src/lib_string.c

   1 /*
   2 ** String library.
   3 ** Copyright (C) 2005-2012 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   int depth;
 152   struct {
 153     const char *init;
 154     ptrdiff_t len;
 155   } capture[LUA_MAXCAPTURES];
 156 } MatchState;
 157
 158 #define L_ESC           '%'
 159 #define SPECIALS        "^$*+?.([%-"
 160
 161 static int check_capture(MatchState *ms, int l)
 162 {
 163   l -= '1';
 164   if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
 165     lj_err_caller(ms->L, LJ_ERR_STRCAPI);
 166   return l;
 167 }
 168
 169 static int capture_to_close(MatchState *ms)
 170 {
 171   int level = ms->level;
 172   for (level--; level>=0; level--)
 173     if (ms->capture[level].len == CAP_UNFINISHED) return level;
 174   lj_err_caller(ms->L, LJ_ERR_STRPATC);
 175   return 0;  /* unreachable */
 176 }
 177
 178 static const char *classend(MatchState *ms, const char *p)
 179 {
 180   switch (*p++) {
 181   case L_ESC:
 182     if (*p == '\0')
 183       lj_err_caller(ms->L, LJ_ERR_STRPATE);
 184     return p+1;
 185   case '[':
 186     if (*p == '^') p++;
 187     do {  /* look for a `]' */
 188       if (*p == '\0')
 189         lj_err_caller(ms->L, LJ_ERR_STRPATM);
 190       if (*(p++) == L_ESC && *p != '\0')
 191         p++;  /* skip escapes (e.g. `%]') */
 192     } while (*p != ']');
 193     return p+1;
 194   default:
 195     return p;
 196   }
 197 }
 198
 199 static const unsigned char match_class_map[32] = {
 200   0,LJ_CHAR_ALPHA,0,LJ_CHAR_CNTRL,LJ_CHAR_DIGIT,0,0,LJ_CHAR_GRAPH,0,0,0,0,
 201   LJ_CHAR_LOWER,0,0,0,LJ_CHAR_PUNCT,0,0,LJ_CHAR_SPACE,0,
 202   LJ_CHAR_UPPER,0,LJ_CHAR_ALNUM,LJ_CHAR_XDIGIT,0,0,0,0,0,0,0
 203 };
 204
 205 static int match_class(int c, int cl)
 206 {
 207   if ((cl & 0xc0) == 0x40) {
 208     int t = match_class_map[(cl&0x1f)];
 209     if (t) {
 210       t = lj_char_isa(c, t);
 211       return (cl & 0x20) ? t : !t;
 212     }
 213     if (cl == 'z') return c == 0;
 214     if (cl == 'Z') return c != 0;
 215   }
 216   return (cl == c);
 217 }
 218
 219 static int matchbracketclass(int c, const char *p, const char *ec)
 220 {
 221   int sig = 1;
 222   if (*(p+1) == '^') {
 223     sig = 0;
 224     p++;  /* skip the `^' */
 225   }
 226   while (++p < ec) {
 227     if (*p == L_ESC) {
 228       p++;
 229       if (match_class(c, uchar(*p)))
 230         return sig;
 231     }
 232     else if ((*(p+1) == '-') && (p+2 < ec)) {
 233       p+=2;
 234       if (uchar(*(p-2)) <= c && c <= uchar(*p))
 235         return sig;
 236     }
 237     else if (uchar(*p) == c) return sig;
 238   }
 239   return !sig;
 240 }
 241
 242 static int singlematch(int c, const char *p, const char *ep)
 243 {
 244   switch (*p) {
 245   case '.': return 1;  /* matches any char */
 246   case L_ESC: return match_class(c, uchar(*(p+1)));
 247   case '[': return matchbracketclass(c, p, ep-1);
 248   default:  return (uchar(*p) == c);
 249   }
 250 }
 251
 252 static const char *match(MatchState *ms, const char *s, const char *p);
 253
 254 static const char *matchbalance(MatchState *ms, const char *s, const char *p)
 255 {
 256   if (*p == 0 || *(p+1) == 0)
 257     lj_err_caller(ms->L, LJ_ERR_STRPATU);
 258   if (*s != *p) {
 259     return NULL;
 260   } else {
 261     int b = *p;
 262     int e = *(p+1);
 263     int cont = 1;
 264     while (++s < ms->src_end) {
 265       if (*s == e) {
 266         if (--cont == 0) return s+1;
 267       } else if (*s == b) {
 268         cont++;
 269       }
 270     }
 271   }
 272   return NULL;  /* string ends out of balance */
 273 }
 274
 275 static const char *max_expand(MatchState *ms, const char *s,
 276                               const char *p, const char *ep)
 277 {
 278   ptrdiff_t i = 0;  /* counts maximum expand for item */
 279   while ((s+i)<ms->src_end && singlematch(uchar(*(s+i)), p, ep))
 280     i++;
 281   /* keeps trying to match with the maximum repetitions */
 282   while (i>=0) {
 283     const char *res = match(ms, (s+i), ep+1);
 284     if (res) return res;
 285     i--;  /* else didn't match; reduce 1 repetition to try again */
 286   }
 287   return NULL;
 288 }
 289
 290 static const char *min_expand(MatchState *ms, const char *s,
 291                               const char *p, const char *ep)
 292 {
 293   for (;;) {
 294     const char *res = match(ms, s, ep+1);
 295     if (res != NULL)
 296       return res;
 297     else if (s<ms->src_end && singlematch(uchar(*s), p, ep))
 298       s++;  /* try with one more repetition */
 299     else
 300       return NULL;
 301   }
 302 }
 303
 304 static const char *start_capture(MatchState *ms, const char *s,
 305                                  const char *p, int what)
 306 {
 307   const char *res;
 308   int level = ms->level;
 309   if (level >= LUA_MAXCAPTURES) lj_err_caller(ms->L, LJ_ERR_STRCAPN);
 310   ms->capture[level].init = s;
 311   ms->capture[level].len = what;
 312   ms->level = level+1;
 313   if ((res=match(ms, s, p)) == NULL)  /* match failed? */
 314     ms->level--;  /* undo capture */
 315   return res;
 316 }
 317
 318 static const char *end_capture(MatchState *ms, const char *s,
 319                                const char *p)
 320 {
 321   int l = capture_to_close(ms);
 322   const char *res;
 323   ms->capture[l].len = s - ms->capture[l].init;  /* close capture */
 324   if ((res = match(ms, s, p)) == NULL)  /* match failed? */
 325     ms->capture[l].len = CAP_UNFINISHED;  /* undo capture */
 326   return res;
 327 }
 328
 329 static const char *match_capture(MatchState *ms, const char *s, int l)
 330 {
 331   size_t len;
 332   l = check_capture(ms, l);
 333   len = (size_t)ms->capture[l].len;
 334   if ((size_t)(ms->src_end-s) >= len &&
 335       memcmp(ms->capture[l].init, s, len) == 0)
 336     return s+len;
 337   else
 338     return NULL;
 339 }
 340
 341 static const char *match(MatchState *ms, const char *s, const char *p)
 342 {
 343   if (++ms->depth > LJ_MAX_XLEVEL)
 344     lj_err_caller(ms->L, LJ_ERR_STRPATX);
 345   init: /* using goto's to optimize tail recursion */
 346   switch (*p) {
 347   case '(':  /* start capture */
 348     if (*(p+1) == ')')  /* position capture? */
 349       s = start_capture(ms, s, p+2, CAP_POSITION);
 350     else
 351       s = start_capture(ms, s, p+1, CAP_UNFINISHED);
 352     break;
 353   case ')':  /* end capture */
 354     s = end_capture(ms, s, p+1);
 355     break;
 356   case L_ESC:
 357     switch (*(p+1)) {
 358     case 'b':  /* balanced string? */
 359       s = matchbalance(ms, s, p+2);
 360       if (s == NULL) break;
 361       p+=4;
 362       goto init;  /* else s = match(ms, s, p+4); */
 363     case 'f': {  /* frontier? */
 364       const char *ep; char previous;
 365       p += 2;
 366       if (*p != '[')
 367         lj_err_caller(ms->L, LJ_ERR_STRPATB);
 368       ep = classend(ms, p);  /* points to what is next */
 369       previous = (s == ms->src_init) ? '\0' : *(s-1);
 370       if (matchbracketclass(uchar(previous), p, ep-1) ||
 371          !matchbracketclass(uchar(*s), p, ep-1)) { s = NULL; break; }
 372       p=ep;
 373       goto init;  /* else s = match(ms, s, ep); */
 374       }
 375     default:
 376       if (lj_char_isdigit(uchar(*(p+1)))) {  /* capture results (%0-%9)? */
 377         s = match_capture(ms, s, uchar(*(p+1)));
 378         if (s == NULL) break;
 379         p+=2;
 380         goto init;  /* else s = match(ms, s, p+2) */
 381       }
 382       goto dflt;  /* case default */
 383     }
 384     break;
 385   case '\0':  /* end of pattern */
 386     break;  /* match succeeded */
 387   case '$':
 388     /* is the `$' the last char in pattern? */
 389     if (*(p+1) != '\0') goto dflt;
 390     if (s != ms->src_end) s = NULL;  /* check end of string */
 391     break;
 392   default: dflt: {  /* it is a pattern item */
 393     const char *ep = classend(ms, p);  /* points to what is next */
 394     int m = s<ms->src_end && singlematch(uchar(*s), p, ep);
 395     switch (*ep) {
 396     case '?': {  /* optional */
 397       const char *res;
 398       if (m && ((res=match(ms, s+1, ep+1)) != NULL)) {
 399         s = res;
 400         break;
 401       }
 402       p=ep+1;
 403       goto init;  /* else s = match(ms, s, ep+1); */
 404       }
 405     case '*':  /* 0 or more repetitions */
 406       s = max_expand(ms, s, p, ep);
 407       break;
 408     case '+':  /* 1 or more repetitions */
 409       s = (m ? max_expand(ms, s+1, p, ep) : NULL);
 410       break;
 411     case '-':  /* 0 or more repetitions (minimum) */
 412       s = min_expand(ms, s, p, ep);
 413       break;
 414     default:
 415       if (m) { s++; p=ep; goto init; }  /* else s = match(ms, s+1, ep); */
 416       s = NULL;
 417       break;
 418     }
 419     break;
 420     }
 421   }
 422   ms->depth--;
 423   return s;
 424 }
 425
 426 static const char *lmemfind(const char *s1, size_t l1,
 427                             const char *s2, size_t l2)
 428 {
 429   if (l2 == 0) {
 430     return s1;  /* empty strings are everywhere */
 431   } else if (l2 > l1) {
 432     return NULL;  /* avoids a negative `l1' */
 433   } else {
 434     const char *init;  /* to search for a `*s2' inside `s1' */
 435     l2--;  /* 1st char will be checked by `memchr' */
 436     l1 = l1-l2;  /* `s2' cannot be found after that */
 437     while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
 438       init++;   /* 1st char is already checked */
 439       if (memcmp(init, s2+1, l2) == 0) {
 440         return init-1;
 441       } else {  /* correct `l1' and `s1' to try again */
 442         l1 -= (size_t)(init-s1);
 443         s1 = init;
 444       }
 445     }
 446     return NULL;  /* not found */
 447   }
 448 }
 449
 450 static void push_onecapture(MatchState *ms, int i, const char *s, const char *e)
 451 {
 452   if (i >= ms->level) {
 453     if (i == 0)  /* ms->level == 0, too */
 454       lua_pushlstring(ms->L, s, (size_t)(e - s));  /* add whole match */
 455     else
 456       lj_err_caller(ms->L, LJ_ERR_STRCAPI);
 457   } else {
 458     ptrdiff_t l = ms->capture[i].len;
 459     if (l == CAP_UNFINISHED) lj_err_caller(ms->L, LJ_ERR_STRCAPU);
 460     if (l == CAP_POSITION)
 461       lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
 462     else
 463       lua_pushlstring(ms->L, ms->capture[i].init, (size_t)l);
 464   }
 465 }
 466
 467 static int push_captures(MatchState *ms, const char *s, const char *e)
 468 {
 469   int i;
 470   int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
 471   luaL_checkstack(ms->L, nlevels, "too many captures");
 472   for (i = 0; i < nlevels; i++)
 473     push_onecapture(ms, i, s, e);
 474   return nlevels;  /* number of strings pushed */
 475 }
 476
 477 static ptrdiff_t posrelat(ptrdiff_t pos, size_t len)
 478 {
 479   /* relative string position: negative means back from end */
 480   if (pos < 0) pos += (ptrdiff_t)len + 1;
 481   return (pos >= 0) ? pos : 0;
 482 }
 483
 484 static int str_find_aux(lua_State *L, int find)
 485 {
 486   size_t l1, l2;
 487   const char *s = luaL_checklstring(L, 1, &l1);
 488   const char *p = luaL_checklstring(L, 2, &l2);
 489   ptrdiff_t init = posrelat(luaL_optinteger(L, 3, 1), l1) - 1;
 490   if (init < 0)
 491     init = 0;
 492   else if ((size_t)(init) > l1)
 493     init = (ptrdiff_t)l1;
 494   if (find && (lua_toboolean(L, 4) ||  /* explicit request? */
 495       strpbrk(p, SPECIALS) == NULL)) {  /* or no special characters? */
 496     /* do a plain search */
 497     const char *s2 = lmemfind(s+init, l1-(size_t)init, p, l2);
 498     if (s2) {
 499       lua_pushinteger(L, s2-s+1);
 500       lua_pushinteger(L, s2-s+(ptrdiff_t)l2);
 501       return 2;
 502     }
 503   } else {
 504     MatchState ms;
 505     int anchor = (*p == '^') ? (p++, 1) : 0;
 506     const char *s1=s+init;
 507     ms.L = L;
 508     ms.src_init = s;
 509     ms.src_end = s+l1;
 510     do {
 511       const char *res;
 512       ms.level = ms.depth = 0;
 513       if ((res=match(&ms, s1, p)) != NULL) {
 514         if (find) {
 515           lua_pushinteger(L, s1-s+1);  /* start */
 516           lua_pushinteger(L, res-s);   /* end */
 517           return push_captures(&ms, NULL, 0) + 2;
 518         } else {
 519           return push_captures(&ms, s1, res);
 520         }
 521       }
 522     } while (s1++ < ms.src_end && !anchor);
 523   }
 524   lua_pushnil(L);  /* not found */
 525   return 1;
 526 }
 527
 528 LJLIB_CF(string_find)
 529 {
 530   return str_find_aux(L, 1);
 531 }
 532
 533 LJLIB_CF(string_match)
 534 {
 535   return str_find_aux(L, 0);
 536 }
 537
 538 LJLIB_NOREG LJLIB_CF(string_gmatch_aux)
 539 {
 540   const char *p = strVdata(lj_lib_upvalue(L, 2));
 541   GCstr *str = strV(lj_lib_upvalue(L, 1));
 542   const char *s = strdata(str);
 543   TValue *tvpos = lj_lib_upvalue(L, 3);
 544   const char *src = s + tvpos->u32.lo;
 545   MatchState ms;
 546   ms.L = L;
 547   ms.src_init = s;
 548   ms.src_end = s + str->len;
 549   for (; src <= ms.src_end; src++) {
 550     const char *e;
 551     ms.level = ms.depth = 0;
 552     if ((e = match(&ms, src, p)) != NULL) {
 553       int32_t pos = (int32_t)(e - s);
 554       if (e == src) pos++;  /* Ensure progress for empty match. */
 555       tvpos->u32.lo = (uint32_t)pos;
 556       return push_captures(&ms, src, e);
 557     }
 558   }
 559   return 0;  /* not found */
 560 }
 561
 562 LJLIB_CF(string_gmatch)
 563 {
 564   lj_lib_checkstr(L, 1);
 565   lj_lib_checkstr(L, 2);
 566   L->top = L->base+3;
 567   (L->top-1)->u64 = 0;
 568   lj_lib_pushcc(L, lj_cf_string_gmatch_aux, FF_string_gmatch_aux, 3);
 569   return 1;
 570 }
 571
 572 static void add_s(MatchState *ms, luaL_Buffer *b, const char *s, const char *e)
 573 {
 574   size_t l, i;
 575   const char *news = lua_tolstring(ms->L, 3, &l);
 576   for (i = 0; i < l; i++) {
 577     if (news[i] != L_ESC) {
 578       luaL_addchar(b, news[i]);
 579     } else {
 580       i++;  /* skip ESC */
 581       if (!lj_char_isdigit(uchar(news[i]))) {
 582         luaL_addchar(b, news[i]);
 583       } else if (news[i] == '0') {
 584         luaL_addlstring(b, s, (size_t)(e - s));
 585       } else {
 586         push_onecapture(ms, news[i] - '1', s, e);
 587         luaL_addvalue(b);  /* add capture to accumulated result */
 588       }
 589     }
 590   }
 591 }
 592
 593 static void add_value(MatchState *ms, luaL_Buffer *b,
 594                       const char *s, const char *e)
 595 {
 596   lua_State *L = ms->L;
 597   switch (lua_type(L, 3)) {
 598     case LUA_TNUMBER:
 599     case LUA_TSTRING: {
 600       add_s(ms, b, s, e);
 601       return;
 602     }
 603     case LUA_TFUNCTION: {
 604       int n;
 605       lua_pushvalue(L, 3);
 606       n = push_captures(ms, s, e);
 607       lua_call(L, n, 1);
 608       break;
 609     }
 610     case LUA_TTABLE: {
 611       push_onecapture(ms, 0, s, e);
 612       lua_gettable(L, 3);
 613       break;
 614     }
 615   }
 616   if (!lua_toboolean(L, -1)) {  /* nil or false? */
 617     lua_pop(L, 1);
 618     lua_pushlstring(L, s, (size_t)(e - s));  /* keep original text */
 619   } else if (!lua_isstring(L, -1)) {
 620     lj_err_callerv(L, LJ_ERR_STRGSRV, luaL_typename(L, -1));
 621   }
 622   luaL_addvalue(b);  /* add result to accumulator */
 623 }
 624
 625 LJLIB_CF(string_gsub)
 626 {
 627   size_t srcl;
 628   const char *src = luaL_checklstring(L, 1, &srcl);
 629   const char *p = luaL_checkstring(L, 2);
 630   int  tr = lua_type(L, 3);
 631   int max_s = luaL_optint(L, 4, (int)(srcl+1));
 632   int anchor = (*p == '^') ? (p++, 1) : 0;
 633   int n = 0;
 634   MatchState ms;
 635   luaL_Buffer b;
 636   if (!(tr == LUA_TNUMBER || tr == LUA_TSTRING ||
 637         tr == LUA_TFUNCTION || tr == LUA_TTABLE))
 638     lj_err_arg(L, 3, LJ_ERR_NOSFT);
 639   luaL_buffinit(L, &b);
 640   ms.L = L;
 641   ms.src_init = src;
 642   ms.src_end = src+srcl;
 643   while (n < max_s) {
 644     const char *e;
 645     ms.level = ms.depth = 0;
 646     e = match(&ms, src, p);
 647     if (e) {
 648       n++;
 649       add_value(&ms, &b, src, e);
 650     }
 651     if (e && e>src) /* non empty match? */
 652       src = e;  /* skip it */
 653     else if (src < ms.src_end)
 654       luaL_addchar(&b, *src++);
 655     else
 656       break;
 657     if (anchor)
 658       break;
 659   }
 660   luaL_addlstring(&b, src, (size_t)(ms.src_end-src));
 661   luaL_pushresult(&b);
 662   lua_pushinteger(L, n);  /* number of substitutions */
 663   return 2;
 664 }
 665
 666 /* ------------------------------------------------------------------------ */
 667
 668 /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
 669 #define MAX_FMTITEM     512
 670 /* valid flags in a format specification */
 671 #define FMT_FLAGS       "-+ #0"
 672 /*
 673 ** maximum size of each format specification (such as '%-099.99d')
 674 ** (+10 accounts for %99.99x plus margin of error)
 675 */
 676 #define MAX_FMTSPEC     (sizeof(FMT_FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
 677
 678 static void addquoted(lua_State *L, luaL_Buffer *b, int arg)
 679 {
 680   GCstr *str = lj_lib_checkstr(L, arg);
 681   int32_t len = (int32_t)str->len;
 682   const char *s = strdata(str);
 683   luaL_addchar(b, '"');
 684   while (len--) {
 685     uint32_t c = uchar(*s);
 686     if (c == '"' || c == '\\' || c == '\n') {
 687       luaL_addchar(b, '\\');
 688     } else if (lj_char_iscntrl(c)) {  /* This can only be 0-31 or 127. */
 689       uint32_t d;
 690       luaL_addchar(b, '\\');
 691       if (c >= 100 || lj_char_isdigit(uchar(s[1]))) {
 692         luaL_addchar(b, '0'+(c >= 100)); if (c >= 100) c -= 100;
 693         goto tens;
 694       } else if (c >= 10) {
 695       tens:
 696         d = (c * 205) >> 11; c -= d * 10; luaL_addchar(b, '0'+d);
 697       }
 698       c += '0';
 699     }
 700     luaL_addchar(b, c);
 701     s++;
 702   }
 703   luaL_addchar(b, '"');
 704 }
 705
 706 static const char *scanformat(lua_State *L, const char *strfrmt, char *form)
 707 {
 708   const char *p = strfrmt;
 709   while (*p != '\0' && strchr(FMT_FLAGS, *p) != NULL) p++;  /* skip flags */
 710   if ((size_t)(p - strfrmt) >= sizeof(FMT_FLAGS))
 711     lj_err_caller(L, LJ_ERR_STRFMTR);
 712   if (lj_char_isdigit(uchar(*p))) p++;  /* skip width */
 713   if (lj_char_isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 714   if (*p == '.') {
 715     p++;
 716     if (lj_char_isdigit(uchar(*p))) p++;  /* skip precision */
 717     if (lj_char_isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 718   }
 719   if (lj_char_isdigit(uchar(*p)))
 720     lj_err_caller(L, LJ_ERR_STRFMTW);
 721   *(form++) = '%';
 722   strncpy(form, strfrmt, (size_t)(p - strfrmt + 1));
 723   form += p - strfrmt + 1;
 724   *form = '\0';
 725   return p;
 726 }
 727
 728 static void addintlen(char *form)
 729 {
 730   size_t l = strlen(form);
 731   char spec = form[l - 1];
 732   strcpy(form + l - 1, LUA_INTFRMLEN);
 733   form[l + sizeof(LUA_INTFRMLEN) - 2] = spec;
 734   form[l + sizeof(LUA_INTFRMLEN) - 1] = '\0';
 735 }
 736
 737 static unsigned LUA_INTFRM_T num2intfrm(lua_State *L, int arg)
 738 {
 739   if (sizeof(LUA_INTFRM_T) == 4) {
 740     return (LUA_INTFRM_T)lj_lib_checkbit(L, arg);
 741   } else {
 742     cTValue *o;
 743     lj_lib_checknumber(L, arg);
 744     o = L->base+arg-1;
 745     if (tvisint(o))
 746       return (LUA_INTFRM_T)intV(o);
 747     else
 748       return (LUA_INTFRM_T)numV(o);
 749   }
 750 }
 751
 752 static unsigned LUA_INTFRM_T num2uintfrm(lua_State *L, int arg)
 753 {
 754   if (sizeof(LUA_INTFRM_T) == 4) {
 755     return (unsigned LUA_INTFRM_T)lj_lib_checkbit(L, arg);
 756   } else {
 757     cTValue *o;
 758     lj_lib_checknumber(L, arg);
 759     o = L->base+arg-1;
 760     if (tvisint(o))
 761       return (unsigned LUA_INTFRM_T)intV(o);
 762     else if ((int32_t)o->u32.hi < 0)
 763       return (unsigned LUA_INTFRM_T)(LUA_INTFRM_T)numV(o);
 764     else
 765       return (unsigned LUA_INTFRM_T)numV(o);
 766   }
 767 }
 768
 769 LJLIB_CF(string_format)
 770 {
 771   int arg = 1, top = (int)(L->top - L->base);
 772   GCstr *fmt = lj_lib_checkstr(L, arg);
 773   const char *strfrmt = strdata(fmt);
 774   const char *strfrmt_end = strfrmt + fmt->len;
 775   luaL_Buffer b;
 776   luaL_buffinit(L, &b);
 777   while (strfrmt < strfrmt_end) {
 778     if (*strfrmt != L_ESC) {
 779       luaL_addchar(&b, *strfrmt++);
 780     } else if (*++strfrmt == L_ESC) {
 781       luaL_addchar(&b, *strfrmt++);  /* %% */
 782     } else { /* format item */
 783       char form[MAX_FMTSPEC];  /* to store the format (`%...') */
 784       char buff[MAX_FMTITEM];  /* to store the formatted item */
 785       if (++arg > top)
 786         luaL_argerror(L, arg, lj_obj_typename[0]);
 787       strfrmt = scanformat(L, strfrmt, form);
 788       switch (*strfrmt++) {
 789       case 'c':
 790         sprintf(buff, form, lj_lib_checkint(L, arg));
 791         break;
 792       case 'd':  case 'i':
 793         addintlen(form);
 794         sprintf(buff, form, num2intfrm(L, arg));
 795         break;
 796       case 'o':  case 'u':  case 'x':  case 'X':
 797         addintlen(form);
 798         sprintf(buff, form, num2uintfrm(L, arg));
 799         break;
 800       case 'e':  case 'E': case 'f': case 'g': case 'G': case 'a': case 'A': {
 801         TValue tv;
 802         tv.n = lj_lib_checknum(L, arg);
 803         if (LJ_UNLIKELY((tv.u32.hi << 1) >= 0xffe00000)) {
 804           /* Canonicalize output of non-finite values. */
 805           char *p, nbuf[LJ_STR_NUMBUF];
 806           size_t len = lj_str_bufnum(nbuf, &tv);
 807           if (strfrmt[-1] < 'a') {
 808             nbuf[len-3] = nbuf[len-3] - 0x20;
 809             nbuf[len-2] = nbuf[len-2] - 0x20;
 810             nbuf[len-1] = nbuf[len-1] - 0x20;
 811           }
 812           nbuf[len] = '\0';
 813           for (p = form; *p < 'A' && *p != '.'; p++) ;
 814           *p++ = 's'; *p = '\0';
 815           sprintf(buff, form, nbuf);
 816           break;
 817         }
 818         sprintf(buff, form, (double)tv.n);
 819         break;
 820         }
 821       case 'q':
 822         addquoted(L, &b, arg);
 823         continue;
 824       case 'p':
 825         lj_str_pushf(L, "%p", lua_topointer(L, arg));
 826         luaL_addvalue(&b);
 827         continue;
 828       case 's': {
 829         GCstr *str = lj_lib_checkstr(L, arg);
 830         if (!strchr(form, '.') && str->len >= 100) {
 831           /* no precision and string is too long to be formatted;
 832              keep original string */
 833           setstrV(L, L->top++, str);
 834           luaL_addvalue(&b);
 835           continue;
 836         }
 837         sprintf(buff, form, strdata(str));
 838         break;
 839         }
 840       default:
 841         lj_err_callerv(L, LJ_ERR_STRFMTO, *(strfrmt -1));
 842         break;
 843       }
 844       luaL_addlstring(&b, buff, strlen(buff));
 845     }
 846   }
 847   luaL_pushresult(&b);
 848   return 1;
 849 }
 850
 851 /* ------------------------------------------------------------------------ */
 852
 853 #include "lj_libdef.h"
 854
 855 LUALIB_API int luaopen_string(lua_State *L)
 856 {
 857   GCtab *mt;
 858   global_State *g;
 859   LJ_LIB_REG(L, LUA_STRLIBNAME, string);
 860 #if defined(LUA_COMPAT_GFIND) && !LJ_52
 861   lua_getfield(L, -1, "gmatch");
 862   lua_setfield(L, -2, "gfind");
 863 #endif
 864   mt = lj_tab_new(L, 0, 1);
 865   /* NOBARRIER: basemt is a GC root. */
 866   g = G(L);
 867   setgcref(basemt_it(g, LJ_TSTR), obj2gco(mt));
 868   settabV(L, lj_tab_setstr(L, mt, mmname_str(g, MM_index)), tabV(L->top-1));
 869   mt->nomm = (uint8_t)(~(1u<<MM_index));
 870   return 1;
 871 }
 872