src/lib_string.c

   1 /*
   2 ** String library.
   3 ** Copyright (C) 2005-2015 Mike Pall. See Copyright Notice in luajit.h
   4 **
   5 ** Major portions taken verbatim or adapted from the Lua interpreter.
   6 ** Copyright (C) 1994-2008 Lua.org, PUC-Rio. See Copyright Notice in lua.h
   7 */
   8
   9 #include <stdio.h>
  10
  11 #define lib_string_c
  12 #define LUA_LIB
  13
  14 #include "lua.h"
  15 #include "lauxlib.h"
  16 #include "lualib.h"
  17
  18 #include "lj_obj.h"
  19 #include "lj_gc.h"
  20 #include "lj_err.h"
  21 #include "lj_str.h"
  22 #include "lj_tab.h"
  23 #include "lj_meta.h"
  24 #include "lj_state.h"
  25 #include "lj_ff.h"
  26 #include "lj_bcdump.h"
  27 #include "lj_char.h"
  28 #include "lj_lib.h"
  29
  30 /* ------------------------------------------------------------------------ */
  31
  32 #define LJLIB_MODULE_string
  33
  34 LJLIB_ASM(string_len)           LJLIB_REC(.)
  35 {
  36   lj_lib_checkstr(L, 1);
  37   return FFH_RETRY;
  38 }
  39
  40 LJLIB_ASM(string_byte)          LJLIB_REC(string_range 0)
  41 {
  42   GCstr *s = lj_lib_checkstr(L, 1);
  43   int32_t len = (int32_t)s->len;
  44   int32_t start = lj_lib_optint(L, 2, 1);
  45   int32_t stop = lj_lib_optint(L, 3, start);
  46   int32_t n, i;
  47   const unsigned char *p;
  48   if (stop < 0) stop += len+1;
  49   if (start < 0) start += len+1;
  50   if (start <= 0) start = 1;
  51   if (stop > len) stop = len;
  52   if (start > stop) return FFH_RES(0);  /* Empty interval: return no results. */
  53   start--;
  54   n = stop - start;
  55   if ((uint32_t)n > LUAI_MAXCSTACK)
  56     lj_err_caller(L, LJ_ERR_STRSLC);
  57   lj_state_checkstack(L, (MSize)n);
  58   p = (const unsigned char *)strdata(s) + start;
  59   for (i = 0; i < n; i++)
  60     setintV(L->base + i-1, p[i]);
  61   return FFH_RES(n);
  62 }
  63
  64 LJLIB_ASM(string_char)
  65 {
  66   int i, nargs = (int)(L->top - L->base);
  67   char *buf = lj_str_needbuf(L, &G(L)->tmpbuf, (MSize)nargs);
  68   for (i = 1; i <= nargs; i++) {
  69     int32_t k = lj_lib_checkint(L, i);
  70     if (!checku8(k))
  71       lj_err_arg(L, i, LJ_ERR_BADVAL);
  72     buf[i-1] = (char)k;
  73   }
  74   setstrV(L, L->base-1, lj_str_new(L, buf, (size_t)nargs));
  75   return FFH_RES(1);
  76 }
  77
  78 LJLIB_ASM(string_sub)           LJLIB_REC(string_range 1)
  79 {
  80   lj_lib_checkstr(L, 1);
  81   lj_lib_checkint(L, 2);
  82   setintV(L->base+2, lj_lib_optint(L, 3, -1));
  83   return FFH_RETRY;
  84 }
  85
  86 LJLIB_ASM(string_rep)
  87 {
  88   GCstr *s = lj_lib_checkstr(L, 1);
  89   int32_t k = lj_lib_checkint(L, 2);
  90   GCstr *sep = lj_lib_optstr(L, 3);
  91   int32_t len = (int32_t)s->len;
  92   global_State *g = G(L);
  93   int64_t tlen;
  94   const char *src;
  95   char *buf;
  96   if (k <= 0) {
  97   empty:
  98     setstrV(L, L->base-1, &g->strempty);
  99     return FFH_RES(1);
 100   }
 101   if (sep) {
 102     tlen = (int64_t)len + sep->len;
 103     if (tlen > LJ_MAX_STR)
 104       lj_err_caller(L, LJ_ERR_STROV);
 105     tlen *= k;
 106     if (tlen > LJ_MAX_STR)
 107       lj_err_caller(L, LJ_ERR_STROV);
 108   } else {
 109     tlen = (int64_t)k * len;
 110     if (tlen > LJ_MAX_STR)
 111       lj_err_caller(L, LJ_ERR_STROV);
 112   }
 113   if (tlen == 0) goto empty;
 114   buf = lj_str_needbuf(L, &g->tmpbuf, (MSize)tlen);
 115   src = strdata(s);
 116   if (sep) {
 117     tlen -= sep->len;  /* Ignore trailing separator. */
 118     if (k > 1) {  /* Paste one string and one separator. */
 119       int32_t i;
 120       i = 0; while (i < len) *buf++ = src[i++];
 121       src = strdata(sep); len = sep->len;
 122       i = 0; while (i < len) *buf++ = src[i++];
 123       src = g->tmpbuf.buf; len += s->len; k--;  /* Now copy that k-1 times. */
 124     }
 125   }
 126   do {
 127     int32_t i = 0;
 128     do { *buf++ = src[i++]; } while (i < len);
 129   } while (--k > 0);
 130   setstrV(L, L->base-1, lj_str_new(L, g->tmpbuf.buf, (size_t)tlen));
 131   return FFH_RES(1);
 132 }
 133
 134 LJLIB_ASM(string_reverse)
 135 {
 136   GCstr *s = lj_lib_checkstr(L, 1);
 137   lj_str_needbuf(L, &G(L)->tmpbuf, s->len);
 138   return FFH_RETRY;
 139 }
 140 LJLIB_ASM_(string_lower)
 141 LJLIB_ASM_(string_upper)
 142
 143 /* ------------------------------------------------------------------------ */
 144
 145 static int writer_buf(lua_State *L, const void *p, size_t size, void *b)
 146 {
 147   luaL_addlstring((luaL_Buffer *)b, (const char *)p, size);
 148   UNUSED(L);
 149   return 0;
 150 }
 151
 152 LJLIB_CF(string_dump)
 153 {
 154   GCfunc *fn = lj_lib_checkfunc(L, 1);
 155   int strip = L->base+1 < L->top && tvistruecond(L->base+1);
 156   luaL_Buffer b;
 157   L->top = L->base+1;
 158   luaL_buffinit(L, &b);
 159   if (!isluafunc(fn) || lj_bcwrite(L, funcproto(fn), writer_buf, &b, strip))
 160     lj_err_caller(L, LJ_ERR_STRDUMP);
 161   luaL_pushresult(&b);
 162   return 1;
 163 }
 164
 165 /* ------------------------------------------------------------------------ */
 166
 167 /* macro to `unsign' a character */
 168 #define uchar(c)        ((unsigned char)(c))
 169
 170 #define CAP_UNFINISHED  (-1)
 171 #define CAP_POSITION    (-2)
 172
 173 typedef struct MatchState {
 174   const char *src_init;  /* init of source string */
 175   const char *src_end;  /* end (`\0') of source string */
 176   lua_State *L;
 177   int level;  /* total number of captures (finished or unfinished) */
 178   int depth;
 179   struct {
 180     const char *init;
 181     ptrdiff_t len;
 182   } capture[LUA_MAXCAPTURES];
 183 } MatchState;
 184
 185 #define L_ESC           '%'
 186 #define SPECIALS        "^$*+?.([%-"
 187
 188 static int check_capture(MatchState *ms, int l)
 189 {
 190   l -= '1';
 191   if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
 192     lj_err_caller(ms->L, LJ_ERR_STRCAPI);
 193   return l;
 194 }
 195
 196 static int capture_to_close(MatchState *ms)
 197 {
 198   int level = ms->level;
 199   for (level--; level>=0; level--)
 200     if (ms->capture[level].len == CAP_UNFINISHED) return level;
 201   lj_err_caller(ms->L, LJ_ERR_STRPATC);
 202   return 0;  /* unreachable */
 203 }
 204
 205 static const char *classend(MatchState *ms, const char *p)
 206 {
 207   switch (*p++) {
 208   case L_ESC:
 209     if (*p == '\0')
 210       lj_err_caller(ms->L, LJ_ERR_STRPATE);
 211     return p+1;
 212   case '[':
 213     if (*p == '^') p++;
 214     do {  /* look for a `]' */
 215       if (*p == '\0')
 216         lj_err_caller(ms->L, LJ_ERR_STRPATM);
 217       if (*(p++) == L_ESC && *p != '\0')
 218         p++;  /* skip escapes (e.g. `%]') */
 219     } while (*p != ']');
 220     return p+1;
 221   default:
 222     return p;
 223   }
 224 }
 225
 226 static const unsigned char match_class_map[32] = {
 227   0,LJ_CHAR_ALPHA,0,LJ_CHAR_CNTRL,LJ_CHAR_DIGIT,0,0,LJ_CHAR_GRAPH,0,0,0,0,
 228   LJ_CHAR_LOWER,0,0,0,LJ_CHAR_PUNCT,0,0,LJ_CHAR_SPACE,0,
 229   LJ_CHAR_UPPER,0,LJ_CHAR_ALNUM,LJ_CHAR_XDIGIT,0,0,0,0,0,0,0
 230 };
 231
 232 static int match_class(int c, int cl)
 233 {
 234   if ((cl & 0xc0) == 0x40) {
 235     int t = match_class_map[(cl&0x1f)];
 236     if (t) {
 237       t = lj_char_isa(c, t);
 238       return (cl & 0x20) ? t : !t;
 239     }
 240     if (cl == 'z') return c == 0;
 241     if (cl == 'Z') return c != 0;
 242   }
 243   return (cl == c);
 244 }
 245
 246 static int matchbracketclass(int c, const char *p, const char *ec)
 247 {
 248   int sig = 1;
 249   if (*(p+1) == '^') {
 250     sig = 0;
 251     p++;  /* skip the `^' */
 252   }
 253   while (++p < ec) {
 254     if (*p == L_ESC) {
 255       p++;
 256       if (match_class(c, uchar(*p)))
 257         return sig;
 258     }
 259     else if ((*(p+1) == '-') && (p+2 < ec)) {
 260       p+=2;
 261       if (uchar(*(p-2)) <= c && c <= uchar(*p))
 262         return sig;
 263     }
 264     else if (uchar(*p) == c) return sig;
 265   }
 266   return !sig;
 267 }
 268
 269 static int singlematch(int c, const char *p, const char *ep)
 270 {
 271   switch (*p) {
 272   case '.': return 1;  /* matches any char */
 273   case L_ESC: return match_class(c, uchar(*(p+1)));
 274   case '[': return matchbracketclass(c, p, ep-1);
 275   default:  return (uchar(*p) == c);
 276   }
 277 }
 278
 279 static const char *match(MatchState *ms, const char *s, const char *p);
 280
 281 static const char *matchbalance(MatchState *ms, const char *s, const char *p)
 282 {
 283   if (*p == 0 || *(p+1) == 0)
 284     lj_err_caller(ms->L, LJ_ERR_STRPATU);
 285   if (*s != *p) {
 286     return NULL;
 287   } else {
 288     int b = *p;
 289     int e = *(p+1);
 290     int cont = 1;
 291     while (++s < ms->src_end) {
 292       if (*s == e) {
 293         if (--cont == 0) return s+1;
 294       } else if (*s == b) {
 295         cont++;
 296       }
 297     }
 298   }
 299   return NULL;  /* string ends out of balance */
 300 }
 301
 302 static const char *max_expand(MatchState *ms, const char *s,
 303                               const char *p, const char *ep)
 304 {
 305   ptrdiff_t i = 0;  /* counts maximum expand for item */
 306   while ((s+i)<ms->src_end && singlematch(uchar(*(s+i)), p, ep))
 307     i++;
 308   /* keeps trying to match with the maximum repetitions */
 309   while (i>=0) {
 310     const char *res = match(ms, (s+i), ep+1);
 311     if (res) return res;
 312     i--;  /* else didn't match; reduce 1 repetition to try again */
 313   }
 314   return NULL;
 315 }
 316
 317 static const char *min_expand(MatchState *ms, const char *s,
 318                               const char *p, const char *ep)
 319 {
 320   for (;;) {
 321     const char *res = match(ms, s, ep+1);
 322     if (res != NULL)
 323       return res;
 324     else if (s<ms->src_end && singlematch(uchar(*s), p, ep))
 325       s++;  /* try with one more repetition */
 326     else
 327       return NULL;
 328   }
 329 }
 330
 331 static const char *start_capture(MatchState *ms, const char *s,
 332                                  const char *p, int what)
 333 {
 334   const char *res;
 335   int level = ms->level;
 336   if (level >= LUA_MAXCAPTURES) lj_err_caller(ms->L, LJ_ERR_STRCAPN);
 337   ms->capture[level].init = s;
 338   ms->capture[level].len = what;
 339   ms->level = level+1;
 340   if ((res=match(ms, s, p)) == NULL)  /* match failed? */
 341     ms->level--;  /* undo capture */
 342   return res;
 343 }
 344
 345 static const char *end_capture(MatchState *ms, const char *s,
 346                                const char *p)
 347 {
 348   int l = capture_to_close(ms);
 349   const char *res;
 350   ms->capture[l].len = s - ms->capture[l].init;  /* close capture */
 351   if ((res = match(ms, s, p)) == NULL)  /* match failed? */
 352     ms->capture[l].len = CAP_UNFINISHED;  /* undo capture */
 353   return res;
 354 }
 355
 356 static const char *match_capture(MatchState *ms, const char *s, int l)
 357 {
 358   size_t len;
 359   l = check_capture(ms, l);
 360   len = (size_t)ms->capture[l].len;
 361   if ((size_t)(ms->src_end-s) >= len &&
 362       memcmp(ms->capture[l].init, s, len) == 0)
 363     return s+len;
 364   else
 365     return NULL;
 366 }
 367
 368 static const char *match(MatchState *ms, const char *s, const char *p)
 369 {
 370   if (++ms->depth > LJ_MAX_XLEVEL)
 371     lj_err_caller(ms->L, LJ_ERR_STRPATX);
 372   init: /* using goto's to optimize tail recursion */
 373   switch (*p) {
 374   case '(':  /* start capture */
 375     if (*(p+1) == ')')  /* position capture? */
 376       s = start_capture(ms, s, p+2, CAP_POSITION);
 377     else
 378       s = start_capture(ms, s, p+1, CAP_UNFINISHED);
 379     break;
 380   case ')':  /* end capture */
 381     s = end_capture(ms, s, p+1);
 382     break;
 383   case L_ESC:
 384     switch (*(p+1)) {
 385     case 'b':  /* balanced string? */
 386       s = matchbalance(ms, s, p+2);
 387       if (s == NULL) break;
 388       p+=4;
 389       goto init;  /* else s = match(ms, s, p+4); */
 390     case 'f': {  /* frontier? */
 391       const char *ep; char previous;
 392       p += 2;
 393       if (*p != '[')
 394         lj_err_caller(ms->L, LJ_ERR_STRPATB);
 395       ep = classend(ms, p);  /* points to what is next */
 396       previous = (s == ms->src_init) ? '\0' : *(s-1);
 397       if (matchbracketclass(uchar(previous), p, ep-1) ||
 398          !matchbracketclass(uchar(*s), p, ep-1)) { s = NULL; break; }
 399       p=ep;
 400       goto init;  /* else s = match(ms, s, ep); */
 401       }
 402     default:
 403       if (lj_char_isdigit(uchar(*(p+1)))) {  /* capture results (%0-%9)? */
 404         s = match_capture(ms, s, uchar(*(p+1)));
 405         if (s == NULL) break;
 406         p+=2;
 407         goto init;  /* else s = match(ms, s, p+2) */
 408       }
 409       goto dflt;  /* case default */
 410     }
 411     break;
 412   case '\0':  /* end of pattern */
 413     break;  /* match succeeded */
 414   case '$':
 415     /* is the `$' the last char in pattern? */
 416     if (*(p+1) != '\0') goto dflt;
 417     if (s != ms->src_end) s = NULL;  /* check end of string */
 418     break;
 419   default: dflt: {  /* it is a pattern item */
 420     const char *ep = classend(ms, p);  /* points to what is next */
 421     int m = s<ms->src_end && singlematch(uchar(*s), p, ep);
 422     switch (*ep) {
 423     case '?': {  /* optional */
 424       const char *res;
 425       if (m && ((res=match(ms, s+1, ep+1)) != NULL)) {
 426         s = res;
 427         break;
 428       }
 429       p=ep+1;
 430       goto init;  /* else s = match(ms, s, ep+1); */
 431       }
 432     case '*':  /* 0 or more repetitions */
 433       s = max_expand(ms, s, p, ep);
 434       break;
 435     case '+':  /* 1 or more repetitions */
 436       s = (m ? max_expand(ms, s+1, p, ep) : NULL);
 437       break;
 438     case '-':  /* 0 or more repetitions (minimum) */
 439       s = min_expand(ms, s, p, ep);
 440       break;
 441     default:
 442       if (m) { s++; p=ep; goto init; }  /* else s = match(ms, s+1, ep); */
 443       s = NULL;
 444       break;
 445     }
 446     break;
 447     }
 448   }
 449   ms->depth--;
 450   return s;
 451 }
 452
 453 static const char *lmemfind(const char *s1, size_t l1,
 454                             const char *s2, size_t l2)
 455 {
 456   if (l2 == 0) {
 457     return s1;  /* empty strings are everywhere */
 458   } else if (l2 > l1) {
 459     return NULL;  /* avoids a negative `l1' */
 460   } else {
 461     const char *init;  /* to search for a `*s2' inside `s1' */
 462     l2--;  /* 1st char will be checked by `memchr' */
 463     l1 = l1-l2;  /* `s2' cannot be found after that */
 464     while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
 465       init++;   /* 1st char is already checked */
 466       if (memcmp(init, s2+1, l2) == 0) {
 467         return init-1;
 468       } else {  /* correct `l1' and `s1' to try again */
 469         l1 -= (size_t)(init-s1);
 470         s1 = init;
 471       }
 472     }
 473     return NULL;  /* not found */
 474   }
 475 }
 476
 477 static void push_onecapture(MatchState *ms, int i, const char *s, const char *e)
 478 {
 479   if (i >= ms->level) {
 480     if (i == 0)  /* ms->level == 0, too */
 481       lua_pushlstring(ms->L, s, (size_t)(e - s));  /* add whole match */
 482     else
 483       lj_err_caller(ms->L, LJ_ERR_STRCAPI);
 484   } else {
 485     ptrdiff_t l = ms->capture[i].len;
 486     if (l == CAP_UNFINISHED) lj_err_caller(ms->L, LJ_ERR_STRCAPU);
 487     if (l == CAP_POSITION)
 488       lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
 489     else
 490       lua_pushlstring(ms->L, ms->capture[i].init, (size_t)l);
 491   }
 492 }
 493
 494 static int push_captures(MatchState *ms, const char *s, const char *e)
 495 {
 496   int i;
 497   int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
 498   luaL_checkstack(ms->L, nlevels, "too many captures");
 499   for (i = 0; i < nlevels; i++)
 500     push_onecapture(ms, i, s, e);
 501   return nlevels;  /* number of strings pushed */
 502 }
 503
 504 static ptrdiff_t posrelat(ptrdiff_t pos, size_t len)
 505 {
 506   /* relative string position: negative means back from end */
 507   if (pos < 0) pos += (ptrdiff_t)len + 1;
 508   return (pos >= 0) ? pos : 0;
 509 }
 510
 511 static int str_find_aux(lua_State *L, int find)
 512 {
 513   size_t l1, l2;
 514   const char *s = luaL_checklstring(L, 1, &l1);
 515   const char *p = luaL_checklstring(L, 2, &l2);
 516   ptrdiff_t init = posrelat(luaL_optinteger(L, 3, 1), l1) - 1;
 517   if (init < 0) {
 518     init = 0;
 519   } else if ((size_t)(init) > l1) {
 520 #if LJ_52
 521     setnilV(L->top-1);
 522     return 1;
 523 #else
 524     init = (ptrdiff_t)l1;
 525 #endif
 526   }
 527   if (find && (lua_toboolean(L, 4) ||  /* explicit request? */
 528       strpbrk(p, SPECIALS) == NULL)) {  /* or no special characters? */
 529     /* do a plain search */
 530     const char *s2 = lmemfind(s+init, l1-(size_t)init, p, l2);
 531     if (s2) {
 532       lua_pushinteger(L, s2-s+1);
 533       lua_pushinteger(L, s2-s+(ptrdiff_t)l2);
 534       return 2;
 535     }
 536   } else {
 537     MatchState ms;
 538     int anchor = (*p == '^') ? (p++, 1) : 0;
 539     const char *s1=s+init;
 540     ms.L = L;
 541     ms.src_init = s;
 542     ms.src_end = s+l1;
 543     do {
 544       const char *res;
 545       ms.level = ms.depth = 0;
 546       if ((res=match(&ms, s1, p)) != NULL) {
 547         if (find) {
 548           lua_pushinteger(L, s1-s+1);  /* start */
 549           lua_pushinteger(L, res-s);   /* end */
 550           return push_captures(&ms, NULL, 0) + 2;
 551         } else {
 552           return push_captures(&ms, s1, res);
 553         }
 554       }
 555     } while (s1++ < ms.src_end && !anchor);
 556   }
 557   lua_pushnil(L);  /* not found */
 558   return 1;
 559 }
 560
 561 LJLIB_CF(string_find)
 562 {
 563   return str_find_aux(L, 1);
 564 }
 565
 566 LJLIB_CF(string_match)
 567 {
 568   return str_find_aux(L, 0);
 569 }
 570
 571 LJLIB_NOREG LJLIB_CF(string_gmatch_aux)
 572 {
 573   const char *p = strVdata(lj_lib_upvalue(L, 2));
 574   GCstr *str = strV(lj_lib_upvalue(L, 1));
 575   const char *s = strdata(str);
 576   TValue *tvpos = lj_lib_upvalue(L, 3);
 577   const char *src = s + tvpos->u32.lo;
 578   MatchState ms;
 579   ms.L = L;
 580   ms.src_init = s;
 581   ms.src_end = s + str->len;
 582   for (; src <= ms.src_end; src++) {
 583     const char *e;
 584     ms.level = ms.depth = 0;
 585     if ((e = match(&ms, src, p)) != NULL) {
 586       int32_t pos = (int32_t)(e - s);
 587       if (e == src) pos++;  /* Ensure progress for empty match. */
 588       tvpos->u32.lo = (uint32_t)pos;
 589       return push_captures(&ms, src, e);
 590     }
 591   }
 592   return 0;  /* not found */
 593 }
 594
 595 LJLIB_CF(string_gmatch)
 596 {
 597   lj_lib_checkstr(L, 1);
 598   lj_lib_checkstr(L, 2);
 599   L->top = L->base+3;
 600   (L->top-1)->u64 = 0;
 601   lj_lib_pushcc(L, lj_cf_string_gmatch_aux, FF_string_gmatch_aux, 3);
 602   return 1;
 603 }
 604
 605 static void add_s(MatchState *ms, luaL_Buffer *b, const char *s, const char *e)
 606 {
 607   size_t l, i;
 608   const char *news = lua_tolstring(ms->L, 3, &l);
 609   for (i = 0; i < l; i++) {
 610     if (news[i] != L_ESC) {
 611       luaL_addchar(b, news[i]);
 612     } else {
 613       i++;  /* skip ESC */
 614       if (!lj_char_isdigit(uchar(news[i]))) {
 615         luaL_addchar(b, news[i]);
 616       } else if (news[i] == '0') {
 617         luaL_addlstring(b, s, (size_t)(e - s));
 618       } else {
 619         push_onecapture(ms, news[i] - '1', s, e);
 620         luaL_addvalue(b);  /* add capture to accumulated result */
 621       }
 622     }
 623   }
 624 }
 625
 626 static void add_value(MatchState *ms, luaL_Buffer *b,
 627                       const char *s, const char *e)
 628 {
 629   lua_State *L = ms->L;
 630   switch (lua_type(L, 3)) {
 631     case LUA_TNUMBER:
 632     case LUA_TSTRING: {
 633       add_s(ms, b, s, e);
 634       return;
 635     }
 636     case LUA_TFUNCTION: {
 637       int n;
 638       lua_pushvalue(L, 3);
 639       n = push_captures(ms, s, e);
 640       lua_call(L, n, 1);
 641       break;
 642     }
 643     case LUA_TTABLE: {
 644       push_onecapture(ms, 0, s, e);
 645       lua_gettable(L, 3);
 646       break;
 647     }
 648   }
 649   if (!lua_toboolean(L, -1)) {  /* nil or false? */
 650     lua_pop(L, 1);
 651     lua_pushlstring(L, s, (size_t)(e - s));  /* keep original text */
 652   } else if (!lua_isstring(L, -1)) {
 653     lj_err_callerv(L, LJ_ERR_STRGSRV, luaL_typename(L, -1));
 654   }
 655   luaL_addvalue(b);  /* add result to accumulator */
 656 }
 657
 658 LJLIB_CF(string_gsub)
 659 {
 660   size_t srcl;
 661   const char *src = luaL_checklstring(L, 1, &srcl);
 662   const char *p = luaL_checkstring(L, 2);
 663   int  tr = lua_type(L, 3);
 664   int max_s = luaL_optint(L, 4, (int)(srcl+1));
 665   int anchor = (*p == '^') ? (p++, 1) : 0;
 666   int n = 0;
 667   MatchState ms;
 668   luaL_Buffer b;
 669   if (!(tr == LUA_TNUMBER || tr == LUA_TSTRING ||
 670         tr == LUA_TFUNCTION || tr == LUA_TTABLE))
 671     lj_err_arg(L, 3, LJ_ERR_NOSFT);
 672   luaL_buffinit(L, &b);
 673   ms.L = L;
 674   ms.src_init = src;
 675   ms.src_end = src+srcl;
 676   while (n < max_s) {
 677     const char *e;
 678     ms.level = ms.depth = 0;
 679     e = match(&ms, src, p);
 680     if (e) {
 681       n++;
 682       add_value(&ms, &b, src, e);
 683     }
 684     if (e && e>src) /* non empty match? */
 685       src = e;  /* skip it */
 686     else if (src < ms.src_end)
 687       luaL_addchar(&b, *src++);
 688     else
 689       break;
 690     if (anchor)
 691       break;
 692   }
 693   luaL_addlstring(&b, src, (size_t)(ms.src_end-src));
 694   luaL_pushresult(&b);
 695   lua_pushinteger(L, n);  /* number of substitutions */
 696   return 2;
 697 }
 698
 699 /* ------------------------------------------------------------------------ */
 700
 701 /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
 702 #define MAX_FMTITEM     512
 703 /* valid flags in a format specification */
 704 #define FMT_FLAGS       "-+ #0"
 705 /*
 706 ** maximum size of each format specification (such as '%-099.99d')
 707 ** (+10 accounts for %99.99x plus margin of error)
 708 */
 709 #define MAX_FMTSPEC     (sizeof(FMT_FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
 710
 711 static void addquoted(lua_State *L, luaL_Buffer *b, int arg)
 712 {
 713   GCstr *str = lj_lib_checkstr(L, arg);
 714   int32_t len = (int32_t)str->len;
 715   const char *s = strdata(str);
 716   luaL_addchar(b, '"');
 717   while (len--) {
 718     uint32_t c = uchar(*s);
 719     if (c == '"' || c == '\\' || c == '\n') {
 720       luaL_addchar(b, '\\');
 721     } else if (lj_char_iscntrl(c)) {  /* This can only be 0-31 or 127. */
 722       uint32_t d;
 723       luaL_addchar(b, '\\');
 724       if (c >= 100 || lj_char_isdigit(uchar(s[1]))) {
 725         luaL_addchar(b, '0'+(c >= 100)); if (c >= 100) c -= 100;
 726         goto tens;
 727       } else if (c >= 10) {
 728       tens:
 729         d = (c * 205) >> 11; c -= d * 10; luaL_addchar(b, '0'+d);
 730       }
 731       c += '0';
 732     }
 733     luaL_addchar(b, c);
 734     s++;
 735   }
 736   luaL_addchar(b, '"');
 737 }
 738
 739 static const char *scanformat(lua_State *L, const char *strfrmt, char *form)
 740 {
 741   const char *p = strfrmt;
 742   while (*p != '\0' && strchr(FMT_FLAGS, *p) != NULL) p++;  /* skip flags */
 743   if ((size_t)(p - strfrmt) >= sizeof(FMT_FLAGS))
 744     lj_err_caller(L, LJ_ERR_STRFMTR);
 745   if (lj_char_isdigit(uchar(*p))) p++;  /* skip width */
 746   if (lj_char_isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 747   if (*p == '.') {
 748     p++;
 749     if (lj_char_isdigit(uchar(*p))) p++;  /* skip precision */
 750     if (lj_char_isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 751   }
 752   if (lj_char_isdigit(uchar(*p)))
 753     lj_err_caller(L, LJ_ERR_STRFMTW);
 754   *(form++) = '%';
 755   strncpy(form, strfrmt, (size_t)(p - strfrmt + 1));
 756   form += p - strfrmt + 1;
 757   *form = '\0';
 758   return p;
 759 }
 760
 761 static void addintlen(char *form)
 762 {
 763   size_t l = strlen(form);
 764   char spec = form[l - 1];
 765   strcpy(form + l - 1, LUA_INTFRMLEN);
 766   form[l + sizeof(LUA_INTFRMLEN) - 2] = spec;
 767   form[l + sizeof(LUA_INTFRMLEN) - 1] = '\0';
 768 }
 769
 770 static unsigned LUA_INTFRM_T num2intfrm(lua_State *L, int arg)
 771 {
 772   if (sizeof(LUA_INTFRM_T) == 4) {
 773     return (LUA_INTFRM_T)lj_lib_checkbit(L, arg);
 774   } else {
 775     cTValue *o;
 776     lj_lib_checknumber(L, arg);
 777     o = L->base+arg-1;
 778     if (tvisint(o))
 779       return (LUA_INTFRM_T)intV(o);
 780     else
 781       return (LUA_INTFRM_T)numV(o);
 782   }
 783 }
 784
 785 static unsigned LUA_INTFRM_T num2uintfrm(lua_State *L, int arg)
 786 {
 787   if (sizeof(LUA_INTFRM_T) == 4) {
 788     return (unsigned LUA_INTFRM_T)lj_lib_checkbit(L, arg);
 789   } else {
 790     cTValue *o;
 791     lj_lib_checknumber(L, arg);
 792     o = L->base+arg-1;
 793     if (tvisint(o))
 794       return (unsigned LUA_INTFRM_T)intV(o);
 795     else if ((int32_t)o->u32.hi < 0)
 796       return (unsigned LUA_INTFRM_T)(LUA_INTFRM_T)numV(o);
 797     else
 798       return (unsigned LUA_INTFRM_T)numV(o);
 799   }
 800 }
 801
 802 static GCstr *meta_tostring(lua_State *L, int arg)
 803 {
 804   TValue *o = L->base+arg-1;
 805   cTValue *mo;
 806   lua_assert(o < L->top);  /* Caller already checks for existence. */
 807   if (LJ_LIKELY(tvisstr(o)))
 808     return strV(o);
 809   if (!tvisnil(mo = lj_meta_lookup(L, o, MM_tostring))) {
 810     copyTV(L, L->top++, mo);
 811     copyTV(L, L->top++, o);
 812     lua_call(L, 1, 1);
 813     L->top--;
 814     if (tvisstr(L->top))
 815       return strV(L->top);
 816     o = L->base+arg-1;
 817     copyTV(L, o, L->top);
 818   }
 819   if (tvisnumber(o)) {
 820     return lj_str_fromnumber(L, o);
 821   } else if (tvisnil(o)) {
 822     return lj_str_newlit(L, "nil");
 823   } else if (tvisfalse(o)) {
 824     return lj_str_newlit(L, "false");
 825   } else if (tvistrue(o)) {
 826     return lj_str_newlit(L, "true");
 827   } else {
 828     if (tvisfunc(o) && isffunc(funcV(o)))
 829       lj_str_pushf(L, "function: builtin#%d", funcV(o)->c.ffid);
 830     else
 831       lj_str_pushf(L, "%s: %p", lj_typename(o), lua_topointer(L, arg));
 832     L->top--;
 833     return strV(L->top);
 834   }
 835 }
 836
 837 LJLIB_CF(string_format)
 838 {
 839   int arg = 1, top = (int)(L->top - L->base);
 840   GCstr *fmt = lj_lib_checkstr(L, arg);
 841   const char *strfrmt = strdata(fmt);
 842   const char *strfrmt_end = strfrmt + fmt->len;
 843   luaL_Buffer b;
 844   luaL_buffinit(L, &b);
 845   while (strfrmt < strfrmt_end) {
 846     if (*strfrmt != L_ESC) {
 847       luaL_addchar(&b, *strfrmt++);
 848     } else if (*++strfrmt == L_ESC) {
 849       luaL_addchar(&b, *strfrmt++);  /* %% */
 850     } else { /* format item */
 851       char form[MAX_FMTSPEC];  /* to store the format (`%...') */
 852       char buff[MAX_FMTITEM];  /* to store the formatted item */
 853       if (++arg > top)
 854         luaL_argerror(L, arg, lj_obj_typename[0]);
 855       strfrmt = scanformat(L, strfrmt, form);
 856       switch (*strfrmt++) {
 857       case 'c':
 858         sprintf(buff, form, lj_lib_checkint(L, arg));
 859         break;
 860       case 'd':  case 'i':
 861         addintlen(form);
 862         sprintf(buff, form, num2intfrm(L, arg));
 863         break;
 864       case 'o':  case 'u':  case 'x':  case 'X':
 865         addintlen(form);
 866         sprintf(buff, form, num2uintfrm(L, arg));
 867         break;
 868       case 'e':  case 'E': case 'f': case 'g': case 'G': case 'a': case 'A': {
 869         TValue tv;
 870         tv.n = lj_lib_checknum(L, arg);
 871         if (LJ_UNLIKELY((tv.u32.hi << 1) >= 0xffe00000)) {
 872           /* Canonicalize output of non-finite values. */
 873           char *p, nbuf[LJ_STR_NUMBUF];
 874           size_t len = lj_str_bufnum(nbuf, &tv);
 875           if (strfrmt[-1] < 'a') {
 876             nbuf[len-3] = nbuf[len-3] - 0x20;
 877             nbuf[len-2] = nbuf[len-2] - 0x20;
 878             nbuf[len-1] = nbuf[len-1] - 0x20;
 879           }
 880           nbuf[len] = '\0';
 881           for (p = form; *p < 'A' && *p != '.'; p++) ;
 882           *p++ = 's'; *p = '\0';
 883           sprintf(buff, form, nbuf);
 884           break;
 885         }
 886         sprintf(buff, form, (double)tv.n);
 887         break;
 888         }
 889       case 'q':
 890         addquoted(L, &b, arg);
 891         continue;
 892       case 'p':
 893         lj_str_pushf(L, "%p", lua_topointer(L, arg));
 894         luaL_addvalue(&b);
 895         continue;
 896       case 's': {
 897         GCstr *str = meta_tostring(L, arg);
 898         if (!strchr(form, '.') && str->len >= 100) {
 899           /* no precision and string is too long to be formatted;
 900              keep original string */
 901           setstrV(L, L->top++, str);
 902           luaL_addvalue(&b);
 903           continue;
 904         }
 905         sprintf(buff, form, strdata(str));
 906         break;
 907         }
 908       default:
 909         lj_err_callerv(L, LJ_ERR_STRFMTO, *(strfrmt -1));
 910         break;
 911       }
 912       luaL_addlstring(&b, buff, strlen(buff));
 913     }
 914   }
 915   luaL_pushresult(&b);
 916   return 1;
 917 }
 918
 919 /* ------------------------------------------------------------------------ */
 920
 921 #include "lj_libdef.h"
 922
 923 LUALIB_API int luaopen_string(lua_State *L)
 924 {
 925   GCtab *mt;
 926   global_State *g;
 927   LJ_LIB_REG(L, LUA_STRLIBNAME, string);
 928 #if defined(LUA_COMPAT_GFIND) && !LJ_52
 929   lua_getfield(L, -1, "gmatch");
 930   lua_setfield(L, -2, "gfind");
 931 #endif
 932   mt = lj_tab_new(L, 0, 1);
 933   /* NOBARRIER: basemt is a GC root. */
 934   g = G(L);
 935   setgcref(basemt_it(g, LJ_TSTR), obj2gco(mt));
 936   settabV(L, lj_tab_setstr(L, mt, mmname_str(g, MM_index)), tabV(L->top-1));
 937   mt->nomm = (uint8_t)(~(1u<<MM_index));
 938   return 1;
 939 }
 940