client/deps/liblua/lstrlib.c

   1 /*
   2 ** $Id: lstrlib.c,v 1.178 2012/08/14 18:12:34 roberto Exp $
   3 ** Standard library for string operations and pattern-matching
   4 ** See Copyright Notice in lua.h
   5 */
   6
   7
   8 #include <ctype.h>
   9 #include <stddef.h>
  10 #include <stdio.h>
  11 #include <stdlib.h>
  12 #include <string.h>
  13
  14 #define lstrlib_c
  15 #define LUA_LIB
  16
  17 #include "lua.h"
  18
  19 #include "lauxlib.h"
  20 #include "lualib.h"
  21
  22
  23 /*
  24 ** maximum number of captures that a pattern can do during
  25 ** pattern-matching. This limit is arbitrary.
  26 */
  27 #if !defined(LUA_MAXCAPTURES)
  28 #define LUA_MAXCAPTURES 32
  29 #endif
  30
  31
  32 /* macro to `unsign' a character */
  33 #define uchar(c) ((unsigned char)(c))
  34
  35
  36
  37 static int str_len(lua_State *L) {
  38     size_t l;
  39     luaL_checklstring(L, 1, &l);
  40     lua_pushinteger(L, (lua_Integer)l);
  41     return 1;
  42 }
  43
  44
  45 /* translate a relative string position: negative means back from end */
  46 static size_t posrelat(ptrdiff_t pos, size_t len) {
  47     if (pos >= 0) return (size_t)pos;
  48     else if (0u - (size_t)pos > len) return 0;
  49     else return len - ((size_t) - pos) + 1;
  50 }
  51
  52
  53 static int str_sub(lua_State *L) {
  54     size_t l;
  55     const char *s = luaL_checklstring(L, 1, &l);
  56     size_t start = posrelat(luaL_checkinteger(L, 2), l);
  57     size_t end = posrelat(luaL_optinteger(L, 3, -1), l);
  58     if (start < 1) start = 1;
  59     if (end > l) end = l;
  60     if (start <= end)
  61         lua_pushlstring(L, s + start - 1, end - start + 1);
  62     else lua_pushliteral(L, "");
  63     return 1;
  64 }
  65
  66
  67 static int str_reverse(lua_State *L) {
  68     size_t l, i;
  69     luaL_Buffer b;
  70     const char *s = luaL_checklstring(L, 1, &l);
  71     char *p = luaL_buffinitsize(L, &b, l);
  72     for (i = 0; i < l; i++)
  73         p[i] = s[l - i - 1];
  74     luaL_pushresultsize(&b, l);
  75     return 1;
  76 }
  77
  78
  79 static int str_lower(lua_State *L) {
  80     size_t l;
  81     size_t i;
  82     luaL_Buffer b;
  83     const char *s = luaL_checklstring(L, 1, &l);
  84     char *p = luaL_buffinitsize(L, &b, l);
  85     for (i = 0; i < l; i++)
  86         p[i] = tolower(uchar(s[i]));
  87     luaL_pushresultsize(&b, l);
  88     return 1;
  89 }
  90
  91
  92 static int str_upper(lua_State *L) {
  93     size_t l;
  94     size_t i;
  95     luaL_Buffer b;
  96     const char *s = luaL_checklstring(L, 1, &l);
  97     char *p = luaL_buffinitsize(L, &b, l);
  98     for (i = 0; i < l; i++)
  99         p[i] = toupper(uchar(s[i]));
 100     luaL_pushresultsize(&b, l);
 101     return 1;
 102 }
 103
 104
 105 /* reasonable limit to avoid arithmetic overflow */
 106 #define MAXSIZE ((~(size_t)0) >> 1)
 107
 108 static int str_rep(lua_State *L) {
 109     size_t l, lsep;
 110     const char *s = luaL_checklstring(L, 1, &l);
 111     int n = luaL_checkint(L, 2);
 112     const char *sep = luaL_optlstring(L, 3, "", &lsep);
 113     if (n <= 0) lua_pushliteral(L, "");
 114     else if (l + lsep < l || l + lsep >= MAXSIZE / n)  /* may overflow? */
 115         return luaL_error(L, "resulting string too large");
 116     else {
 117         size_t totallen = n * l + (n - 1) * lsep;
 118         luaL_Buffer b;
 119         char *p = luaL_buffinitsize(L, &b, totallen);
 120         while (n-- > 1) {  /* first n-1 copies (followed by separator) */
 121             memcpy(p, s, l * sizeof(char));
 122             p += l;
 123             if (lsep > 0) {  /* avoid empty 'memcpy' (may be expensive) */
 124                 memcpy(p, sep, lsep * sizeof(char));
 125                 p += lsep;
 126             }
 127         }
 128         memcpy(p, s, l * sizeof(char));  /* last copy (not followed by separator) */
 129         luaL_pushresultsize(&b, totallen);
 130     }
 131     return 1;
 132 }
 133
 134
 135 static int str_byte(lua_State *L) {
 136     size_t l;
 137     const char *s = luaL_checklstring(L, 1, &l);
 138     size_t posi = posrelat(luaL_optinteger(L, 2, 1), l);
 139     size_t pose = posrelat(luaL_optinteger(L, 3, posi), l);
 140     int n, i;
 141     if (posi < 1) posi = 1;
 142     if (pose > l) pose = l;
 143     if (posi > pose) return 0;  /* empty interval; return no values */
 144     n = (int)(pose -  posi + 1);
 145     if (posi + n <= pose)  /* (size_t -> int) overflow? */
 146         return luaL_error(L, "string slice too long");
 147     luaL_checkstack(L, n, "string slice too long");
 148     for (i = 0; i < n; i++)
 149         lua_pushinteger(L, uchar(s[posi + i - 1]));
 150     return n;
 151 }
 152
 153
 154 static int str_char(lua_State *L) {
 155     int n = lua_gettop(L);  /* number of arguments */
 156     int i;
 157     luaL_Buffer b;
 158     char *p = luaL_buffinitsize(L, &b, n);
 159     for (i = 1; i <= n; i++) {
 160         int c = luaL_checkint(L, i);
 161         luaL_argcheck(L, uchar(c) == c, i, "value out of range");
 162         p[i - 1] = uchar(c);
 163     }
 164     luaL_pushresultsize(&b, n);
 165     return 1;
 166 }
 167
 168
 169 static int writer(lua_State *L, const void *b, size_t size, void *B) {
 170     (void)L;
 171     luaL_addlstring((luaL_Buffer *) B, (const char *)b, size);
 172     return 0;
 173 }
 174
 175
 176 static int str_dump(lua_State *L) {
 177     luaL_Buffer b;
 178     luaL_checktype(L, 1, LUA_TFUNCTION);
 179     lua_settop(L, 1);
 180     luaL_buffinit(L, &b);
 181     if (lua_dump(L, writer, &b) != 0)
 182         return luaL_error(L, "unable to dump given function");
 183     luaL_pushresult(&b);
 184     return 1;
 185 }
 186
 187
 188
 189 /*
 190 ** {======================================================
 191 ** PATTERN MATCHING
 192 ** =======================================================
 193 */
 194
 195
 196 #define CAP_UNFINISHED (-1)
 197 #define CAP_POSITION   (-2)
 198
 199
 200 typedef struct MatchState {
 201     int matchdepth;  /* control for recursive depth (to avoid C stack overflow) */
 202     const char *src_init;  /* init of source string */
 203     const char *src_end;  /* end ('\0') of source string */
 204     const char *p_end;  /* end ('\0') of pattern */
 205     lua_State *L;
 206     int level;  /* total number of captures (finished or unfinished) */
 207     struct {
 208         const char *init;
 209         ptrdiff_t len;
 210     } capture[LUA_MAXCAPTURES];
 211 } MatchState;
 212
 213
 214 /* recursive function */
 215 static const char *match(MatchState *ms, const char *s, const char *p);
 216
 217
 218 /* maximum recursion depth for 'match' */
 219 #if !defined(MAXCCALLS)
 220 #define MAXCCALLS 200
 221 #endif
 222
 223
 224 #define L_ESC '%'
 225 #define SPECIALS "^$*+?.([%-"
 226
 227
 228 static int check_capture(MatchState *ms, int l) {
 229     l -= '1';
 230     if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
 231         return luaL_error(ms->L, "invalid capture index %%%d", l + 1);
 232     return l;
 233 }
 234
 235
 236 static int capture_to_close(MatchState *ms) {
 237     int level = ms->level;
 238     for (level--; level >= 0; level--)
 239         if (ms->capture[level].len == CAP_UNFINISHED) return level;
 240     return luaL_error(ms->L, "invalid pattern capture");
 241 }
 242
 243
 244 static const char *classend(MatchState *ms, const char *p) {
 245     switch (*p++) {
 246         case L_ESC: {
 247             if (p == ms->p_end)
 248                 luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");
 249             return p + 1;
 250         }
 251         case '[': {
 252             if (*p == '^') p++;
 253             do {  /* look for a `]' */
 254                 if (p == ms->p_end)
 255                     luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");
 256                 if (*(p++) == L_ESC && p < ms->p_end)
 257                     p++;  /* skip escapes (e.g. `%]') */
 258             } while (*p != ']');
 259             return p + 1;
 260         }
 261         default: {
 262             return p;
 263         }
 264     }
 265 }
 266
 267
 268 static int match_class(int c, int cl) {
 269     int res;
 270     switch (tolower(cl)) {
 271         case 'a' :
 272             res = isalpha(c);
 273             break;
 274         case 'c' :
 275             res = iscntrl(c);
 276             break;
 277         case 'd' :
 278             res = isdigit(c);
 279             break;
 280         case 'g' :
 281             res = isgraph(c);
 282             break;
 283         case 'l' :
 284             res = islower(c);
 285             break;
 286         case 'p' :
 287             res = ispunct(c);
 288             break;
 289         case 's' :
 290             res = isspace(c);
 291             break;
 292         case 'u' :
 293             res = isupper(c);
 294             break;
 295         case 'w' :
 296             res = isalnum(c);
 297             break;
 298         case 'x' :
 299             res = isxdigit(c);
 300             break;
 301         case 'z' :
 302             res = (c == 0);
 303             break;  /* deprecated option */
 304         default:
 305             return (cl == c);
 306     }
 307     return (islower(cl) ? res : !res);
 308 }
 309
 310
 311 static int matchbracketclass(int c, const char *p, const char *ec) {
 312     int sig = 1;
 313     if (*(p + 1) == '^') {
 314         sig = 0;
 315         p++;  /* skip the `^' */
 316     }
 317     while (++p < ec) {
 318         if (*p == L_ESC) {
 319             p++;
 320             if (match_class(c, uchar(*p)))
 321                 return sig;
 322         } else if ((*(p + 1) == '-') && (p + 2 < ec)) {
 323             p += 2;
 324             if (uchar(*(p - 2)) <= c && c <= uchar(*p))
 325                 return sig;
 326         } else if (uchar(*p) == c) return sig;
 327     }
 328     return !sig;
 329 }
 330
 331
 332 static int singlematch(MatchState *ms, const char *s, const char *p,
 333                        const char *ep) {
 334     if (s >= ms->src_end)
 335         return 0;
 336     else {
 337         int c = uchar(*s);
 338         switch (*p) {
 339             case '.':
 340                 return 1;  /* matches any char */
 341             case L_ESC:
 342                 return match_class(c, uchar(*(p + 1)));
 343             case '[':
 344                 return matchbracketclass(c, p, ep - 1);
 345             default:
 346                 return (uchar(*p) == c);
 347         }
 348     }
 349 }
 350
 351
 352 static const char *matchbalance(MatchState *ms, const char *s,
 353                                 const char *p) {
 354     if (p >= ms->p_end - 1)
 355         luaL_error(ms->L, "malformed pattern "
 356                    "(missing arguments to " LUA_QL("%%b") ")");
 357     if (*s != *p) return NULL;
 358     else {
 359         int b = *p;
 360         int e = *(p + 1);
 361         int cont = 1;
 362         while (++s < ms->src_end) {
 363             if (*s == e) {
 364                 if (--cont == 0) return s + 1;
 365             } else if (*s == b) cont++;
 366         }
 367     }
 368     return NULL;  /* string ends out of balance */
 369 }
 370
 371
 372 static const char *max_expand(MatchState *ms, const char *s,
 373                               const char *p, const char *ep) {
 374     ptrdiff_t i = 0;  /* counts maximum expand for item */
 375     while (singlematch(ms, s + i, p, ep))
 376         i++;
 377     /* keeps trying to match with the maximum repetitions */
 378     while (i >= 0) {
 379         const char *res = match(ms, (s + i), ep + 1);
 380         if (res) return res;
 381         i--;  /* else didn't match; reduce 1 repetition to try again */
 382     }
 383     return NULL;
 384 }
 385
 386
 387 static const char *min_expand(MatchState *ms, const char *s,
 388                               const char *p, const char *ep) {
 389     for (;;) {
 390         const char *res = match(ms, s, ep + 1);
 391         if (res != NULL)
 392             return res;
 393         else if (singlematch(ms, s, p, ep))
 394             s++;  /* try with one more repetition */
 395         else return NULL;
 396     }
 397 }
 398
 399
 400 static const char *start_capture(MatchState *ms, const char *s,
 401                                  const char *p, int what) {
 402     const char *res;
 403     int level = ms->level;
 404     if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
 405     ms->capture[level].init = s;
 406     ms->capture[level].len = what;
 407     ms->level = level + 1;
 408     if ((res = match(ms, s, p)) == NULL) /* match failed? */
 409         ms->level--;  /* undo capture */
 410     return res;
 411 }
 412
 413
 414 static const char *end_capture(MatchState *ms, const char *s,
 415                                const char *p) {
 416     int l = capture_to_close(ms);
 417     const char *res;
 418     ms->capture[l].len = s - ms->capture[l].init;  /* close capture */
 419     if ((res = match(ms, s, p)) == NULL)  /* match failed? */
 420         ms->capture[l].len = CAP_UNFINISHED;  /* undo capture */
 421     return res;
 422 }
 423
 424
 425 static const char *match_capture(MatchState *ms, const char *s, int l) {
 426     size_t len;
 427     l = check_capture(ms, l);
 428     len = ms->capture[l].len;
 429     if ((size_t)(ms->src_end - s) >= len &&
 430             memcmp(ms->capture[l].init, s, len) == 0)
 431         return s + len;
 432     else return NULL;
 433 }
 434
 435
 436 static const char *match(MatchState *ms, const char *s, const char *p) {
 437     if (ms->matchdepth-- == 0)
 438         luaL_error(ms->L, "pattern too complex");
 439 init: /* using goto's to optimize tail recursion */
 440     if (p != ms->p_end) {  /* end of pattern? */
 441         switch (*p) {
 442             case '(': {  /* start capture */
 443                 if (*(p + 1) == ')')  /* position capture? */
 444                     s = start_capture(ms, s, p + 2, CAP_POSITION);
 445                 else
 446                     s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
 447                 break;
 448             }
 449             case ')': {  /* end capture */
 450                 s = end_capture(ms, s, p + 1);
 451                 break;
 452             }
 453             case '$': {
 454                 if ((p + 1) != ms->p_end)  /* is the `$' the last char in pattern? */
 455                     goto dflt;  /* no; go to default */
 456                 s = (s == ms->src_end) ? s : NULL;  /* check end of string */
 457                 break;
 458             }
 459             case L_ESC: {  /* escaped sequences not in the format class[*+?-]? */
 460                 switch (*(p + 1)) {
 461                     case 'b': {  /* balanced string? */
 462                         s = matchbalance(ms, s, p + 2);
 463                         if (s != NULL) {
 464                             p += 4;
 465                             goto init;  /* return match(ms, s, p + 4); */
 466                         }  /* else fail (s == NULL) */
 467                         break;
 468                     }
 469                     case 'f': {  /* frontier? */
 470                         const char *ep;
 471                         char previous;
 472                         p += 2;
 473                         if (*p != '[')
 474                             luaL_error(ms->L, "missing " LUA_QL("[") " after "
 475                                        LUA_QL("%%f") " in pattern");
 476                         ep = classend(ms, p);  /* points to what is next */
 477                         previous = (s == ms->src_init) ? '\0' : *(s - 1);
 478                         if (!matchbracketclass(uchar(previous), p, ep - 1) &&
 479                                 matchbracketclass(uchar(*s), p, ep - 1)) {
 480                             p = ep;
 481                             goto init;  /* return match(ms, s, ep); */
 482                         }
 483                         s = NULL;  /* match failed */
 484                         break;
 485                     }
 486                     case '0':
 487                     case '1':
 488                     case '2':
 489                     case '3':
 490                     case '4':
 491                     case '5':
 492                     case '6':
 493                     case '7':
 494                     case '8':
 495                     case '9': {  /* capture results (%0-%9)? */
 496                         s = match_capture(ms, s, uchar(*(p + 1)));
 497                         if (s != NULL) {
 498                             p += 2;
 499                             goto init;  /* return match(ms, s, p + 2) */
 500                         }
 501                         break;
 502                     }
 503                     default:
 504                         goto dflt;
 505                 }
 506                 break;
 507             }
 508             default:
 509 dflt: {  /* pattern class plus optional suffix */
 510                     const char *ep = classend(ms, p);  /* points to optional suffix */
 511                     /* does not match at least once? */
 512                     if (!singlematch(ms, s, p, ep)) {
 513                         if (*ep == '*' || *ep == '?' || *ep == '-') {  /* accept empty? */
 514                             p = ep + 1;
 515                             goto init;  /* return match(ms, s, ep + 1); */
 516                         } else /* '+' or no suffix */
 517                             s = NULL;  /* fail */
 518                     } else { /* matched once */
 519                         switch (*ep) {  /* handle optional suffix */
 520                             case '?': {  /* optional */
 521                                 const char *res;
 522                                 if ((res = match(ms, s + 1, ep + 1)) != NULL)
 523                                     s = res;
 524                                 else {
 525                                     p = ep + 1;
 526                                     goto init;  /* else return match(ms, s, ep + 1); */
 527                                 }
 528                                 break;
 529                             }
 530                             case '+':  /* 1 or more repetitions */
 531                                 s++;  /* 1 match already done */
 532                             /* go through */
 533                             case '*':  /* 0 or more repetitions */
 534                                 s = max_expand(ms, s, p, ep);
 535                                 break;
 536                             case '-':  /* 0 or more repetitions (minimum) */
 537                                 s = min_expand(ms, s, p, ep);
 538                                 break;
 539                             default:  /* no suffix */
 540                                 s++;
 541                                 p = ep;
 542                                 goto init;  /* return match(ms, s + 1, ep); */
 543                         }
 544                     }
 545                     break;
 546                 }
 547         }
 548     }
 549     ms->matchdepth++;
 550     return s;
 551 }
 552
 553
 554
 555 static const char *lmemfind(const char *s1, size_t l1,
 556                             const char *s2, size_t l2) {
 557     if (l2 == 0) return s1;  /* empty strings are everywhere */
 558     else if (l2 > l1) return NULL;  /* avoids a negative `l1' */
 559     else {
 560         const char *init;  /* to search for a `*s2' inside `s1' */
 561         l2--;  /* 1st char will be checked by `memchr' */
 562         l1 = l1 - l2; /* `s2' cannot be found after that */
 563         while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
 564             init++;   /* 1st char is already checked */
 565             if (memcmp(init, s2 + 1, l2) == 0)
 566                 return init - 1;
 567             else {  /* correct `l1' and `s1' to try again */
 568                 l1 -= init - s1;
 569                 s1 = init;
 570             }
 571         }
 572         return NULL;  /* not found */
 573     }
 574 }
 575
 576
 577 static void push_onecapture(MatchState *ms, int i, const char *s,
 578                             const char *e) {
 579     if (i >= ms->level) {
 580         if (i == 0)  /* ms->level == 0, too */
 581             lua_pushlstring(ms->L, s, e - s);  /* add whole match */
 582         else
 583             luaL_error(ms->L, "invalid capture index");
 584     } else {
 585         ptrdiff_t l = ms->capture[i].len;
 586         if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
 587         if (l == CAP_POSITION)
 588             lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
 589         else
 590             lua_pushlstring(ms->L, ms->capture[i].init, l);
 591     }
 592 }
 593
 594
 595 static int push_captures(MatchState *ms, const char *s, const char *e) {
 596     int i;
 597     int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
 598     luaL_checkstack(ms->L, nlevels, "too many captures");
 599     for (i = 0; i < nlevels; i++)
 600         push_onecapture(ms, i, s, e);
 601     return nlevels;  /* number of strings pushed */
 602 }
 603
 604
 605 /* check whether pattern has no special characters */
 606 static int nospecials(const char *p, size_t l) {
 607     size_t upto = 0;
 608     do {
 609         if (strpbrk(p + upto, SPECIALS))
 610             return 0;  /* pattern has a special character */
 611         upto += strlen(p + upto) + 1;  /* may have more after \0 */
 612     } while (upto <= l);
 613     return 1;  /* no special chars found */
 614 }
 615
 616
 617 static int str_find_aux(lua_State *L, int find) {
 618     size_t ls, lp;
 619     const char *s = luaL_checklstring(L, 1, &ls);
 620     const char *p = luaL_checklstring(L, 2, &lp);
 621     size_t init = posrelat(luaL_optinteger(L, 3, 1), ls);
 622     if (init < 1) init = 1;
 623     else if (init > ls + 1) {  /* start after string's end? */
 624         lua_pushnil(L);  /* cannot find anything */
 625         return 1;
 626     }
 627     /* explicit request or no special characters? */
 628     if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {
 629         /* do a plain search */
 630         const char *s2 = lmemfind(s + init - 1, ls - init + 1, p, lp);
 631         if (s2) {
 632             lua_pushinteger(L, s2 - s + 1);
 633             lua_pushinteger(L, s2 - s + lp);
 634             return 2;
 635         }
 636     } else {
 637         MatchState ms;
 638         const char *s1 = s + init - 1;
 639         int anchor = (*p == '^');
 640         if (anchor) {
 641             p++;
 642             lp--;  /* skip anchor character */
 643         }
 644         ms.L = L;
 645         ms.matchdepth = MAXCCALLS;
 646         ms.src_init = s;
 647         ms.src_end = s + ls;
 648         ms.p_end = p + lp;
 649         do {
 650             const char *res;
 651             ms.level = 0;
 652             lua_assert(ms.matchdepth == MAXCCALLS);
 653             if ((res = match(&ms, s1, p)) != NULL) {
 654                 if (find) {
 655                     lua_pushinteger(L, s1 - s + 1);  /* start */
 656                     lua_pushinteger(L, res - s);   /* end */
 657                     return push_captures(&ms, NULL, 0) + 2;
 658                 } else
 659                     return push_captures(&ms, s1, res);
 660             }
 661         } while (s1++ < ms.src_end && !anchor);
 662     }
 663     lua_pushnil(L);  /* not found */
 664     return 1;
 665 }
 666
 667
 668 static int str_find(lua_State *L) {
 669     return str_find_aux(L, 1);
 670 }
 671
 672
 673 static int str_match(lua_State *L) {
 674     return str_find_aux(L, 0);
 675 }
 676
 677
 678 static int gmatch_aux(lua_State *L) {
 679     MatchState ms;
 680     size_t ls, lp;
 681     const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
 682     const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp);
 683     const char *src;
 684     ms.L = L;
 685     ms.matchdepth = MAXCCALLS;
 686     ms.src_init = s;
 687     ms.src_end = s + ls;
 688     ms.p_end = p + lp;
 689     for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
 690             src <= ms.src_end;
 691             src++) {
 692         const char *e;
 693         ms.level = 0;
 694         lua_assert(ms.matchdepth == MAXCCALLS);
 695         if ((e = match(&ms, src, p)) != NULL) {
 696             lua_Integer newstart = e - s;
 697             if (e == src) newstart++;  /* empty match? go at least one position */
 698             lua_pushinteger(L, newstart);
 699             lua_replace(L, lua_upvalueindex(3));
 700             return push_captures(&ms, src, e);
 701         }
 702     }
 703     return 0;  /* not found */
 704 }
 705
 706
 707 static int gmatch(lua_State *L) {
 708     luaL_checkstring(L, 1);
 709     luaL_checkstring(L, 2);
 710     lua_settop(L, 2);
 711     lua_pushinteger(L, 0);
 712     lua_pushcclosure(L, gmatch_aux, 3);
 713     return 1;
 714 }
 715
 716
 717 static void add_s(MatchState *ms, luaL_Buffer *b, const char *s,
 718                   const char *e) {
 719     size_t l, i;
 720     const char *news = lua_tolstring(ms->L, 3, &l);
 721     for (i = 0; i < l; i++) {
 722         if (news[i] != L_ESC)
 723             luaL_addchar(b, news[i]);
 724         else {
 725             i++;  /* skip ESC */
 726             if (!isdigit(uchar(news[i]))) {
 727                 if (news[i] != L_ESC)
 728                     luaL_error(ms->L, "invalid use of " LUA_QL("%c")
 729                                " in replacement string", L_ESC);
 730                 luaL_addchar(b, news[i]);
 731             } else if (news[i] == '0')
 732                 luaL_addlstring(b, s, e - s);
 733             else {
 734                 push_onecapture(ms, news[i] - '1', s, e);
 735                 luaL_addvalue(b);  /* add capture to accumulated result */
 736             }
 737         }
 738     }
 739 }
 740
 741
 742 static void add_value(MatchState *ms, luaL_Buffer *b, const char *s,
 743                       const char *e, int tr) {
 744     lua_State *L = ms->L;
 745     switch (tr) {
 746         case LUA_TFUNCTION: {
 747             int n;
 748             lua_pushvalue(L, 3);
 749             n = push_captures(ms, s, e);
 750             lua_call(L, n, 1);
 751             break;
 752         }
 753         case LUA_TTABLE: {
 754             push_onecapture(ms, 0, s, e);
 755             lua_gettable(L, 3);
 756             break;
 757         }
 758         default: {  /* LUA_TNUMBER or LUA_TSTRING */
 759             add_s(ms, b, s, e);
 760             return;
 761         }
 762     }
 763     if (!lua_toboolean(L, -1)) {  /* nil or false? */
 764         lua_pop(L, 1);
 765         lua_pushlstring(L, s, e - s);  /* keep original text */
 766     } else if (!lua_isstring(L, -1))
 767         luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
 768     luaL_addvalue(b);  /* add result to accumulator */
 769 }
 770
 771
 772 static int str_gsub(lua_State *L) {
 773     size_t srcl, lp;
 774     const char *src = luaL_checklstring(L, 1, &srcl);
 775     const char *p = luaL_checklstring(L, 2, &lp);
 776     int tr = lua_type(L, 3);
 777     size_t max_s = luaL_optinteger(L, 4, srcl + 1);
 778     int anchor = (*p == '^');
 779     size_t n = 0;
 780     MatchState ms;
 781     luaL_Buffer b;
 782     luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
 783                   tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
 784                   "string/function/table expected");
 785     luaL_buffinit(L, &b);
 786     if (anchor) {
 787         p++;
 788         lp--;  /* skip anchor character */
 789     }
 790     ms.L = L;
 791     ms.matchdepth = MAXCCALLS;
 792     ms.src_init = src;
 793     ms.src_end = src + srcl;
 794     ms.p_end = p + lp;
 795     while (n < max_s) {
 796         const char *e;
 797         ms.level = 0;
 798         lua_assert(ms.matchdepth == MAXCCALLS);
 799         e = match(&ms, src, p);
 800         if (e) {
 801             n++;
 802             add_value(&ms, &b, src, e, tr);
 803         }
 804         if (e && e > src) /* non empty match? */
 805             src = e;  /* skip it */
 806         else if (src < ms.src_end)
 807             luaL_addchar(&b, *src++);
 808         else break;
 809         if (anchor) break;
 810     }
 811     luaL_addlstring(&b, src, ms.src_end - src);
 812     luaL_pushresult(&b);
 813     lua_pushinteger(L, n);  /* number of substitutions */
 814     return 2;
 815 }
 816
 817 /* }====================================================== */
 818
 819
 820
 821 /*
 822 ** {======================================================
 823 ** STRING FORMAT
 824 ** =======================================================
 825 */
 826
 827 /*
 828 ** LUA_INTFRMLEN is the length modifier for integer conversions in
 829 ** 'string.format'; LUA_INTFRM_T is the integer type corresponding to
 830 ** the previous length
 831 */
 832 #if !defined(LUA_INTFRMLEN) /* { */
 833 #if defined(LUA_USE_LONGLONG)
 834
 835 #define LUA_INTFRMLEN "ll"
 836 #define LUA_INTFRM_T long long
 837
 838 #else
 839
 840 #define LUA_INTFRMLEN "l"
 841 #define LUA_INTFRM_T long
 842
 843 #endif
 844 #endif                      /* } */
 845
 846
 847 /*
 848 ** LUA_FLTFRMLEN is the length modifier for float conversions in
 849 ** 'string.format'; LUA_FLTFRM_T is the float type corresponding to
 850 ** the previous length
 851 */
 852 #if !defined(LUA_FLTFRMLEN)
 853
 854 #define LUA_FLTFRMLEN ""
 855 #define LUA_FLTFRM_T double
 856
 857 #endif
 858
 859
 860 /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
 861 #define MAX_ITEM 512
 862 /* valid flags in a format specification */
 863 #define FLAGS "-+ #0"
 864 /*
 865 ** maximum size of each format specification (such as '%-099.99d')
 866 ** (+10 accounts for %99.99x plus margin of error)
 867 */
 868 #define MAX_FORMAT (sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
 869
 870
 871 static void addquoted(lua_State *L, luaL_Buffer *b, int arg) {
 872     size_t l;
 873     const char *s = luaL_checklstring(L, arg, &l);
 874     luaL_addchar(b, '"');
 875     while (l--) {
 876         if (*s == '"' || *s == '\\' || *s == '\n') {
 877             luaL_addchar(b, '\\');
 878             luaL_addchar(b, *s);
 879         } else if (*s == '\0' || iscntrl(uchar(*s))) {
 880             char buff[10];
 881             if (!isdigit(uchar(*(s + 1))))
 882                 sprintf(buff, "\\%d", (int)uchar(*s));
 883             else
 884                 sprintf(buff, "\\%03d", (int)uchar(*s));
 885             luaL_addstring(b, buff);
 886         } else
 887             luaL_addchar(b, *s);
 888         s++;
 889     }
 890     luaL_addchar(b, '"');
 891 }
 892
 893 static const char *scanformat(lua_State *L, const char *strfrmt, char *form) {
 894     const char *p = strfrmt;
 895     while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++;  /* skip flags */
 896     if ((size_t)(p - strfrmt) >= sizeof(FLAGS) / sizeof(char))
 897         luaL_error(L, "invalid format (repeated flags)");
 898     if (isdigit(uchar(*p))) p++;  /* skip width */
 899     if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 900     if (*p == '.') {
 901         p++;
 902         if (isdigit(uchar(*p))) p++;  /* skip precision */
 903         if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 904     }
 905     if (isdigit(uchar(*p)))
 906         luaL_error(L, "invalid format (width or precision too long)");
 907     *(form++) = '%';
 908     memcpy(form, strfrmt, (p - strfrmt + 1) * sizeof(char));
 909     form += p - strfrmt + 1;
 910     *form = '\0';
 911     return p;
 912 }
 913
 914
 915 /*
 916 ** add length modifier into formats
 917 */
 918 static void addlenmod(char *form, const char *lenmod) {
 919     size_t l = strlen(form);
 920     size_t lm = strlen(lenmod);
 921     char spec = form[l - 1];
 922     strcpy(form + l - 1, lenmod);
 923     form[l + lm - 1] = spec;
 924     form[l + lm] = '\0';
 925 }
 926
 927
 928 static int str_format(lua_State *L) {
 929     int top = lua_gettop(L);
 930     int arg = 1;
 931     size_t sfl;
 932     const char *strfrmt = luaL_checklstring(L, arg, &sfl);
 933     const char *strfrmt_end = strfrmt + sfl;
 934     luaL_Buffer b;
 935     luaL_buffinit(L, &b);
 936     while (strfrmt < strfrmt_end) {
 937         if (*strfrmt != L_ESC)
 938             luaL_addchar(&b, *strfrmt++);
 939         else if (*++strfrmt == L_ESC)
 940             luaL_addchar(&b, *strfrmt++);  /* %% */
 941         else { /* format item */
 942             char form[MAX_FORMAT];  /* to store the format (`%...') */
 943             char *buff = luaL_prepbuffsize(&b, MAX_ITEM);  /* to put formatted item */
 944             int nb = 0;  /* number of bytes in added item */
 945             if (++arg > top)
 946                 luaL_argerror(L, arg, "no value");
 947             strfrmt = scanformat(L, strfrmt, form);
 948             switch (*strfrmt++) {
 949                 case 'c': {
 950                     nb = sprintf(buff, form, luaL_checkint(L, arg));
 951                     break;
 952                 }
 953                 case 'd':
 954                 case 'i': {
 955                     lua_Number n = luaL_checknumber(L, arg);
 956                     LUA_INTFRM_T ni = (LUA_INTFRM_T)n;
 957                     lua_Number diff = n - (lua_Number)ni;
 958                     luaL_argcheck(L, -1 < diff && diff < 1, arg,
 959                                   "not a number in proper range");
 960                     addlenmod(form, LUA_INTFRMLEN);
 961                     nb = sprintf(buff, form, ni);
 962                     break;
 963                 }
 964                 case 'o':
 965                 case 'u':
 966                 case 'x':
 967                 case 'X': {
 968                     lua_Number n = luaL_checknumber(L, arg);
 969                     unsigned LUA_INTFRM_T ni = (unsigned LUA_INTFRM_T)n;
 970                     lua_Number diff = n - (lua_Number)ni;
 971                     luaL_argcheck(L, -1 < diff && diff < 1, arg,
 972                                   "not a non-negative number in proper range");
 973                     addlenmod(form, LUA_INTFRMLEN);
 974                     nb = sprintf(buff, form, ni);
 975                     break;
 976                 }
 977                 case 'e':
 978                 case 'E':
 979                 case 'f':
 980 #if defined(LUA_USE_AFORMAT)
 981                 case 'a':
 982                 case 'A':
 983 #endif
 984                 case 'g':
 985                 case 'G': {
 986                     addlenmod(form, LUA_FLTFRMLEN);
 987                     nb = sprintf(buff, form, (LUA_FLTFRM_T)luaL_checknumber(L, arg));
 988                     break;
 989                 }
 990                 case 'q': {
 991                     addquoted(L, &b, arg);
 992                     break;
 993                 }
 994                 case 's': {
 995                     size_t l;
 996                     const char *s = luaL_tolstring(L, arg, &l);
 997                     if (!strchr(form, '.') && l >= 100) {
 998                         /* no precision and string is too long to be formatted;
 999                            keep original string */
1000                         luaL_addvalue(&b);
1001                         break;
1002                     } else {
1003                         nb = sprintf(buff, form, s);
1004                         lua_pop(L, 1);  /* remove result from 'luaL_tolstring' */
1005                         break;
1006                     }
1007                 }
1008                 default: {  /* also treat cases `pnLlh' */
1009                     return luaL_error(L, "invalid option " LUA_QL("%%%c") " to "
1010                                       LUA_QL("format"), *(strfrmt - 1));
1011                 }
1012             }
1013             luaL_addsize(&b, nb);
1014         }
1015     }
1016     luaL_pushresult(&b);
1017     return 1;
1018 }
1019
1020 /* }====================================================== */
1021
1022
1023 static const luaL_Reg strlib[] = {
1024     {"byte", str_byte},
1025     {"char", str_char},
1026     {"dump", str_dump},
1027     {"find", str_find},
1028     {"format", str_format},
1029     {"gmatch", gmatch},
1030     {"gsub", str_gsub},
1031     {"len", str_len},
1032     {"lower", str_lower},
1033     {"match", str_match},
1034     {"rep", str_rep},
1035     {"reverse", str_reverse},
1036     {"sub", str_sub},
1037     {"upper", str_upper},
1038     {NULL, NULL}
1039 };
1040
1041
1042 static void createmetatable(lua_State *L) {
1043     lua_createtable(L, 0, 1);  /* table to be metatable for strings */
1044     lua_pushliteral(L, "");  /* dummy string */
1045     lua_pushvalue(L, -2);  /* copy table */
1046     lua_setmetatable(L, -2);  /* set table as metatable for strings */
1047     lua_pop(L, 1);  /* pop dummy string */
1048     lua_pushvalue(L, -2);  /* get string library */
1049     lua_setfield(L, -2, "__index");  /* metatable.__index = string */
1050     lua_pop(L, 1);  /* pop metatable */
1051 }
1052
1053
1054 /*
1055 ** Open string library
1056 */
1057 LUAMOD_API int luaopen_string(lua_State *L) {
1058     luaL_newlib(L, strlib);
1059     createmetatable(L);
1060     return 1;
1061 }
1062