usr/src/uts/common/fs/zfs/lua/lstrlib.c

   1 /*
   2 ** $Id: lstrlib.c,v 1.178.1.1 2013/04/12 18:48:47 roberto Exp $
   3 ** Standard library for string operations and pattern-matching
   4 ** See Copyright Notice in lua.h
   5 */
   6
   7
   8 #include <sys/ctype.h>
   9 #include <sys/zfs_context.h>
  10
  11 #define lstrlib_c
  12 #define LUA_LIB
  13
  14 #include "lua.h"
  15
  16 #include "lauxlib.h"
  17 #include "lualib.h"
  18
  19
  20 /*
  21 ** maximum number of captures that a pattern can do during
  22 ** pattern-matching. This limit is arbitrary.
  23 */
  24 #if !defined(LUA_MAXCAPTURES)
  25 #define LUA_MAXCAPTURES         32
  26 #endif
  27
  28
  29 /* macro to `unsign' a character */
  30 #define uchar(c)        ((unsigned char)(c))
  31
  32 /*
  33  * PATCHED: add missing character macros.
  34  */
  35 #define tolower(C)      (((C) >= 'A' && (C) <= 'Z') ? (C) - 'A' + 'a' : (C))
  36 #define toupper(C)      (((C) >= 'a' && (C) <= 'z') ? (C) - 'a' + 'A': (C))
  37 #define iscntrl(C)      ((((C) >= 0) && ((C) <= 0x1f)) || ((C) == 0x7f))
  38 #define isgraph(C)      ((C) >= 0x21 && (C) <= 0x7E)
  39 #define ispunct(C)      (((C) >= 0x21 && (C) <= 0x2F) || \
  40     ((C) >= 0x3A && (C) <= 0x40) || \
  41     ((C) >= 0x5B && (C) <= 0x60) || \
  42     ((C) >= 0x7B && (C) <= 0x7E))
  43
  44 /*
  45  * The provided version of sprintf returns a char *, but str_format expects
  46  * it to return the number of characters printed. This version has the expected
  47  * behavior.
  48  */
  49 static size_t str_sprintf(char *buf, const char *fmt, ...) {
  50   va_list args;
  51   size_t len;
  52
  53   va_start(args, fmt);
  54   len = vsnprintf(buf, INT_MAX, fmt, args);
  55   va_end(args);
  56
  57   return len;
  58 }
  59
  60
  61 static int str_len (lua_State *L) {
  62   size_t l;
  63   luaL_checklstring(L, 1, &l);
  64   lua_pushinteger(L, (lua_Integer)l);
  65   return 1;
  66 }
  67
  68
  69 /* translate a relative string position: negative means back from end */
  70 static size_t posrelat (ptrdiff_t pos, size_t len) {
  71   if (pos >= 0) return (size_t)pos;
  72   else if (0u - (size_t)pos > len) return 0;
  73   else return len - ((size_t)-pos) + 1;
  74 }
  75
  76
  77 static int str_sub (lua_State *L) {
  78   size_t l;
  79   const char *s = luaL_checklstring(L, 1, &l);
  80   size_t start = posrelat(luaL_checkinteger(L, 2), l);
  81   size_t end = posrelat(luaL_optinteger(L, 3, -1), l);
  82   if (start < 1) start = 1;
  83   if (end > l) end = l;
  84   if (start <= end)
  85     lua_pushlstring(L, s + start - 1, end - start + 1);
  86   else lua_pushliteral(L, "");
  87   return 1;
  88 }
  89
  90
  91 static int str_reverse (lua_State *L) {
  92   size_t l, i;
  93   luaL_Buffer b;
  94   const char *s = luaL_checklstring(L, 1, &l);
  95   char *p = luaL_buffinitsize(L, &b, l);
  96   for (i = 0; i < l; i++)
  97     p[i] = s[l - i - 1];
  98   luaL_pushresultsize(&b, l);
  99   return 1;
 100 }
 101
 102
 103 static int str_lower (lua_State *L) {
 104   size_t l;
 105   size_t i;
 106   luaL_Buffer b;
 107   const char *s = luaL_checklstring(L, 1, &l);
 108   char *p = luaL_buffinitsize(L, &b, l);
 109   for (i=0; i<l; i++)
 110     p[i] = tolower(uchar(s[i]));
 111   luaL_pushresultsize(&b, l);
 112   return 1;
 113 }
 114
 115
 116 static int str_upper (lua_State *L) {
 117   size_t l;
 118   size_t i;
 119   luaL_Buffer b;
 120   const char *s = luaL_checklstring(L, 1, &l);
 121   char *p = luaL_buffinitsize(L, &b, l);
 122   for (i=0; i<l; i++)
 123     p[i] = toupper(uchar(s[i]));
 124   luaL_pushresultsize(&b, l);
 125   return 1;
 126 }
 127
 128
 129 /* reasonable limit to avoid arithmetic overflow */
 130 #define MAXSIZE         ((~(size_t)0) >> 1)
 131
 132 static int str_rep (lua_State *L) {
 133   size_t l, lsep;
 134   const char *s = luaL_checklstring(L, 1, &l);
 135   int n = luaL_checkint(L, 2);
 136   const char *sep = luaL_optlstring(L, 3, "", &lsep);
 137   if (n <= 0) lua_pushliteral(L, "");
 138   else if (l + lsep < l || l + lsep >= MAXSIZE / n)  /* may overflow? */
 139     return luaL_error(L, "resulting string too large");
 140   else {
 141     size_t totallen = n * l + (n - 1) * lsep;
 142     luaL_Buffer b;
 143     char *p = luaL_buffinitsize(L, &b, totallen);
 144     while (n-- > 1) {  /* first n-1 copies (followed by separator) */
 145       memcpy(p, s, l * sizeof(char)); p += l;
 146       if (lsep > 0) {  /* avoid empty 'memcpy' (may be expensive) */
 147         memcpy(p, sep, lsep * sizeof(char)); p += lsep;
 148       }
 149     }
 150     memcpy(p, s, l * sizeof(char));  /* last copy (not followed by separator) */
 151     luaL_pushresultsize(&b, totallen);
 152   }
 153   return 1;
 154 }
 155
 156
 157 static int str_byte (lua_State *L) {
 158   size_t l;
 159   const char *s = luaL_checklstring(L, 1, &l);
 160   size_t posi = posrelat(luaL_optinteger(L, 2, 1), l);
 161   size_t pose = posrelat(luaL_optinteger(L, 3, posi), l);
 162   int n, i;
 163   if (posi < 1) posi = 1;
 164   if (pose > l) pose = l;
 165   if (posi > pose) return 0;  /* empty interval; return no values */
 166   n = (int)(pose -  posi + 1);
 167   if (posi + n <= pose)  /* (size_t -> int) overflow? */
 168     return luaL_error(L, "string slice too long");
 169   luaL_checkstack(L, n, "string slice too long");
 170   for (i=0; i<n; i++)
 171     lua_pushinteger(L, uchar(s[posi+i-1]));
 172   return n;
 173 }
 174
 175
 176 static int str_char (lua_State *L) {
 177   int n = lua_gettop(L);  /* number of arguments */
 178   int i;
 179   luaL_Buffer b;
 180   char *p = luaL_buffinitsize(L, &b, n);
 181   for (i=1; i<=n; i++) {
 182     int c = luaL_checkint(L, i);
 183     luaL_argcheck(L, uchar(c) == c, i, "value out of range");
 184     p[i - 1] = uchar(c);
 185   }
 186   luaL_pushresultsize(&b, n);
 187   return 1;
 188 }
 189
 190
 191 static int writer (lua_State *L, const void* b, size_t size, void* B) {
 192   (void)L;
 193   luaL_addlstring((luaL_Buffer*) B, (const char *)b, size);
 194   return 0;
 195 }
 196
 197
 198 static int str_dump (lua_State *L) {
 199   luaL_Buffer b;
 200   luaL_checktype(L, 1, LUA_TFUNCTION);
 201   lua_settop(L, 1);
 202   luaL_buffinit(L,&b);
 203   if (lua_dump(L, writer, &b) != 0)
 204     return luaL_error(L, "unable to dump given function");
 205   luaL_pushresult(&b);
 206   return 1;
 207 }
 208
 209
 210
 211 /*
 212 ** {======================================================
 213 ** PATTERN MATCHING
 214 ** =======================================================
 215 */
 216
 217
 218 #define CAP_UNFINISHED  (-1)
 219 #define CAP_POSITION    (-2)
 220
 221
 222 typedef struct MatchState {
 223   int matchdepth;  /* control for recursive depth (to avoid C stack overflow) */
 224   const char *src_init;  /* init of source string */
 225   const char *src_end;  /* end ('\0') of source string */
 226   const char *p_end;  /* end ('\0') of pattern */
 227   lua_State *L;
 228   int level;  /* total number of captures (finished or unfinished) */
 229   struct {
 230     const char *init;
 231     ptrdiff_t len;
 232   } capture[LUA_MAXCAPTURES];
 233 } MatchState;
 234
 235
 236 /* recursive function */
 237 static const char *match (MatchState *ms, const char *s, const char *p);
 238
 239
 240 /* maximum recursion depth for 'match' */
 241 #if !defined(MAXCCALLS)
 242 #define MAXCCALLS       200
 243 #endif
 244
 245
 246 #define L_ESC           '%'
 247 #define SPECIALS        "^$*+?.([%-"
 248
 249
 250 static int check_capture (MatchState *ms, int l) {
 251   l -= '1';
 252   if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
 253     return luaL_error(ms->L, "invalid capture index %%%d", l + 1);
 254   return l;
 255 }
 256
 257
 258 static int capture_to_close (MatchState *ms) {
 259   int level = ms->level;
 260   for (level--; level>=0; level--)
 261     if (ms->capture[level].len == CAP_UNFINISHED) return level;
 262   return luaL_error(ms->L, "invalid pattern capture");
 263 }
 264
 265
 266 static const char *classend (MatchState *ms, const char *p) {
 267   switch (*p++) {
 268     case L_ESC: {
 269       if (p == ms->p_end)
 270         luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");
 271       return p+1;
 272     }
 273     case '[': {
 274       if (*p == '^') p++;
 275       do {  /* look for a `]' */
 276         if (p == ms->p_end)
 277           luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");
 278         if (*(p++) == L_ESC && p < ms->p_end)
 279           p++;  /* skip escapes (e.g. `%]') */
 280       } while (*p != ']');
 281       return p+1;
 282     }
 283     default: {
 284       return p;
 285     }
 286   }
 287 }
 288
 289
 290 static int match_class (int c, int cl) {
 291   int res;
 292   switch (tolower(cl)) {
 293     case 'a' : res = isalpha(c); break;
 294     case 'c' : res = iscntrl(c); break;
 295     case 'd' : res = isdigit(c); break;
 296     case 'g' : res = isgraph(c); break;
 297     case 'l' : res = islower(c); break;
 298     case 'p' : res = ispunct(c); break;
 299     case 's' : res = isspace(c); break;
 300     case 'u' : res = isupper(c); break;
 301     case 'w' : res = isalnum(c); break;
 302     case 'x' : res = isxdigit(c); break;
 303     case 'z' : res = (c == 0); break;  /* deprecated option */
 304     default: return (cl == c);
 305   }
 306   return (islower(cl) ? res : !res);
 307 }
 308
 309
 310 static int matchbracketclass (int c, const char *p, const char *ec) {
 311   int sig = 1;
 312   if (*(p+1) == '^') {
 313     sig = 0;
 314     p++;  /* skip the `^' */
 315   }
 316   while (++p < ec) {
 317     if (*p == L_ESC) {
 318       p++;
 319       if (match_class(c, uchar(*p)))
 320         return sig;
 321     }
 322     else if ((*(p+1) == '-') && (p+2 < ec)) {
 323       p+=2;
 324       if (uchar(*(p-2)) <= c && c <= uchar(*p))
 325         return sig;
 326     }
 327     else if (uchar(*p) == c) return sig;
 328   }
 329   return !sig;
 330 }
 331
 332
 333 static int singlematch (MatchState *ms, const char *s, const char *p,
 334                         const char *ep) {
 335   if (s >= ms->src_end)
 336     return 0;
 337   else {
 338     int c = uchar(*s);
 339     switch (*p) {
 340       case '.': return 1;  /* matches any char */
 341       case L_ESC: return match_class(c, uchar(*(p+1)));
 342       case '[': return matchbracketclass(c, p, ep-1);
 343       default:  return (uchar(*p) == c);
 344     }
 345   }
 346 }
 347
 348
 349 static const char *matchbalance (MatchState *ms, const char *s,
 350                                    const char *p) {
 351   if (p >= ms->p_end - 1)
 352     luaL_error(ms->L, "malformed pattern "
 353                       "(missing arguments to " LUA_QL("%%b") ")");
 354   if (*s != *p) return NULL;
 355   else {
 356     int b = *p;
 357     int e = *(p+1);
 358     int cont = 1;
 359     while (++s < ms->src_end) {
 360       if (*s == e) {
 361         if (--cont == 0) return s+1;
 362       }
 363       else if (*s == b) cont++;
 364     }
 365   }
 366   return NULL;  /* string ends out of balance */
 367 }
 368
 369
 370 static const char *max_expand (MatchState *ms, const char *s,
 371                                  const char *p, const char *ep) {
 372   ptrdiff_t i = 0;  /* counts maximum expand for item */
 373   while (singlematch(ms, s + i, p, ep))
 374     i++;
 375   /* keeps trying to match with the maximum repetitions */
 376   while (i>=0) {
 377     const char *res = match(ms, (s+i), ep+1);
 378     if (res) return res;
 379     i--;  /* else didn't match; reduce 1 repetition to try again */
 380   }
 381   return NULL;
 382 }
 383
 384
 385 static const char *min_expand (MatchState *ms, const char *s,
 386                                  const char *p, const char *ep) {
 387   for (;;) {
 388     const char *res = match(ms, s, ep+1);
 389     if (res != NULL)
 390       return res;
 391     else if (singlematch(ms, s, p, ep))
 392       s++;  /* try with one more repetition */
 393     else return NULL;
 394   }
 395 }
 396
 397
 398 static const char *start_capture (MatchState *ms, const char *s,
 399                                     const char *p, int what) {
 400   const char *res;
 401   int level = ms->level;
 402   if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
 403   ms->capture[level].init = s;
 404   ms->capture[level].len = what;
 405   ms->level = level+1;
 406   if ((res=match(ms, s, p)) == NULL)  /* match failed? */
 407     ms->level--;  /* undo capture */
 408   return res;
 409 }
 410
 411
 412 static const char *end_capture (MatchState *ms, const char *s,
 413                                   const char *p) {
 414   int l = capture_to_close(ms);
 415   const char *res;
 416   ms->capture[l].len = s - ms->capture[l].init;  /* close capture */
 417   if ((res = match(ms, s, p)) == NULL)  /* match failed? */
 418     ms->capture[l].len = CAP_UNFINISHED;  /* undo capture */
 419   return res;
 420 }
 421
 422
 423 static const char *match_capture (MatchState *ms, const char *s, int l) {
 424   size_t len;
 425   l = check_capture(ms, l);
 426   len = ms->capture[l].len;
 427   if ((size_t)(ms->src_end-s) >= len &&
 428       memcmp(ms->capture[l].init, s, len) == 0)
 429     return s+len;
 430   else return NULL;
 431 }
 432
 433
 434 static const char *match (MatchState *ms, const char *s, const char *p) {
 435   if (ms->matchdepth-- == 0)
 436     luaL_error(ms->L, "pattern too complex");
 437   init: /* using goto's to optimize tail recursion */
 438   if (p != ms->p_end) {  /* end of pattern? */
 439     switch (*p) {
 440       case '(': {  /* start capture */
 441         if (*(p + 1) == ')')  /* position capture? */
 442           s = start_capture(ms, s, p + 2, CAP_POSITION);
 443         else
 444           s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
 445         break;
 446       }
 447       case ')': {  /* end capture */
 448         s = end_capture(ms, s, p + 1);
 449         break;
 450       }
 451       case '$': {
 452         if ((p + 1) != ms->p_end)  /* is the `$' the last char in pattern? */
 453           goto dflt;  /* no; go to default */
 454         s = (s == ms->src_end) ? s : NULL;  /* check end of string */
 455         break;
 456       }
 457       case L_ESC: {  /* escaped sequences not in the format class[*+?-]? */
 458         switch (*(p + 1)) {
 459           case 'b': {  /* balanced string? */
 460             s = matchbalance(ms, s, p + 2);
 461             if (s != NULL) {
 462               p += 4; goto init;  /* return match(ms, s, p + 4); */
 463             }  /* else fail (s == NULL) */
 464             break;
 465           }
 466           case 'f': {  /* frontier? */
 467             const char *ep; char previous;
 468             p += 2;
 469             if (*p != '[')
 470               luaL_error(ms->L, "missing " LUA_QL("[") " after "
 471                                  LUA_QL("%%f") " in pattern");
 472             ep = classend(ms, p);  /* points to what is next */
 473             previous = (s == ms->src_init) ? '\0' : *(s - 1);
 474             if (!matchbracketclass(uchar(previous), p, ep - 1) &&
 475                matchbracketclass(uchar(*s), p, ep - 1)) {
 476               p = ep; goto init;  /* return match(ms, s, ep); */
 477             }
 478             s = NULL;  /* match failed */
 479             break;
 480           }
 481           case '0': case '1': case '2': case '3':
 482           case '4': case '5': case '6': case '7':
 483           case '8': case '9': {  /* capture results (%0-%9)? */
 484             s = match_capture(ms, s, uchar(*(p + 1)));
 485             if (s != NULL) {
 486               p += 2; goto init;  /* return match(ms, s, p + 2) */
 487             }
 488             break;
 489           }
 490           default: goto dflt;
 491         }
 492         break;
 493       }
 494       default: dflt: {  /* pattern class plus optional suffix */
 495         const char *ep = classend(ms, p);  /* points to optional suffix */
 496         /* does not match at least once? */
 497         if (!singlematch(ms, s, p, ep)) {
 498           if (*ep == '*' || *ep == '?' || *ep == '-') {  /* accept empty? */
 499             p = ep + 1; goto init;  /* return match(ms, s, ep + 1); */
 500           }
 501           else  /* '+' or no suffix */
 502             s = NULL;  /* fail */
 503         }
 504         else {  /* matched once */
 505           switch (*ep) {  /* handle optional suffix */
 506             case '?': {  /* optional */
 507               const char *res;
 508               if ((res = match(ms, s + 1, ep + 1)) != NULL)
 509                 s = res;
 510               else {
 511                 p = ep + 1; goto init;  /* else return match(ms, s, ep + 1); */
 512               }
 513               break;
 514             }
 515             case '+':  /* 1 or more repetitions */
 516               s++;  /* 1 match already done */
 517               /* go through */
 518             case '*':  /* 0 or more repetitions */
 519               s = max_expand(ms, s, p, ep);
 520               break;
 521             case '-':  /* 0 or more repetitions (minimum) */
 522               s = min_expand(ms, s, p, ep);
 523               break;
 524             default:  /* no suffix */
 525               s++; p = ep; goto init;  /* return match(ms, s + 1, ep); */
 526           }
 527         }
 528         break;
 529       }
 530     }
 531   }
 532   ms->matchdepth++;
 533   return s;
 534 }
 535
 536
 537
 538 static const char *lmemfind (const char *s1, size_t l1,
 539                                const char *s2, size_t l2) {
 540   if (l2 == 0) return s1;  /* empty strings are everywhere */
 541   else if (l2 > l1) return NULL;  /* avoids a negative `l1' */
 542   else {
 543     const char *init;  /* to search for a `*s2' inside `s1' */
 544     l2--;  /* 1st char will be checked by `memchr' */
 545     l1 = l1-l2;  /* `s2' cannot be found after that */
 546     while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
 547       init++;   /* 1st char is already checked */
 548       if (memcmp(init, s2+1, l2) == 0)
 549         return init-1;
 550       else {  /* correct `l1' and `s1' to try again */
 551         l1 -= init-s1;
 552         s1 = init;
 553       }
 554     }
 555     return NULL;  /* not found */
 556   }
 557 }
 558
 559
 560 static void push_onecapture (MatchState *ms, int i, const char *s,
 561                                                     const char *e) {
 562   if (i >= ms->level) {
 563     if (i == 0)  /* ms->level == 0, too */
 564       lua_pushlstring(ms->L, s, e - s);  /* add whole match */
 565     else
 566       luaL_error(ms->L, "invalid capture index");
 567   }
 568   else {
 569     ptrdiff_t l = ms->capture[i].len;
 570     if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
 571     if (l == CAP_POSITION)
 572       lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
 573     else
 574       lua_pushlstring(ms->L, ms->capture[i].init, l);
 575   }
 576 }
 577
 578
 579 static int push_captures (MatchState *ms, const char *s, const char *e) {
 580   int i;
 581   int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
 582   luaL_checkstack(ms->L, nlevels, "too many captures");
 583   for (i = 0; i < nlevels; i++)
 584     push_onecapture(ms, i, s, e);
 585   return nlevels;  /* number of strings pushed */
 586 }
 587
 588
 589 /* check whether pattern has no special characters */
 590 static int nospecials (const char *p, size_t l) {
 591   size_t upto = 0;
 592   do {
 593     if (strpbrk(p + upto, SPECIALS))
 594       return 0;  /* pattern has a special character */
 595     upto += strlen(p + upto) + 1;  /* may have more after \0 */
 596   } while (upto <= l);
 597   return 1;  /* no special chars found */
 598 }
 599
 600
 601 static int str_find_aux (lua_State *L, int find) {
 602   size_t ls, lp;
 603   const char *s = luaL_checklstring(L, 1, &ls);
 604   const char *p = luaL_checklstring(L, 2, &lp);
 605   size_t init = posrelat(luaL_optinteger(L, 3, 1), ls);
 606   if (init < 1) init = 1;
 607   else if (init > ls + 1) {  /* start after string's end? */
 608     lua_pushnil(L);  /* cannot find anything */
 609     return 1;
 610   }
 611   /* explicit request or no special characters? */
 612   if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {
 613     /* do a plain search */
 614     const char *s2 = lmemfind(s + init - 1, ls - init + 1, p, lp);
 615     if (s2) {
 616       lua_pushinteger(L, s2 - s + 1);
 617       lua_pushinteger(L, s2 - s + lp);
 618       return 2;
 619     }
 620   }
 621   else {
 622     MatchState ms;
 623     const char *s1 = s + init - 1;
 624     int anchor = (*p == '^');
 625     if (anchor) {
 626       p++; lp--;  /* skip anchor character */
 627     }
 628     ms.L = L;
 629     ms.matchdepth = MAXCCALLS;
 630     ms.src_init = s;
 631     ms.src_end = s + ls;
 632     ms.p_end = p + lp;
 633     do {
 634       const char *res;
 635       ms.level = 0;
 636       lua_assert(ms.matchdepth == MAXCCALLS);
 637       if ((res=match(&ms, s1, p)) != NULL) {
 638         if (find) {
 639           lua_pushinteger(L, s1 - s + 1);  /* start */
 640           lua_pushinteger(L, res - s);   /* end */
 641           return push_captures(&ms, NULL, 0) + 2;
 642         }
 643         else
 644           return push_captures(&ms, s1, res);
 645       }
 646     } while (s1++ < ms.src_end && !anchor);
 647   }
 648   lua_pushnil(L);  /* not found */
 649   return 1;
 650 }
 651
 652
 653 static int str_find (lua_State *L) {
 654   return str_find_aux(L, 1);
 655 }
 656
 657
 658 static int str_match (lua_State *L) {
 659   return str_find_aux(L, 0);
 660 }
 661
 662
 663 static int gmatch_aux (lua_State *L) {
 664   MatchState ms;
 665   size_t ls, lp;
 666   const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
 667   const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp);
 668   const char *src;
 669   ms.L = L;
 670   ms.matchdepth = MAXCCALLS;
 671   ms.src_init = s;
 672   ms.src_end = s+ls;
 673   ms.p_end = p + lp;
 674   for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
 675        src <= ms.src_end;
 676        src++) {
 677     const char *e;
 678     ms.level = 0;
 679     lua_assert(ms.matchdepth == MAXCCALLS);
 680     if ((e = match(&ms, src, p)) != NULL) {
 681       lua_Integer newstart = e-s;
 682       if (e == src) newstart++;  /* empty match? go at least one position */
 683       lua_pushinteger(L, newstart);
 684       lua_replace(L, lua_upvalueindex(3));
 685       return push_captures(&ms, src, e);
 686     }
 687   }
 688   return 0;  /* not found */
 689 }
 690
 691
 692 static int str_gmatch (lua_State *L) {
 693   luaL_checkstring(L, 1);
 694   luaL_checkstring(L, 2);
 695   lua_settop(L, 2);
 696   lua_pushinteger(L, 0);
 697   lua_pushcclosure(L, gmatch_aux, 3);
 698   return 1;
 699 }
 700
 701
 702 static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
 703                                                    const char *e) {
 704   size_t l, i;
 705   const char *news = lua_tolstring(ms->L, 3, &l);
 706   for (i = 0; i < l; i++) {
 707     if (news[i] != L_ESC)
 708       luaL_addchar(b, news[i]);
 709     else {
 710       i++;  /* skip ESC */
 711       if (!isdigit(uchar(news[i]))) {
 712         if (news[i] != L_ESC)
 713           luaL_error(ms->L, "invalid use of " LUA_QL("%c")
 714                            " in replacement string", L_ESC);
 715         luaL_addchar(b, news[i]);
 716       }
 717       else if (news[i] == '0')
 718           luaL_addlstring(b, s, e - s);
 719       else {
 720         push_onecapture(ms, news[i] - '1', s, e);
 721         luaL_addvalue(b);  /* add capture to accumulated result */
 722       }
 723     }
 724   }
 725 }
 726
 727
 728 static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
 729                                        const char *e, int tr) {
 730   lua_State *L = ms->L;
 731   switch (tr) {
 732     case LUA_TFUNCTION: {
 733       int n;
 734       lua_pushvalue(L, 3);
 735       n = push_captures(ms, s, e);
 736       lua_call(L, n, 1);
 737       break;
 738     }
 739     case LUA_TTABLE: {
 740       push_onecapture(ms, 0, s, e);
 741       lua_gettable(L, 3);
 742       break;
 743     }
 744     default: {  /* LUA_TNUMBER or LUA_TSTRING */
 745       add_s(ms, b, s, e);
 746       return;
 747     }
 748   }
 749   if (!lua_toboolean(L, -1)) {  /* nil or false? */
 750     lua_pop(L, 1);
 751     lua_pushlstring(L, s, e - s);  /* keep original text */
 752   }
 753   else if (!lua_isstring(L, -1))
 754     luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
 755   luaL_addvalue(b);  /* add result to accumulator */
 756 }
 757
 758
 759 static int str_gsub (lua_State *L) {
 760   size_t srcl, lp;
 761   const char *src = luaL_checklstring(L, 1, &srcl);
 762   const char *p = luaL_checklstring(L, 2, &lp);
 763   int tr = lua_type(L, 3);
 764   size_t max_s = luaL_optinteger(L, 4, srcl+1);
 765   int anchor = (*p == '^');
 766   size_t n = 0;
 767   MatchState ms;
 768   luaL_Buffer b;
 769   luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
 770                    tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
 771                       "string/function/table expected");
 772   luaL_buffinit(L, &b);
 773   if (anchor) {
 774     p++; lp--;  /* skip anchor character */
 775   }
 776   ms.L = L;
 777   ms.matchdepth = MAXCCALLS;
 778   ms.src_init = src;
 779   ms.src_end = src+srcl;
 780   ms.p_end = p + lp;
 781   while (n < max_s) {
 782     const char *e;
 783     ms.level = 0;
 784     lua_assert(ms.matchdepth == MAXCCALLS);
 785     e = match(&ms, src, p);
 786     if (e) {
 787       n++;
 788       add_value(&ms, &b, src, e, tr);
 789     }
 790     if (e && e>src) /* non empty match? */
 791       src = e;  /* skip it */
 792     else if (src < ms.src_end)
 793       luaL_addchar(&b, *src++);
 794     else break;
 795     if (anchor) break;
 796   }
 797   luaL_addlstring(&b, src, ms.src_end-src);
 798   luaL_pushresult(&b);
 799   lua_pushinteger(L, n);  /* number of substitutions */
 800   return 2;
 801 }
 802
 803 /* }====================================================== */
 804
 805
 806
 807 /*
 808 ** {======================================================
 809 ** STRING FORMAT
 810 ** =======================================================
 811 */
 812
 813 /*
 814 ** LUA_INTFRMLEN is the length modifier for integer conversions in
 815 ** 'string.format'; LUA_INTFRM_T is the integer type corresponding to
 816 ** the previous length
 817 */
 818 #if !defined(LUA_INTFRMLEN)     /* { */
 819 #if defined(LUA_USE_LONGLONG)
 820
 821 #define LUA_INTFRMLEN           "ll"
 822 #define LUA_INTFRM_T            long long
 823
 824 #else
 825
 826 #define LUA_INTFRMLEN           "l"
 827 #define LUA_INTFRM_T            long
 828
 829 #endif
 830 #endif                          /* } */
 831
 832
 833 /*
 834 ** LUA_FLTFRMLEN is the length modifier for float conversions in
 835 ** 'string.format'; LUA_FLTFRM_T is the float type corresponding to
 836 ** the previous length
 837 */
 838 #if !defined(LUA_FLTFRMLEN)
 839
 840 #define LUA_FLTFRMLEN           ""
 841 #define LUA_FLTFRM_T            double
 842
 843 #endif
 844
 845
 846 /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
 847 #define MAX_ITEM        512
 848 /* valid flags in a format specification */
 849 #define FLAGS   "-+ #0"
 850 /*
 851 ** maximum size of each format specification (such as '%-099.99d')
 852 ** (+10 accounts for %99.99x plus margin of error)
 853 */
 854 #define MAX_FORMAT      (sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
 855
 856
 857 static void addquoted (lua_State *L, luaL_Buffer *b, int arg) {
 858   size_t l;
 859   const char *s = luaL_checklstring(L, arg, &l);
 860   luaL_addchar(b, '"');
 861   while (l--) {
 862     if (*s == '"' || *s == '\\' || *s == '\n') {
 863       luaL_addchar(b, '\\');
 864       luaL_addchar(b, *s);
 865     }
 866     else if (*s == '\0' || iscntrl(uchar(*s))) {
 867       char buff[10];
 868       if (!isdigit(uchar(*(s+1))))
 869         sprintf(buff, "\\%d", (int)uchar(*s));
 870       else
 871         sprintf(buff, "\\%03d", (int)uchar(*s));
 872       luaL_addstring(b, buff);
 873     }
 874     else
 875       luaL_addchar(b, *s);
 876     s++;
 877   }
 878   luaL_addchar(b, '"');
 879 }
 880
 881 static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
 882   const char *p = strfrmt;
 883   while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++;  /* skip flags */
 884   if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char))
 885     luaL_error(L, "invalid format (repeated flags)");
 886   if (isdigit(uchar(*p))) p++;  /* skip width */
 887   if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 888   if (*p == '.') {
 889     p++;
 890     if (isdigit(uchar(*p))) p++;  /* skip precision */
 891     if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 892   }
 893   if (isdigit(uchar(*p)))
 894     luaL_error(L, "invalid format (width or precision too long)");
 895   *(form++) = '%';
 896   memcpy(form, strfrmt, (p - strfrmt + 1) * sizeof(char));
 897   form += p - strfrmt + 1;
 898   *form = '\0';
 899   return p;
 900 }
 901
 902
 903 /*
 904 ** add length modifier into formats
 905 */
 906 static void addlenmod (char *form, const char *lenmod) {
 907   size_t l = strlen(form);
 908   size_t lm = strlen(lenmod);
 909   char spec = form[l - 1];
 910   strcpy(form + l - 1, lenmod);
 911   form[l + lm - 1] = spec;
 912   form[l + lm] = '\0';
 913 }
 914
 915
 916 static int str_format (lua_State *L) {
 917   int top = lua_gettop(L);
 918   int arg = 1;
 919   size_t sfl;
 920   const char *strfrmt = luaL_checklstring(L, arg, &sfl);
 921   const char *strfrmt_end = strfrmt+sfl;
 922   luaL_Buffer b;
 923   luaL_buffinit(L, &b);
 924   while (strfrmt < strfrmt_end) {
 925     if (*strfrmt != L_ESC)
 926       luaL_addchar(&b, *strfrmt++);
 927     else if (*++strfrmt == L_ESC)
 928       luaL_addchar(&b, *strfrmt++);  /* %% */
 929     else { /* format item */
 930       char form[MAX_FORMAT];  /* to store the format (`%...') */
 931       char *buff = luaL_prepbuffsize(&b, MAX_ITEM);  /* to put formatted item */
 932       int nb = 0;  /* number of bytes in added item */
 933       if (++arg > top)
 934         luaL_argerror(L, arg, "no value");
 935       strfrmt = scanformat(L, strfrmt, form);
 936       switch (*strfrmt++) {
 937         case 'c': {
 938           nb = str_sprintf(buff, form, luaL_checkint(L, arg));
 939           break;
 940         }
 941         case 'd': case 'i': {
 942           lua_Number n = luaL_checknumber(L, arg);
 943           LUA_INTFRM_T ni = (LUA_INTFRM_T)n;
 944           lua_Number diff = n - (lua_Number)ni;
 945           luaL_argcheck(L, -1 < diff && diff < 1, arg,
 946                         "not a number in proper range");
 947           addlenmod(form, LUA_INTFRMLEN);
 948           nb = str_sprintf(buff, form, ni);
 949           break;
 950         }
 951         case 'o': case 'u': case 'x': case 'X': {
 952           lua_Number n = luaL_checknumber(L, arg);
 953           unsigned LUA_INTFRM_T ni = (unsigned LUA_INTFRM_T)n;
 954           lua_Number diff = n - (lua_Number)ni;
 955           luaL_argcheck(L, -1 < diff && diff < 1, arg,
 956                         "not a non-negative number in proper range");
 957           addlenmod(form, LUA_INTFRMLEN);
 958           nb = str_sprintf(buff, form, ni);
 959           break;
 960         }
 961 #if defined(LUA_USE_FLOAT_FORMATS)
 962         case 'e': case 'E': case 'f':
 963 #if defined(LUA_USE_AFORMAT)
 964         case 'a': case 'A':
 965 #endif
 966         case 'g': case 'G': {
 967           addlenmod(form, LUA_FLTFRMLEN);
 968           nb = str_sprintf(buff, form, (LUA_FLTFRM_T)luaL_checknumber(L, arg));
 969           break;
 970         }
 971 #endif
 972         case 'q': {
 973           addquoted(L, &b, arg);
 974           break;
 975         }
 976         case 's': {
 977           size_t l;
 978           const char *s = luaL_tolstring(L, arg, &l);
 979           if (!strchr(form, '.') && l >= 100) {
 980             /* no precision and string is too long to be formatted;
 981                keep original string */
 982             luaL_addvalue(&b);
 983             break;
 984           }
 985           else {
 986             nb = str_sprintf(buff, form, s);
 987             lua_pop(L, 1);  /* remove result from 'luaL_tolstring' */
 988             break;
 989           }
 990         }
 991         default: {  /* also treat cases `pnLlh' */
 992           return luaL_error(L, "invalid option " LUA_QL("%%%c") " to "
 993                                LUA_QL("format"), *(strfrmt - 1));
 994         }
 995       }
 996       luaL_addsize(&b, nb);
 997     }
 998   }
 999   luaL_pushresult(&b);
1000   return 1;
1001 }
1002
1003 /* }====================================================== */
1004
1005
1006 static const luaL_Reg strlib[] = {
1007   {"byte", str_byte},
1008   {"char", str_char},
1009   {"dump", str_dump},
1010   {"find", str_find},
1011   {"format", str_format},
1012   {"gmatch", str_gmatch},
1013   {"gsub", str_gsub},
1014   {"len", str_len},
1015   {"lower", str_lower},
1016   {"match", str_match},
1017   {"rep", str_rep},
1018   {"reverse", str_reverse},
1019   {"sub", str_sub},
1020   {"upper", str_upper},
1021   {NULL, NULL}
1022 };
1023
1024
1025 static void createmetatable (lua_State *L) {
1026   lua_createtable(L, 0, 1);  /* table to be metatable for strings */
1027   lua_pushliteral(L, "");  /* dummy string */
1028   lua_pushvalue(L, -2);  /* copy table */
1029   lua_setmetatable(L, -2);  /* set table as metatable for strings */
1030   lua_pop(L, 1);  /* pop dummy string */
1031   lua_pushvalue(L, -2);  /* get string library */
1032   lua_setfield(L, -2, "__index");  /* metatable.__index = string */
1033   lua_pop(L, 1);  /* pop metatable */
1034 }
1035
1036
1037 /*
1038 ** Open string library
1039 */
1040 LUAMOD_API int luaopen_string (lua_State *L) {
1041   luaL_newlib(L, strlib);
1042   createmetatable(L);
1043   return 1;
1044 }
1045