |
| 104 | lua_Integer pose = u_posrelat(luaL_optinteger(L, 3, posi), len); |
| 105 | int n; |
| 106 | const char *se; |
| 107 | luaL_argcheck(L, posi >= 1, 2, "out of range" ); |
| 108 | luaL_argcheck(L, pose <= (lua_Integer)len, 3, "out of range" ); |
| 109 | if (posi > pose) return 0; /* empty interval; return no values */ |
| 110 | if (pose - posi >= INT_MAX) /* (lua_Integer -> int) overflow? */ |
| 111 | return luaL_error(L, "string slice too long" ); |
| 112 | n = (int)(pose - posi) + 1; |
| 113 | luaL_checkstack(L, n, "string slice too long" ); |
| 114 | n = 0; |
| 115 | se = s + pose; |
| 116 | for (s += posi - 1; s < se;) { |
| 117 | int code; |
| 118 | s = utf8_decode(s, &code); |
| 119 | if (s == NULL) |
| 120 | return luaL_error(L, "invalid UTF-8 code" ); |
| 121 | lua_pushinteger(L, code); |
| 122 | n++; |
| 123 | } |
| 124 | return n; |
| 125 | } |
| 126 | |
| 127 | |
| 128 | static void pushutfchar (lua_State *L, int arg) { |
| 129 | lua_Integer code = luaL_checkinteger(L, arg); |
| 130 | luaL_argcheck(L, 0 <= code && code <= MAXUNICODE, arg, "value out of range" ); |
| 131 | lua_pushfstring(L, "%U" , (long)code); |
| 132 | } |
| 133 | |
| 134 | |
| 135 | /* |
| 136 | ** utfchar(n1, n2, ...) -> char(n1)..char(n2)... |
| 137 | */ |
| 138 | static int utfchar (lua_State *L) { |
| 139 | int n = lua_gettop(L); /* number of arguments */ |
| 140 | if (n == 1) /* optimize common case of single char */ |
| 141 | pushutfchar(L, 1); |
| 142 | else { |
| 143 | int i; |
| 144 | luaL_Buffer b; |
| 145 | luaL_buffinit(L, &b); |
| 146 | for (i = 1; i <= n; i++) { |
| 147 | pushutfchar(L, i); |
| 148 | luaL_addvalue(&b); |
| 149 | } |
| 150 | luaL_pushresult(&b); |
| 151 | } |
| 152 | return 1; |
| 153 | } |
| 154 | |
| 155 | |
| 156 | /* |
| 157 | ** offset(s, n, [i]) -> index where n-th character counting from |
| 158 | ** position 'i' starts; 0 means character at 'i'. |
| 159 | */ |
| 160 | static int byteoffset (lua_State *L) { |
| 161 | size_t len; |
| 162 | const char *s = luaL_checklstring(L, 1, &len); |
| 163 | lua_Integer n = luaL_checkinteger(L, 2); |
| 164 | lua_Integer posi = (n >= 0) ? 1 : len + 1; |
| 165 | posi = u_posrelat(luaL_optinteger(L, 3, posi), len); |
| 166 | luaL_argcheck(L, 1 <= posi && --posi <= (lua_Integer)len, 3, |
| 167 | "position out of range" ); |
| 168 | if (n == 0) { |
| 169 | /* find beginning of current byte sequence */ |
| 170 | while (posi > 0 && iscont(s + posi)) posi--; |
| 171 | } |
| 172 | else { |
| 173 | if (iscont(s + posi)) |
| 174 | return luaL_error(L, "initial position is a continuation byte" ); |
| 175 | if (n < 0) { |
| 176 | while (n < 0 && posi > 0) { /* move back */ |
| 177 | do { /* find beginning of previous character */ |
| 178 | posi--; |
| 179 | } while (posi > 0 && iscont(s + posi)); |
| 180 | n++; |
| 181 | } |
| 182 | } |
| 183 | else { |
| 184 | n--; /* do not move for 1st character */ |
| 185 | while (n > 0 && posi < (lua_Integer)len) { |
| 186 | do { /* find beginning of next character */ |
| 187 | posi++; |
| 188 | } while (iscont(s + posi)); /* (cannot pass final '\0') */ |
| 189 | n--; |
| 190 | } |
| 191 | } |
| 192 | } |
| 193 | if (n == 0) /* did it find given character? */ |
| 194 | lua_pushinteger(L, posi + 1); |
| 195 | else /* no such character */ |
| 196 | lua_pushnil(L); |
| 197 | return 1; |
| 198 | } |
| 199 | |
| 200 | |
| 201 | static int iter_aux (lua_State *L) { |
| 202 | size_t len; |
| 203 | const char *s = luaL_checklstring(L, 1, &len); |
| 204 | lua_Integer n = lua_tointeger(L, 2) - 1; |
| 205 | if (n < 0) /* first iteration? */ |
| 206 | n = 0; /* start from here */ |
| 207 | else if (n < (lua_Integer)len) { |
| 208 | n++; /* skip current byte */ |
| 209 | while (iscont(s + n)) n++; /* and its continuations */ |
| 210 | } |
| 211 | if (n >= (lua_Integer)len) |
| 212 | return 0; /* no more codepoints */ |
| 213 | else { |
| 214 | int code; |
| 215 | const char *next = utf8_decode(s + n, &code); |
| 216 | if (next == NULL || iscont(next)) |
| 217 | return luaL_error(L, "invalid UTF-8 code" ); |
| 218 | lua_pushinteger(L, n + 1); |
| 219 | lua_pushinteger(L, code); |
| 220 | return 2; |
| 221 | } |
| 222 | } |
| 223 | |
| 224 | |
| 225 | static int iter_codes (lua_State *L) { |
| 226 | luaL_checkstring(L, 1); |
| 227 | lua_pushcfunction(L, iter_aux); |
| 228 | lua_pushvalue(L, 1); |
| 229 | lua_pushinteger(L, 0); |
| 230 | return 3; |
| 231 | } |
| 232 | |
| 233 | |
| 234 | /* pattern to match a single UTF-8 character */ |
| 235 | #define UTF8PATT "[\0-\x7F\xC2-\xF4][\x80-\xBF]*" |
| 236 | |
| 237 | |
| 238 | static const luaL_Reg funcs[] = { |
| 239 | {"offset" , byteoffset}, |
| 240 | {"codepoint" , codepoint}, |
| 241 | {"char" , utfchar}, |
| 242 | {"len" , utflen}, |
| 243 | {"codes" , iter_codes}, |
| 244 | /* placeholders */ |
| 245 | {"charpattern" , NULL}, |
| 246 | {NULL, NULL} |
| 247 | }; |
| 248 | |
| 249 | |
| 250 | LUAMOD_API int luaopen_utf8 (lua_State *L) { |
| 251 | luaL_newlib(L, funcs); |
| 252 | lua_pushlstring(L, UTF8PATT, sizeof(UTF8PATT)/sizeof(char) - 1); |
| 253 | lua_setfield(L, -2, "charpattern" ); |
| 254 | return 1; |
| 255 | } |
| 256 | |
| 257 | |