| Line | Branch | Decision | Exec | Source |
|---|---|---|---|---|
| 1 | /* | |||
| 2 | ** $Id: lstrlib.c $ | |||
| 3 | ** Standard library for string operations and pattern-matching | |||
| 4 | ** See Copyright Notice in lua.h | |||
| 5 | */ | |||
| 6 | ||||
| 7 | #define lstrlib_c | |||
| 8 | #define LUA_LIB | |||
| 9 | ||||
| 10 | #include "lprefix.h" | |||
| 11 | ||||
| 12 | ||||
| 13 | #include <ctype.h> | |||
| 14 | #include <float.h> | |||
| 15 | #include <limits.h> | |||
| 16 | #include <locale.h> | |||
| 17 | #include <math.h> | |||
| 18 | #include <stddef.h> | |||
| 19 | #include <stdio.h> | |||
| 20 | #include <stdlib.h> | |||
| 21 | #include <string.h> | |||
| 22 | ||||
| 23 | #include "lua.h" | |||
| 24 | ||||
| 25 | #include "lauxlib.h" | |||
| 26 | #include "lualib.h" | |||
| 27 | ||||
| 28 | ||||
| 29 | /* | |||
| 30 | ** maximum number of captures that a pattern can do during | |||
| 31 | ** pattern-matching. This limit is arbitrary, but must fit in | |||
| 32 | ** an unsigned char. | |||
| 33 | */ | |||
| 34 | #if !defined(LUA_MAXCAPTURES) | |||
| 35 | #define LUA_MAXCAPTURES 32 | |||
| 36 | #endif | |||
| 37 | ||||
| 38 | ||||
| 39 | /* macro to 'unsign' a character */ | |||
| 40 | #define uchar(c) ((unsigned char)(c)) | |||
| 41 | ||||
| 42 | ||||
| 43 | /* | |||
| 44 | ** Some sizes are better limited to fit in 'int', but must also fit in | |||
| 45 | ** 'size_t'. (We assume that 'lua_Integer' cannot be smaller than 'int'.) | |||
| 46 | */ | |||
| 47 | #define MAX_SIZET ((size_t)(~(size_t)0)) | |||
| 48 | ||||
| 49 | #define MAXSIZE \ | |||
| 50 | (sizeof(size_t) < sizeof(int) ? MAX_SIZET : (size_t)(INT_MAX)) | |||
| 51 | ||||
| 52 | ||||
| 53 | ||||
| 54 | ||||
| 55 | ✗ | static int str_len (lua_State *L) { | ||
| 56 | size_t l; | |||
| 57 | ✗ | luaL_checklstring(L, 1, &l); | ||
| 58 | ✗ | lua_pushinteger(L, (lua_Integer)l); | ||
| 59 | ✗ | return 1; | ||
| 60 | } | |||
| 61 | ||||
| 62 | ||||
| 63 | /* | |||
| 64 | ** translate a relative initial string position | |||
| 65 | ** (negative means back from end): clip result to [1, inf). | |||
| 66 | ** The length of any string in Lua must fit in a lua_Integer, | |||
| 67 | ** so there are no overflows in the casts. | |||
| 68 | ** The inverted comparison avoids a possible overflow | |||
| 69 | ** computing '-pos'. | |||
| 70 | */ | |||
| 71 | ✗ | static size_t posrelatI (lua_Integer pos, size_t len) { | ||
| 72 | ✗ | if (pos > 0) | ||
| 73 | ✗ | return (size_t)pos; | ||
| 74 | ✗ | else if (pos == 0) | ||
| 75 | ✗ | return 1; | ||
| 76 | ✗ | else if (pos < -(lua_Integer)len) /* inverted comparison */ | ||
| 77 | ✗ | return 1; /* clip to 1 */ | ||
| 78 | ✗ | else return len + (size_t)pos + 1; | ||
| 79 | } | |||
| 80 | ||||
| 81 | ||||
| 82 | /* | |||
| 83 | ** Gets an optional ending string position from argument 'arg', | |||
| 84 | ** with default value 'def'. | |||
| 85 | ** Negative means back from end: clip result to [0, len] | |||
| 86 | */ | |||
| 87 | ✗ | static size_t getendpos (lua_State *L, int arg, lua_Integer def, | ||
| 88 | size_t len) { | |||
| 89 | ✗ | lua_Integer pos = luaL_optinteger(L, arg, def); | ||
| 90 | ✗ | if (pos > (lua_Integer)len) | ||
| 91 | ✗ | return len; | ||
| 92 | ✗ | else if (pos >= 0) | ||
| 93 | ✗ | return (size_t)pos; | ||
| 94 | ✗ | else if (pos < -(lua_Integer)len) | ||
| 95 | ✗ | return 0; | ||
| 96 | ✗ | else return len + (size_t)pos + 1; | ||
| 97 | } | |||
| 98 | ||||
| 99 | ||||
| 100 | ✗ | static int str_sub (lua_State *L) { | ||
| 101 | size_t l; | |||
| 102 | ✗ | const char *s = luaL_checklstring(L, 1, &l); | ||
| 103 | ✗ | size_t start = posrelatI(luaL_checkinteger(L, 2), l); | ||
| 104 | ✗ | size_t end = getendpos(L, 3, -1, l); | ||
| 105 | ✗ | if (start <= end) | ||
| 106 | ✗ | lua_pushlstring(L, s + start - 1, (end - start) + 1); | ||
| 107 | ✗ | else lua_pushliteral(L, ""); | ||
| 108 | ✗ | return 1; | ||
| 109 | } | |||
| 110 | ||||
| 111 | ||||
| 112 | ✗ | static int str_reverse (lua_State *L) { | ||
| 113 | size_t l, i; | |||
| 114 | luaL_Buffer b; | |||
| 115 | ✗ | const char *s = luaL_checklstring(L, 1, &l); | ||
| 116 | ✗ | char *p = luaL_buffinitsize(L, &b, l); | ||
| 117 | ✗ | for (i = 0; i < l; i++) | ||
| 118 | ✗ | p[i] = s[l - i - 1]; | ||
| 119 | ✗ | luaL_pushresultsize(&b, l); | ||
| 120 | ✗ | return 1; | ||
| 121 | } | |||
| 122 | ||||
| 123 | ||||
| 124 | ✗ | static int str_lower (lua_State *L) { | ||
| 125 | size_t l; | |||
| 126 | size_t i; | |||
| 127 | luaL_Buffer b; | |||
| 128 | ✗ | const char *s = luaL_checklstring(L, 1, &l); | ||
| 129 | ✗ | char *p = luaL_buffinitsize(L, &b, l); | ||
| 130 | ✗ | for (i=0; i<l; i++) | ||
| 131 | ✗ | p[i] = tolower(uchar(s[i])); | ||
| 132 | ✗ | luaL_pushresultsize(&b, l); | ||
| 133 | ✗ | return 1; | ||
| 134 | } | |||
| 135 | ||||
| 136 | ||||
| 137 | ✗ | static int str_upper (lua_State *L) { | ||
| 138 | size_t l; | |||
| 139 | size_t i; | |||
| 140 | luaL_Buffer b; | |||
| 141 | ✗ | const char *s = luaL_checklstring(L, 1, &l); | ||
| 142 | ✗ | char *p = luaL_buffinitsize(L, &b, l); | ||
| 143 | ✗ | for (i=0; i<l; i++) | ||
| 144 | ✗ | p[i] = toupper(uchar(s[i])); | ||
| 145 | ✗ | luaL_pushresultsize(&b, l); | ||
| 146 | ✗ | return 1; | ||
| 147 | } | |||
| 148 | ||||
| 149 | ||||
| 150 | ✗ | static int str_rep (lua_State *L) { | ||
| 151 | size_t l, lsep; | |||
| 152 | ✗ | const char *s = luaL_checklstring(L, 1, &l); | ||
| 153 | ✗ | lua_Integer n = luaL_checkinteger(L, 2); | ||
| 154 | ✗ | const char *sep = luaL_optlstring(L, 3, "", &lsep); | ||
| 155 | ✗ | if (n <= 0) | ||
| 156 | ✗ | lua_pushliteral(L, ""); | ||
| 157 | ✗ | else if (l_unlikely(l + lsep < l || l + lsep > MAXSIZE / n)) | ||
| 158 | ✗ | return luaL_error(L, "resulting string too large"); | ||
| 159 | else { | |||
| 160 | ✗ | size_t totallen = (size_t)n * l + (size_t)(n - 1) * lsep; | ||
| 161 | luaL_Buffer b; | |||
| 162 | ✗ | char *p = luaL_buffinitsize(L, &b, totallen); | ||
| 163 | ✗ | while (n-- > 1) { /* first n-1 copies (followed by separator) */ | ||
| 164 | ✗ | memcpy(p, s, l * sizeof(char)); p += l; | ||
| 165 | ✗ | if (lsep > 0) { /* empty 'memcpy' is not that cheap */ | ||
| 166 | ✗ | memcpy(p, sep, lsep * sizeof(char)); | ||
| 167 | ✗ | p += lsep; | ||
| 168 | } | |||
| 169 | } | |||
| 170 | ✗ | memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */ | ||
| 171 | ✗ | luaL_pushresultsize(&b, totallen); | ||
| 172 | } | |||
| 173 | ✗ | return 1; | ||
| 174 | } | |||
| 175 | ||||
| 176 | ||||
| 177 | ✗ | static int str_byte (lua_State *L) { | ||
| 178 | size_t l; | |||
| 179 | ✗ | const char *s = luaL_checklstring(L, 1, &l); | ||
| 180 | ✗ | lua_Integer pi = luaL_optinteger(L, 2, 1); | ||
| 181 | ✗ | size_t posi = posrelatI(pi, l); | ||
| 182 | ✗ | size_t pose = getendpos(L, 3, pi, l); | ||
| 183 | int n, i; | |||
| 184 | ✗ | if (posi > pose) return 0; /* empty interval; return no values */ | ||
| 185 | ✗ | if (l_unlikely(pose - posi >= (size_t)INT_MAX)) /* arithmetic overflow? */ | ||
| 186 | ✗ | return luaL_error(L, "string slice too long"); | ||
| 187 | ✗ | n = (int)(pose - posi) + 1; | ||
| 188 | ✗ | luaL_checkstack(L, n, "string slice too long"); | ||
| 189 | ✗ | for (i=0; i<n; i++) | ||
| 190 | ✗ | lua_pushinteger(L, uchar(s[posi+i-1])); | ||
| 191 | ✗ | return n; | ||
| 192 | } | |||
| 193 | ||||
| 194 | ||||
| 195 | ✗ | static int str_char (lua_State *L) { | ||
| 196 | ✗ | int n = lua_gettop(L); /* number of arguments */ | ||
| 197 | int i; | |||
| 198 | luaL_Buffer b; | |||
| 199 | ✗ | char *p = luaL_buffinitsize(L, &b, n); | ||
| 200 | ✗ | for (i=1; i<=n; i++) { | ||
| 201 | ✗ | lua_Unsigned c = (lua_Unsigned)luaL_checkinteger(L, i); | ||
| 202 | ✗ | luaL_argcheck(L, c <= (lua_Unsigned)UCHAR_MAX, i, "value out of range"); | ||
| 203 | ✗ | p[i - 1] = uchar(c); | ||
| 204 | } | |||
| 205 | ✗ | luaL_pushresultsize(&b, n); | ||
| 206 | ✗ | return 1; | ||
| 207 | } | |||
| 208 | ||||
| 209 | ||||
| 210 | /* | |||
| 211 | ** Buffer to store the result of 'string.dump'. It must be initialized | |||
| 212 | ** after the call to 'lua_dump', to ensure that the function is on the | |||
| 213 | ** top of the stack when 'lua_dump' is called. ('luaL_buffinit' might | |||
| 214 | ** push stuff.) | |||
| 215 | */ | |||
| 216 | struct str_Writer { | |||
| 217 | int init; /* true iff buffer has been initialized */ | |||
| 218 | luaL_Buffer B; | |||
| 219 | }; | |||
| 220 | ||||
| 221 | ||||
| 222 | ✗ | static int writer (lua_State *L, const void *b, size_t size, void *ud) { | ||
| 223 | ✗ | struct str_Writer *state = (struct str_Writer *)ud; | ||
| 224 | ✗ | if (!state->init) { | ||
| 225 | ✗ | state->init = 1; | ||
| 226 | ✗ | luaL_buffinit(L, &state->B); | ||
| 227 | } | |||
| 228 | ✗ | luaL_addlstring(&state->B, (const char *)b, size); | ||
| 229 | ✗ | return 0; | ||
| 230 | } | |||
| 231 | ||||
| 232 | ||||
| 233 | ✗ | static int str_dump (lua_State *L) { | ||
| 234 | struct str_Writer state; | |||
| 235 | ✗ | int strip = lua_toboolean(L, 2); | ||
| 236 | ✗ | luaL_checktype(L, 1, LUA_TFUNCTION); | ||
| 237 | ✗ | lua_settop(L, 1); /* ensure function is on the top of the stack */ | ||
| 238 | ✗ | state.init = 0; | ||
| 239 | ✗ | if (l_unlikely(lua_dump(L, writer, &state, strip) != 0)) | ||
| 240 | ✗ | return luaL_error(L, "unable to dump given function"); | ||
| 241 | ✗ | luaL_pushresult(&state.B); | ||
| 242 | ✗ | return 1; | ||
| 243 | } | |||
| 244 | ||||
| 245 | ||||
| 246 | ||||
| 247 | /* | |||
| 248 | ** {====================================================== | |||
| 249 | ** METAMETHODS | |||
| 250 | ** ======================================================= | |||
| 251 | */ | |||
| 252 | ||||
| 253 | #if defined(LUA_NOCVTS2N) /* { */ | |||
| 254 | ||||
| 255 | /* no coercion from strings to numbers */ | |||
| 256 | ||||
| 257 | static const luaL_Reg stringmetamethods[] = { | |||
| 258 | {"__index", NULL}, /* placeholder */ | |||
| 259 | {NULL, NULL} | |||
| 260 | }; | |||
| 261 | ||||
| 262 | #else /* }{ */ | |||
| 263 | ||||
| 264 | ✗ | static int tonum (lua_State *L, int arg) { | ||
| 265 | ✗ | if (lua_type(L, arg) == LUA_TNUMBER) { /* already a number? */ | ||
| 266 | ✗ | lua_pushvalue(L, arg); | ||
| 267 | ✗ | return 1; | ||
| 268 | } | |||
| 269 | else { /* check whether it is a numerical string */ | |||
| 270 | size_t len; | |||
| 271 | ✗ | const char *s = lua_tolstring(L, arg, &len); | ||
| 272 | ✗ | return (s != NULL && lua_stringtonumber(L, s) == len + 1); | ||
| 273 | } | |||
| 274 | } | |||
| 275 | ||||
| 276 | ||||
| 277 | ✗ | static void trymt (lua_State *L, const char *mtname) { | ||
| 278 | ✗ | lua_settop(L, 2); /* back to the original arguments */ | ||
| 279 | ✗ | if (l_unlikely(lua_type(L, 2) == LUA_TSTRING || | ||
| 280 | !luaL_getmetafield(L, 2, mtname))) | |||
| 281 | ✗ | luaL_error(L, "attempt to %s a '%s' with a '%s'", mtname + 2, | ||
| 282 | luaL_typename(L, -2), luaL_typename(L, -1)); | |||
| 283 | ✗ | lua_insert(L, -3); /* put metamethod before arguments */ | ||
| 284 | ✗ | lua_call(L, 2, 1); /* call metamethod */ | ||
| 285 | ✗ | } | ||
| 286 | ||||
| 287 | ||||
| 288 | ✗ | static int arith (lua_State *L, int op, const char *mtname) { | ||
| 289 | ✗ | if (tonum(L, 1) && tonum(L, 2)) | ||
| 290 | ✗ | lua_arith(L, op); /* result will be on the top */ | ||
| 291 | else | |||
| 292 | ✗ | trymt(L, mtname); | ||
| 293 | ✗ | return 1; | ||
| 294 | } | |||
| 295 | ||||
| 296 | ||||
| 297 | ✗ | static int arith_add (lua_State *L) { | ||
| 298 | ✗ | return arith(L, LUA_OPADD, "__add"); | ||
| 299 | } | |||
| 300 | ||||
| 301 | ✗ | static int arith_sub (lua_State *L) { | ||
| 302 | ✗ | return arith(L, LUA_OPSUB, "__sub"); | ||
| 303 | } | |||
| 304 | ||||
| 305 | ✗ | static int arith_mul (lua_State *L) { | ||
| 306 | ✗ | return arith(L, LUA_OPMUL, "__mul"); | ||
| 307 | } | |||
| 308 | ||||
| 309 | ✗ | static int arith_mod (lua_State *L) { | ||
| 310 | ✗ | return arith(L, LUA_OPMOD, "__mod"); | ||
| 311 | } | |||
| 312 | ||||
| 313 | ✗ | static int arith_pow (lua_State *L) { | ||
| 314 | ✗ | return arith(L, LUA_OPPOW, "__pow"); | ||
| 315 | } | |||
| 316 | ||||
| 317 | ✗ | static int arith_div (lua_State *L) { | ||
| 318 | ✗ | return arith(L, LUA_OPDIV, "__div"); | ||
| 319 | } | |||
| 320 | ||||
| 321 | ✗ | static int arith_idiv (lua_State *L) { | ||
| 322 | ✗ | return arith(L, LUA_OPIDIV, "__idiv"); | ||
| 323 | } | |||
| 324 | ||||
| 325 | ✗ | static int arith_unm (lua_State *L) { | ||
| 326 | ✗ | return arith(L, LUA_OPUNM, "__unm"); | ||
| 327 | } | |||
| 328 | ||||
| 329 | ||||
| 330 | static const luaL_Reg stringmetamethods[] = { | |||
| 331 | {"__add", arith_add}, | |||
| 332 | {"__sub", arith_sub}, | |||
| 333 | {"__mul", arith_mul}, | |||
| 334 | {"__mod", arith_mod}, | |||
| 335 | {"__pow", arith_pow}, | |||
| 336 | {"__div", arith_div}, | |||
| 337 | {"__idiv", arith_idiv}, | |||
| 338 | {"__unm", arith_unm}, | |||
| 339 | {"__index", NULL}, /* placeholder */ | |||
| 340 | {NULL, NULL} | |||
| 341 | }; | |||
| 342 | ||||
| 343 | #endif /* } */ | |||
| 344 | ||||
| 345 | /* }====================================================== */ | |||
| 346 | ||||
| 347 | /* | |||
| 348 | ** {====================================================== | |||
| 349 | ** PATTERN MATCHING | |||
| 350 | ** ======================================================= | |||
| 351 | */ | |||
| 352 | ||||
| 353 | ||||
| 354 | #define CAP_UNFINISHED (-1) | |||
| 355 | #define CAP_POSITION (-2) | |||
| 356 | ||||
| 357 | ||||
| 358 | typedef struct MatchState { | |||
| 359 | const char *src_init; /* init of source string */ | |||
| 360 | const char *src_end; /* end ('\0') of source string */ | |||
| 361 | const char *p_end; /* end ('\0') of pattern */ | |||
| 362 | lua_State *L; | |||
| 363 | int matchdepth; /* control for recursive depth (to avoid C stack overflow) */ | |||
| 364 | unsigned char level; /* total number of captures (finished or unfinished) */ | |||
| 365 | struct { | |||
| 366 | const char *init; | |||
| 367 | ptrdiff_t len; | |||
| 368 | } capture[LUA_MAXCAPTURES]; | |||
| 369 | } MatchState; | |||
| 370 | ||||
| 371 | ||||
| 372 | /* recursive function */ | |||
| 373 | static const char *match (MatchState *ms, const char *s, const char *p); | |||
| 374 | ||||
| 375 | ||||
| 376 | /* maximum recursion depth for 'match' */ | |||
| 377 | #if !defined(MAXCCALLS) | |||
| 378 | #define MAXCCALLS 200 | |||
| 379 | #endif | |||
| 380 | ||||
| 381 | ||||
| 382 | #define L_ESC '%' | |||
| 383 | #define SPECIALS "^$*+?.([%-" | |||
| 384 | ||||
| 385 | ||||
| 386 | ✗ | static int check_capture (MatchState *ms, int l) { | ||
| 387 | ✗ | l -= '1'; | ||
| 388 | ✗ | if (l_unlikely(l < 0 || l >= ms->level || | ||
| 389 | ms->capture[l].len == CAP_UNFINISHED)) | |||
| 390 | ✗ | return luaL_error(ms->L, "invalid capture index %%%d", l + 1); | ||
| 391 | ✗ | return l; | ||
| 392 | } | |||
| 393 | ||||
| 394 | ||||
| 395 | ✗ | static int capture_to_close (MatchState *ms) { | ||
| 396 | ✗ | int level = ms->level; | ||
| 397 | ✗ | for (level--; level>=0; level--) | ||
| 398 | ✗ | if (ms->capture[level].len == CAP_UNFINISHED) return level; | ||
| 399 | ✗ | return luaL_error(ms->L, "invalid pattern capture"); | ||
| 400 | } | |||
| 401 | ||||
| 402 | ||||
| 403 | ✗ | static const char *classend (MatchState *ms, const char *p) { | ||
| 404 | ✗ | switch (*p++) { | ||
| 405 | ✗ | case L_ESC: { | ||
| 406 | ✗ | if (l_unlikely(p == ms->p_end)) | ||
| 407 | ✗ | luaL_error(ms->L, "malformed pattern (ends with '%%')"); | ||
| 408 | ✗ | return p+1; | ||
| 409 | } | |||
| 410 | ✗ | case '[': { | ||
| 411 | ✗ | if (*p == '^') p++; | ||
| 412 | do { /* look for a ']' */ | |||
| 413 | ✗ | if (l_unlikely(p == ms->p_end)) | ||
| 414 | ✗ | luaL_error(ms->L, "malformed pattern (missing ']')"); | ||
| 415 | ✗ | if (*(p++) == L_ESC && p < ms->p_end) | ||
| 416 | ✗ | p++; /* skip escapes (e.g. '%]') */ | ||
| 417 | ✗ | } while (*p != ']'); | ||
| 418 | ✗ | return p+1; | ||
| 419 | } | |||
| 420 | ✗ | default: { | ||
| 421 | ✗ | return p; | ||
| 422 | } | |||
| 423 | } | |||
| 424 | } | |||
| 425 | ||||
| 426 | ||||
| 427 | ✗ | static int match_class (int c, int cl) { | ||
| 428 | int res; | |||
| 429 | ✗ | switch (tolower(cl)) { | ||
| 430 | ✗ | case 'a' : res = isalpha(c); break; | ||
| 431 | ✗ | case 'c' : res = iscntrl(c); break; | ||
| 432 | ✗ | case 'd' : res = isdigit(c); break; | ||
| 433 | ✗ | case 'g' : res = isgraph(c); break; | ||
| 434 | ✗ | case 'l' : res = islower(c); break; | ||
| 435 | ✗ | case 'p' : res = ispunct(c); break; | ||
| 436 | ✗ | case 's' : res = isspace(c); break; | ||
| 437 | ✗ | case 'u' : res = isupper(c); break; | ||
| 438 | ✗ | case 'w' : res = isalnum(c); break; | ||
| 439 | ✗ | case 'x' : res = isxdigit(c); break; | ||
| 440 | ✗ | case 'z' : res = (c == 0); break; /* deprecated option */ | ||
| 441 | ✗ | default: return (cl == c); | ||
| 442 | } | |||
| 443 | ✗ | return (islower(cl) ? res : !res); | ||
| 444 | } | |||
| 445 | ||||
| 446 | ||||
| 447 | ✗ | static int matchbracketclass (int c, const char *p, const char *ec) { | ||
| 448 | ✗ | int sig = 1; | ||
| 449 | ✗ | if (*(p+1) == '^') { | ||
| 450 | ✗ | sig = 0; | ||
| 451 | ✗ | p++; /* skip the '^' */ | ||
| 452 | } | |||
| 453 | ✗ | while (++p < ec) { | ||
| 454 | ✗ | if (*p == L_ESC) { | ||
| 455 | ✗ | p++; | ||
| 456 | ✗ | if (match_class(c, uchar(*p))) | ||
| 457 | ✗ | return sig; | ||
| 458 | } | |||
| 459 | ✗ | else if ((*(p+1) == '-') && (p+2 < ec)) { | ||
| 460 | ✗ | p+=2; | ||
| 461 | ✗ | if (uchar(*(p-2)) <= c && c <= uchar(*p)) | ||
| 462 | ✗ | return sig; | ||
| 463 | } | |||
| 464 | ✗ | else if (uchar(*p) == c) return sig; | ||
| 465 | } | |||
| 466 | ✗ | return !sig; | ||
| 467 | } | |||
| 468 | ||||
| 469 | ||||
| 470 | ✗ | static int singlematch (MatchState *ms, const char *s, const char *p, | ||
| 471 | const char *ep) { | |||
| 472 | ✗ | if (s >= ms->src_end) | ||
| 473 | ✗ | return 0; | ||
| 474 | else { | |||
| 475 | ✗ | int c = uchar(*s); | ||
| 476 | ✗ | switch (*p) { | ||
| 477 | ✗ | case '.': return 1; /* matches any char */ | ||
| 478 | ✗ | case L_ESC: return match_class(c, uchar(*(p+1))); | ||
| 479 | ✗ | case '[': return matchbracketclass(c, p, ep-1); | ||
| 480 | ✗ | default: return (uchar(*p) == c); | ||
| 481 | } | |||
| 482 | } | |||
| 483 | } | |||
| 484 | ||||
| 485 | ||||
| 486 | ✗ | static const char *matchbalance (MatchState *ms, const char *s, | ||
| 487 | const char *p) { | |||
| 488 | ✗ | if (l_unlikely(p >= ms->p_end - 1)) | ||
| 489 | ✗ | luaL_error(ms->L, "malformed pattern (missing arguments to '%%b')"); | ||
| 490 | ✗ | if (*s != *p) return NULL; | ||
| 491 | else { | |||
| 492 | ✗ | int b = *p; | ||
| 493 | ✗ | int e = *(p+1); | ||
| 494 | ✗ | int cont = 1; | ||
| 495 | ✗ | while (++s < ms->src_end) { | ||
| 496 | ✗ | if (*s == e) { | ||
| 497 | ✗ | if (--cont == 0) return s+1; | ||
| 498 | } | |||
| 499 | ✗ | else if (*s == b) cont++; | ||
| 500 | } | |||
| 501 | } | |||
| 502 | ✗ | return NULL; /* string ends out of balance */ | ||
| 503 | } | |||
| 504 | ||||
| 505 | ||||
| 506 | ✗ | static const char *max_expand (MatchState *ms, const char *s, | ||
| 507 | const char *p, const char *ep) { | |||
| 508 | ✗ | ptrdiff_t i = 0; /* counts maximum expand for item */ | ||
| 509 | ✗ | while (singlematch(ms, s + i, p, ep)) | ||
| 510 | ✗ | i++; | ||
| 511 | /* keeps trying to match with the maximum repetitions */ | |||
| 512 | ✗ | while (i>=0) { | ||
| 513 | ✗ | const char *res = match(ms, (s+i), ep+1); | ||
| 514 | ✗ | if (res) return res; | ||
| 515 | ✗ | i--; /* else didn't match; reduce 1 repetition to try again */ | ||
| 516 | } | |||
| 517 | ✗ | return NULL; | ||
| 518 | } | |||
| 519 | ||||
| 520 | ||||
| 521 | ✗ | static const char *min_expand (MatchState *ms, const char *s, | ||
| 522 | const char *p, const char *ep) { | |||
| 523 | ✗ | for (;;) { | ||
| 524 | ✗ | const char *res = match(ms, s, ep+1); | ||
| 525 | ✗ | if (res != NULL) | ||
| 526 | ✗ | return res; | ||
| 527 | ✗ | else if (singlematch(ms, s, p, ep)) | ||
| 528 | ✗ | s++; /* try with one more repetition */ | ||
| 529 | ✗ | else return NULL; | ||
| 530 | } | |||
| 531 | } | |||
| 532 | ||||
| 533 | ||||
| 534 | ✗ | static const char *start_capture (MatchState *ms, const char *s, | ||
| 535 | const char *p, int what) { | |||
| 536 | const char *res; | |||
| 537 | ✗ | int level = ms->level; | ||
| 538 | ✗ | if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures"); | ||
| 539 | ✗ | ms->capture[level].init = s; | ||
| 540 | ✗ | ms->capture[level].len = what; | ||
| 541 | ✗ | ms->level = level+1; | ||
| 542 | ✗ | if ((res=match(ms, s, p)) == NULL) /* match failed? */ | ||
| 543 | ✗ | ms->level--; /* undo capture */ | ||
| 544 | ✗ | return res; | ||
| 545 | } | |||
| 546 | ||||
| 547 | ||||
| 548 | ✗ | static const char *end_capture (MatchState *ms, const char *s, | ||
| 549 | const char *p) { | |||
| 550 | ✗ | int l = capture_to_close(ms); | ||
| 551 | const char *res; | |||
| 552 | ✗ | ms->capture[l].len = s - ms->capture[l].init; /* close capture */ | ||
| 553 | ✗ | if ((res = match(ms, s, p)) == NULL) /* match failed? */ | ||
| 554 | ✗ | ms->capture[l].len = CAP_UNFINISHED; /* undo capture */ | ||
| 555 | ✗ | return res; | ||
| 556 | } | |||
| 557 | ||||
| 558 | ||||
| 559 | ✗ | static const char *match_capture (MatchState *ms, const char *s, int l) { | ||
| 560 | size_t len; | |||
| 561 | ✗ | l = check_capture(ms, l); | ||
| 562 | ✗ | len = ms->capture[l].len; | ||
| 563 | ✗ | if ((size_t)(ms->src_end-s) >= len && | ||
| 564 | ✗ | memcmp(ms->capture[l].init, s, len) == 0) | ||
| 565 | ✗ | return s+len; | ||
| 566 | ✗ | else return NULL; | ||
| 567 | } | |||
| 568 | ||||
| 569 | ||||
| 570 | ✗ | static const char *match (MatchState *ms, const char *s, const char *p) { | ||
| 571 | ✗ | if (l_unlikely(ms->matchdepth-- == 0)) | ||
| 572 | ✗ | luaL_error(ms->L, "pattern too complex"); | ||
| 573 | ✗ | init: /* using goto to optimize tail recursion */ | ||
| 574 | ✗ | if (p != ms->p_end) { /* end of pattern? */ | ||
| 575 | ✗ | switch (*p) { | ||
| 576 | ✗ | case '(': { /* start capture */ | ||
| 577 | ✗ | if (*(p + 1) == ')') /* position capture? */ | ||
| 578 | ✗ | s = start_capture(ms, s, p + 2, CAP_POSITION); | ||
| 579 | else | |||
| 580 | ✗ | s = start_capture(ms, s, p + 1, CAP_UNFINISHED); | ||
| 581 | ✗ | break; | ||
| 582 | } | |||
| 583 | ✗ | case ')': { /* end capture */ | ||
| 584 | ✗ | s = end_capture(ms, s, p + 1); | ||
| 585 | ✗ | break; | ||
| 586 | } | |||
| 587 | ✗ | case '$': { | ||
| 588 | ✗ | if ((p + 1) != ms->p_end) /* is the '$' the last char in pattern? */ | ||
| 589 | ✗ | goto dflt; /* no; go to default */ | ||
| 590 | ✗ | s = (s == ms->src_end) ? s : NULL; /* check end of string */ | ||
| 591 | ✗ | break; | ||
| 592 | } | |||
| 593 | ✗ | case L_ESC: { /* escaped sequences not in the format class[*+?-]? */ | ||
| 594 | ✗ | switch (*(p + 1)) { | ||
| 595 | ✗ | case 'b': { /* balanced string? */ | ||
| 596 | ✗ | s = matchbalance(ms, s, p + 2); | ||
| 597 | ✗ | if (s != NULL) { | ||
| 598 | ✗ | p += 4; goto init; /* return match(ms, s, p + 4); */ | ||
| 599 | } /* else fail (s == NULL) */ | |||
| 600 | ✗ | break; | ||
| 601 | } | |||
| 602 | ✗ | case 'f': { /* frontier? */ | ||
| 603 | const char *ep; char previous; | |||
| 604 | ✗ | p += 2; | ||
| 605 | ✗ | if (l_unlikely(*p != '[')) | ||
| 606 | ✗ | luaL_error(ms->L, "missing '[' after '%%f' in pattern"); | ||
| 607 | ✗ | ep = classend(ms, p); /* points to what is next */ | ||
| 608 | ✗ | previous = (s == ms->src_init) ? '\0' : *(s - 1); | ||
| 609 | ✗ | if (!matchbracketclass(uchar(previous), p, ep - 1) && | ||
| 610 | ✗ | matchbracketclass(uchar(*s), p, ep - 1)) { | ||
| 611 | ✗ | p = ep; goto init; /* return match(ms, s, ep); */ | ||
| 612 | } | |||
| 613 | ✗ | s = NULL; /* match failed */ | ||
| 614 | ✗ | break; | ||
| 615 | } | |||
| 616 | ✗ | case '0': case '1': case '2': case '3': | ||
| 617 | case '4': case '5': case '6': case '7': | |||
| 618 | case '8': case '9': { /* capture results (%0-%9)? */ | |||
| 619 | ✗ | s = match_capture(ms, s, uchar(*(p + 1))); | ||
| 620 | ✗ | if (s != NULL) { | ||
| 621 | ✗ | p += 2; goto init; /* return match(ms, s, p + 2) */ | ||
| 622 | } | |||
| 623 | ✗ | break; | ||
| 624 | } | |||
| 625 | ✗ | default: goto dflt; | ||
| 626 | } | |||
| 627 | ✗ | break; | ||
| 628 | } | |||
| 629 | ✗ | default: dflt: { /* pattern class plus optional suffix */ | ||
| 630 | ✗ | const char *ep = classend(ms, p); /* points to optional suffix */ | ||
| 631 | /* does not match at least once? */ | |||
| 632 | ✗ | if (!singlematch(ms, s, p, ep)) { | ||
| 633 | ✗ | if (*ep == '*' || *ep == '?' || *ep == '-') { /* accept empty? */ | ||
| 634 | ✗ | p = ep + 1; goto init; /* return match(ms, s, ep + 1); */ | ||
| 635 | } | |||
| 636 | else /* '+' or no suffix */ | |||
| 637 | ✗ | s = NULL; /* fail */ | ||
| 638 | } | |||
| 639 | else { /* matched once */ | |||
| 640 | ✗ | switch (*ep) { /* handle optional suffix */ | ||
| 641 | ✗ | case '?': { /* optional */ | ||
| 642 | const char *res; | |||
| 643 | ✗ | if ((res = match(ms, s + 1, ep + 1)) != NULL) | ||
| 644 | ✗ | s = res; | ||
| 645 | else { | |||
| 646 | ✗ | p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */ | ||
| 647 | } | |||
| 648 | ✗ | break; | ||
| 649 | } | |||
| 650 | ✗ | case '+': /* 1 or more repetitions */ | ||
| 651 | ✗ | s++; /* 1 match already done */ | ||
| 652 | /* FALLTHROUGH */ | |||
| 653 | ✗ | case '*': /* 0 or more repetitions */ | ||
| 654 | ✗ | s = max_expand(ms, s, p, ep); | ||
| 655 | ✗ | break; | ||
| 656 | ✗ | case '-': /* 0 or more repetitions (minimum) */ | ||
| 657 | ✗ | s = min_expand(ms, s, p, ep); | ||
| 658 | ✗ | break; | ||
| 659 | ✗ | default: /* no suffix */ | ||
| 660 | ✗ | s++; p = ep; goto init; /* return match(ms, s + 1, ep); */ | ||
| 661 | } | |||
| 662 | } | |||
| 663 | ✗ | break; | ||
| 664 | } | |||
| 665 | } | |||
| 666 | } | |||
| 667 | ✗ | ms->matchdepth++; | ||
| 668 | ✗ | return s; | ||
| 669 | } | |||
| 670 | ||||
| 671 | ||||
| 672 | ||||
| 673 | ✗ | static const char *lmemfind (const char *s1, size_t l1, | ||
| 674 | const char *s2, size_t l2) { | |||
| 675 | ✗ | if (l2 == 0) return s1; /* empty strings are everywhere */ | ||
| 676 | ✗ | else if (l2 > l1) return NULL; /* avoids a negative 'l1' */ | ||
| 677 | else { | |||
| 678 | const char *init; /* to search for a '*s2' inside 's1' */ | |||
| 679 | ✗ | l2--; /* 1st char will be checked by 'memchr' */ | ||
| 680 | ✗ | l1 = l1-l2; /* 's2' cannot be found after that */ | ||
| 681 | ✗ | while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) { | ||
| 682 | ✗ | init++; /* 1st char is already checked */ | ||
| 683 | ✗ | if (memcmp(init, s2+1, l2) == 0) | ||
| 684 | ✗ | return init-1; | ||
| 685 | else { /* correct 'l1' and 's1' to try again */ | |||
| 686 | ✗ | l1 -= init-s1; | ||
| 687 | ✗ | s1 = init; | ||
| 688 | } | |||
| 689 | } | |||
| 690 | ✗ | return NULL; /* not found */ | ||
| 691 | } | |||
| 692 | } | |||
| 693 | ||||
| 694 | ||||
| 695 | /* | |||
| 696 | ** get information about the i-th capture. If there are no captures | |||
| 697 | ** and 'i==0', return information about the whole match, which | |||
| 698 | ** is the range 's'..'e'. If the capture is a string, return | |||
| 699 | ** its length and put its address in '*cap'. If it is an integer | |||
| 700 | ** (a position), push it on the stack and return CAP_POSITION. | |||
| 701 | */ | |||
| 702 | ✗ | static size_t get_onecapture (MatchState *ms, int i, const char *s, | ||
| 703 | const char *e, const char **cap) { | |||
| 704 | ✗ | if (i >= ms->level) { | ||
| 705 | ✗ | if (l_unlikely(i != 0)) | ||
| 706 | ✗ | luaL_error(ms->L, "invalid capture index %%%d", i + 1); | ||
| 707 | ✗ | *cap = s; | ||
| 708 | ✗ | return e - s; | ||
| 709 | } | |||
| 710 | else { | |||
| 711 | ✗ | ptrdiff_t capl = ms->capture[i].len; | ||
| 712 | ✗ | *cap = ms->capture[i].init; | ||
| 713 | ✗ | if (l_unlikely(capl == CAP_UNFINISHED)) | ||
| 714 | ✗ | luaL_error(ms->L, "unfinished capture"); | ||
| 715 | ✗ | else if (capl == CAP_POSITION) | ||
| 716 | ✗ | lua_pushinteger(ms->L, (ms->capture[i].init - ms->src_init) + 1); | ||
| 717 | ✗ | return capl; | ||
| 718 | } | |||
| 719 | } | |||
| 720 | ||||
| 721 | ||||
| 722 | /* | |||
| 723 | ** Push the i-th capture on the stack. | |||
| 724 | */ | |||
| 725 | ✗ | static void push_onecapture (MatchState *ms, int i, const char *s, | ||
| 726 | const char *e) { | |||
| 727 | const char *cap; | |||
| 728 | ✗ | ptrdiff_t l = get_onecapture(ms, i, s, e, &cap); | ||
| 729 | ✗ | if (l != CAP_POSITION) | ||
| 730 | ✗ | lua_pushlstring(ms->L, cap, l); | ||
| 731 | /* else position was already pushed */ | |||
| 732 | ✗ | } | ||
| 733 | ||||
| 734 | ||||
| 735 | ✗ | static int push_captures (MatchState *ms, const char *s, const char *e) { | ||
| 736 | int i; | |||
| 737 | ✗ | int nlevels = (ms->level == 0 && s) ? 1 : ms->level; | ||
| 738 | ✗ | luaL_checkstack(ms->L, nlevels, "too many captures"); | ||
| 739 | ✗ | for (i = 0; i < nlevels; i++) | ||
| 740 | ✗ | push_onecapture(ms, i, s, e); | ||
| 741 | ✗ | return nlevels; /* number of strings pushed */ | ||
| 742 | } | |||
| 743 | ||||
| 744 | ||||
| 745 | /* check whether pattern has no special characters */ | |||
| 746 | ✗ | static int nospecials (const char *p, size_t l) { | ||
| 747 | ✗ | size_t upto = 0; | ||
| 748 | do { | |||
| 749 | ✗ | if (strpbrk(p + upto, SPECIALS)) | ||
| 750 | ✗ | return 0; /* pattern has a special character */ | ||
| 751 | ✗ | upto += strlen(p + upto) + 1; /* may have more after \0 */ | ||
| 752 | ✗ | } while (upto <= l); | ||
| 753 | ✗ | return 1; /* no special chars found */ | ||
| 754 | } | |||
| 755 | ||||
| 756 | ||||
| 757 | ✗ | static void prepstate (MatchState *ms, lua_State *L, | ||
| 758 | const char *s, size_t ls, const char *p, size_t lp) { | |||
| 759 | ✗ | ms->L = L; | ||
| 760 | ✗ | ms->matchdepth = MAXCCALLS; | ||
| 761 | ✗ | ms->src_init = s; | ||
| 762 | ✗ | ms->src_end = s + ls; | ||
| 763 | ✗ | ms->p_end = p + lp; | ||
| 764 | ✗ | } | ||
| 765 | ||||
| 766 | ||||
| 767 | ✗ | static void reprepstate (MatchState *ms) { | ||
| 768 | ✗ | ms->level = 0; | ||
| 769 | lua_assert(ms->matchdepth == MAXCCALLS); | |||
| 770 | ✗ | } | ||
| 771 | ||||
| 772 | ||||
| 773 | ✗ | static int str_find_aux (lua_State *L, int find) { | ||
| 774 | size_t ls, lp; | |||
| 775 | ✗ | const char *s = luaL_checklstring(L, 1, &ls); | ||
| 776 | ✗ | const char *p = luaL_checklstring(L, 2, &lp); | ||
| 777 | ✗ | size_t init = posrelatI(luaL_optinteger(L, 3, 1), ls) - 1; | ||
| 778 | ✗ | if (init > ls) { /* start after string's end? */ | ||
| 779 | ✗ | luaL_pushfail(L); /* cannot find anything */ | ||
| 780 | ✗ | return 1; | ||
| 781 | } | |||
| 782 | /* explicit request or no special characters? */ | |||
| 783 | ✗ | if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) { | ||
| 784 | /* do a plain search */ | |||
| 785 | ✗ | const char *s2 = lmemfind(s + init, ls - init, p, lp); | ||
| 786 | ✗ | if (s2) { | ||
| 787 | ✗ | lua_pushinteger(L, (s2 - s) + 1); | ||
| 788 | ✗ | lua_pushinteger(L, (s2 - s) + lp); | ||
| 789 | ✗ | return 2; | ||
| 790 | } | |||
| 791 | } | |||
| 792 | else { | |||
| 793 | MatchState ms; | |||
| 794 | ✗ | const char *s1 = s + init; | ||
| 795 | ✗ | int anchor = (*p == '^'); | ||
| 796 | ✗ | if (anchor) { | ||
| 797 | ✗ | p++; lp--; /* skip anchor character */ | ||
| 798 | } | |||
| 799 | ✗ | prepstate(&ms, L, s, ls, p, lp); | ||
| 800 | do { | |||
| 801 | const char *res; | |||
| 802 | ✗ | reprepstate(&ms); | ||
| 803 | ✗ | if ((res=match(&ms, s1, p)) != NULL) { | ||
| 804 | ✗ | if (find) { | ||
| 805 | ✗ | lua_pushinteger(L, (s1 - s) + 1); /* start */ | ||
| 806 | ✗ | lua_pushinteger(L, res - s); /* end */ | ||
| 807 | ✗ | return push_captures(&ms, NULL, 0) + 2; | ||
| 808 | } | |||
| 809 | else | |||
| 810 | ✗ | return push_captures(&ms, s1, res); | ||
| 811 | } | |||
| 812 | ✗ | } while (s1++ < ms.src_end && !anchor); | ||
| 813 | } | |||
| 814 | ✗ | luaL_pushfail(L); /* not found */ | ||
| 815 | ✗ | return 1; | ||
| 816 | } | |||
| 817 | ||||
| 818 | ||||
| 819 | ✗ | static int str_find (lua_State *L) { | ||
| 820 | ✗ | return str_find_aux(L, 1); | ||
| 821 | } | |||
| 822 | ||||
| 823 | ||||
| 824 | ✗ | static int str_match (lua_State *L) { | ||
| 825 | ✗ | return str_find_aux(L, 0); | ||
| 826 | } | |||
| 827 | ||||
| 828 | ||||
| 829 | /* state for 'gmatch' */ | |||
| 830 | typedef struct GMatchState { | |||
| 831 | const char *src; /* current position */ | |||
| 832 | const char *p; /* pattern */ | |||
| 833 | const char *lastmatch; /* end of last match */ | |||
| 834 | MatchState ms; /* match state */ | |||
| 835 | } GMatchState; | |||
| 836 | ||||
| 837 | ||||
| 838 | ✗ | static int gmatch_aux (lua_State *L) { | ||
| 839 | ✗ | GMatchState *gm = (GMatchState *)lua_touserdata(L, lua_upvalueindex(3)); | ||
| 840 | const char *src; | |||
| 841 | ✗ | gm->ms.L = L; | ||
| 842 | ✗ | for (src = gm->src; src <= gm->ms.src_end; src++) { | ||
| 843 | const char *e; | |||
| 844 | ✗ | reprepstate(&gm->ms); | ||
| 845 | ✗ | if ((e = match(&gm->ms, src, gm->p)) != NULL && e != gm->lastmatch) { | ||
| 846 | ✗ | gm->src = gm->lastmatch = e; | ||
| 847 | ✗ | return push_captures(&gm->ms, src, e); | ||
| 848 | } | |||
| 849 | } | |||
| 850 | ✗ | return 0; /* not found */ | ||
| 851 | } | |||
| 852 | ||||
| 853 | ||||
| 854 | ✗ | static int gmatch (lua_State *L) { | ||
| 855 | size_t ls, lp; | |||
| 856 | ✗ | const char *s = luaL_checklstring(L, 1, &ls); | ||
| 857 | ✗ | const char *p = luaL_checklstring(L, 2, &lp); | ||
| 858 | ✗ | size_t init = posrelatI(luaL_optinteger(L, 3, 1), ls) - 1; | ||
| 859 | GMatchState *gm; | |||
| 860 | ✗ | lua_settop(L, 2); /* keep strings on closure to avoid being collected */ | ||
| 861 | ✗ | gm = (GMatchState *)lua_newuserdatauv(L, sizeof(GMatchState), 0); | ||
| 862 | ✗ | if (init > ls) /* start after string's end? */ | ||
| 863 | ✗ | init = ls + 1; /* avoid overflows in 's + init' */ | ||
| 864 | ✗ | prepstate(&gm->ms, L, s, ls, p, lp); | ||
| 865 | ✗ | gm->src = s + init; gm->p = p; gm->lastmatch = NULL; | ||
| 866 | ✗ | lua_pushcclosure(L, gmatch_aux, 3); | ||
| 867 | ✗ | return 1; | ||
| 868 | } | |||
| 869 | ||||
| 870 | ||||
| 871 | ✗ | static void add_s (MatchState *ms, luaL_Buffer *b, const char *s, | ||
| 872 | const char *e) { | |||
| 873 | size_t l; | |||
| 874 | ✗ | lua_State *L = ms->L; | ||
| 875 | ✗ | const char *news = lua_tolstring(L, 3, &l); | ||
| 876 | const char *p; | |||
| 877 | ✗ | while ((p = (char *)memchr(news, L_ESC, l)) != NULL) { | ||
| 878 | ✗ | luaL_addlstring(b, news, p - news); | ||
| 879 | ✗ | p++; /* skip ESC */ | ||
| 880 | ✗ | if (*p == L_ESC) /* '%%' */ | ||
| 881 | ✗ | luaL_addchar(b, *p); | ||
| 882 | ✗ | else if (*p == '0') /* '%0' */ | ||
| 883 | ✗ | luaL_addlstring(b, s, e - s); | ||
| 884 | ✗ | else if (isdigit(uchar(*p))) { /* '%n' */ | ||
| 885 | const char *cap; | |||
| 886 | ✗ | ptrdiff_t resl = get_onecapture(ms, *p - '1', s, e, &cap); | ||
| 887 | ✗ | if (resl == CAP_POSITION) | ||
| 888 | ✗ | luaL_addvalue(b); /* add position to accumulated result */ | ||
| 889 | else | |||
| 890 | ✗ | luaL_addlstring(b, cap, resl); | ||
| 891 | } | |||
| 892 | else | |||
| 893 | ✗ | luaL_error(L, "invalid use of '%c' in replacement string", L_ESC); | ||
| 894 | ✗ | l -= p + 1 - news; | ||
| 895 | ✗ | news = p + 1; | ||
| 896 | } | |||
| 897 | ✗ | luaL_addlstring(b, news, l); | ||
| 898 | ✗ | } | ||
| 899 | ||||
| 900 | ||||
| 901 | /* | |||
| 902 | ** Add the replacement value to the string buffer 'b'. | |||
| 903 | ** Return true if the original string was changed. (Function calls and | |||
| 904 | ** table indexing resulting in nil or false do not change the subject.) | |||
| 905 | */ | |||
| 906 | ✗ | static int add_value (MatchState *ms, luaL_Buffer *b, const char *s, | ||
| 907 | const char *e, int tr) { | |||
| 908 | ✗ | lua_State *L = ms->L; | ||
| 909 | ✗ | switch (tr) { | ||
| 910 | ✗ | case LUA_TFUNCTION: { /* call the function */ | ||
| 911 | int n; | |||
| 912 | ✗ | lua_pushvalue(L, 3); /* push the function */ | ||
| 913 | ✗ | n = push_captures(ms, s, e); /* all captures as arguments */ | ||
| 914 | ✗ | lua_call(L, n, 1); /* call it */ | ||
| 915 | ✗ | break; | ||
| 916 | } | |||
| 917 | ✗ | case LUA_TTABLE: { /* index the table */ | ||
| 918 | ✗ | push_onecapture(ms, 0, s, e); /* first capture is the index */ | ||
| 919 | ✗ | lua_gettable(L, 3); | ||
| 920 | ✗ | break; | ||
| 921 | } | |||
| 922 | ✗ | default: { /* LUA_TNUMBER or LUA_TSTRING */ | ||
| 923 | ✗ | add_s(ms, b, s, e); /* add value to the buffer */ | ||
| 924 | ✗ | return 1; /* something changed */ | ||
| 925 | } | |||
| 926 | } | |||
| 927 | ✗ | if (!lua_toboolean(L, -1)) { /* nil or false? */ | ||
| 928 | ✗ | lua_pop(L, 1); /* remove value */ | ||
| 929 | ✗ | luaL_addlstring(b, s, e - s); /* keep original text */ | ||
| 930 | ✗ | return 0; /* no changes */ | ||
| 931 | } | |||
| 932 | ✗ | else if (l_unlikely(!lua_isstring(L, -1))) | ||
| 933 | ✗ | return luaL_error(L, "invalid replacement value (a %s)", | ||
| 934 | luaL_typename(L, -1)); | |||
| 935 | else { | |||
| 936 | ✗ | luaL_addvalue(b); /* add result to accumulator */ | ||
| 937 | ✗ | return 1; /* something changed */ | ||
| 938 | } | |||
| 939 | } | |||
| 940 | ||||
| 941 | ||||
| 942 | ✗ | static int str_gsub (lua_State *L) { | ||
| 943 | size_t srcl, lp; | |||
| 944 | ✗ | const char *src = luaL_checklstring(L, 1, &srcl); /* subject */ | ||
| 945 | ✗ | const char *p = luaL_checklstring(L, 2, &lp); /* pattern */ | ||
| 946 | ✗ | const char *lastmatch = NULL; /* end of last match */ | ||
| 947 | ✗ | int tr = lua_type(L, 3); /* replacement type */ | ||
| 948 | ✗ | lua_Integer max_s = luaL_optinteger(L, 4, srcl + 1); /* max replacements */ | ||
| 949 | ✗ | int anchor = (*p == '^'); | ||
| 950 | ✗ | lua_Integer n = 0; /* replacement count */ | ||
| 951 | ✗ | int changed = 0; /* change flag */ | ||
| 952 | MatchState ms; | |||
| 953 | luaL_Buffer b; | |||
| 954 | ✗ | luaL_argexpected(L, tr == LUA_TNUMBER || tr == LUA_TSTRING || | ||
| 955 | tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3, | |||
| 956 | "string/function/table"); | |||
| 957 | ✗ | luaL_buffinit(L, &b); | ||
| 958 | ✗ | if (anchor) { | ||
| 959 | ✗ | p++; lp--; /* skip anchor character */ | ||
| 960 | } | |||
| 961 | ✗ | prepstate(&ms, L, src, srcl, p, lp); | ||
| 962 | ✗ | while (n < max_s) { | ||
| 963 | const char *e; | |||
| 964 | ✗ | reprepstate(&ms); /* (re)prepare state for new match */ | ||
| 965 | ✗ | if ((e = match(&ms, src, p)) != NULL && e != lastmatch) { /* match? */ | ||
| 966 | ✗ | n++; | ||
| 967 | ✗ | changed = add_value(&ms, &b, src, e, tr) | changed; | ||
| 968 | ✗ | src = lastmatch = e; | ||
| 969 | } | |||
| 970 | ✗ | else if (src < ms.src_end) /* otherwise, skip one character */ | ||
| 971 | ✗ | luaL_addchar(&b, *src++); | ||
| 972 | ✗ | else break; /* end of subject */ | ||
| 973 | ✗ | if (anchor) break; | ||
| 974 | } | |||
| 975 | ✗ | if (!changed) /* no changes? */ | ||
| 976 | ✗ | lua_pushvalue(L, 1); /* return original string */ | ||
| 977 | else { /* something changed */ | |||
| 978 | ✗ | luaL_addlstring(&b, src, ms.src_end-src); | ||
| 979 | ✗ | luaL_pushresult(&b); /* create and return new string */ | ||
| 980 | } | |||
| 981 | ✗ | lua_pushinteger(L, n); /* number of substitutions */ | ||
| 982 | ✗ | return 2; | ||
| 983 | } | |||
| 984 | ||||
| 985 | /* }====================================================== */ | |||
| 986 | ||||
| 987 | ||||
| 988 | ||||
| 989 | /* | |||
| 990 | ** {====================================================== | |||
| 991 | ** STRING FORMAT | |||
| 992 | ** ======================================================= | |||
| 993 | */ | |||
| 994 | ||||
| 995 | #if !defined(lua_number2strx) /* { */ | |||
| 996 | ||||
| 997 | /* | |||
| 998 | ** Hexadecimal floating-point formatter | |||
| 999 | */ | |||
| 1000 | ||||
| 1001 | #define SIZELENMOD (sizeof(LUA_NUMBER_FRMLEN)/sizeof(char)) | |||
| 1002 | ||||
| 1003 | ||||
| 1004 | /* | |||
| 1005 | ** Number of bits that goes into the first digit. It can be any value | |||
| 1006 | ** between 1 and 4; the following definition tries to align the number | |||
| 1007 | ** to nibble boundaries by making what is left after that first digit a | |||
| 1008 | ** multiple of 4. | |||
| 1009 | */ | |||
| 1010 | #define L_NBFD ((l_floatatt(MANT_DIG) - 1)%4 + 1) | |||
| 1011 | ||||
| 1012 | ||||
| 1013 | /* | |||
| 1014 | ** Add integer part of 'x' to buffer and return new 'x' | |||
| 1015 | */ | |||
| 1016 | static lua_Number adddigit (char *buff, int n, lua_Number x) { | |||
| 1017 | lua_Number dd = l_mathop(floor)(x); /* get integer part from 'x' */ | |||
| 1018 | int d = (int)dd; | |||
| 1019 | buff[n] = (d < 10 ? d + '0' : d - 10 + 'a'); /* add to buffer */ | |||
| 1020 | return x - dd; /* return what is left */ | |||
| 1021 | } | |||
| 1022 | ||||
| 1023 | ||||
| 1024 | static int num2straux (char *buff, int sz, lua_Number x) { | |||
| 1025 | /* if 'inf' or 'NaN', format it like '%g' */ | |||
| 1026 | if (x != x || x == (lua_Number)HUGE_VAL || x == -(lua_Number)HUGE_VAL) | |||
| 1027 | return l_sprintf(buff, sz, LUA_NUMBER_FMT, (LUAI_UACNUMBER)x); | |||
| 1028 | else if (x == 0) { /* can be -0... */ | |||
| 1029 | /* create "0" or "-0" followed by exponent */ | |||
| 1030 | return l_sprintf(buff, sz, LUA_NUMBER_FMT "x0p+0", (LUAI_UACNUMBER)x); | |||
| 1031 | } | |||
| 1032 | else { | |||
| 1033 | int e; | |||
| 1034 | lua_Number m = l_mathop(frexp)(x, &e); /* 'x' fraction and exponent */ | |||
| 1035 | int n = 0; /* character count */ | |||
| 1036 | if (m < 0) { /* is number negative? */ | |||
| 1037 | buff[n++] = '-'; /* add sign */ | |||
| 1038 | m = -m; /* make it positive */ | |||
| 1039 | } | |||
| 1040 | buff[n++] = '0'; buff[n++] = 'x'; /* add "0x" */ | |||
| 1041 | m = adddigit(buff, n++, m * (1 << L_NBFD)); /* add first digit */ | |||
| 1042 | e -= L_NBFD; /* this digit goes before the radix point */ | |||
| 1043 | if (m > 0) { /* more digits? */ | |||
| 1044 | buff[n++] = lua_getlocaledecpoint(); /* add radix point */ | |||
| 1045 | do { /* add as many digits as needed */ | |||
| 1046 | m = adddigit(buff, n++, m * 16); | |||
| 1047 | } while (m > 0); | |||
| 1048 | } | |||
| 1049 | n += l_sprintf(buff + n, sz - n, "p%+d", e); /* add exponent */ | |||
| 1050 | lua_assert(n < sz); | |||
| 1051 | return n; | |||
| 1052 | } | |||
| 1053 | } | |||
| 1054 | ||||
| 1055 | ||||
| 1056 | static int lua_number2strx (lua_State *L, char *buff, int sz, | |||
| 1057 | const char *fmt, lua_Number x) { | |||
| 1058 | int n = num2straux(buff, sz, x); | |||
| 1059 | if (fmt[SIZELENMOD] == 'A') { | |||
| 1060 | int i; | |||
| 1061 | for (i = 0; i < n; i++) | |||
| 1062 | buff[i] = toupper(uchar(buff[i])); | |||
| 1063 | } | |||
| 1064 | else if (l_unlikely(fmt[SIZELENMOD] != 'a')) | |||
| 1065 | return luaL_error(L, "modifiers for format '%%a'/'%%A' not implemented"); | |||
| 1066 | return n; | |||
| 1067 | } | |||
| 1068 | ||||
| 1069 | #endif /* } */ | |||
| 1070 | ||||
| 1071 | ||||
| 1072 | /* | |||
| 1073 | ** Maximum size for items formatted with '%f'. This size is produced | |||
| 1074 | ** by format('%.99f', -maxfloat), and is equal to 99 + 3 ('-', '.', | |||
| 1075 | ** and '\0') + number of decimal digits to represent maxfloat (which | |||
| 1076 | ** is maximum exponent + 1). (99+3+1, adding some extra, 110) | |||
| 1077 | */ | |||
| 1078 | #define MAX_ITEMF (110 + l_floatatt(MAX_10_EXP)) | |||
| 1079 | ||||
| 1080 | ||||
| 1081 | /* | |||
| 1082 | ** All formats except '%f' do not need that large limit. The other | |||
| 1083 | ** float formats use exponents, so that they fit in the 99 limit for | |||
| 1084 | ** significant digits; 's' for large strings and 'q' add items directly | |||
| 1085 | ** to the buffer; all integer formats also fit in the 99 limit. The | |||
| 1086 | ** worst case are floats: they may need 99 significant digits, plus | |||
| 1087 | ** '0x', '-', '.', 'e+XXXX', and '\0'. Adding some extra, 120. | |||
| 1088 | */ | |||
| 1089 | #define MAX_ITEM 120 | |||
| 1090 | ||||
| 1091 | ||||
| 1092 | /* valid flags in a format specification */ | |||
| 1093 | #if !defined(L_FMTFLAGSF) | |||
| 1094 | ||||
| 1095 | /* valid flags for a, A, e, E, f, F, g, and G conversions */ | |||
| 1096 | #define L_FMTFLAGSF "-+#0 " | |||
| 1097 | ||||
| 1098 | /* valid flags for o, x, and X conversions */ | |||
| 1099 | #define L_FMTFLAGSX "-#0" | |||
| 1100 | ||||
| 1101 | /* valid flags for d and i conversions */ | |||
| 1102 | #define L_FMTFLAGSI "-+0 " | |||
| 1103 | ||||
| 1104 | /* valid flags for u conversions */ | |||
| 1105 | #define L_FMTFLAGSU "-0" | |||
| 1106 | ||||
| 1107 | /* valid flags for c, p, and s conversions */ | |||
| 1108 | #define L_FMTFLAGSC "-" | |||
| 1109 | ||||
| 1110 | #endif | |||
| 1111 | ||||
| 1112 | ||||
| 1113 | /* | |||
| 1114 | ** Maximum size of each format specification (such as "%-099.99d"): | |||
| 1115 | ** Initial '%', flags (up to 5), width (2), period, precision (2), | |||
| 1116 | ** length modifier (8), conversion specifier, and final '\0', plus some | |||
| 1117 | ** extra. | |||
| 1118 | */ | |||
| 1119 | #define MAX_FORMAT 32 | |||
| 1120 | ||||
| 1121 | ||||
| 1122 | ✗ | static void addquoted (luaL_Buffer *b, const char *s, size_t len) { | ||
| 1123 | ✗ | luaL_addchar(b, '"'); | ||
| 1124 | ✗ | while (len--) { | ||
| 1125 | ✗ | if (*s == '"' || *s == '\\' || *s == '\n') { | ||
| 1126 | ✗ | luaL_addchar(b, '\\'); | ||
| 1127 | ✗ | luaL_addchar(b, *s); | ||
| 1128 | } | |||
| 1129 | ✗ | else if (iscntrl(uchar(*s))) { | ||
| 1130 | char buff[10]; | |||
| 1131 | ✗ | if (!isdigit(uchar(*(s+1)))) | ||
| 1132 | ✗ | l_sprintf(buff, sizeof(buff), "\\%d", (int)uchar(*s)); | ||
| 1133 | else | |||
| 1134 | ✗ | l_sprintf(buff, sizeof(buff), "\\%03d", (int)uchar(*s)); | ||
| 1135 | ✗ | luaL_addstring(b, buff); | ||
| 1136 | } | |||
| 1137 | else | |||
| 1138 | ✗ | luaL_addchar(b, *s); | ||
| 1139 | ✗ | s++; | ||
| 1140 | } | |||
| 1141 | ✗ | luaL_addchar(b, '"'); | ||
| 1142 | ✗ | } | ||
| 1143 | ||||
| 1144 | ||||
| 1145 | /* | |||
| 1146 | ** Serialize a floating-point number in such a way that it can be | |||
| 1147 | ** scanned back by Lua. Use hexadecimal format for "common" numbers | |||
| 1148 | ** (to preserve precision); inf, -inf, and NaN are handled separately. | |||
| 1149 | ** (NaN cannot be expressed as a numeral, so we write '(0/0)' for it.) | |||
| 1150 | */ | |||
| 1151 | ✗ | static int quotefloat (lua_State *L, char *buff, lua_Number n) { | ||
| 1152 | const char *s; /* for the fixed representations */ | |||
| 1153 | ✗ | if (n == (lua_Number)HUGE_VAL) /* inf? */ | ||
| 1154 | ✗ | s = "1e9999"; | ||
| 1155 | ✗ | else if (n == -(lua_Number)HUGE_VAL) /* -inf? */ | ||
| 1156 | ✗ | s = "-1e9999"; | ||
| 1157 | ✗ | else if (n != n) /* NaN? */ | ||
| 1158 | ✗ | s = "(0/0)"; | ||
| 1159 | else { /* format number as hexadecimal */ | |||
| 1160 | ✗ | int nb = lua_number2strx(L, buff, MAX_ITEM, | ||
| 1161 | "%" LUA_NUMBER_FRMLEN "a", n); | |||
| 1162 | /* ensures that 'buff' string uses a dot as the radix character */ | |||
| 1163 | ✗ | if (memchr(buff, '.', nb) == NULL) { /* no dot? */ | ||
| 1164 | ✗ | char point = lua_getlocaledecpoint(); /* try locale point */ | ||
| 1165 | ✗ | char *ppoint = (char *)memchr(buff, point, nb); | ||
| 1166 | ✗ | if (ppoint) *ppoint = '.'; /* change it to a dot */ | ||
| 1167 | } | |||
| 1168 | ✗ | return nb; | ||
| 1169 | } | |||
| 1170 | /* for the fixed representations */ | |||
| 1171 | ✗ | return l_sprintf(buff, MAX_ITEM, "%s", s); | ||
| 1172 | } | |||
| 1173 | ||||
| 1174 | ||||
| 1175 | ✗ | static void addliteral (lua_State *L, luaL_Buffer *b, int arg) { | ||
| 1176 | ✗ | switch (lua_type(L, arg)) { | ||
| 1177 | ✗ | case LUA_TSTRING: { | ||
| 1178 | size_t len; | |||
| 1179 | ✗ | const char *s = lua_tolstring(L, arg, &len); | ||
| 1180 | ✗ | addquoted(b, s, len); | ||
| 1181 | ✗ | break; | ||
| 1182 | } | |||
| 1183 | ✗ | case LUA_TNUMBER: { | ||
| 1184 | ✗ | char *buff = luaL_prepbuffsize(b, MAX_ITEM); | ||
| 1185 | int nb; | |||
| 1186 | ✗ | if (!lua_isinteger(L, arg)) /* float? */ | ||
| 1187 | ✗ | nb = quotefloat(L, buff, lua_tonumber(L, arg)); | ||
| 1188 | else { /* integers */ | |||
| 1189 | ✗ | lua_Integer n = lua_tointeger(L, arg); | ||
| 1190 | ✗ | const char *format = (n == LUA_MININTEGER) /* corner case? */ | ||
| 1191 | ? "0x%" LUA_INTEGER_FRMLEN "x" /* use hex */ | |||
| 1192 | ✗ | : LUA_INTEGER_FMT; /* else use default format */ | ||
| 1193 | ✗ | nb = l_sprintf(buff, MAX_ITEM, format, (LUAI_UACINT)n); | ||
| 1194 | } | |||
| 1195 | ✗ | luaL_addsize(b, nb); | ||
| 1196 | ✗ | break; | ||
| 1197 | } | |||
| 1198 | ✗ | case LUA_TNIL: case LUA_TBOOLEAN: { | ||
| 1199 | ✗ | luaL_tolstring(L, arg, NULL); | ||
| 1200 | ✗ | luaL_addvalue(b); | ||
| 1201 | ✗ | break; | ||
| 1202 | } | |||
| 1203 | ✗ | default: { | ||
| 1204 | ✗ | luaL_argerror(L, arg, "value has no literal form"); | ||
| 1205 | } | |||
| 1206 | } | |||
| 1207 | ✗ | } | ||
| 1208 | ||||
| 1209 | ||||
| 1210 | ✗ | static const char *get2digits (const char *s) { | ||
| 1211 | ✗ | if (isdigit(uchar(*s))) { | ||
| 1212 | ✗ | s++; | ||
| 1213 | ✗ | if (isdigit(uchar(*s))) s++; /* (2 digits at most) */ | ||
| 1214 | } | |||
| 1215 | ✗ | return s; | ||
| 1216 | } | |||
| 1217 | ||||
| 1218 | ||||
| 1219 | /* | |||
| 1220 | ** Check whether a conversion specification is valid. When called, | |||
| 1221 | ** first character in 'form' must be '%' and last character must | |||
| 1222 | ** be a valid conversion specifier. 'flags' are the accepted flags; | |||
| 1223 | ** 'precision' signals whether to accept a precision. | |||
| 1224 | */ | |||
| 1225 | ✗ | static void checkformat (lua_State *L, const char *form, const char *flags, | ||
| 1226 | int precision) { | |||
| 1227 | ✗ | const char *spec = form + 1; /* skip '%' */ | ||
| 1228 | ✗ | spec += strspn(spec, flags); /* skip flags */ | ||
| 1229 | ✗ | if (*spec != '0') { /* a width cannot start with '0' */ | ||
| 1230 | ✗ | spec = get2digits(spec); /* skip width */ | ||
| 1231 | ✗ | if (*spec == '.' && precision) { | ||
| 1232 | ✗ | spec++; | ||
| 1233 | ✗ | spec = get2digits(spec); /* skip precision */ | ||
| 1234 | } | |||
| 1235 | } | |||
| 1236 | ✗ | if (!isalpha(uchar(*spec))) /* did not go to the end? */ | ||
| 1237 | ✗ | luaL_error(L, "invalid conversion specification: '%s'", form); | ||
| 1238 | ✗ | } | ||
| 1239 | ||||
| 1240 | ||||
| 1241 | /* | |||
| 1242 | ** Get a conversion specification and copy it to 'form'. | |||
| 1243 | ** Return the address of its last character. | |||
| 1244 | */ | |||
| 1245 | ✗ | static const char *getformat (lua_State *L, const char *strfrmt, | ||
| 1246 | char *form) { | |||
| 1247 | /* spans flags, width, and precision ('0' is included as a flag) */ | |||
| 1248 | ✗ | size_t len = strspn(strfrmt, L_FMTFLAGSF "123456789."); | ||
| 1249 | ✗ | len++; /* adds following character (should be the specifier) */ | ||
| 1250 | /* still needs space for '%', '\0', plus a length modifier */ | |||
| 1251 | ✗ | if (len >= MAX_FORMAT - 10) | ||
| 1252 | ✗ | luaL_error(L, "invalid format (too long)"); | ||
| 1253 | ✗ | *(form++) = '%'; | ||
| 1254 | ✗ | memcpy(form, strfrmt, len * sizeof(char)); | ||
| 1255 | ✗ | *(form + len) = '\0'; | ||
| 1256 | ✗ | return strfrmt + len - 1; | ||
| 1257 | } | |||
| 1258 | ||||
| 1259 | ||||
| 1260 | /* | |||
| 1261 | ** add length modifier into formats | |||
| 1262 | */ | |||
| 1263 | ✗ | static void addlenmod (char *form, const char *lenmod) { | ||
| 1264 | ✗ | size_t l = strlen(form); | ||
| 1265 | ✗ | size_t lm = strlen(lenmod); | ||
| 1266 | ✗ | char spec = form[l - 1]; | ||
| 1267 | ✗ | strcpy(form + l - 1, lenmod); | ||
| 1268 | ✗ | form[l + lm - 1] = spec; | ||
| 1269 | ✗ | form[l + lm] = '\0'; | ||
| 1270 | ✗ | } | ||
| 1271 | ||||
| 1272 | ||||
| 1273 | ✗ | static int str_format (lua_State *L) { | ||
| 1274 | ✗ | int top = lua_gettop(L); | ||
| 1275 | ✗ | int arg = 1; | ||
| 1276 | size_t sfl; | |||
| 1277 | ✗ | const char *strfrmt = luaL_checklstring(L, arg, &sfl); | ||
| 1278 | ✗ | const char *strfrmt_end = strfrmt+sfl; | ||
| 1279 | const char *flags; | |||
| 1280 | luaL_Buffer b; | |||
| 1281 | ✗ | luaL_buffinit(L, &b); | ||
| 1282 | ✗ | while (strfrmt < strfrmt_end) { | ||
| 1283 | ✗ | if (*strfrmt != L_ESC) | ||
| 1284 | ✗ | luaL_addchar(&b, *strfrmt++); | ||
| 1285 | ✗ | else if (*++strfrmt == L_ESC) | ||
| 1286 | ✗ | luaL_addchar(&b, *strfrmt++); /* %% */ | ||
| 1287 | else { /* format item */ | |||
| 1288 | char form[MAX_FORMAT]; /* to store the format ('%...') */ | |||
| 1289 | ✗ | int maxitem = MAX_ITEM; /* maximum length for the result */ | ||
| 1290 | ✗ | char *buff = luaL_prepbuffsize(&b, maxitem); /* to put result */ | ||
| 1291 | ✗ | int nb = 0; /* number of bytes in result */ | ||
| 1292 | ✗ | if (++arg > top) | ||
| 1293 | ✗ | return luaL_argerror(L, arg, "no value"); | ||
| 1294 | ✗ | strfrmt = getformat(L, strfrmt, form); | ||
| 1295 | ✗ | switch (*strfrmt++) { | ||
| 1296 | ✗ | case 'c': { | ||
| 1297 | ✗ | checkformat(L, form, L_FMTFLAGSC, 0); | ||
| 1298 | ✗ | nb = l_sprintf(buff, maxitem, form, (int)luaL_checkinteger(L, arg)); | ||
| 1299 | ✗ | break; | ||
| 1300 | } | |||
| 1301 | ✗ | case 'd': case 'i': | ||
| 1302 | ✗ | flags = L_FMTFLAGSI; | ||
| 1303 | ✗ | goto intcase; | ||
| 1304 | ✗ | case 'u': | ||
| 1305 | ✗ | flags = L_FMTFLAGSU; | ||
| 1306 | ✗ | goto intcase; | ||
| 1307 | ✗ | case 'o': case 'x': case 'X': | ||
| 1308 | ✗ | flags = L_FMTFLAGSX; | ||
| 1309 | ✗ | intcase: { | ||
| 1310 | ✗ | lua_Integer n = luaL_checkinteger(L, arg); | ||
| 1311 | ✗ | checkformat(L, form, flags, 1); | ||
| 1312 | ✗ | addlenmod(form, LUA_INTEGER_FRMLEN); | ||
| 1313 | ✗ | nb = l_sprintf(buff, maxitem, form, (LUAI_UACINT)n); | ||
| 1314 | ✗ | break; | ||
| 1315 | } | |||
| 1316 | ✗ | case 'a': case 'A': | ||
| 1317 | ✗ | checkformat(L, form, L_FMTFLAGSF, 1); | ||
| 1318 | ✗ | addlenmod(form, LUA_NUMBER_FRMLEN); | ||
| 1319 | ✗ | nb = lua_number2strx(L, buff, maxitem, form, | ||
| 1320 | luaL_checknumber(L, arg)); | |||
| 1321 | ✗ | break; | ||
| 1322 | ✗ | case 'f': | ||
| 1323 | ✗ | maxitem = MAX_ITEMF; /* extra space for '%f' */ | ||
| 1324 | ✗ | buff = luaL_prepbuffsize(&b, maxitem); | ||
| 1325 | /* FALLTHROUGH */ | |||
| 1326 | ✗ | case 'e': case 'E': case 'g': case 'G': { | ||
| 1327 | ✗ | lua_Number n = luaL_checknumber(L, arg); | ||
| 1328 | ✗ | checkformat(L, form, L_FMTFLAGSF, 1); | ||
| 1329 | ✗ | addlenmod(form, LUA_NUMBER_FRMLEN); | ||
| 1330 | ✗ | nb = l_sprintf(buff, maxitem, form, (LUAI_UACNUMBER)n); | ||
| 1331 | ✗ | break; | ||
| 1332 | } | |||
| 1333 | ✗ | case 'p': { | ||
| 1334 | ✗ | const void *p = lua_topointer(L, arg); | ||
| 1335 | ✗ | checkformat(L, form, L_FMTFLAGSC, 0); | ||
| 1336 | ✗ | if (p == NULL) { /* avoid calling 'printf' with argument NULL */ | ||
| 1337 | ✗ | p = "(null)"; /* result */ | ||
| 1338 | ✗ | form[strlen(form) - 1] = 's'; /* format it as a string */ | ||
| 1339 | } | |||
| 1340 | ✗ | nb = l_sprintf(buff, maxitem, form, p); | ||
| 1341 | ✗ | break; | ||
| 1342 | } | |||
| 1343 | ✗ | case 'q': { | ||
| 1344 | ✗ | if (form[2] != '\0') /* modifiers? */ | ||
| 1345 | ✗ | return luaL_error(L, "specifier '%%q' cannot have modifiers"); | ||
| 1346 | ✗ | addliteral(L, &b, arg); | ||
| 1347 | ✗ | break; | ||
| 1348 | } | |||
| 1349 | ✗ | case 's': { | ||
| 1350 | size_t l; | |||
| 1351 | ✗ | const char *s = luaL_tolstring(L, arg, &l); | ||
| 1352 | ✗ | if (form[2] == '\0') /* no modifiers? */ | ||
| 1353 | ✗ | luaL_addvalue(&b); /* keep entire string */ | ||
| 1354 | else { | |||
| 1355 | ✗ | luaL_argcheck(L, l == strlen(s), arg, "string contains zeros"); | ||
| 1356 | ✗ | checkformat(L, form, L_FMTFLAGSC, 1); | ||
| 1357 | ✗ | if (strchr(form, '.') == NULL && l >= 100) { | ||
| 1358 | /* no precision and string is too long to be formatted */ | |||
| 1359 | ✗ | luaL_addvalue(&b); /* keep entire string */ | ||
| 1360 | } | |||
| 1361 | else { /* format the string into 'buff' */ | |||
| 1362 | ✗ | nb = l_sprintf(buff, maxitem, form, s); | ||
| 1363 | ✗ | lua_pop(L, 1); /* remove result from 'luaL_tolstring' */ | ||
| 1364 | } | |||
| 1365 | } | |||
| 1366 | ✗ | break; | ||
| 1367 | } | |||
| 1368 | ✗ | default: { /* also treat cases 'pnLlh' */ | ||
| 1369 | ✗ | return luaL_error(L, "invalid conversion '%s' to 'format'", form); | ||
| 1370 | } | |||
| 1371 | } | |||
| 1372 | lua_assert(nb < maxitem); | |||
| 1373 | ✗ | luaL_addsize(&b, nb); | ||
| 1374 | } | |||
| 1375 | } | |||
| 1376 | ✗ | luaL_pushresult(&b); | ||
| 1377 | ✗ | return 1; | ||
| 1378 | } | |||
| 1379 | ||||
| 1380 | /* }====================================================== */ | |||
| 1381 | ||||
| 1382 | ||||
| 1383 | /* | |||
| 1384 | ** {====================================================== | |||
| 1385 | ** PACK/UNPACK | |||
| 1386 | ** ======================================================= | |||
| 1387 | */ | |||
| 1388 | ||||
| 1389 | ||||
| 1390 | /* value used for padding */ | |||
| 1391 | #if !defined(LUAL_PACKPADBYTE) | |||
| 1392 | #define LUAL_PACKPADBYTE 0x00 | |||
| 1393 | #endif | |||
| 1394 | ||||
| 1395 | /* maximum size for the binary representation of an integer */ | |||
| 1396 | #define MAXINTSIZE 16 | |||
| 1397 | ||||
| 1398 | /* number of bits in a character */ | |||
| 1399 | #define NB CHAR_BIT | |||
| 1400 | ||||
| 1401 | /* mask for one character (NB 1's) */ | |||
| 1402 | #define MC ((1 << NB) - 1) | |||
| 1403 | ||||
| 1404 | /* size of a lua_Integer */ | |||
| 1405 | #define SZINT ((int)sizeof(lua_Integer)) | |||
| 1406 | ||||
| 1407 | ||||
| 1408 | /* dummy union to get native endianness */ | |||
| 1409 | static const union { | |||
| 1410 | int dummy; | |||
| 1411 | char little; /* true iff machine is little endian */ | |||
| 1412 | } nativeendian = {1}; | |||
| 1413 | ||||
| 1414 | ||||
| 1415 | /* | |||
| 1416 | ** information to pack/unpack stuff | |||
| 1417 | */ | |||
| 1418 | typedef struct Header { | |||
| 1419 | lua_State *L; | |||
| 1420 | int islittle; | |||
| 1421 | int maxalign; | |||
| 1422 | } Header; | |||
| 1423 | ||||
| 1424 | ||||
| 1425 | /* | |||
| 1426 | ** options for pack/unpack | |||
| 1427 | */ | |||
| 1428 | typedef enum KOption { | |||
| 1429 | Kint, /* signed integers */ | |||
| 1430 | Kuint, /* unsigned integers */ | |||
| 1431 | Kfloat, /* single-precision floating-point numbers */ | |||
| 1432 | Knumber, /* Lua "native" floating-point numbers */ | |||
| 1433 | Kdouble, /* double-precision floating-point numbers */ | |||
| 1434 | Kchar, /* fixed-length strings */ | |||
| 1435 | Kstring, /* strings with prefixed length */ | |||
| 1436 | Kzstr, /* zero-terminated strings */ | |||
| 1437 | Kpadding, /* padding */ | |||
| 1438 | Kpaddalign, /* padding for alignment */ | |||
| 1439 | Knop /* no-op (configuration or spaces) */ | |||
| 1440 | } KOption; | |||
| 1441 | ||||
| 1442 | ||||
| 1443 | /* | |||
| 1444 | ** Read an integer numeral from string 'fmt' or return 'df' if | |||
| 1445 | ** there is no numeral | |||
| 1446 | */ | |||
| 1447 | ✗ | static int digit (int c) { return '0' <= c && c <= '9'; } | ||
| 1448 | ||||
| 1449 | ✗ | static int getnum (const char **fmt, int df) { | ||
| 1450 | ✗ | if (!digit(**fmt)) /* no number? */ | ||
| 1451 | ✗ | return df; /* return default value */ | ||
| 1452 | else { | |||
| 1453 | ✗ | int a = 0; | ||
| 1454 | do { | |||
| 1455 | ✗ | a = a*10 + (*((*fmt)++) - '0'); | ||
| 1456 | ✗ | } while (digit(**fmt) && a <= ((int)MAXSIZE - 9)/10); | ||
| 1457 | ✗ | return a; | ||
| 1458 | } | |||
| 1459 | } | |||
| 1460 | ||||
| 1461 | ||||
| 1462 | /* | |||
| 1463 | ** Read an integer numeral and raises an error if it is larger | |||
| 1464 | ** than the maximum size for integers. | |||
| 1465 | */ | |||
| 1466 | ✗ | static int getnumlimit (Header *h, const char **fmt, int df) { | ||
| 1467 | ✗ | int sz = getnum(fmt, df); | ||
| 1468 | ✗ | if (l_unlikely(sz > MAXINTSIZE || sz <= 0)) | ||
| 1469 | ✗ | return luaL_error(h->L, "integral size (%d) out of limits [1,%d]", | ||
| 1470 | sz, MAXINTSIZE); | |||
| 1471 | ✗ | return sz; | ||
| 1472 | } | |||
| 1473 | ||||
| 1474 | ||||
| 1475 | /* | |||
| 1476 | ** Initialize Header | |||
| 1477 | */ | |||
| 1478 | ✗ | static void initheader (lua_State *L, Header *h) { | ||
| 1479 | ✗ | h->L = L; | ||
| 1480 | ✗ | h->islittle = nativeendian.little; | ||
| 1481 | ✗ | h->maxalign = 1; | ||
| 1482 | ✗ | } | ||
| 1483 | ||||
| 1484 | ||||
| 1485 | /* | |||
| 1486 | ** Read and classify next option. 'size' is filled with option's size. | |||
| 1487 | */ | |||
| 1488 | ✗ | static KOption getoption (Header *h, const char **fmt, int *size) { | ||
| 1489 | /* dummy structure to get native alignment requirements */ | |||
| 1490 | struct cD { char c; union { LUAI_MAXALIGN; } u; }; | |||
| 1491 | ✗ | int opt = *((*fmt)++); | ||
| 1492 | ✗ | *size = 0; /* default */ | ||
| 1493 | ✗ | switch (opt) { | ||
| 1494 | ✗ | case 'b': *size = sizeof(char); return Kint; | ||
| 1495 | ✗ | case 'B': *size = sizeof(char); return Kuint; | ||
| 1496 | ✗ | case 'h': *size = sizeof(short); return Kint; | ||
| 1497 | ✗ | case 'H': *size = sizeof(short); return Kuint; | ||
| 1498 | ✗ | case 'l': *size = sizeof(long); return Kint; | ||
| 1499 | ✗ | case 'L': *size = sizeof(long); return Kuint; | ||
| 1500 | ✗ | case 'j': *size = sizeof(lua_Integer); return Kint; | ||
| 1501 | ✗ | case 'J': *size = sizeof(lua_Integer); return Kuint; | ||
| 1502 | ✗ | case 'T': *size = sizeof(size_t); return Kuint; | ||
| 1503 | ✗ | case 'f': *size = sizeof(float); return Kfloat; | ||
| 1504 | ✗ | case 'n': *size = sizeof(lua_Number); return Knumber; | ||
| 1505 | ✗ | case 'd': *size = sizeof(double); return Kdouble; | ||
| 1506 | ✗ | case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint; | ||
| 1507 | ✗ | case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint; | ||
| 1508 | ✗ | case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring; | ||
| 1509 | ✗ | case 'c': | ||
| 1510 | ✗ | *size = getnum(fmt, -1); | ||
| 1511 | ✗ | if (l_unlikely(*size == -1)) | ||
| 1512 | ✗ | luaL_error(h->L, "missing size for format option 'c'"); | ||
| 1513 | ✗ | return Kchar; | ||
| 1514 | ✗ | case 'z': return Kzstr; | ||
| 1515 | ✗ | case 'x': *size = 1; return Kpadding; | ||
| 1516 | ✗ | case 'X': return Kpaddalign; | ||
| 1517 | ✗ | case ' ': break; | ||
| 1518 | ✗ | case '<': h->islittle = 1; break; | ||
| 1519 | ✗ | case '>': h->islittle = 0; break; | ||
| 1520 | ✗ | case '=': h->islittle = nativeendian.little; break; | ||
| 1521 | ✗ | case '!': { | ||
| 1522 | ✗ | const int maxalign = offsetof(struct cD, u); | ||
| 1523 | ✗ | h->maxalign = getnumlimit(h, fmt, maxalign); | ||
| 1524 | ✗ | break; | ||
| 1525 | } | |||
| 1526 | ✗ | default: luaL_error(h->L, "invalid format option '%c'", opt); | ||
| 1527 | } | |||
| 1528 | ✗ | return Knop; | ||
| 1529 | } | |||
| 1530 | ||||
| 1531 | ||||
| 1532 | /* | |||
| 1533 | ** Read, classify, and fill other details about the next option. | |||
| 1534 | ** 'psize' is filled with option's size, 'notoalign' with its | |||
| 1535 | ** alignment requirements. | |||
| 1536 | ** Local variable 'size' gets the size to be aligned. (Kpadal option | |||
| 1537 | ** always gets its full alignment, other options are limited by | |||
| 1538 | ** the maximum alignment ('maxalign'). Kchar option needs no alignment | |||
| 1539 | ** despite its size. | |||
| 1540 | */ | |||
| 1541 | ✗ | static KOption getdetails (Header *h, size_t totalsize, | ||
| 1542 | const char **fmt, int *psize, int *ntoalign) { | |||
| 1543 | ✗ | KOption opt = getoption(h, fmt, psize); | ||
| 1544 | ✗ | int align = *psize; /* usually, alignment follows size */ | ||
| 1545 | ✗ | if (opt == Kpaddalign) { /* 'X' gets alignment from following option */ | ||
| 1546 | ✗ | if (**fmt == '\0' || getoption(h, fmt, &align) == Kchar || align == 0) | ||
| 1547 | ✗ | luaL_argerror(h->L, 1, "invalid next option for option 'X'"); | ||
| 1548 | } | |||
| 1549 | ✗ | if (align <= 1 || opt == Kchar) /* need no alignment? */ | ||
| 1550 | ✗ | *ntoalign = 0; | ||
| 1551 | else { | |||
| 1552 | ✗ | if (align > h->maxalign) /* enforce maximum alignment */ | ||
| 1553 | ✗ | align = h->maxalign; | ||
| 1554 | ✗ | if (l_unlikely((align & (align - 1)) != 0)) /* not a power of 2? */ | ||
| 1555 | ✗ | luaL_argerror(h->L, 1, "format asks for alignment not power of 2"); | ||
| 1556 | ✗ | *ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1); | ||
| 1557 | } | |||
| 1558 | ✗ | return opt; | ||
| 1559 | } | |||
| 1560 | ||||
| 1561 | ||||
| 1562 | /* | |||
| 1563 | ** Pack integer 'n' with 'size' bytes and 'islittle' endianness. | |||
| 1564 | ** The final 'if' handles the case when 'size' is larger than | |||
| 1565 | ** the size of a Lua integer, correcting the extra sign-extension | |||
| 1566 | ** bytes if necessary (by default they would be zeros). | |||
| 1567 | */ | |||
| 1568 | ✗ | static void packint (luaL_Buffer *b, lua_Unsigned n, | ||
| 1569 | int islittle, int size, int neg) { | |||
| 1570 | ✗ | char *buff = luaL_prepbuffsize(b, size); | ||
| 1571 | int i; | |||
| 1572 | ✗ | buff[islittle ? 0 : size - 1] = (char)(n & MC); /* first byte */ | ||
| 1573 | ✗ | for (i = 1; i < size; i++) { | ||
| 1574 | ✗ | n >>= NB; | ||
| 1575 | ✗ | buff[islittle ? i : size - 1 - i] = (char)(n & MC); | ||
| 1576 | } | |||
| 1577 | ✗ | if (neg && size > SZINT) { /* negative number need sign extension? */ | ||
| 1578 | ✗ | for (i = SZINT; i < size; i++) /* correct extra bytes */ | ||
| 1579 | ✗ | buff[islittle ? i : size - 1 - i] = (char)MC; | ||
| 1580 | } | |||
| 1581 | ✗ | luaL_addsize(b, size); /* add result to buffer */ | ||
| 1582 | ✗ | } | ||
| 1583 | ||||
| 1584 | ||||
| 1585 | /* | |||
| 1586 | ** Copy 'size' bytes from 'src' to 'dest', correcting endianness if | |||
| 1587 | ** given 'islittle' is different from native endianness. | |||
| 1588 | */ | |||
| 1589 | ✗ | static void copywithendian (char *dest, const char *src, | ||
| 1590 | int size, int islittle) { | |||
| 1591 | ✗ | if (islittle == nativeendian.little) | ||
| 1592 | ✗ | memcpy(dest, src, size); | ||
| 1593 | else { | |||
| 1594 | ✗ | dest += size - 1; | ||
| 1595 | ✗ | while (size-- != 0) | ||
| 1596 | ✗ | *(dest--) = *(src++); | ||
| 1597 | } | |||
| 1598 | ✗ | } | ||
| 1599 | ||||
| 1600 | ||||
| 1601 | ✗ | static int str_pack (lua_State *L) { | ||
| 1602 | luaL_Buffer b; | |||
| 1603 | Header h; | |||
| 1604 | ✗ | const char *fmt = luaL_checkstring(L, 1); /* format string */ | ||
| 1605 | ✗ | int arg = 1; /* current argument to pack */ | ||
| 1606 | ✗ | size_t totalsize = 0; /* accumulate total size of result */ | ||
| 1607 | ✗ | initheader(L, &h); | ||
| 1608 | ✗ | lua_pushnil(L); /* mark to separate arguments from string buffer */ | ||
| 1609 | ✗ | luaL_buffinit(L, &b); | ||
| 1610 | ✗ | while (*fmt != '\0') { | ||
| 1611 | int size, ntoalign; | |||
| 1612 | ✗ | KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign); | ||
| 1613 | ✗ | totalsize += ntoalign + size; | ||
| 1614 | ✗ | while (ntoalign-- > 0) | ||
| 1615 | ✗ | luaL_addchar(&b, LUAL_PACKPADBYTE); /* fill alignment */ | ||
| 1616 | ✗ | arg++; | ||
| 1617 | ✗ | switch (opt) { | ||
| 1618 | ✗ | case Kint: { /* signed integers */ | ||
| 1619 | ✗ | lua_Integer n = luaL_checkinteger(L, arg); | ||
| 1620 | ✗ | if (size < SZINT) { /* need overflow check? */ | ||
| 1621 | ✗ | lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1); | ||
| 1622 | ✗ | luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow"); | ||
| 1623 | } | |||
| 1624 | ✗ | packint(&b, (lua_Unsigned)n, h.islittle, size, (n < 0)); | ||
| 1625 | ✗ | break; | ||
| 1626 | } | |||
| 1627 | ✗ | case Kuint: { /* unsigned integers */ | ||
| 1628 | ✗ | lua_Integer n = luaL_checkinteger(L, arg); | ||
| 1629 | ✗ | if (size < SZINT) /* need overflow check? */ | ||
| 1630 | ✗ | luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)), | ||
| 1631 | arg, "unsigned overflow"); | |||
| 1632 | ✗ | packint(&b, (lua_Unsigned)n, h.islittle, size, 0); | ||
| 1633 | ✗ | break; | ||
| 1634 | } | |||
| 1635 | ✗ | case Kfloat: { /* C float */ | ||
| 1636 | ✗ | float f = (float)luaL_checknumber(L, arg); /* get argument */ | ||
| 1637 | ✗ | char *buff = luaL_prepbuffsize(&b, sizeof(f)); | ||
| 1638 | /* move 'f' to final result, correcting endianness if needed */ | |||
| 1639 | ✗ | copywithendian(buff, (char *)&f, sizeof(f), h.islittle); | ||
| 1640 | ✗ | luaL_addsize(&b, size); | ||
| 1641 | ✗ | break; | ||
| 1642 | } | |||
| 1643 | ✗ | case Knumber: { /* Lua float */ | ||
| 1644 | ✗ | lua_Number f = luaL_checknumber(L, arg); /* get argument */ | ||
| 1645 | ✗ | char *buff = luaL_prepbuffsize(&b, sizeof(f)); | ||
| 1646 | /* move 'f' to final result, correcting endianness if needed */ | |||
| 1647 | ✗ | copywithendian(buff, (char *)&f, sizeof(f), h.islittle); | ||
| 1648 | ✗ | luaL_addsize(&b, size); | ||
| 1649 | ✗ | break; | ||
| 1650 | } | |||
| 1651 | ✗ | case Kdouble: { /* C double */ | ||
| 1652 | ✗ | double f = (double)luaL_checknumber(L, arg); /* get argument */ | ||
| 1653 | ✗ | char *buff = luaL_prepbuffsize(&b, sizeof(f)); | ||
| 1654 | /* move 'f' to final result, correcting endianness if needed */ | |||
| 1655 | ✗ | copywithendian(buff, (char *)&f, sizeof(f), h.islittle); | ||
| 1656 | ✗ | luaL_addsize(&b, size); | ||
| 1657 | ✗ | break; | ||
| 1658 | } | |||
| 1659 | ✗ | case Kchar: { /* fixed-size string */ | ||
| 1660 | size_t len; | |||
| 1661 | ✗ | const char *s = luaL_checklstring(L, arg, &len); | ||
| 1662 | ✗ | luaL_argcheck(L, len <= (size_t)size, arg, | ||
| 1663 | "string longer than given size"); | |||
| 1664 | ✗ | luaL_addlstring(&b, s, len); /* add string */ | ||
| 1665 | ✗ | while (len++ < (size_t)size) /* pad extra space */ | ||
| 1666 | ✗ | luaL_addchar(&b, LUAL_PACKPADBYTE); | ||
| 1667 | ✗ | break; | ||
| 1668 | } | |||
| 1669 | ✗ | case Kstring: { /* strings with length count */ | ||
| 1670 | size_t len; | |||
| 1671 | ✗ | const char *s = luaL_checklstring(L, arg, &len); | ||
| 1672 | ✗ | luaL_argcheck(L, size >= (int)sizeof(size_t) || | ||
| 1673 | len < ((size_t)1 << (size * NB)), | |||
| 1674 | arg, "string length does not fit in given size"); | |||
| 1675 | ✗ | packint(&b, (lua_Unsigned)len, h.islittle, size, 0); /* pack length */ | ||
| 1676 | ✗ | luaL_addlstring(&b, s, len); | ||
| 1677 | ✗ | totalsize += len; | ||
| 1678 | ✗ | break; | ||
| 1679 | } | |||
| 1680 | ✗ | case Kzstr: { /* zero-terminated string */ | ||
| 1681 | size_t len; | |||
| 1682 | ✗ | const char *s = luaL_checklstring(L, arg, &len); | ||
| 1683 | ✗ | luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros"); | ||
| 1684 | ✗ | luaL_addlstring(&b, s, len); | ||
| 1685 | ✗ | luaL_addchar(&b, '\0'); /* add zero at the end */ | ||
| 1686 | ✗ | totalsize += len + 1; | ||
| 1687 | ✗ | break; | ||
| 1688 | } | |||
| 1689 | ✗ | case Kpadding: luaL_addchar(&b, LUAL_PACKPADBYTE); /* FALLTHROUGH */ | ||
| 1690 | ✗ | case Kpaddalign: case Knop: | ||
| 1691 | ✗ | arg--; /* undo increment */ | ||
| 1692 | ✗ | break; | ||
| 1693 | } | |||
| 1694 | } | |||
| 1695 | ✗ | luaL_pushresult(&b); | ||
| 1696 | ✗ | return 1; | ||
| 1697 | } | |||
| 1698 | ||||
| 1699 | ||||
| 1700 | ✗ | static int str_packsize (lua_State *L) { | ||
| 1701 | Header h; | |||
| 1702 | ✗ | const char *fmt = luaL_checkstring(L, 1); /* format string */ | ||
| 1703 | ✗ | size_t totalsize = 0; /* accumulate total size of result */ | ||
| 1704 | ✗ | initheader(L, &h); | ||
| 1705 | ✗ | while (*fmt != '\0') { | ||
| 1706 | int size, ntoalign; | |||
| 1707 | ✗ | KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign); | ||
| 1708 | ✗ | luaL_argcheck(L, opt != Kstring && opt != Kzstr, 1, | ||
| 1709 | "variable-length format"); | |||
| 1710 | ✗ | size += ntoalign; /* total space used by option */ | ||
| 1711 | ✗ | luaL_argcheck(L, totalsize <= MAXSIZE - size, 1, | ||
| 1712 | "format result too large"); | |||
| 1713 | ✗ | totalsize += size; | ||
| 1714 | } | |||
| 1715 | ✗ | lua_pushinteger(L, (lua_Integer)totalsize); | ||
| 1716 | ✗ | return 1; | ||
| 1717 | } | |||
| 1718 | ||||
| 1719 | ||||
| 1720 | /* | |||
| 1721 | ** Unpack an integer with 'size' bytes and 'islittle' endianness. | |||
| 1722 | ** If size is smaller than the size of a Lua integer and integer | |||
| 1723 | ** is signed, must do sign extension (propagating the sign to the | |||
| 1724 | ** higher bits); if size is larger than the size of a Lua integer, | |||
| 1725 | ** it must check the unread bytes to see whether they do not cause an | |||
| 1726 | ** overflow. | |||
| 1727 | */ | |||
| 1728 | ✗ | static lua_Integer unpackint (lua_State *L, const char *str, | ||
| 1729 | int islittle, int size, int issigned) { | |||
| 1730 | ✗ | lua_Unsigned res = 0; | ||
| 1731 | int i; | |||
| 1732 | ✗ | int limit = (size <= SZINT) ? size : SZINT; | ||
| 1733 | ✗ | for (i = limit - 1; i >= 0; i--) { | ||
| 1734 | ✗ | res <<= NB; | ||
| 1735 | ✗ | res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i]; | ||
| 1736 | } | |||
| 1737 | ✗ | if (size < SZINT) { /* real size smaller than lua_Integer? */ | ||
| 1738 | ✗ | if (issigned) { /* needs sign extension? */ | ||
| 1739 | ✗ | lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1); | ||
| 1740 | ✗ | res = ((res ^ mask) - mask); /* do sign extension */ | ||
| 1741 | } | |||
| 1742 | } | |||
| 1743 | ✗ | else if (size > SZINT) { /* must check unread bytes */ | ||
| 1744 | ✗ | int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC; | ||
| 1745 | ✗ | for (i = limit; i < size; i++) { | ||
| 1746 | ✗ | if (l_unlikely((unsigned char)str[islittle ? i : size - 1 - i] != mask)) | ||
| 1747 | ✗ | luaL_error(L, "%d-byte integer does not fit into Lua Integer", size); | ||
| 1748 | } | |||
| 1749 | } | |||
| 1750 | ✗ | return (lua_Integer)res; | ||
| 1751 | } | |||
| 1752 | ||||
| 1753 | ||||
| 1754 | ✗ | static int str_unpack (lua_State *L) { | ||
| 1755 | Header h; | |||
| 1756 | ✗ | const char *fmt = luaL_checkstring(L, 1); | ||
| 1757 | size_t ld; | |||
| 1758 | ✗ | const char *data = luaL_checklstring(L, 2, &ld); | ||
| 1759 | ✗ | size_t pos = posrelatI(luaL_optinteger(L, 3, 1), ld) - 1; | ||
| 1760 | ✗ | int n = 0; /* number of results */ | ||
| 1761 | ✗ | luaL_argcheck(L, pos <= ld, 3, "initial position out of string"); | ||
| 1762 | ✗ | initheader(L, &h); | ||
| 1763 | ✗ | while (*fmt != '\0') { | ||
| 1764 | int size, ntoalign; | |||
| 1765 | ✗ | KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign); | ||
| 1766 | ✗ | luaL_argcheck(L, (size_t)ntoalign + size <= ld - pos, 2, | ||
| 1767 | "data string too short"); | |||
| 1768 | ✗ | pos += ntoalign; /* skip alignment */ | ||
| 1769 | /* stack space for item + next position */ | |||
| 1770 | ✗ | luaL_checkstack(L, 2, "too many results"); | ||
| 1771 | ✗ | n++; | ||
| 1772 | ✗ | switch (opt) { | ||
| 1773 | ✗ | case Kint: | ||
| 1774 | case Kuint: { | |||
| 1775 | ✗ | lua_Integer res = unpackint(L, data + pos, h.islittle, size, | ||
| 1776 | (opt == Kint)); | |||
| 1777 | ✗ | lua_pushinteger(L, res); | ||
| 1778 | ✗ | break; | ||
| 1779 | } | |||
| 1780 | ✗ | case Kfloat: { | ||
| 1781 | float f; | |||
| 1782 | ✗ | copywithendian((char *)&f, data + pos, sizeof(f), h.islittle); | ||
| 1783 | ✗ | lua_pushnumber(L, (lua_Number)f); | ||
| 1784 | ✗ | break; | ||
| 1785 | } | |||
| 1786 | ✗ | case Knumber: { | ||
| 1787 | lua_Number f; | |||
| 1788 | ✗ | copywithendian((char *)&f, data + pos, sizeof(f), h.islittle); | ||
| 1789 | ✗ | lua_pushnumber(L, f); | ||
| 1790 | ✗ | break; | ||
| 1791 | } | |||
| 1792 | ✗ | case Kdouble: { | ||
| 1793 | double f; | |||
| 1794 | ✗ | copywithendian((char *)&f, data + pos, sizeof(f), h.islittle); | ||
| 1795 | ✗ | lua_pushnumber(L, (lua_Number)f); | ||
| 1796 | ✗ | break; | ||
| 1797 | } | |||
| 1798 | ✗ | case Kchar: { | ||
| 1799 | ✗ | lua_pushlstring(L, data + pos, size); | ||
| 1800 | ✗ | break; | ||
| 1801 | } | |||
| 1802 | ✗ | case Kstring: { | ||
| 1803 | ✗ | size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0); | ||
| 1804 | ✗ | luaL_argcheck(L, len <= ld - pos - size, 2, "data string too short"); | ||
| 1805 | ✗ | lua_pushlstring(L, data + pos + size, len); | ||
| 1806 | ✗ | pos += len; /* skip string */ | ||
| 1807 | ✗ | break; | ||
| 1808 | } | |||
| 1809 | ✗ | case Kzstr: { | ||
| 1810 | ✗ | size_t len = strlen(data + pos); | ||
| 1811 | ✗ | luaL_argcheck(L, pos + len < ld, 2, | ||
| 1812 | "unfinished string for format 'z'"); | |||
| 1813 | ✗ | lua_pushlstring(L, data + pos, len); | ||
| 1814 | ✗ | pos += len + 1; /* skip string plus final '\0' */ | ||
| 1815 | ✗ | break; | ||
| 1816 | } | |||
| 1817 | ✗ | case Kpaddalign: case Kpadding: case Knop: | ||
| 1818 | ✗ | n--; /* undo increment */ | ||
| 1819 | ✗ | break; | ||
| 1820 | } | |||
| 1821 | ✗ | pos += size; | ||
| 1822 | } | |||
| 1823 | ✗ | lua_pushinteger(L, pos + 1); /* next position */ | ||
| 1824 | ✗ | return n + 1; | ||
| 1825 | } | |||
| 1826 | ||||
| 1827 | /* }====================================================== */ | |||
| 1828 | ||||
| 1829 | ||||
| 1830 | static const luaL_Reg strlib[] = { | |||
| 1831 | {"byte", str_byte}, | |||
| 1832 | {"char", str_char}, | |||
| 1833 | {"dump", str_dump}, | |||
| 1834 | {"find", str_find}, | |||
| 1835 | {"format", str_format}, | |||
| 1836 | {"gmatch", gmatch}, | |||
| 1837 | {"gsub", str_gsub}, | |||
| 1838 | {"len", str_len}, | |||
| 1839 | {"lower", str_lower}, | |||
| 1840 | {"match", str_match}, | |||
| 1841 | {"rep", str_rep}, | |||
| 1842 | {"reverse", str_reverse}, | |||
| 1843 | {"sub", str_sub}, | |||
| 1844 | {"upper", str_upper}, | |||
| 1845 | {"pack", str_pack}, | |||
| 1846 | {"packsize", str_packsize}, | |||
| 1847 | {"unpack", str_unpack}, | |||
| 1848 | {NULL, NULL} | |||
| 1849 | }; | |||
| 1850 | ||||
| 1851 | ||||
| 1852 | ✗ | static void createmetatable (lua_State *L) { | ||
| 1853 | /* table to be metatable for strings */ | |||
| 1854 | ✗ | luaL_newlibtable(L, stringmetamethods); | ||
| 1855 | ✗ | luaL_setfuncs(L, stringmetamethods, 0); | ||
| 1856 | ✗ | lua_pushliteral(L, ""); /* dummy string */ | ||
| 1857 | ✗ | lua_pushvalue(L, -2); /* copy table */ | ||
| 1858 | ✗ | lua_setmetatable(L, -2); /* set table as metatable for strings */ | ||
| 1859 | ✗ | lua_pop(L, 1); /* pop dummy string */ | ||
| 1860 | ✗ | lua_pushvalue(L, -2); /* get string library */ | ||
| 1861 | ✗ | lua_setfield(L, -2, "__index"); /* metatable.__index = string */ | ||
| 1862 | ✗ | lua_pop(L, 1); /* pop metatable */ | ||
| 1863 | ✗ | } | ||
| 1864 | ||||
| 1865 | ||||
| 1866 | /* | |||
| 1867 | ** Open string library | |||
| 1868 | */ | |||
| 1869 | ✗ | LUAMOD_API int luaopen_string (lua_State *L) { | ||
| 1870 | ✗ | luaL_newlib(L, strlib); | ||
| 1871 | ✗ | createmetatable(L); | ||
| 1872 | ✗ | return 1; | ||
| 1873 | } | |||
| 1874 | ||||
| 1875 |