1 /* $NetBSD: lstrlib.c,v 1.18 2018/08/04 17:30:01 alnsn Exp $ */ 2 3 /* 4 ** Id: lstrlib.c,v 1.254.1.1 2017/04/19 17:29:57 roberto Exp 5 ** Standard library for string operations and pattern-matching 6 ** See Copyright Notice in lua.h 7 */ 8 9 #define lstrlib_c 10 #define LUA_LIB 11 12 #include "lprefix.h" 13 14 15 #ifndef _KERNEL 16 #include <ctype.h> 17 #include <float.h> 18 #include <limits.h> 19 #include <locale.h> 20 #include <stddef.h> 21 #include <stdio.h> 22 #include <stdlib.h> 23 #include <string.h> 24 #endif /* _KERNEL */ 25 26 #include "lua.h" 27 28 #include "lauxlib.h" 29 #include "lualib.h" 30 31 32 /* 33 ** maximum number of captures that a pattern can do during 34 ** pattern-matching. This limit is arbitrary, but must fit in 35 ** an unsigned char. 36 */ 37 #if !defined(LUA_MAXCAPTURES) 38 #define LUA_MAXCAPTURES 32 39 #endif 40 41 42 /* macro to 'unsign' a character */ 43 #define uchar(c) ((unsigned char)(c)) 44 45 46 /* 47 ** Some sizes are better limited to fit in 'int', but must also fit in 48 ** 'size_t'. (We assume that 'lua_Integer' cannot be smaller than 'int'.) 49 */ 50 #define MAX_SIZET ((size_t)(~(size_t)0)) 51 52 #define MAXSIZE \ 53 (sizeof(size_t) < sizeof(int) ? MAX_SIZET : (size_t)(INT_MAX)) 54 55 56 57 58 static int str_len (lua_State *L) { 59 size_t l; 60 luaL_checklstring(L, 1, &l); 61 lua_pushinteger(L, (lua_Integer)l); 62 return 1; 63 } 64 65 66 /* translate a relative string position: negative means back from end */ 67 static lua_Integer posrelat (lua_Integer pos, size_t len) { 68 if (pos >= 0) return pos; 69 else if (0u - (size_t)pos > len) return 0; 70 else return (lua_Integer)len + pos + 1; 71 } 72 73 74 static int str_sub (lua_State *L) { 75 size_t l; 76 const char *s = luaL_checklstring(L, 1, &l); 77 lua_Integer start = posrelat(luaL_checkinteger(L, 2), l); 78 lua_Integer end = posrelat(luaL_optinteger(L, 3, -1), l); 79 if (start < 1) start = 1; 80 if (end > (lua_Integer)l) end = l; 81 if (start <= end) 82 lua_pushlstring(L, s + start - 1, (size_t)(end - start) + 1); 83 else lua_pushliteral(L, ""); 84 return 1; 85 } 86 87 88 static int str_reverse (lua_State *L) { 89 size_t l, i; 90 luaL_Buffer b; 91 const char *s = luaL_checklstring(L, 1, &l); 92 char *p = luaL_buffinitsize(L, &b, l); 93 for (i = 0; i < l; i++) 94 p[i] = s[l - i - 1]; 95 luaL_pushresultsize(&b, l); 96 return 1; 97 } 98 99 100 static int str_lower (lua_State *L) { 101 size_t l; 102 size_t i; 103 luaL_Buffer b; 104 const char *s = luaL_checklstring(L, 1, &l); 105 char *p = luaL_buffinitsize(L, &b, l); 106 for (i=0; i<l; i++) 107 p[i] = tolower(uchar(s[i])); 108 luaL_pushresultsize(&b, l); 109 return 1; 110 } 111 112 113 static int str_upper (lua_State *L) { 114 size_t l; 115 size_t i; 116 luaL_Buffer b; 117 const char *s = luaL_checklstring(L, 1, &l); 118 char *p = luaL_buffinitsize(L, &b, l); 119 for (i=0; i<l; i++) 120 p[i] = toupper(uchar(s[i])); 121 luaL_pushresultsize(&b, l); 122 return 1; 123 } 124 125 126 static int str_rep (lua_State *L) { 127 size_t l, lsep; 128 const char *s = luaL_checklstring(L, 1, &l); 129 lua_Integer n = luaL_checkinteger(L, 2); 130 const char *sep = luaL_optlstring(L, 3, "", &lsep); 131 if (n <= 0) lua_pushliteral(L, ""); 132 else if (l + lsep < l || l + lsep > MAXSIZE / n) /* may overflow? */ 133 return luaL_error(L, "resulting string too large"); 134 else { 135 size_t totallen = (size_t)n * l + (size_t)(n - 1) * lsep; 136 luaL_Buffer b; 137 char *p = luaL_buffinitsize(L, &b, totallen); 138 while (n-- > 1) { /* first n-1 copies (followed by separator) */ 139 memcpy(p, s, l * sizeof(char)); p += l; 140 if (lsep > 0) { /* empty 'memcpy' is not that cheap */ 141 memcpy(p, sep, lsep * sizeof(char)); 142 p += lsep; 143 } 144 } 145 memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */ 146 luaL_pushresultsize(&b, totallen); 147 } 148 return 1; 149 } 150 151 152 static int str_byte (lua_State *L) { 153 size_t l; 154 const char *s = luaL_checklstring(L, 1, &l); 155 lua_Integer posi = posrelat(luaL_optinteger(L, 2, 1), l); 156 lua_Integer pose = posrelat(luaL_optinteger(L, 3, posi), l); 157 int n, i; 158 if (posi < 1) posi = 1; 159 if (pose > (lua_Integer)l) pose = l; 160 if (posi > pose) return 0; /* empty interval; return no values */ 161 if (pose - posi >= INT_MAX) /* arithmetic overflow? */ 162 return luaL_error(L, "string slice too long"); 163 n = (int)(pose - posi) + 1; 164 luaL_checkstack(L, n, "string slice too long"); 165 for (i=0; i<n; i++) 166 lua_pushinteger(L, uchar(s[posi+i-1])); 167 return n; 168 } 169 170 171 static int str_char (lua_State *L) { 172 int n = lua_gettop(L); /* number of arguments */ 173 int i; 174 luaL_Buffer b; 175 char *p = luaL_buffinitsize(L, &b, n); 176 for (i=1; i<=n; i++) { 177 lua_Integer c = luaL_checkinteger(L, i); 178 luaL_argcheck(L, uchar(c) == c, i, "value out of range"); 179 p[i - 1] = uchar(c); 180 } 181 luaL_pushresultsize(&b, n); 182 return 1; 183 } 184 185 186 static int writer (lua_State *L, const void *b, size_t size, void *B) { 187 (void)L; 188 luaL_addlstring((luaL_Buffer *) B, (const char *)b, size); 189 return 0; 190 } 191 192 193 static int str_dump (lua_State *L) { 194 luaL_Buffer b; 195 int strip = lua_toboolean(L, 2); 196 luaL_checktype(L, 1, LUA_TFUNCTION); 197 lua_settop(L, 1); 198 luaL_buffinit(L,&b); 199 if (lua_dump(L, writer, &b, strip) != 0) 200 return luaL_error(L, "unable to dump given function"); 201 luaL_pushresult(&b); 202 return 1; 203 } 204 205 206 207 /* 208 ** {====================================================== 209 ** PATTERN MATCHING 210 ** ======================================================= 211 */ 212 213 214 #define CAP_UNFINISHED (-1) 215 #define CAP_POSITION (-2) 216 217 218 typedef struct MatchState { 219 const char *src_init; /* init of source string */ 220 const char *src_end; /* end ('\0') of source string */ 221 const char *p_end; /* end ('\0') of pattern */ 222 lua_State *L; 223 int matchdepth; /* control for recursive depth (to avoid C stack overflow) */ 224 unsigned char level; /* total number of captures (finished or unfinished) */ 225 struct { 226 const char *init; 227 ptrdiff_t len; 228 } capture[LUA_MAXCAPTURES]; 229 } MatchState; 230 231 232 /* recursive function */ 233 static const char *match (MatchState *ms, const char *s, const char *p); 234 235 236 /* maximum recursion depth for 'match' */ 237 #if !defined(MAXCCALLS) 238 #define MAXCCALLS 200 239 #endif 240 241 242 #define L_ESC '%' 243 #define SPECIALS "^$*+?.([%-" 244 245 246 static int check_capture (MatchState *ms, int l) { 247 l -= '1'; 248 if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED) 249 return luaL_error(ms->L, "invalid capture index %%%d", l + 1); 250 return l; 251 } 252 253 254 static int capture_to_close (MatchState *ms) { 255 int level = ms->level; 256 for (level--; level>=0; level--) 257 if (ms->capture[level].len == CAP_UNFINISHED) return level; 258 return luaL_error(ms->L, "invalid pattern capture"); 259 } 260 261 262 static const char *classend (MatchState *ms, const char *p) { 263 switch (*p++) { 264 case L_ESC: { 265 if (p == ms->p_end) 266 luaL_error(ms->L, "malformed pattern (ends with '%%')"); 267 return p+1; 268 } 269 case '[': { 270 if (*p == '^') p++; 271 do { /* look for a ']' */ 272 if (p == ms->p_end) 273 luaL_error(ms->L, "malformed pattern (missing ']')"); 274 if (*(p++) == L_ESC && p < ms->p_end) 275 p++; /* skip escapes (e.g. '%]') */ 276 } while (*p != ']'); 277 return p+1; 278 } 279 default: { 280 return p; 281 } 282 } 283 } 284 285 286 static int match_class (int c, int cl) { 287 int res; 288 switch (tolower(cl)) { 289 case 'a' : res = isalpha(c); break; 290 case 'c' : res = iscntrl(c); break; 291 case 'd' : res = isdigit(c); break; 292 case 'g' : res = isgraph(c); break; 293 case 'l' : res = islower(c); break; 294 case 'p' : res = ispunct(c); break; 295 case 's' : res = isspace(c); break; 296 case 'u' : res = isupper(c); break; 297 case 'w' : res = isalnum(c); break; 298 case 'x' : res = isxdigit(c); break; 299 case 'z' : res = (c == 0); break; /* deprecated option */ 300 default: return (cl == c); 301 } 302 return (islower(cl) ? res : !res); 303 } 304 305 306 static int matchbracketclass (int c, const char *p, const char *ec) { 307 int sig = 1; 308 if (*(p+1) == '^') { 309 sig = 0; 310 p++; /* skip the '^' */ 311 } 312 while (++p < ec) { 313 if (*p == L_ESC) { 314 p++; 315 if (match_class(c, uchar(*p))) 316 return sig; 317 } 318 else if ((*(p+1) == '-') && (p+2 < ec)) { 319 p+=2; 320 if (uchar(*(p-2)) <= c && c <= uchar(*p)) 321 return sig; 322 } 323 else if (uchar(*p) == c) return sig; 324 } 325 return !sig; 326 } 327 328 329 static int singlematch (MatchState *ms, const char *s, const char *p, 330 const char *ep) { 331 if (s >= ms->src_end) 332 return 0; 333 else { 334 int c = uchar(*s); 335 switch (*p) { 336 case '.': return 1; /* matches any char */ 337 case L_ESC: return match_class(c, uchar(*(p+1))); 338 case '[': return matchbracketclass(c, p, ep-1); 339 default: return (uchar(*p) == c); 340 } 341 } 342 } 343 344 345 static const char *matchbalance (MatchState *ms, const char *s, 346 const char *p) { 347 if (p >= ms->p_end - 1) 348 luaL_error(ms->L, "malformed pattern (missing arguments to '%%b')"); 349 if (*s != *p) return NULL; 350 else { 351 int b = *p; 352 int e = *(p+1); 353 int cont = 1; 354 while (++s < ms->src_end) { 355 if (*s == e) { 356 if (--cont == 0) return s+1; 357 } 358 else if (*s == b) cont++; 359 } 360 } 361 return NULL; /* string ends out of balance */ 362 } 363 364 365 static const char *max_expand (MatchState *ms, const char *s, 366 const char *p, const char *ep) { 367 ptrdiff_t i = 0; /* counts maximum expand for item */ 368 while (singlematch(ms, s + i, p, ep)) 369 i++; 370 /* keeps trying to match with the maximum repetitions */ 371 while (i>=0) { 372 const char *res = match(ms, (s+i), ep+1); 373 if (res) return res; 374 i--; /* else didn't match; reduce 1 repetition to try again */ 375 } 376 return NULL; 377 } 378 379 380 static const char *min_expand (MatchState *ms, const char *s, 381 const char *p, const char *ep) { 382 for (;;) { 383 const char *res = match(ms, s, ep+1); 384 if (res != NULL) 385 return res; 386 else if (singlematch(ms, s, p, ep)) 387 s++; /* try with one more repetition */ 388 else return NULL; 389 } 390 } 391 392 393 static const char *start_capture (MatchState *ms, const char *s, 394 const char *p, int what) { 395 const char *res; 396 int level = ms->level; 397 if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures"); 398 ms->capture[level].init = s; 399 ms->capture[level].len = what; 400 ms->level = level+1; 401 if ((res=match(ms, s, p)) == NULL) /* match failed? */ 402 ms->level--; /* undo capture */ 403 return res; 404 } 405 406 407 static const char *end_capture (MatchState *ms, const char *s, 408 const char *p) { 409 int l = capture_to_close(ms); 410 const char *res; 411 ms->capture[l].len = s - ms->capture[l].init; /* close capture */ 412 if ((res = match(ms, s, p)) == NULL) /* match failed? */ 413 ms->capture[l].len = CAP_UNFINISHED; /* undo capture */ 414 return res; 415 } 416 417 418 static const char *match_capture (MatchState *ms, const char *s, int l) { 419 size_t len; 420 l = check_capture(ms, l); 421 len = ms->capture[l].len; 422 if ((size_t)(ms->src_end-s) >= len && 423 memcmp(ms->capture[l].init, s, len) == 0) 424 return s+len; 425 else return NULL; 426 } 427 428 429 static const char *match (MatchState *ms, const char *s, const char *p) { 430 if (ms->matchdepth-- == 0) 431 luaL_error(ms->L, "pattern too complex"); 432 init: /* using goto's to optimize tail recursion */ 433 if (p != ms->p_end) { /* end of pattern? */ 434 switch (*p) { 435 case '(': { /* start capture */ 436 if (*(p + 1) == ')') /* position capture? */ 437 s = start_capture(ms, s, p + 2, CAP_POSITION); 438 else 439 s = start_capture(ms, s, p + 1, CAP_UNFINISHED); 440 break; 441 } 442 case ')': { /* end capture */ 443 s = end_capture(ms, s, p + 1); 444 break; 445 } 446 case '$': { 447 if ((p + 1) != ms->p_end) /* is the '$' the last char in pattern? */ 448 goto dflt; /* no; go to default */ 449 s = (s == ms->src_end) ? s : NULL; /* check end of string */ 450 break; 451 } 452 case L_ESC: { /* escaped sequences not in the format class[*+?-]? */ 453 switch (*(p + 1)) { 454 case 'b': { /* balanced string? */ 455 s = matchbalance(ms, s, p + 2); 456 if (s != NULL) { 457 p += 4; goto init; /* return match(ms, s, p + 4); */ 458 } /* else fail (s == NULL) */ 459 break; 460 } 461 case 'f': { /* frontier? */ 462 const char *ep; char previous; 463 p += 2; 464 if (*p != '[') 465 luaL_error(ms->L, "missing '[' after '%%f' in pattern"); 466 ep = classend(ms, p); /* points to what is next */ 467 previous = (s == ms->src_init) ? '\0' : *(s - 1); 468 if (!matchbracketclass(uchar(previous), p, ep - 1) && 469 matchbracketclass(uchar(*s), p, ep - 1)) { 470 p = ep; goto init; /* return match(ms, s, ep); */ 471 } 472 s = NULL; /* match failed */ 473 break; 474 } 475 case '0': case '1': case '2': case '3': 476 case '4': case '5': case '6': case '7': 477 case '8': case '9': { /* capture results (%0-%9)? */ 478 s = match_capture(ms, s, uchar(*(p + 1))); 479 if (s != NULL) { 480 p += 2; goto init; /* return match(ms, s, p + 2) */ 481 } 482 break; 483 } 484 default: goto dflt; 485 } 486 break; 487 } 488 default: dflt: { /* pattern class plus optional suffix */ 489 const char *ep = classend(ms, p); /* points to optional suffix */ 490 /* does not match at least once? */ 491 if (!singlematch(ms, s, p, ep)) { 492 if (*ep == '*' || *ep == '?' || *ep == '-') { /* accept empty? */ 493 p = ep + 1; goto init; /* return match(ms, s, ep + 1); */ 494 } 495 else /* '+' or no suffix */ 496 s = NULL; /* fail */ 497 } 498 else { /* matched once */ 499 switch (*ep) { /* handle optional suffix */ 500 case '?': { /* optional */ 501 const char *res; 502 if ((res = match(ms, s + 1, ep + 1)) != NULL) 503 s = res; 504 else { 505 p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */ 506 } 507 break; 508 } 509 case '+': /* 1 or more repetitions */ 510 s++; /* 1 match already done */ 511 /* FALLTHROUGH */ 512 case '*': /* 0 or more repetitions */ 513 s = max_expand(ms, s, p, ep); 514 break; 515 case '-': /* 0 or more repetitions (minimum) */ 516 s = min_expand(ms, s, p, ep); 517 break; 518 default: /* no suffix */ 519 s++; p = ep; goto init; /* return match(ms, s + 1, ep); */ 520 } 521 } 522 break; 523 } 524 } 525 } 526 ms->matchdepth++; 527 return s; 528 } 529 530 531 532 static const char *lmemfind (const char *s1, size_t l1, 533 const char *s2, size_t l2) { 534 if (l2 == 0) return s1; /* empty strings are everywhere */ 535 else if (l2 > l1) return NULL; /* avoids a negative 'l1' */ 536 else { 537 const char *init; /* to search for a '*s2' inside 's1' */ 538 l2--; /* 1st char will be checked by 'memchr' */ 539 l1 = l1-l2; /* 's2' cannot be found after that */ 540 while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) { 541 init++; /* 1st char is already checked */ 542 if (memcmp(init, s2+1, l2) == 0) 543 return init-1; 544 else { /* correct 'l1' and 's1' to try again */ 545 l1 -= init-s1; 546 s1 = init; 547 } 548 } 549 return NULL; /* not found */ 550 } 551 } 552 553 554 static void push_onecapture (MatchState *ms, int i, const char *s, 555 const char *e) { 556 if (i >= ms->level) { 557 if (i == 0) /* ms->level == 0, too */ 558 lua_pushlstring(ms->L, s, e - s); /* add whole match */ 559 else 560 luaL_error(ms->L, "invalid capture index %%%d", i + 1); 561 } 562 else { 563 ptrdiff_t l = ms->capture[i].len; 564 if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture"); 565 if (l == CAP_POSITION) 566 lua_pushinteger(ms->L, (ms->capture[i].init - ms->src_init) + 1); 567 else 568 lua_pushlstring(ms->L, ms->capture[i].init, l); 569 } 570 } 571 572 573 static int push_captures (MatchState *ms, const char *s, const char *e) { 574 int i; 575 int nlevels = (ms->level == 0 && s) ? 1 : ms->level; 576 luaL_checkstack(ms->L, nlevels, "too many captures"); 577 for (i = 0; i < nlevels; i++) 578 push_onecapture(ms, i, s, e); 579 return nlevels; /* number of strings pushed */ 580 } 581 582 583 /* check whether pattern has no special characters */ 584 static int nospecials (const char *p, size_t l) { 585 size_t upto = 0; 586 do { 587 if (strpbrk(p + upto, SPECIALS)) 588 return 0; /* pattern has a special character */ 589 upto += strlen(p + upto) + 1; /* may have more after \0 */ 590 } while (upto <= l); 591 return 1; /* no special chars found */ 592 } 593 594 595 static void prepstate (MatchState *ms, lua_State *L, 596 const char *s, size_t ls, const char *p, size_t lp) { 597 ms->L = L; 598 ms->matchdepth = MAXCCALLS; 599 ms->src_init = s; 600 ms->src_end = s + ls; 601 ms->p_end = p + lp; 602 } 603 604 605 static void reprepstate (MatchState *ms) { 606 ms->level = 0; 607 lua_assert(ms->matchdepth == MAXCCALLS); 608 } 609 610 611 static int str_find_aux (lua_State *L, int find) { 612 size_t ls, lp; 613 const char *s = luaL_checklstring(L, 1, &ls); 614 const char *p = luaL_checklstring(L, 2, &lp); 615 lua_Integer init = posrelat(luaL_optinteger(L, 3, 1), ls); 616 if (init < 1) init = 1; 617 else if (init > (lua_Integer)ls + 1) { /* start after string's end? */ 618 lua_pushnil(L); /* cannot find anything */ 619 return 1; 620 } 621 /* explicit request or no special characters? */ 622 if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) { 623 /* do a plain search */ 624 const char *s2 = lmemfind(s + init - 1, ls - (size_t)init + 1, p, lp); 625 if (s2) { 626 lua_pushinteger(L, (s2 - s) + 1); 627 lua_pushinteger(L, (s2 - s) + lp); 628 return 2; 629 } 630 } 631 else { 632 MatchState ms; 633 const char *s1 = s + init - 1; 634 int anchor = (*p == '^'); 635 if (anchor) { 636 p++; lp--; /* skip anchor character */ 637 } 638 prepstate(&ms, L, s, ls, p, lp); 639 do { 640 const char *res; 641 reprepstate(&ms); 642 if ((res=match(&ms, s1, p)) != NULL) { 643 if (find) { 644 lua_pushinteger(L, (s1 - s) + 1); /* start */ 645 lua_pushinteger(L, res - s); /* end */ 646 return push_captures(&ms, NULL, 0) + 2; 647 } 648 else 649 return push_captures(&ms, s1, res); 650 } 651 } while (s1++ < ms.src_end && !anchor); 652 } 653 lua_pushnil(L); /* not found */ 654 return 1; 655 } 656 657 658 static int str_find (lua_State *L) { 659 return str_find_aux(L, 1); 660 } 661 662 663 static int str_match (lua_State *L) { 664 return str_find_aux(L, 0); 665 } 666 667 668 /* state for 'gmatch' */ 669 typedef struct GMatchState { 670 const char *src; /* current position */ 671 const char *p; /* pattern */ 672 const char *lastmatch; /* end of last match */ 673 MatchState ms; /* match state */ 674 } GMatchState; 675 676 677 static int gmatch_aux (lua_State *L) { 678 GMatchState *gm = (GMatchState *)lua_touserdata(L, lua_upvalueindex(3)); 679 const char *src; 680 gm->ms.L = L; 681 for (src = gm->src; src <= gm->ms.src_end; src++) { 682 const char *e; 683 reprepstate(&gm->ms); 684 if ((e = match(&gm->ms, src, gm->p)) != NULL && e != gm->lastmatch) { 685 gm->src = gm->lastmatch = e; 686 return push_captures(&gm->ms, src, e); 687 } 688 } 689 return 0; /* not found */ 690 } 691 692 693 static int gmatch (lua_State *L) { 694 size_t ls, lp; 695 const char *s = luaL_checklstring(L, 1, &ls); 696 const char *p = luaL_checklstring(L, 2, &lp); 697 GMatchState *gm; 698 lua_settop(L, 2); /* keep them on closure to avoid being collected */ 699 gm = (GMatchState *)lua_newuserdata(L, sizeof(GMatchState)); 700 prepstate(&gm->ms, L, s, ls, p, lp); 701 gm->src = s; gm->p = p; gm->lastmatch = NULL; 702 lua_pushcclosure(L, gmatch_aux, 3); 703 return 1; 704 } 705 706 707 static void add_s (MatchState *ms, luaL_Buffer *b, const char *s, 708 const char *e) { 709 size_t l, i; 710 lua_State *L = ms->L; 711 const char *news = lua_tolstring(L, 3, &l); 712 for (i = 0; i < l; i++) { 713 if (news[i] != L_ESC) 714 luaL_addchar(b, news[i]); 715 else { 716 i++; /* skip ESC */ 717 if (!isdigit(uchar(news[i]))) { 718 if (news[i] != L_ESC) 719 luaL_error(L, "invalid use of '%c' in replacement string", L_ESC); 720 luaL_addchar(b, news[i]); 721 } 722 else if (news[i] == '0') 723 luaL_addlstring(b, s, e - s); 724 else { 725 push_onecapture(ms, news[i] - '1', s, e); 726 luaL_tolstring(L, -1, NULL); /* if number, convert it to string */ 727 lua_remove(L, -2); /* remove original value */ 728 luaL_addvalue(b); /* add capture to accumulated result */ 729 } 730 } 731 } 732 } 733 734 735 static void add_value (MatchState *ms, luaL_Buffer *b, const char *s, 736 const char *e, int tr) { 737 lua_State *L = ms->L; 738 switch (tr) { 739 case LUA_TFUNCTION: { 740 int n; 741 lua_pushvalue(L, 3); 742 n = push_captures(ms, s, e); 743 lua_call(L, n, 1); 744 break; 745 } 746 case LUA_TTABLE: { 747 push_onecapture(ms, 0, s, e); 748 lua_gettable(L, 3); 749 break; 750 } 751 default: { /* LUA_TNUMBER or LUA_TSTRING */ 752 add_s(ms, b, s, e); 753 return; 754 } 755 } 756 if (!lua_toboolean(L, -1)) { /* nil or false? */ 757 lua_pop(L, 1); 758 lua_pushlstring(L, s, e - s); /* keep original text */ 759 } 760 else if (!lua_isstring(L, -1)) 761 luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1)); 762 luaL_addvalue(b); /* add result to accumulator */ 763 } 764 765 766 static int str_gsub (lua_State *L) { 767 size_t srcl, lp; 768 const char *src = luaL_checklstring(L, 1, &srcl); /* subject */ 769 const char *p = luaL_checklstring(L, 2, &lp); /* pattern */ 770 const char *lastmatch = NULL; /* end of last match */ 771 int tr = lua_type(L, 3); /* replacement type */ 772 lua_Integer max_s = luaL_optinteger(L, 4, srcl + 1); /* max replacements */ 773 int anchor = (*p == '^'); 774 lua_Integer n = 0; /* replacement count */ 775 MatchState ms; 776 luaL_Buffer b; 777 luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING || 778 tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3, 779 "string/function/table expected"); 780 luaL_buffinit(L, &b); 781 if (anchor) { 782 p++; lp--; /* skip anchor character */ 783 } 784 prepstate(&ms, L, src, srcl, p, lp); 785 while (n < max_s) { 786 const char *e; 787 reprepstate(&ms); /* (re)prepare state for new match */ 788 if ((e = match(&ms, src, p)) != NULL && e != lastmatch) { /* match? */ 789 n++; 790 add_value(&ms, &b, src, e, tr); /* add replacement to buffer */ 791 src = lastmatch = e; 792 } 793 else if (src < ms.src_end) /* otherwise, skip one character */ 794 luaL_addchar(&b, *src++); 795 else break; /* end of subject */ 796 if (anchor) break; 797 } 798 luaL_addlstring(&b, src, ms.src_end-src); 799 luaL_pushresult(&b); 800 lua_pushinteger(L, n); /* number of substitutions */ 801 return 2; 802 } 803 804 /* }====================================================== */ 805 806 807 808 /* 809 ** {====================================================== 810 ** STRING FORMAT 811 ** ======================================================= 812 */ 813 814 #if !defined(lua_number2strx) /* { */ 815 816 /* 817 ** Hexadecimal floating-point formatter 818 */ 819 820 #include <math.h> 821 822 #define SIZELENMOD (sizeof(LUA_NUMBER_FRMLEN)/sizeof(char)) 823 824 825 /* 826 ** Number of bits that goes into the first digit. It can be any value 827 ** between 1 and 4; the following definition tries to align the number 828 ** to nibble boundaries by making what is left after that first digit a 829 ** multiple of 4. 830 */ 831 #define L_NBFD ((l_mathlim(MANT_DIG) - 1)%4 + 1) 832 833 834 /* 835 ** Add integer part of 'x' to buffer and return new 'x' 836 */ 837 static lua_Number adddigit (char *buff, int n, lua_Number x) { 838 lua_Number dd = l_mathop(floor)(x); /* get integer part from 'x' */ 839 int d = (int)dd; 840 buff[n] = (d < 10 ? d + '0' : d - 10 + 'a'); /* add to buffer */ 841 return x - dd; /* return what is left */ 842 } 843 844 845 static int num2straux (char *buff, int sz, lua_Number x) { 846 /* if 'inf' or 'NaN', format it like '%g' */ 847 if (x != x || x == (lua_Number)HUGE_VAL || x == -(lua_Number)HUGE_VAL) 848 return l_sprintf(buff, sz, LUA_NUMBER_FMT, (LUAI_UACNUMBER)x); 849 else if (x == 0) { /* can be -0... */ 850 /* create "0" or "-0" followed by exponent */ 851 return l_sprintf(buff, sz, LUA_NUMBER_FMT "x0p+0", (LUAI_UACNUMBER)x); 852 } 853 else { 854 int e; 855 lua_Number m = l_mathop(frexp)(x, &e); /* 'x' fraction and exponent */ 856 int n = 0; /* character count */ 857 if (m < 0) { /* is number negative? */ 858 buff[n++] = '-'; /* add signal */ 859 m = -m; /* make it positive */ 860 } 861 buff[n++] = '0'; buff[n++] = 'x'; /* add "0x" */ 862 m = adddigit(buff, n++, m * (1 << L_NBFD)); /* add first digit */ 863 e -= L_NBFD; /* this digit goes before the radix point */ 864 if (m > 0) { /* more digits? */ 865 buff[n++] = lua_getlocaledecpoint(); /* add radix point */ 866 do { /* add as many digits as needed */ 867 m = adddigit(buff, n++, m * 16); 868 } while (m > 0); 869 } 870 n += l_sprintf(buff + n, sz - n, "p%+d", e); /* add exponent */ 871 lua_assert(n < sz); 872 return n; 873 } 874 } 875 876 877 static int lua_number2strx (lua_State *L, char *buff, int sz, 878 const char *fmt, lua_Number x) { 879 int n = num2straux(buff, sz, x); 880 if (fmt[SIZELENMOD] == 'A') { 881 int i; 882 for (i = 0; i < n; i++) 883 buff[i] = toupper(uchar(buff[i])); 884 } 885 else if (fmt[SIZELENMOD] != 'a') 886 return luaL_error(L, "modifiers for format '%%a'/'%%A' not implemented"); 887 return n; 888 } 889 890 #endif /* } */ 891 892 893 /* 894 ** Maximum size of each formatted item. This maximum size is produced 895 ** by format('%.99f', -maxfloat), and is equal to 99 + 3 ('-', '.', 896 ** and '\0') + number of decimal digits to represent maxfloat (which 897 ** is maximum exponent + 1). (99+3+1 then rounded to 120 for "extra 898 ** expenses", such as locale-dependent stuff) 899 */ 900 #define MAX_ITEM (120 + l_mathlim(MAX_10_EXP)) 901 902 903 /* valid flags in a format specification */ 904 #define FLAGS "-+ #0" 905 906 /* 907 ** maximum size of each format specification (such as "%-099.99d") 908 */ 909 #define MAX_FORMAT 32 910 911 912 static void addquoted (luaL_Buffer *b, const char *s, size_t len) { 913 luaL_addchar(b, '"'); 914 while (len--) { 915 if (*s == '"' || *s == '\\' || *s == '\n') { 916 luaL_addchar(b, '\\'); 917 luaL_addchar(b, *s); 918 } 919 else if (iscntrl(uchar(*s))) { 920 char buff[10]; 921 if (!isdigit(uchar(*(s+1)))) 922 l_sprintf(buff, sizeof(buff), "\\%d", (int)uchar(*s)); 923 else 924 l_sprintf(buff, sizeof(buff), "\\%03d", (int)uchar(*s)); 925 luaL_addstring(b, buff); 926 } 927 else 928 luaL_addchar(b, *s); 929 s++; 930 } 931 luaL_addchar(b, '"'); 932 } 933 934 935 #ifndef _KERNEL 936 /* 937 ** Ensures the 'buff' string uses a dot as the radix character. 938 */ 939 static void checkdp (char *buff, int nb) { 940 if (memchr(buff, '.', nb) == NULL) { /* no dot? */ 941 char point = lua_getlocaledecpoint(); /* try locale point */ 942 char *ppoint = (char *)memchr(buff, point, nb); 943 if (ppoint) *ppoint = '.'; /* change it to a dot */ 944 } 945 } 946 #endif /* _KERNEL */ 947 948 949 static void addliteral (lua_State *L, luaL_Buffer *b, int arg) { 950 switch (lua_type(L, arg)) { 951 case LUA_TSTRING: { 952 size_t len; 953 const char *s = lua_tolstring(L, arg, &len); 954 addquoted(b, s, len); 955 break; 956 } 957 case LUA_TNUMBER: { 958 char *buff = luaL_prepbuffsize(b, MAX_ITEM); 959 int nb; 960 if (!lua_isinteger(L, arg)) { /* float? */ 961 #ifndef _KERNEL 962 lua_Number n = lua_tonumber(L, arg); /* write as hexa ('%a') */ 963 nb = lua_number2strx(L, buff, MAX_ITEM, "%" LUA_NUMBER_FRMLEN "a", n); 964 checkdp(buff, nb); /* ensure it uses a dot */ 965 #endif /* _KERNEL */ 966 } 967 else { /* integers */ 968 lua_Integer n = lua_tointeger(L, arg); 969 const char *format = (n == LUA_MININTEGER) /* corner case? */ 970 ? "0x%" LUA_INTEGER_FRMLEN "x" /* use hexa */ 971 : LUA_INTEGER_FMT; /* else use default format */ 972 nb = l_sprintf(buff, MAX_ITEM, format, (LUAI_UACINT)n); 973 } 974 luaL_addsize(b, nb); 975 break; 976 } 977 case LUA_TNIL: case LUA_TBOOLEAN: { 978 luaL_tolstring(L, arg, NULL); 979 luaL_addvalue(b); 980 break; 981 } 982 default: { 983 luaL_argerror(L, arg, "value has no literal form"); 984 } 985 } 986 } 987 988 989 static const char *scanformat (lua_State *L, const char *strfrmt, char *form) { 990 const char *p = strfrmt; 991 while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */ 992 if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char)) 993 luaL_error(L, "invalid format (repeated flags)"); 994 if (isdigit(uchar(*p))) p++; /* skip width */ 995 if (isdigit(uchar(*p))) p++; /* (2 digits at most) */ 996 if (*p == '.') { 997 p++; 998 if (isdigit(uchar(*p))) p++; /* skip precision */ 999 if (isdigit(uchar(*p))) p++; /* (2 digits at most) */ 1000 } 1001 if (isdigit(uchar(*p))) 1002 luaL_error(L, "invalid format (width or precision too long)"); 1003 *(form++) = '%'; 1004 memcpy(form, strfrmt, ((p - strfrmt) + 1) * sizeof(char)); 1005 form += (p - strfrmt) + 1; 1006 *form = '\0'; 1007 return p; 1008 } 1009 1010 1011 /* 1012 ** add length modifier into formats 1013 */ 1014 static void addlenmod (char *form, const char *lenmod) { 1015 size_t l = strlen(form); 1016 size_t lm = strlen(lenmod); 1017 char spec = form[l - 1]; 1018 strcpy(form + l - 1, lenmod); 1019 form[l + lm - 1] = spec; 1020 form[l + lm] = '\0'; 1021 } 1022 1023 1024 static int str_format (lua_State *L) { 1025 int top = lua_gettop(L); 1026 int arg = 1; 1027 size_t sfl; 1028 const char *strfrmt = luaL_checklstring(L, arg, &sfl); 1029 const char *strfrmt_end = strfrmt+sfl; 1030 luaL_Buffer b; 1031 luaL_buffinit(L, &b); 1032 while (strfrmt < strfrmt_end) { 1033 if (*strfrmt != L_ESC) 1034 luaL_addchar(&b, *strfrmt++); 1035 else if (*++strfrmt == L_ESC) 1036 luaL_addchar(&b, *strfrmt++); /* %% */ 1037 else { /* format item */ 1038 char form[MAX_FORMAT]; /* to store the format ('%...') */ 1039 char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */ 1040 int nb = 0; /* number of bytes in added item */ 1041 if (++arg > top) 1042 luaL_argerror(L, arg, "no value"); 1043 strfrmt = scanformat(L, strfrmt, form); 1044 switch (*strfrmt++) { 1045 case 'c': { 1046 nb = l_sprintf(buff, MAX_ITEM, form, (int)luaL_checkinteger(L, arg)); 1047 break; 1048 } 1049 case 'd': case 'i': 1050 case 'o': case 'u': case 'x': case 'X': { 1051 lua_Integer n = luaL_checkinteger(L, arg); 1052 addlenmod(form, LUA_INTEGER_FRMLEN); 1053 nb = l_sprintf(buff, MAX_ITEM, form, (LUAI_UACINT)n); 1054 break; 1055 } 1056 #ifndef _KERNEL 1057 case 'a': case 'A': 1058 addlenmod(form, LUA_NUMBER_FRMLEN); 1059 nb = lua_number2strx(L, buff, MAX_ITEM, form, 1060 luaL_checknumber(L, arg)); 1061 break; 1062 case 'e': case 'E': case 'f': 1063 case 'g': case 'G': { 1064 lua_Number n = luaL_checknumber(L, arg); 1065 addlenmod(form, LUA_NUMBER_FRMLEN); 1066 nb = l_sprintf(buff, MAX_ITEM, form, (LUAI_UACNUMBER)n); 1067 break; 1068 } 1069 #endif /* _KERNEL */ 1070 case 'q': { 1071 addliteral(L, &b, arg); 1072 break; 1073 } 1074 case 's': { 1075 size_t l; 1076 const char *s = luaL_tolstring(L, arg, &l); 1077 if (form[2] == '\0') /* no modifiers? */ 1078 luaL_addvalue(&b); /* keep entire string */ 1079 else { 1080 luaL_argcheck(L, l == strlen(s), arg, "string contains zeros"); 1081 if (!strchr(form, '.') && l >= 100) { 1082 /* no precision and string is too long to be formatted */ 1083 luaL_addvalue(&b); /* keep entire string */ 1084 } 1085 else { /* format the string into 'buff' */ 1086 nb = l_sprintf(buff, MAX_ITEM, form, s); 1087 lua_pop(L, 1); /* remove result from 'luaL_tolstring' */ 1088 } 1089 } 1090 break; 1091 } 1092 default: { /* also treat cases 'pnLlh' */ 1093 return luaL_error(L, "invalid option '%%%c' to 'format'", 1094 *(strfrmt - 1)); 1095 } 1096 } 1097 lua_assert(nb < MAX_ITEM); 1098 luaL_addsize(&b, nb); 1099 } 1100 } 1101 luaL_pushresult(&b); 1102 return 1; 1103 } 1104 1105 /* }====================================================== */ 1106 1107 1108 /* 1109 ** {====================================================== 1110 ** PACK/UNPACK 1111 ** ======================================================= 1112 */ 1113 1114 1115 /* value used for padding */ 1116 #if !defined(LUAL_PACKPADBYTE) 1117 #define LUAL_PACKPADBYTE 0x00 1118 #endif 1119 1120 /* maximum size for the binary representation of an integer */ 1121 #define MAXINTSIZE 16 1122 1123 /* number of bits in a character */ 1124 #define NB CHAR_BIT 1125 1126 /* mask for one character (NB 1's) */ 1127 #define MC ((1 << NB) - 1) 1128 1129 /* size of a lua_Integer */ 1130 #define SZINT ((int)sizeof(lua_Integer)) 1131 1132 1133 /* dummy union to get native endianness */ 1134 static const union { 1135 int dummy; 1136 char little; /* true iff machine is little endian */ 1137 } nativeendian = {1}; 1138 1139 1140 /* dummy structure to get native alignment requirements */ 1141 struct cD { 1142 char c; 1143 #ifndef _KERNEL 1144 union { double d; void *p; lua_Integer i; lua_Number n; } u; 1145 #else /* _KERNEL */ 1146 union { void *p; lua_Integer i; lua_Number n; } u; 1147 #endif /* _KERNEL */ 1148 }; 1149 1150 #define MAXALIGN (offsetof(struct cD, u)) 1151 1152 1153 #ifndef _KERNEL 1154 /* 1155 ** Union for serializing floats 1156 */ 1157 typedef union Ftypes { 1158 float f; 1159 double d; 1160 lua_Number n; 1161 char buff[5 * sizeof(lua_Number)]; /* enough for any float type */ 1162 } Ftypes; 1163 #endif /* _KERNEL */ 1164 1165 1166 /* 1167 ** information to pack/unpack stuff 1168 */ 1169 typedef struct Header { 1170 lua_State *L; 1171 int islittle; 1172 int maxalign; 1173 } Header; 1174 1175 1176 /* 1177 ** options for pack/unpack 1178 */ 1179 typedef enum KOption { 1180 Kint, /* signed integers */ 1181 Kuint, /* unsigned integers */ 1182 #ifndef _KERNEL 1183 Kfloat, /* floating-point numbers */ 1184 #endif /* _KERNEL */ 1185 Kchar, /* fixed-length strings */ 1186 Kstring, /* strings with prefixed length */ 1187 Kzstr, /* zero-terminated strings */ 1188 Kpadding, /* padding */ 1189 Kpaddalign, /* padding for alignment */ 1190 Knop /* no-op (configuration or spaces) */ 1191 } KOption; 1192 1193 1194 /* 1195 ** Read an integer numeral from string 'fmt' or return 'df' if 1196 ** there is no numeral 1197 */ 1198 static int digit (int c) { return '0' <= c && c <= '9'; } 1199 1200 static int getnum (const char **fmt, int df) { 1201 if (!digit(**fmt)) /* no number? */ 1202 return df; /* return default value */ 1203 else { 1204 int a = 0; 1205 do { 1206 a = a*10 + (*((*fmt)++) - '0'); 1207 } while (digit(**fmt) && a <= ((int)MAXSIZE - 9)/10); 1208 return a; 1209 } 1210 } 1211 1212 1213 /* 1214 ** Read an integer numeral and raises an error if it is larger 1215 ** than the maximum size for integers. 1216 */ 1217 static int getnumlimit (Header *h, const char **fmt, int df) { 1218 int sz = getnum(fmt, df); 1219 if (sz > MAXINTSIZE || sz <= 0) 1220 return luaL_error(h->L, "integral size (%d) out of limits [1,%d]", 1221 sz, MAXINTSIZE); 1222 return sz; 1223 } 1224 1225 1226 /* 1227 ** Initialize Header 1228 */ 1229 static void initheader (lua_State *L, Header *h) { 1230 h->L = L; 1231 h->islittle = nativeendian.little; 1232 h->maxalign = 1; 1233 } 1234 1235 1236 /* 1237 ** Read and classify next option. 'size' is filled with option's size. 1238 */ 1239 static KOption getoption (Header *h, const char **fmt, int *size) { 1240 int opt = *((*fmt)++); 1241 *size = 0; /* default */ 1242 switch (opt) { 1243 case 'b': *size = sizeof(char); return Kint; 1244 case 'B': *size = sizeof(char); return Kuint; 1245 case 'h': *size = sizeof(short); return Kint; 1246 case 'H': *size = sizeof(short); return Kuint; 1247 case 'l': *size = sizeof(long); return Kint; 1248 case 'L': *size = sizeof(long); return Kuint; 1249 case 'j': *size = sizeof(lua_Integer); return Kint; 1250 case 'J': *size = sizeof(lua_Integer); return Kuint; 1251 case 'T': *size = sizeof(size_t); return Kuint; 1252 #ifndef _KERNEL 1253 case 'f': *size = sizeof(float); return Kfloat; 1254 case 'd': *size = sizeof(double); return Kfloat; 1255 case 'n': *size = sizeof(lua_Number); return Kfloat; 1256 #else /* _KERNEL */ 1257 case 'n': *size = sizeof(lua_Number); return Kint; 1258 #endif /* _KERNEL */ 1259 case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint; 1260 case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint; 1261 case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring; 1262 case 'c': 1263 *size = getnum(fmt, -1); 1264 if (*size == -1) 1265 luaL_error(h->L, "missing size for format option 'c'"); 1266 return Kchar; 1267 case 'z': return Kzstr; 1268 case 'x': *size = 1; return Kpadding; 1269 case 'X': return Kpaddalign; 1270 case ' ': break; 1271 case '<': h->islittle = 1; break; 1272 case '>': h->islittle = 0; break; 1273 case '=': h->islittle = nativeendian.little; break; 1274 case '!': h->maxalign = getnumlimit(h, fmt, MAXALIGN); break; 1275 default: luaL_error(h->L, "invalid format option '%c'", opt); 1276 } 1277 return Knop; 1278 } 1279 1280 1281 /* 1282 ** Read, classify, and fill other details about the next option. 1283 ** 'psize' is filled with option's size, 'notoalign' with its 1284 ** alignment requirements. 1285 ** Local variable 'size' gets the size to be aligned. (Kpadal option 1286 ** always gets its full alignment, other options are limited by 1287 ** the maximum alignment ('maxalign'). Kchar option needs no alignment 1288 ** despite its size. 1289 */ 1290 static KOption getdetails (Header *h, size_t totalsize, 1291 const char **fmt, int *psize, int *ntoalign) { 1292 KOption opt = getoption(h, fmt, psize); 1293 int align = *psize; /* usually, alignment follows size */ 1294 if (opt == Kpaddalign) { /* 'X' gets alignment from following option */ 1295 if (**fmt == '\0' || getoption(h, fmt, &align) == Kchar || align == 0) 1296 luaL_argerror(h->L, 1, "invalid next option for option 'X'"); 1297 } 1298 if (align <= 1 || opt == Kchar) /* need no alignment? */ 1299 *ntoalign = 0; 1300 else { 1301 if (align > h->maxalign) /* enforce maximum alignment */ 1302 align = h->maxalign; 1303 if ((align & (align - 1)) != 0) /* is 'align' not a power of 2? */ 1304 luaL_argerror(h->L, 1, "format asks for alignment not power of 2"); 1305 *ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1); 1306 } 1307 return opt; 1308 } 1309 1310 1311 /* 1312 ** Pack integer 'n' with 'size' bytes and 'islittle' endianness. 1313 ** The final 'if' handles the case when 'size' is larger than 1314 ** the size of a Lua integer, correcting the extra sign-extension 1315 ** bytes if necessary (by default they would be zeros). 1316 */ 1317 static void packint (luaL_Buffer *b, lua_Unsigned n, 1318 int islittle, int size, int neg) { 1319 char *buff = luaL_prepbuffsize(b, size); 1320 int i; 1321 buff[islittle ? 0 : size - 1] = (char)(n & MC); /* first byte */ 1322 for (i = 1; i < size; i++) { 1323 n >>= NB; 1324 buff[islittle ? i : size - 1 - i] = (char)(n & MC); 1325 } 1326 if (neg && size > SZINT) { /* negative number need sign extension? */ 1327 for (i = SZINT; i < size; i++) /* correct extra bytes */ 1328 buff[islittle ? i : size - 1 - i] = (char)MC; 1329 } 1330 luaL_addsize(b, size); /* add result to buffer */ 1331 } 1332 1333 1334 #ifndef _KERNEL 1335 /* 1336 ** Copy 'size' bytes from 'src' to 'dest', correcting endianness if 1337 ** given 'islittle' is different from native endianness. 1338 */ 1339 static void copywithendian (volatile char *dest, volatile const char *src, 1340 int size, int islittle) { 1341 if (islittle == nativeendian.little) { 1342 while (size-- != 0) 1343 *(dest++) = *(src++); 1344 } 1345 else { 1346 dest += size - 1; 1347 while (size-- != 0) 1348 *(dest--) = *(src++); 1349 } 1350 } 1351 #endif /* _KERNEL */ 1352 1353 1354 static int str_pack (lua_State *L) { 1355 luaL_Buffer b; 1356 Header h; 1357 const char *fmt = luaL_checkstring(L, 1); /* format string */ 1358 int arg = 1; /* current argument to pack */ 1359 size_t totalsize = 0; /* accumulate total size of result */ 1360 initheader(L, &h); 1361 lua_pushnil(L); /* mark to separate arguments from string buffer */ 1362 luaL_buffinit(L, &b); 1363 while (*fmt != '\0') { 1364 int size, ntoalign; 1365 KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign); 1366 totalsize += ntoalign + size; 1367 while (ntoalign-- > 0) 1368 luaL_addchar(&b, LUAL_PACKPADBYTE); /* fill alignment */ 1369 arg++; 1370 switch (opt) { 1371 case Kint: { /* signed integers */ 1372 lua_Integer n = luaL_checkinteger(L, arg); 1373 if (size < SZINT) { /* need overflow check? */ 1374 lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1); 1375 luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow"); 1376 } 1377 packint(&b, (lua_Unsigned)n, h.islittle, size, (n < 0)); 1378 break; 1379 } 1380 case Kuint: { /* unsigned integers */ 1381 lua_Integer n = luaL_checkinteger(L, arg); 1382 if (size < SZINT) /* need overflow check? */ 1383 luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)), 1384 arg, "unsigned overflow"); 1385 packint(&b, (lua_Unsigned)n, h.islittle, size, 0); 1386 break; 1387 } 1388 #ifndef _KERNEL 1389 case Kfloat: { /* floating-point options */ 1390 volatile Ftypes u; 1391 char *buff = luaL_prepbuffsize(&b, size); 1392 lua_Number n = luaL_checknumber(L, arg); /* get argument */ 1393 if (size == sizeof(u.f)) u.f = (float)n; /* copy it into 'u' */ 1394 else if (size == sizeof(u.d)) u.d = (double)n; 1395 else u.n = n; 1396 /* move 'u' to final result, correcting endianness if needed */ 1397 copywithendian(buff, u.buff, size, h.islittle); 1398 luaL_addsize(&b, size); 1399 break; 1400 } 1401 #endif /* _KERNEL */ 1402 case Kchar: { /* fixed-size string */ 1403 size_t len; 1404 const char *s = luaL_checklstring(L, arg, &len); 1405 luaL_argcheck(L, len <= (size_t)size, arg, 1406 "string longer than given size"); 1407 luaL_addlstring(&b, s, len); /* add string */ 1408 while (len++ < (size_t)size) /* pad extra space */ 1409 luaL_addchar(&b, LUAL_PACKPADBYTE); 1410 break; 1411 } 1412 case Kstring: { /* strings with length count */ 1413 size_t len; 1414 const char *s = luaL_checklstring(L, arg, &len); 1415 luaL_argcheck(L, size >= (int)sizeof(size_t) || 1416 len < ((size_t)1 << (size * NB)), 1417 arg, "string length does not fit in given size"); 1418 packint(&b, (lua_Unsigned)len, h.islittle, size, 0); /* pack length */ 1419 luaL_addlstring(&b, s, len); 1420 totalsize += len; 1421 break; 1422 } 1423 case Kzstr: { /* zero-terminated string */ 1424 size_t len; 1425 const char *s = luaL_checklstring(L, arg, &len); 1426 luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros"); 1427 luaL_addlstring(&b, s, len); 1428 luaL_addchar(&b, '\0'); /* add zero at the end */ 1429 totalsize += len + 1; 1430 break; 1431 } 1432 case Kpadding: luaL_addchar(&b, LUAL_PACKPADBYTE); /* FALLTHROUGH */ 1433 case Kpaddalign: case Knop: 1434 arg--; /* undo increment */ 1435 break; 1436 } 1437 } 1438 luaL_pushresult(&b); 1439 return 1; 1440 } 1441 1442 1443 static int str_packsize (lua_State *L) { 1444 Header h; 1445 const char *fmt = luaL_checkstring(L, 1); /* format string */ 1446 size_t totalsize = 0; /* accumulate total size of result */ 1447 initheader(L, &h); 1448 while (*fmt != '\0') { 1449 int size, ntoalign; 1450 KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign); 1451 size += ntoalign; /* total space used by option */ 1452 luaL_argcheck(L, totalsize <= MAXSIZE - size, 1, 1453 "format result too large"); 1454 totalsize += size; 1455 switch (opt) { 1456 case Kstring: /* strings with length count */ 1457 case Kzstr: /* zero-terminated string */ 1458 luaL_argerror(L, 1, "variable-length format"); 1459 /* call never return, but to avoid warnings: *//* FALLTHROUGH */ 1460 default: break; 1461 } 1462 } 1463 lua_pushinteger(L, (lua_Integer)totalsize); 1464 return 1; 1465 } 1466 1467 1468 /* 1469 ** Unpack an integer with 'size' bytes and 'islittle' endianness. 1470 ** If size is smaller than the size of a Lua integer and integer 1471 ** is signed, must do sign extension (propagating the sign to the 1472 ** higher bits); if size is larger than the size of a Lua integer, 1473 ** it must check the unread bytes to see whether they do not cause an 1474 ** overflow. 1475 */ 1476 static lua_Integer unpackint (lua_State *L, const char *str, 1477 int islittle, int size, int issigned) { 1478 lua_Unsigned res = 0; 1479 int i; 1480 int limit = (size <= SZINT) ? size : SZINT; 1481 for (i = limit - 1; i >= 0; i--) { 1482 res <<= NB; 1483 res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i]; 1484 } 1485 if (size < SZINT) { /* real size smaller than lua_Integer? */ 1486 if (issigned) { /* needs sign extension? */ 1487 lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1); 1488 res = ((res ^ mask) - mask); /* do sign extension */ 1489 } 1490 } 1491 else if (size > SZINT) { /* must check unread bytes */ 1492 int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC; 1493 for (i = limit; i < size; i++) { 1494 if ((unsigned char)str[islittle ? i : size - 1 - i] != mask) 1495 luaL_error(L, "%d-byte integer does not fit into Lua Integer", size); 1496 } 1497 } 1498 return (lua_Integer)res; 1499 } 1500 1501 1502 static int str_unpack (lua_State *L) { 1503 Header h; 1504 const char *fmt = luaL_checkstring(L, 1); 1505 size_t ld; 1506 const char *data = luaL_checklstring(L, 2, &ld); 1507 size_t pos = (size_t)posrelat(luaL_optinteger(L, 3, 1), ld) - 1; 1508 int n = 0; /* number of results */ 1509 luaL_argcheck(L, pos <= ld, 3, "initial position out of string"); 1510 initheader(L, &h); 1511 while (*fmt != '\0') { 1512 int size, ntoalign; 1513 KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign); 1514 if ((size_t)ntoalign + size > ~pos || pos + ntoalign + size > ld) 1515 luaL_argerror(L, 2, "data string too short"); 1516 pos += ntoalign; /* skip alignment */ 1517 /* stack space for item + next position */ 1518 luaL_checkstack(L, 2, "too many results"); 1519 n++; 1520 switch (opt) { 1521 case Kint: 1522 case Kuint: { 1523 lua_Integer res = unpackint(L, data + pos, h.islittle, size, 1524 (opt == Kint)); 1525 lua_pushinteger(L, res); 1526 break; 1527 } 1528 #ifndef _KERNEL 1529 case Kfloat: { 1530 volatile Ftypes u; 1531 lua_Number num; 1532 copywithendian(u.buff, data + pos, size, h.islittle); 1533 if (size == sizeof(u.f)) num = (lua_Number)u.f; 1534 else if (size == sizeof(u.d)) num = (lua_Number)u.d; 1535 else num = u.n; 1536 lua_pushnumber(L, num); 1537 break; 1538 } 1539 #endif /* _KERNEL */ 1540 case Kchar: { 1541 lua_pushlstring(L, data + pos, size); 1542 break; 1543 } 1544 case Kstring: { 1545 size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0); 1546 luaL_argcheck(L, pos + len + size <= ld, 2, "data string too short"); 1547 lua_pushlstring(L, data + pos + size, len); 1548 pos += len; /* skip string */ 1549 break; 1550 } 1551 case Kzstr: { 1552 size_t len = (int)strlen(data + pos); 1553 lua_pushlstring(L, data + pos, len); 1554 pos += len + 1; /* skip string plus final '\0' */ 1555 break; 1556 } 1557 case Kpaddalign: case Kpadding: case Knop: 1558 n--; /* undo increment */ 1559 break; 1560 } 1561 pos += size; 1562 } 1563 lua_pushinteger(L, pos + 1); /* next position */ 1564 return n + 1; 1565 } 1566 1567 /* }====================================================== */ 1568 1569 1570 static const luaL_Reg strlib[] = { 1571 {"byte", str_byte}, 1572 {"char", str_char}, 1573 {"dump", str_dump}, 1574 {"find", str_find}, 1575 {"format", str_format}, 1576 {"gmatch", gmatch}, 1577 {"gsub", str_gsub}, 1578 {"len", str_len}, 1579 {"lower", str_lower}, 1580 {"match", str_match}, 1581 {"rep", str_rep}, 1582 {"reverse", str_reverse}, 1583 {"sub", str_sub}, 1584 {"upper", str_upper}, 1585 {"pack", str_pack}, 1586 {"packsize", str_packsize}, 1587 {"unpack", str_unpack}, 1588 {NULL, NULL} 1589 }; 1590 1591 1592 static void createmetatable (lua_State *L) { 1593 lua_createtable(L, 0, 1); /* table to be metatable for strings */ 1594 lua_pushliteral(L, ""); /* dummy string */ 1595 lua_pushvalue(L, -2); /* copy table */ 1596 lua_setmetatable(L, -2); /* set table as metatable for strings */ 1597 lua_pop(L, 1); /* pop dummy string */ 1598 lua_pushvalue(L, -2); /* get string library */ 1599 lua_setfield(L, -2, "__index"); /* metatable.__index = string */ 1600 lua_pop(L, 1); /* pop metatable */ 1601 } 1602 1603 1604 /* 1605 ** Open string library 1606 */ 1607 LUAMOD_API int luaopen_string (lua_State *L) { 1608 luaL_newlib(L, strlib); 1609 createmetatable(L); 1610 return 1; 1611 } 1612 1613