src/lib_string.c

   1 /*
   2 ** String library.
   3 ** Copyright (C) 2005-2011 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_state.h"
  24 #include "lj_ff.h"
  25 #include "lj_bcdump.h"
  26 #include "lj_char.h"
  27 #include "lj_lib.h"
  28
  29 /* ------------------------------------------------------------------------ */
  30
  31 #define LJLIB_MODULE_string
  32
  33 LJLIB_ASM(string_len)           LJLIB_REC(.)
  34 {
  35   lj_lib_checkstr(L, 1);
  36   return FFH_RETRY;
  37 }
  38
  39 LJLIB_ASM(string_byte)          LJLIB_REC(string_range 0)
  40 {
  41   GCstr *s = lj_lib_checkstr(L, 1);
  42   int32_t len = (int32_t)s->len;
  43   int32_t start = lj_lib_optint(L, 2, 1);
  44   int32_t stop = lj_lib_optint(L, 3, start);
  45   int32_t n, i;
  46   const unsigned char *p;
  47   if (stop < 0) stop += len+1;
  48   if (start < 0) start += len+1;
  49   if (start <= 0) start = 1;
  50   if (stop > len) stop = len;
  51   if (start > stop) return FFH_RES(0);  /* Empty interval: return no results. */
  52   start--;
  53   n = stop - start;
  54   if ((uint32_t)n > LUAI_MAXCSTACK)
  55     lj_err_caller(L, LJ_ERR_STRSLC);
  56   lj_state_checkstack(L, (MSize)n);
  57   p = (const unsigned char *)strdata(s) + start;
  58   for (i = 0; i < n; i++)
  59     setintV(L->base + i-1, p[i]);
  60   return FFH_RES(n);
  61 }
  62
  63 LJLIB_ASM(string_char)
  64 {
  65   int i, nargs = (int)(L->top - L->base);
  66   char *buf = lj_str_needbuf(L, &G(L)->tmpbuf, (size_t)nargs);
  67   for (i = 1; i <= nargs; i++) {
  68     int32_t k = lj_lib_checkint(L, i);
  69     if (!checku8(k))
  70       lj_err_arg(L, i, LJ_ERR_BADVAL);
  71     buf[i-1] = (char)k;
  72   }
  73   setstrV(L, L->base-1, lj_str_new(L, buf, (size_t)nargs));
  74   return FFH_RES(1);
  75 }
  76
  77 LJLIB_ASM(string_sub)           LJLIB_REC(string_range 1)
  78 {
  79   lj_lib_checkstr(L, 1);
  80   lj_lib_checkint(L, 2);
  81   setintV(L->base+2, lj_lib_optint(L, 3, -1));
  82   return FFH_RETRY;
  83 }
  84
  85 LJLIB_ASM(string_rep)
  86 {
  87   GCstr *s = lj_lib_checkstr(L, 1);
  88   int32_t len = (int32_t)s->len;
  89   int32_t k = lj_lib_checkint(L, 2);
  90   int64_t tlen = (int64_t)k * len;
  91   const char *src;
  92   char *buf;
  93   if (k <= 0) return FFH_RETRY;
  94   if (tlen > LJ_MAX_STR)
  95     lj_err_caller(L, LJ_ERR_STROV);
  96   buf = lj_str_needbuf(L, &G(L)->tmpbuf, (MSize)tlen);
  97   if (len <= 1) return FFH_RETRY;  /* ASM code only needed buffer resize. */
  98   src = strdata(s);
  99   do {
 100     int32_t i = 0;
 101     do { *buf++ = src[i++]; } while (i < len);
 102   } while (--k > 0);
 103   setstrV(L, L->base-1, lj_str_new(L, G(L)->tmpbuf.buf, (size_t)tlen));
 104   return FFH_RES(1);
 105 }
 106
 107 LJLIB_ASM(string_reverse)
 108 {
 109   GCstr *s = lj_lib_checkstr(L, 1);
 110   lj_str_needbuf(L, &G(L)->tmpbuf, s->len);
 111   return FFH_RETRY;
 112 }
 113 LJLIB_ASM_(string_lower)
 114 LJLIB_ASM_(string_upper)
 115
 116 /* ------------------------------------------------------------------------ */
 117
 118 static int writer_buf(lua_State *L, const void *p, size_t size, void *b)
 119 {
 120   luaL_addlstring((luaL_Buffer *)b, (const char *)p, size);
 121   UNUSED(L);
 122   return 0;
 123 }
 124
 125 LJLIB_CF(string_dump)
 126 {
 127   GCfunc *fn = lj_lib_checkfunc(L, 1);
 128   int strip = L->base+1 < L->top && tvistruecond(L->base+1);
 129   luaL_Buffer b;
 130   L->top = L->base+1;
 131   luaL_buffinit(L, &b);
 132   if (!isluafunc(fn) || lj_bcwrite(L, funcproto(fn), writer_buf, &b, strip))
 133     lj_err_caller(L, LJ_ERR_STRDUMP);
 134   luaL_pushresult(&b);
 135   return 1;
 136 }
 137
 138 /* ------------------------------------------------------------------------ */
 139
 140 /* macro to `unsign' a character */
 141 #define uchar(c)        ((unsigned char)(c))
 142
 143 #define CAP_UNFINISHED  (-1)
 144 #define CAP_POSITION    (-2)
 145
 146 typedef struct MatchState {
 147   const char *src_init;  /* init of source string */
 148   const char *src_end;  /* end (`\0') of source string */
 149   lua_State *L;
 150   int level;  /* total number of captures (finished or unfinished) */
 151   struct {
 152     const char *init;
 153     ptrdiff_t len;
 154   } capture[LUA_MAXCAPTURES];
 155 } MatchState;
 156
 157 #define L_ESC           '%'
 158 #define SPECIALS        "^$*+?.([%-"
 159
 160 static int check_capture(MatchState *ms, int l)
 161 {
 162   l -= '1';
 163   if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
 164     lj_err_caller(ms->L, LJ_ERR_STRCAPI);
 165   return l;
 166 }
 167
 168 static int capture_to_close(MatchState *ms)
 169 {
 170   int level = ms->level;
 171   for (level--; level>=0; level--)
 172     if (ms->capture[level].len == CAP_UNFINISHED) return level;
 173   lj_err_caller(ms->L, LJ_ERR_STRPATC);
 174   return 0;  /* unreachable */
 175 }
 176
 177 static const char *classend(MatchState *ms, const char *p)
 178 {
 179   switch (*p++) {
 180   case L_ESC:
 181     if (*p == '\0')
 182       lj_err_caller(ms->L, LJ_ERR_STRPATE);
 183     return p+1;
 184   case '[':
 185     if (*p == '^') p++;
 186     do {  /* look for a `]' */
 187       if (*p == '\0')
 188         lj_err_caller(ms->L, LJ_ERR_STRPATM);
 189       if (*(p++) == L_ESC && *p != '\0')
 190         p++;  /* skip escapes (e.g. `%]') */
 191     } while (*p != ']');
 192     return p+1;
 193   default:
 194     return p;
 195   }
 196 }
 197
 198 static const unsigned char match_class_map[32] = {
 199   0,LJ_CHAR_ALPHA,0,LJ_CHAR_CNTRL,LJ_CHAR_DIGIT,0,0,LJ_CHAR_GRAPH,0,0,0,0,
 200   LJ_CHAR_LOWER,0,0,0,LJ_CHAR_PUNCT,0,0,LJ_CHAR_SPACE,0,
 201   LJ_CHAR_UPPER,0,LJ_CHAR_ALNUM,LJ_CHAR_XDIGIT,0,0,0,0,0,0,0
 202 };
 203
 204 static int match_class(int c, int cl)
 205 {
 206   if ((cl & 0xc0) == 0x40) {
 207     int t = match_class_map[(cl&0x1f)];
 208     if (t) {
 209       t = lj_char_isa(c, t);
 210       return (cl & 0x20) ? t : !t;
 211     }
 212     if (cl == 'z') return c == 0;
 213     if (cl == 'Z') return c != 0;
 214   }
 215   return (cl == c);
 216 }
 217
 218 static int matchbracketclass(int c, const char *p, const char *ec)
 219 {
 220   int sig = 1;
 221   if (*(p+1) == '^') {
 222     sig = 0;
 223     p++;  /* skip the `^' */
 224   }
 225   while (++p < ec) {
 226     if (*p == L_ESC) {
 227       p++;
 228       if (match_class(c, uchar(*p)))
 229         return sig;
 230     }
 231     else if ((*(p+1) == '-') && (p+2 < ec)) {
 232       p+=2;
 233       if (uchar(*(p-2)) <= c && c <= uchar(*p))
 234         return sig;
 235     }
 236     else if (uchar(*p) == c) return sig;
 237   }
 238   return !sig;
 239 }
 240
 241 static int singlematch(int c, const char *p, const char *ep)
 242 {
 243   switch (*p) {
 244   case '.': return 1;  /* matches any char */
 245   case L_ESC: return match_class(c, uchar(*(p+1)));
 246   case '[': return matchbracketclass(c, p, ep-1);
 247   default:  return (uchar(*p) == c);
 248   }
 249 }
 250
 251 static const char *match(MatchState *ms, const char *s, const char *p);
 252
 253 static const char *matchbalance(MatchState *ms, const char *s, const char *p)
 254 {
 255   if (*p == 0 || *(p+1) == 0)
 256     lj_err_caller(ms->L, LJ_ERR_STRPATU);
 257   if (*s != *p) {
 258     return NULL;
 259   } else {
 260     int b = *p;
 261     int e = *(p+1);
 262     int cont = 1;
 263     while (++s < ms->src_end) {
 264       if (*s == e) {
 265         if (--cont == 0) return s+1;
 266       } else if (*s == b) {
 267         cont++;
 268       }
 269     }
 270   }
 271   return NULL;  /* string ends out of balance */
 272 }
 273
 274 static const char *max_expand(MatchState *ms, const char *s,
 275                               const char *p, const char *ep)
 276 {
 277   ptrdiff_t i = 0;  /* counts maximum expand for item */
 278   while ((s+i)<ms->src_end && singlematch(uchar(*(s+i)), p, ep))
 279     i++;
 280   /* keeps trying to match with the maximum repetitions */
 281   while (i>=0) {
 282     const char *res = match(ms, (s+i), ep+1);
 283     if (res) return res;
 284     i--;  /* else didn't match; reduce 1 repetition to try again */
 285   }
 286   return NULL;
 287 }
 288
 289 static const char *min_expand(MatchState *ms, const char *s,
 290                               const char *p, const char *ep)
 291 {
 292   for (;;) {
 293     const char *res = match(ms, s, ep+1);
 294     if (res != NULL)
 295       return res;
 296     else if (s<ms->src_end && singlematch(uchar(*s), p, ep))
 297       s++;  /* try with one more repetition */
 298     else
 299       return NULL;
 300   }
 301 }
 302
 303 static const char *start_capture(MatchState *ms, const char *s,
 304                                  const char *p, int what)
 305 {
 306   const char *res;
 307   int level = ms->level;
 308   if (level >= LUA_MAXCAPTURES) lj_err_caller(ms->L, LJ_ERR_STRCAPN);
 309   ms->capture[level].init = s;
 310   ms->capture[level].len = what;
 311   ms->level = level+1;
 312   if ((res=match(ms, s, p)) == NULL)  /* match failed? */
 313     ms->level--;  /* undo capture */
 314   return res;
 315 }
 316
 317 static const char *end_capture(MatchState *ms, const char *s,
 318                                const char *p)
 319 {
 320   int l = capture_to_close(ms);
 321   const char *res;
 322   ms->capture[l].len = s - ms->capture[l].init;  /* close capture */
 323   if ((res = match(ms, s, p)) == NULL)  /* match failed? */
 324     ms->capture[l].len = CAP_UNFINISHED;  /* undo capture */
 325   return res;
 326 }
 327
 328 static const char *match_capture(MatchState *ms, const char *s, int l)
 329 {
 330   size_t len;
 331   l = check_capture(ms, l);
 332   len = (size_t)ms->capture[l].len;
 333   if ((size_t)(ms->src_end-s) >= len &&
 334       memcmp(ms->capture[l].init, s, len) == 0)
 335     return s+len;
 336   else
 337     return NULL;
 338 }
 339
 340 static const char *match(MatchState *ms, const char *s, const char *p)
 341 {
 342   init: /* using goto's to optimize tail recursion */
 343   switch (*p) {
 344   case '(':  /* start capture */
 345     if (*(p+1) == ')')  /* position capture? */
 346       return start_capture(ms, s, p+2, CAP_POSITION);
 347     else
 348       return start_capture(ms, s, p+1, CAP_UNFINISHED);
 349   case ')':  /* end capture */
 350     return end_capture(ms, s, p+1);
 351   case L_ESC:
 352     switch (*(p+1)) {
 353     case 'b':  /* balanced string? */
 354       s = matchbalance(ms, s, p+2);
 355       if (s == NULL) return NULL;
 356       p+=4;
 357       goto init;  /* else return match(ms, s, p+4); */
 358     case 'f': {  /* frontier? */
 359       const char *ep; char previous;
 360       p += 2;
 361       if (*p != '[')
 362         lj_err_caller(ms->L, LJ_ERR_STRPATB);
 363       ep = classend(ms, p);  /* points to what is next */
 364       previous = (s == ms->src_init) ? '\0' : *(s-1);
 365       if (matchbracketclass(uchar(previous), p, ep-1) ||
 366          !matchbracketclass(uchar(*s), p, ep-1)) return NULL;
 367       p=ep;
 368       goto init;  /* else return match(ms, s, ep); */
 369       }
 370     default:
 371       if (lj_char_isdigit(uchar(*(p+1)))) {  /* capture results (%0-%9)? */
 372         s = match_capture(ms, s, uchar(*(p+1)));
 373         if (s == NULL) return NULL;
 374         p+=2;
 375         goto init;  /* else return match(ms, s, p+2) */
 376       }
 377       goto dflt;  /* case default */
 378     }
 379   case '\0':  /* end of pattern */
 380     return s;  /* match succeeded */
 381   case '$':
 382     if (*(p+1) == '\0')  /* is the `$' the last char in pattern? */
 383       return (s == ms->src_end) ? s : NULL;  /* check end of string */
 384     else
 385       goto dflt;
 386   default: dflt: {  /* it is a pattern item */
 387     const char *ep = classend(ms, p);  /* points to what is next */
 388     int m = s<ms->src_end && singlematch(uchar(*s), p, ep);
 389     switch (*ep) {
 390     case '?': {  /* optional */
 391       const char *res;
 392       if (m && ((res=match(ms, s+1, ep+1)) != NULL))
 393         return res;
 394       p=ep+1;
 395       goto init;  /* else return match(ms, s, ep+1); */
 396       }
 397     case '*':  /* 0 or more repetitions */
 398       return max_expand(ms, s, p, ep);
 399     case '+':  /* 1 or more repetitions */
 400       return (m ? max_expand(ms, s+1, p, ep) : NULL);
 401     case '-':  /* 0 or more repetitions (minimum) */
 402       return min_expand(ms, s, p, ep);
 403     default:
 404       if (!m) return NULL;
 405       s++; p=ep;
 406       goto init;  /* else return match(ms, s+1, ep); */
 407     }
 408     }
 409   }
 410 }
 411
 412 static const char *lmemfind(const char *s1, size_t l1,
 413                             const char *s2, size_t l2)
 414 {
 415   if (l2 == 0) {
 416     return s1;  /* empty strings are everywhere */
 417   } else if (l2 > l1) {
 418     return NULL;  /* avoids a negative `l1' */
 419   } else {
 420     const char *init;  /* to search for a `*s2' inside `s1' */
 421     l2--;  /* 1st char will be checked by `memchr' */
 422     l1 = l1-l2;  /* `s2' cannot be found after that */
 423     while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
 424       init++;   /* 1st char is already checked */
 425       if (memcmp(init, s2+1, l2) == 0) {
 426         return init-1;
 427       } else {  /* correct `l1' and `s1' to try again */
 428         l1 -= (size_t)(init-s1);
 429         s1 = init;
 430       }
 431     }
 432     return NULL;  /* not found */
 433   }
 434 }
 435
 436 static void push_onecapture(MatchState *ms, int i, const char *s, const char *e)
 437 {
 438   if (i >= ms->level) {
 439     if (i == 0)  /* ms->level == 0, too */
 440       lua_pushlstring(ms->L, s, (size_t)(e - s));  /* add whole match */
 441     else
 442       lj_err_caller(ms->L, LJ_ERR_STRCAPI);
 443   } else {
 444     ptrdiff_t l = ms->capture[i].len;
 445     if (l == CAP_UNFINISHED) lj_err_caller(ms->L, LJ_ERR_STRCAPU);
 446     if (l == CAP_POSITION)
 447       lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
 448     else
 449       lua_pushlstring(ms->L, ms->capture[i].init, (size_t)l);
 450   }
 451 }
 452
 453 static int push_captures(MatchState *ms, const char *s, const char *e)
 454 {
 455   int i;
 456   int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
 457   luaL_checkstack(ms->L, nlevels, "too many captures");
 458   for (i = 0; i < nlevels; i++)
 459     push_onecapture(ms, i, s, e);
 460   return nlevels;  /* number of strings pushed */
 461 }
 462
 463 static ptrdiff_t posrelat(ptrdiff_t pos, size_t len)
 464 {
 465   /* relative string position: negative means back from end */
 466   if (pos < 0) pos += (ptrdiff_t)len + 1;
 467   return (pos >= 0) ? pos : 0;
 468 }
 469
 470 static int str_find_aux(lua_State *L, int find)
 471 {
 472   size_t l1, l2;
 473   const char *s = luaL_checklstring(L, 1, &l1);
 474   const char *p = luaL_checklstring(L, 2, &l2);
 475   ptrdiff_t init = posrelat(luaL_optinteger(L, 3, 1), l1) - 1;
 476   if (init < 0)
 477     init = 0;
 478   else if ((size_t)(init) > l1)
 479     init = (ptrdiff_t)l1;
 480   if (find && (lua_toboolean(L, 4) ||  /* explicit request? */
 481       strpbrk(p, SPECIALS) == NULL)) {  /* or no special characters? */
 482     /* do a plain search */
 483     const char *s2 = lmemfind(s+init, l1-(size_t)init, p, l2);
 484     if (s2) {
 485       lua_pushinteger(L, s2-s+1);
 486       lua_pushinteger(L, s2-s+(ptrdiff_t)l2);
 487       return 2;
 488     }
 489   } else {
 490     MatchState ms;
 491     int anchor = (*p == '^') ? (p++, 1) : 0;
 492     const char *s1=s+init;
 493     ms.L = L;
 494     ms.src_init = s;
 495     ms.src_end = s+l1;
 496     do {
 497       const char *res;
 498       ms.level = 0;
 499       if ((res=match(&ms, s1, p)) != NULL) {
 500         if (find) {
 501           lua_pushinteger(L, s1-s+1);  /* start */
 502           lua_pushinteger(L, res-s);   /* end */
 503           return push_captures(&ms, NULL, 0) + 2;
 504         } else {
 505           return push_captures(&ms, s1, res);
 506         }
 507       }
 508     } while (s1++ < ms.src_end && !anchor);
 509   }
 510   lua_pushnil(L);  /* not found */
 511   return 1;
 512 }
 513
 514 LJLIB_CF(string_find)
 515 {
 516   return str_find_aux(L, 1);
 517 }
 518
 519 LJLIB_CF(string_match)
 520 {
 521   return str_find_aux(L, 0);
 522 }
 523
 524 LJLIB_NOREG LJLIB_CF(string_gmatch_aux)
 525 {
 526   const char *p = strVdata(lj_lib_upvalue(L, 2));
 527   GCstr *str = strV(lj_lib_upvalue(L, 1));
 528   const char *s = strdata(str);
 529   TValue *tvpos = lj_lib_upvalue(L, 3);
 530   const char *src = s + tvpos->u32.lo;
 531   MatchState ms;
 532   ms.L = L;
 533   ms.src_init = s;
 534   ms.src_end = s + str->len;
 535   for (; src <= ms.src_end; src++) {
 536     const char *e;
 537     ms.level = 0;
 538     if ((e = match(&ms, src, p)) != NULL) {
 539       int32_t pos = (int32_t)(e - s);
 540       if (e == src) pos++;  /* Ensure progress for empty match. */
 541       tvpos->u32.lo = (uint32_t)pos;
 542       return push_captures(&ms, src, e);
 543     }
 544   }
 545   return 0;  /* not found */
 546 }
 547
 548 LJLIB_CF(string_gmatch)
 549 {
 550   lj_lib_checkstr(L, 1);
 551   lj_lib_checkstr(L, 2);
 552   L->top = L->base+3;
 553   (L->top-1)->u64 = 0;
 554   lj_lib_pushcc(L, lj_cf_string_gmatch_aux, FF_string_gmatch_aux, 3);
 555   return 1;
 556 }
 557
 558 static void add_s(MatchState *ms, luaL_Buffer *b, const char *s, const char *e)
 559 {
 560   size_t l, i;
 561   const char *news = lua_tolstring(ms->L, 3, &l);
 562   for (i = 0; i < l; i++) {
 563     if (news[i] != L_ESC) {
 564       luaL_addchar(b, news[i]);
 565     } else {
 566       i++;  /* skip ESC */
 567       if (!lj_char_isdigit(uchar(news[i]))) {
 568         luaL_addchar(b, news[i]);
 569       } else if (news[i] == '0') {
 570         luaL_addlstring(b, s, (size_t)(e - s));
 571       } else {
 572         push_onecapture(ms, news[i] - '1', s, e);
 573         luaL_addvalue(b);  /* add capture to accumulated result */
 574       }
 575     }
 576   }
 577 }
 578
 579 static void add_value(MatchState *ms, luaL_Buffer *b,
 580                       const char *s, const char *e)
 581 {
 582   lua_State *L = ms->L;
 583   switch (lua_type(L, 3)) {
 584     case LUA_TNUMBER:
 585     case LUA_TSTRING: {
 586       add_s(ms, b, s, e);
 587       return;
 588     }
 589     case LUA_TFUNCTION: {
 590       int n;
 591       lua_pushvalue(L, 3);
 592       n = push_captures(ms, s, e);
 593       lua_call(L, n, 1);
 594       break;
 595     }
 596     case LUA_TTABLE: {
 597       push_onecapture(ms, 0, s, e);
 598       lua_gettable(L, 3);
 599       break;
 600     }
 601   }
 602   if (!lua_toboolean(L, -1)) {  /* nil or false? */
 603     lua_pop(L, 1);
 604     lua_pushlstring(L, s, (size_t)(e - s));  /* keep original text */
 605   } else if (!lua_isstring(L, -1)) {
 606     lj_err_callerv(L, LJ_ERR_STRGSRV, luaL_typename(L, -1));
 607   }
 608   luaL_addvalue(b);  /* add result to accumulator */
 609 }
 610
 611 LJLIB_CF(string_gsub)
 612 {
 613   size_t srcl;
 614   const char *src = luaL_checklstring(L, 1, &srcl);
 615   const char *p = luaL_checkstring(L, 2);
 616   int  tr = lua_type(L, 3);
 617   int max_s = luaL_optint(L, 4, (int)(srcl+1));
 618   int anchor = (*p == '^') ? (p++, 1) : 0;
 619   int n = 0;
 620   MatchState ms;
 621   luaL_Buffer b;
 622   if (!(tr == LUA_TNUMBER || tr == LUA_TSTRING ||
 623         tr == LUA_TFUNCTION || tr == LUA_TTABLE))
 624     lj_err_arg(L, 3, LJ_ERR_NOSFT);
 625   luaL_buffinit(L, &b);
 626   ms.L = L;
 627   ms.src_init = src;
 628   ms.src_end = src+srcl;
 629   while (n < max_s) {
 630     const char *e;
 631     ms.level = 0;
 632     e = match(&ms, src, p);
 633     if (e) {
 634       n++;
 635       add_value(&ms, &b, src, e);
 636     }
 637     if (e && e>src) /* non empty match? */
 638       src = e;  /* skip it */
 639     else if (src < ms.src_end)
 640       luaL_addchar(&b, *src++);
 641     else
 642       break;
 643     if (anchor)
 644       break;
 645   }
 646   luaL_addlstring(&b, src, (size_t)(ms.src_end-src));
 647   luaL_pushresult(&b);
 648   lua_pushinteger(L, n);  /* number of substitutions */
 649   return 2;
 650 }
 651
 652 /* ------------------------------------------------------------------------ */
 653
 654 /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
 655 #define MAX_FMTITEM     512
 656 /* valid flags in a format specification */
 657 #define FMT_FLAGS       "-+ #0"
 658 /*
 659 ** maximum size of each format specification (such as '%-099.99d')
 660 ** (+10 accounts for %99.99x plus margin of error)
 661 */
 662 #define MAX_FMTSPEC     (sizeof(FMT_FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
 663
 664 static void addquoted(lua_State *L, luaL_Buffer *b, int arg)
 665 {
 666   GCstr *str = lj_lib_checkstr(L, arg);
 667   int32_t len = (int32_t)str->len;
 668   const char *s = strdata(str);
 669   luaL_addchar(b, '"');
 670   while (len--) {
 671     if (*s == '"' || *s == '\\' || *s == '\n') {
 672       luaL_addchar(b, '\\');
 673       luaL_addchar(b, *s);
 674     } else if (lj_char_iscntrl(uchar(*s))) {
 675       uint32_t c1, c2, c3;
 676       luaL_addchar(b, '\\');
 677       c1 = uchar(*s); c3 = c1 % 10; c1 /= 10; c2 = c1 % 10; c1 /= 10;
 678       if (c1 + lj_char_isdigit(uchar(s[1]))) luaL_addchar(b, '0' + c1);
 679       if (c2 + (c1 + lj_char_isdigit(uchar(s[1])))) luaL_addchar(b, '0' + c2);
 680       luaL_addchar(b, '0' + c3);
 681     } else {
 682       luaL_addchar(b, *s);
 683     }
 684     s++;
 685   }
 686   luaL_addchar(b, '"');
 687 }
 688
 689 static const char *scanformat(lua_State *L, const char *strfrmt, char *form)
 690 {
 691   const char *p = strfrmt;
 692   while (*p != '\0' && strchr(FMT_FLAGS, *p) != NULL) p++;  /* skip flags */
 693   if ((size_t)(p - strfrmt) >= sizeof(FMT_FLAGS))
 694     lj_err_caller(L, LJ_ERR_STRFMTR);
 695   if (lj_char_isdigit(uchar(*p))) p++;  /* skip width */
 696   if (lj_char_isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 697   if (*p == '.') {
 698     p++;
 699     if (lj_char_isdigit(uchar(*p))) p++;  /* skip precision */
 700     if (lj_char_isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 701   }
 702   if (lj_char_isdigit(uchar(*p)))
 703     lj_err_caller(L, LJ_ERR_STRFMTW);
 704   *(form++) = '%';
 705   strncpy(form, strfrmt, (size_t)(p - strfrmt + 1));
 706   form += p - strfrmt + 1;
 707   *form = '\0';
 708   return p;
 709 }
 710
 711 static void addintlen(char *form)
 712 {
 713   size_t l = strlen(form);
 714   char spec = form[l - 1];
 715   strcpy(form + l - 1, LUA_INTFRMLEN);
 716   form[l + sizeof(LUA_INTFRMLEN) - 2] = spec;
 717   form[l + sizeof(LUA_INTFRMLEN) - 1] = '\0';
 718 }
 719
 720 static unsigned LUA_INTFRM_T num2intfrm(lua_State *L, int arg)
 721 {
 722   if (sizeof(LUA_INTFRM_T) == 4) {
 723     return (LUA_INTFRM_T)lj_lib_checkbit(L, arg);
 724   } else {
 725     cTValue *o;
 726     lj_lib_checknumber(L, arg);
 727     o = L->base+arg-1;
 728     if (tvisint(o))
 729       return (LUA_INTFRM_T)intV(o);
 730     else
 731       return (LUA_INTFRM_T)numV(o);
 732   }
 733 }
 734
 735 static unsigned LUA_INTFRM_T num2uintfrm(lua_State *L, int arg)
 736 {
 737   if (sizeof(LUA_INTFRM_T) == 4) {
 738     return (unsigned LUA_INTFRM_T)lj_lib_checkbit(L, arg);
 739   } else {
 740     cTValue *o;
 741     lj_lib_checknumber(L, arg);
 742     o = L->base+arg-1;
 743     if (tvisint(o))
 744       return (unsigned LUA_INTFRM_T)intV(o);
 745     else if ((int32_t)o->u32.hi < 0)
 746       return (unsigned LUA_INTFRM_T)(LUA_INTFRM_T)numV(o);
 747     else
 748       return (unsigned LUA_INTFRM_T)numV(o);
 749   }
 750 }
 751
 752 LJLIB_CF(string_format)
 753 {
 754   int arg = 1, top = (int)(L->top - L->base);
 755   GCstr *fmt = lj_lib_checkstr(L, arg);
 756   const char *strfrmt = strdata(fmt);
 757   const char *strfrmt_end = strfrmt + fmt->len;
 758   luaL_Buffer b;
 759   luaL_buffinit(L, &b);
 760   while (strfrmt < strfrmt_end) {
 761     if (*strfrmt != L_ESC) {
 762       luaL_addchar(&b, *strfrmt++);
 763     } else if (*++strfrmt == L_ESC) {
 764       luaL_addchar(&b, *strfrmt++);  /* %% */
 765     } else { /* format item */
 766       char form[MAX_FMTSPEC];  /* to store the format (`%...') */
 767       char buff[MAX_FMTITEM];  /* to store the formatted item */
 768       if (++arg > top)
 769         luaL_argerror(L, arg, lj_obj_typename[0]);
 770       strfrmt = scanformat(L, strfrmt, form);
 771       switch (*strfrmt++) {
 772       case 'c':
 773         sprintf(buff, form, lj_lib_checkint(L, arg));
 774         break;
 775       case 'd':  case 'i':
 776         addintlen(form);
 777         sprintf(buff, form, num2intfrm(L, arg));
 778         break;
 779       case 'o':  case 'u':  case 'x':  case 'X':
 780         addintlen(form);
 781         sprintf(buff, form, num2uintfrm(L, arg));
 782         break;
 783       case 'e':  case 'E': case 'f': case 'g': case 'G': {
 784         TValue tv;
 785         tv.n = lj_lib_checknum(L, arg);
 786         if (LJ_UNLIKELY((tv.u32.hi << 1) >= 0xffe00000)) {
 787           /* Canonicalize output of non-finite values. */
 788           char *p, nbuf[LJ_STR_NUMBUF];
 789           size_t len = lj_str_bufnum(nbuf, &tv);
 790           if (strfrmt[-1] == 'E' || strfrmt[-1] == 'G') {
 791             nbuf[len-3] = nbuf[len-3] - 0x20;
 792             nbuf[len-2] = nbuf[len-2] - 0x20;
 793             nbuf[len-1] = nbuf[len-1] - 0x20;
 794           }
 795           nbuf[len] = '\0';
 796           for (p = form; *p < 'e' && *p != '.'; p++) ;
 797           *p++ = 's'; *p = '\0';
 798           sprintf(buff, form, nbuf);
 799           break;
 800         }
 801         sprintf(buff, form, (double)tv.n);
 802         break;
 803         }
 804       case 'q':
 805         addquoted(L, &b, arg);
 806         continue;
 807       case 'p':
 808         lj_str_pushf(L, "%p", lua_topointer(L, arg));
 809         luaL_addvalue(&b);
 810         continue;
 811       case 's': {
 812         GCstr *str = lj_lib_checkstr(L, arg);
 813         if (!strchr(form, '.') && str->len >= 100) {
 814           /* no precision and string is too long to be formatted;
 815              keep original string */
 816           setstrV(L, L->top++, str);
 817           luaL_addvalue(&b);
 818           continue;
 819         }
 820         sprintf(buff, form, strdata(str));
 821         break;
 822         }
 823       default:
 824         lj_err_callerv(L, LJ_ERR_STRFMTO, *(strfrmt -1));
 825         break;
 826       }
 827       luaL_addlstring(&b, buff, strlen(buff));
 828     }
 829   }
 830   luaL_pushresult(&b);
 831   return 1;
 832 }
 833
 834 /* ------------------------------------------------------------------------ */
 835
 836 #include "lj_libdef.h"
 837
 838 LUALIB_API int luaopen_string(lua_State *L)
 839 {
 840   GCtab *mt;
 841   global_State *g;
 842   LJ_LIB_REG(L, LUA_STRLIBNAME, string);
 843 #if defined(LUA_COMPAT_GFIND)
 844   lua_getfield(L, -1, "gmatch");
 845   lua_setfield(L, -2, "gfind");
 846 #endif
 847   mt = lj_tab_new(L, 0, 1);
 848   /* NOBARRIER: basemt is a GC root. */
 849   g = G(L);
 850   setgcref(basemt_it(g, LJ_TSTR), obj2gco(mt));
 851   settabV(L, lj_tab_setstr(L, mt, mmname_str(g, MM_index)), tabV(L->top-1));
 852   mt->nomm = (uint8_t)(~(1u<<MM_index));
 853   return 1;
 854 }
 855