external/mit/lua/dist/src/lstrlib.c

   1 /*      $NetBSD: lstrlib.c,v 1.1.1.2 2012/03/15 00:08:12 alnsn Exp $    */
   2
   3 /*
   4 ** $Id: lstrlib.c,v 1.1.1.2 2012/03/15 00:08:12 alnsn Exp $
   5 ** Standard library for string operations and pattern-matching
   6 ** See Copyright Notice in lua.h
   7 */
   8
   9
  10 #include <ctype.h>
  11 #include <stddef.h>
  12 #include <stdio.h>
  13 #include <stdlib.h>
  14 #include <string.h>
  15
  16 #define lstrlib_c
  17 #define LUA_LIB
  18
  19 #include "lua.h"
  20
  21 #include "lauxlib.h"
  22 #include "lualib.h"
  23
  24
  25 /* macro to `unsign' a character */
  26 #define uchar(c)        ((unsigned char)(c))
  27
  28
  29
  30 static int str_len (lua_State *L) {
  31   size_t l;
  32   luaL_checklstring(L, 1, &l);
  33   lua_pushinteger(L, l);
  34   return 1;
  35 }
  36
  37
  38 static ptrdiff_t posrelat (ptrdiff_t pos, size_t len) {
  39   /* relative string position: negative means back from end */
  40   if (pos < 0) pos += (ptrdiff_t)len + 1;
  41   return (pos >= 0) ? pos : 0;
  42 }
  43
  44
  45 static int str_sub (lua_State *L) {
  46   size_t l;
  47   const char *s = luaL_checklstring(L, 1, &l);
  48   ptrdiff_t start = posrelat(luaL_checkinteger(L, 2), l);
  49   ptrdiff_t end = posrelat(luaL_optinteger(L, 3, -1), l);
  50   if (start < 1) start = 1;
  51   if (end > (ptrdiff_t)l) end = (ptrdiff_t)l;
  52   if (start <= end)
  53     lua_pushlstring(L, s+start-1, end-start+1);
  54   else lua_pushliteral(L, "");
  55   return 1;
  56 }
  57
  58
  59 static int str_reverse (lua_State *L) {
  60   size_t l;
  61   luaL_Buffer b;
  62   const char *s = luaL_checklstring(L, 1, &l);
  63   luaL_buffinit(L, &b);
  64   while (l--) luaL_addchar(&b, s[l]);
  65   luaL_pushresult(&b);
  66   return 1;
  67 }
  68
  69
  70 static int str_lower (lua_State *L) {
  71   size_t l;
  72   size_t i;
  73   luaL_Buffer b;
  74   const char *s = luaL_checklstring(L, 1, &l);
  75   luaL_buffinit(L, &b);
  76   for (i=0; i<l; i++)
  77     luaL_addchar(&b, tolower(uchar(s[i])));
  78   luaL_pushresult(&b);
  79   return 1;
  80 }
  81
  82
  83 static int str_upper (lua_State *L) {
  84   size_t l;
  85   size_t i;
  86   luaL_Buffer b;
  87   const char *s = luaL_checklstring(L, 1, &l);
  88   luaL_buffinit(L, &b);
  89   for (i=0; i<l; i++)
  90     luaL_addchar(&b, toupper(uchar(s[i])));
  91   luaL_pushresult(&b);
  92   return 1;
  93 }
  94
  95 static int str_rep (lua_State *L) {
  96   size_t l;
  97   luaL_Buffer b;
  98   const char *s = luaL_checklstring(L, 1, &l);
  99   int n = luaL_checkint(L, 2);
 100   luaL_buffinit(L, &b);
 101   while (n-- > 0)
 102     luaL_addlstring(&b, s, l);
 103   luaL_pushresult(&b);
 104   return 1;
 105 }
 106
 107
 108 static int str_byte (lua_State *L) {
 109   size_t l;
 110   const char *s = luaL_checklstring(L, 1, &l);
 111   ptrdiff_t posi = posrelat(luaL_optinteger(L, 2, 1), l);
 112   ptrdiff_t pose = posrelat(luaL_optinteger(L, 3, posi), l);
 113   int n, i;
 114   if (posi <= 0) posi = 1;
 115   if ((size_t)pose > l) pose = l;
 116   if (posi > pose) return 0;  /* empty interval; return no values */
 117   n = (int)(pose -  posi + 1);
 118   if (posi + n <= pose)  /* overflow? */
 119     luaL_error(L, "string slice too long");
 120   luaL_checkstack(L, n, "string slice too long");
 121   for (i=0; i<n; i++)
 122     lua_pushinteger(L, uchar(s[posi+i-1]));
 123   return n;
 124 }
 125
 126
 127 static int str_char (lua_State *L) {
 128   int n = lua_gettop(L);  /* number of arguments */
 129   int i;
 130   luaL_Buffer b;
 131   luaL_buffinit(L, &b);
 132   for (i=1; i<=n; i++) {
 133     int c = luaL_checkint(L, i);
 134     luaL_argcheck(L, uchar(c) == c, i, "invalid value");
 135     luaL_addchar(&b, uchar(c));
 136   }
 137   luaL_pushresult(&b);
 138   return 1;
 139 }
 140
 141
 142 static int writer (lua_State *L, const void* b, size_t size, void* B) {
 143   (void)L;
 144   luaL_addlstring((luaL_Buffer*) B, (const char *)b, size);
 145   return 0;
 146 }
 147
 148
 149 static int str_dump (lua_State *L) {
 150   luaL_Buffer b;
 151   luaL_checktype(L, 1, LUA_TFUNCTION);
 152   lua_settop(L, 1);
 153   luaL_buffinit(L,&b);
 154   if (lua_dump(L, writer, &b) != 0)
 155     luaL_error(L, "unable to dump given function");
 156   luaL_pushresult(&b);
 157   return 1;
 158 }
 159
 160
 161
 162 /*
 163 ** {======================================================
 164 ** PATTERN MATCHING
 165 ** =======================================================
 166 */
 167
 168
 169 #define CAP_UNFINISHED  (-1)
 170 #define CAP_POSITION    (-2)
 171
 172 typedef struct MatchState {
 173   const char *src_init;  /* init of source string */
 174   const char *src_end;  /* end (`\0') of source string */
 175   lua_State *L;
 176   int level;  /* total number of captures (finished or unfinished) */
 177   struct {
 178     const char *init;
 179     ptrdiff_t len;
 180   } capture[LUA_MAXCAPTURES];
 181 } MatchState;
 182
 183
 184 #define L_ESC           '%'
 185 #define SPECIALS        "^$*+?.([%-"
 186
 187
 188 static int check_capture (MatchState *ms, int l) {
 189   l -= '1';
 190   if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
 191     return luaL_error(ms->L, "invalid capture index");
 192   return l;
 193 }
 194
 195
 196 static int capture_to_close (MatchState *ms) {
 197   int level = ms->level;
 198   for (level--; level>=0; level--)
 199     if (ms->capture[level].len == CAP_UNFINISHED) return level;
 200   return luaL_error(ms->L, "invalid pattern capture");
 201 }
 202
 203
 204 static const char *classend (MatchState *ms, const char *p) {
 205   switch (*p++) {
 206     case L_ESC: {
 207       if (*p == '\0')
 208         luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");
 209       return p+1;
 210     }
 211     case '[': {
 212       if (*p == '^') p++;
 213       do {  /* look for a `]' */
 214         if (*p == '\0')
 215           luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");
 216         if (*(p++) == L_ESC && *p != '\0')
 217           p++;  /* skip escapes (e.g. `%]') */
 218       } while (*p != ']');
 219       return p+1;
 220     }
 221     default: {
 222       return p;
 223     }
 224   }
 225 }
 226
 227
 228 static int match_class (int c, int cl) {
 229   int res;
 230   switch (tolower(cl)) {
 231     case 'a' : res = isalpha(c); break;
 232     case 'c' : res = iscntrl(c); break;
 233     case 'd' : res = isdigit(c); break;
 234     case 'l' : res = islower(c); break;
 235     case 'p' : res = ispunct(c); break;
 236     case 's' : res = isspace(c); break;
 237     case 'u' : res = isupper(c); break;
 238     case 'w' : res = isalnum(c); break;
 239     case 'x' : res = isxdigit(c); break;
 240     case 'z' : res = (c == 0); break;
 241     default: return (cl == c);
 242   }
 243   return (islower(cl) ? res : !res);
 244 }
 245
 246
 247 static int matchbracketclass (int c, const char *p, const char *ec) {
 248   int sig = 1;
 249   if (*(p+1) == '^') {
 250     sig = 0;
 251     p++;  /* skip the `^' */
 252   }
 253   while (++p < ec) {
 254     if (*p == L_ESC) {
 255       p++;
 256       if (match_class(c, uchar(*p)))
 257         return sig;
 258     }
 259     else if ((*(p+1) == '-') && (p+2 < ec)) {
 260       p+=2;
 261       if (uchar(*(p-2)) <= c && c <= uchar(*p))
 262         return sig;
 263     }
 264     else if (uchar(*p) == c) return sig;
 265   }
 266   return !sig;
 267 }
 268
 269
 270 static int singlematch (int c, const char *p, const char *ep) {
 271   switch (*p) {
 272     case '.': return 1;  /* matches any char */
 273     case L_ESC: return match_class(c, uchar(*(p+1)));
 274     case '[': return matchbracketclass(c, p, ep-1);
 275     default:  return (uchar(*p) == c);
 276   }
 277 }
 278
 279
 280 static const char *match (MatchState *ms, const char *s, const char *p);
 281
 282
 283 static const char *matchbalance (MatchState *ms, const char *s,
 284                                    const char *p) {
 285   if (*p == 0 || *(p+1) == 0)
 286     luaL_error(ms->L, "unbalanced pattern");
 287   if (*s != *p) return NULL;
 288   else {
 289     int b = *p;
 290     int e = *(p+1);
 291     int cont = 1;
 292     while (++s < ms->src_end) {
 293       if (*s == e) {
 294         if (--cont == 0) return s+1;
 295       }
 296       else if (*s == b) cont++;
 297     }
 298   }
 299   return NULL;  /* string ends out of balance */
 300 }
 301
 302
 303 static const char *max_expand (MatchState *ms, const char *s,
 304                                  const char *p, const char *ep) {
 305   ptrdiff_t i = 0;  /* counts maximum expand for item */
 306   while ((s+i)<ms->src_end && singlematch(uchar(*(s+i)), p, ep))
 307     i++;
 308   /* keeps trying to match with the maximum repetitions */
 309   while (i>=0) {
 310     const char *res = match(ms, (s+i), ep+1);
 311     if (res) return res;
 312     i--;  /* else didn't match; reduce 1 repetition to try again */
 313   }
 314   return NULL;
 315 }
 316
 317
 318 static const char *min_expand (MatchState *ms, const char *s,
 319                                  const char *p, const char *ep) {
 320   for (;;) {
 321     const char *res = match(ms, s, ep+1);
 322     if (res != NULL)
 323       return res;
 324     else if (s<ms->src_end && singlematch(uchar(*s), p, ep))
 325       s++;  /* try with one more repetition */
 326     else return NULL;
 327   }
 328 }
 329
 330
 331 static const char *start_capture (MatchState *ms, const char *s,
 332                                     const char *p, int what) {
 333   const char *res;
 334   int level = ms->level;
 335   if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
 336   ms->capture[level].init = s;
 337   ms->capture[level].len = what;
 338   ms->level = level+1;
 339   if ((res=match(ms, s, p)) == NULL)  /* match failed? */
 340     ms->level--;  /* undo capture */
 341   return res;
 342 }
 343
 344
 345 static const char *end_capture (MatchState *ms, const char *s,
 346                                   const char *p) {
 347   int l = capture_to_close(ms);
 348   const char *res;
 349   ms->capture[l].len = s - ms->capture[l].init;  /* close capture */
 350   if ((res = match(ms, s, p)) == NULL)  /* match failed? */
 351     ms->capture[l].len = CAP_UNFINISHED;  /* undo capture */
 352   return res;
 353 }
 354
 355
 356 static const char *match_capture (MatchState *ms, const char *s, int l) {
 357   size_t len;
 358   l = check_capture(ms, l);
 359   len = ms->capture[l].len;
 360   if ((size_t)(ms->src_end-s) >= len &&
 361       memcmp(ms->capture[l].init, s, len) == 0)
 362     return s+len;
 363   else return NULL;
 364 }
 365
 366
 367 static const char *match (MatchState *ms, const char *s, const char *p) {
 368   init: /* using goto's to optimize tail recursion */
 369   switch (*p) {
 370     case '(': {  /* start capture */
 371       if (*(p+1) == ')')  /* position capture? */
 372         return start_capture(ms, s, p+2, CAP_POSITION);
 373       else
 374         return start_capture(ms, s, p+1, CAP_UNFINISHED);
 375     }
 376     case ')': {  /* end capture */
 377       return end_capture(ms, s, p+1);
 378     }
 379     case L_ESC: {
 380       switch (*(p+1)) {
 381         case 'b': {  /* balanced string? */
 382           s = matchbalance(ms, s, p+2);
 383           if (s == NULL) return NULL;
 384           p+=4; goto init;  /* else return match(ms, s, p+4); */
 385         }
 386         case 'f': {  /* frontier? */
 387           const char *ep; char previous;
 388           p += 2;
 389           if (*p != '[')
 390             luaL_error(ms->L, "missing " LUA_QL("[") " after "
 391                                LUA_QL("%%f") " in pattern");
 392           ep = classend(ms, p);  /* points to what is next */
 393           previous = (s == ms->src_init) ? '\0' : *(s-1);
 394           if (matchbracketclass(uchar(previous), p, ep-1) ||
 395              !matchbracketclass(uchar(*s), p, ep-1)) return NULL;
 396           p=ep; goto init;  /* else return match(ms, s, ep); */
 397         }
 398         default: {
 399           if (isdigit(uchar(*(p+1)))) {  /* capture results (%0-%9)? */
 400             s = match_capture(ms, s, uchar(*(p+1)));
 401             if (s == NULL) return NULL;
 402             p+=2; goto init;  /* else return match(ms, s, p+2) */
 403           }
 404           goto dflt;  /* case default */
 405         }
 406       }
 407     }
 408     case '\0': {  /* end of pattern */
 409       return s;  /* match succeeded */
 410     }
 411     case '$': {
 412       if (*(p+1) == '\0')  /* is the `$' the last char in pattern? */
 413         return (s == ms->src_end) ? s : NULL;  /* check end of string */
 414       else goto dflt;
 415     }
 416     default: dflt: {  /* it is a pattern item */
 417       const char *ep = classend(ms, p);  /* points to what is next */
 418       int m = s<ms->src_end && singlematch(uchar(*s), p, ep);
 419       switch (*ep) {
 420         case '?': {  /* optional */
 421           const char *res;
 422           if (m && ((res=match(ms, s+1, ep+1)) != NULL))
 423             return res;
 424           p=ep+1; goto init;  /* else return match(ms, s, ep+1); */
 425         }
 426         case '*': {  /* 0 or more repetitions */
 427           return max_expand(ms, s, p, ep);
 428         }
 429         case '+': {  /* 1 or more repetitions */
 430           return (m ? max_expand(ms, s+1, p, ep) : NULL);
 431         }
 432         case '-': {  /* 0 or more repetitions (minimum) */
 433           return min_expand(ms, s, p, ep);
 434         }
 435         default: {
 436           if (!m) return NULL;
 437           s++; p=ep; goto init;  /* else return match(ms, s+1, ep); */
 438         }
 439       }
 440     }
 441   }
 442 }
 443
 444
 445
 446 static const char *lmemfind (const char *s1, size_t l1,
 447                                const char *s2, size_t l2) {
 448   if (l2 == 0) return s1;  /* empty strings are everywhere */
 449   else if (l2 > l1) return NULL;  /* avoids a negative `l1' */
 450   else {
 451     const char *init;  /* to search for a `*s2' inside `s1' */
 452     l2--;  /* 1st char will be checked by `memchr' */
 453     l1 = l1-l2;  /* `s2' cannot be found after that */
 454     while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
 455       init++;   /* 1st char is already checked */
 456       if (memcmp(init, s2+1, l2) == 0)
 457         return init-1;
 458       else {  /* correct `l1' and `s1' to try again */
 459         l1 -= init-s1;
 460         s1 = init;
 461       }
 462     }
 463     return NULL;  /* not found */
 464   }
 465 }
 466
 467
 468 static void push_onecapture (MatchState *ms, int i, const char *s,
 469                                                     const char *e) {
 470   if (i >= ms->level) {
 471     if (i == 0)  /* ms->level == 0, too */
 472       lua_pushlstring(ms->L, s, e - s);  /* add whole match */
 473     else
 474       luaL_error(ms->L, "invalid capture index");
 475   }
 476   else {
 477     ptrdiff_t l = ms->capture[i].len;
 478     if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
 479     if (l == CAP_POSITION)
 480       lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
 481     else
 482       lua_pushlstring(ms->L, ms->capture[i].init, l);
 483   }
 484 }
 485
 486
 487 static int push_captures (MatchState *ms, const char *s, const char *e) {
 488   int i;
 489   int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
 490   luaL_checkstack(ms->L, nlevels, "too many captures");
 491   for (i = 0; i < nlevels; i++)
 492     push_onecapture(ms, i, s, e);
 493   return nlevels;  /* number of strings pushed */
 494 }
 495
 496
 497 static int str_find_aux (lua_State *L, int find) {
 498   size_t l1, l2;
 499   const char *s = luaL_checklstring(L, 1, &l1);
 500   const char *p = luaL_checklstring(L, 2, &l2);
 501   ptrdiff_t init = posrelat(luaL_optinteger(L, 3, 1), l1) - 1;
 502   if (init < 0) init = 0;
 503   else if ((size_t)(init) > l1) init = (ptrdiff_t)l1;
 504   if (find && (lua_toboolean(L, 4) ||  /* explicit request? */
 505       strpbrk(p, SPECIALS) == NULL)) {  /* or no special characters? */
 506     /* do a plain search */
 507     const char *s2 = lmemfind(s+init, l1-init, p, l2);
 508     if (s2) {
 509       lua_pushinteger(L, s2-s+1);
 510       lua_pushinteger(L, s2-s+l2);
 511       return 2;
 512     }
 513   }
 514   else {
 515     MatchState ms;
 516     int anchor = (*p == '^') ? (p++, 1) : 0;
 517     const char *s1=s+init;
 518     ms.L = L;
 519     ms.src_init = s;
 520     ms.src_end = s+l1;
 521     do {
 522       const char *res;
 523       ms.level = 0;
 524       if ((res=match(&ms, s1, p)) != NULL) {
 525         if (find) {
 526           lua_pushinteger(L, s1-s+1);  /* start */
 527           lua_pushinteger(L, res-s);   /* end */
 528           return push_captures(&ms, NULL, 0) + 2;
 529         }
 530         else
 531           return push_captures(&ms, s1, res);
 532       }
 533     } while (s1++ < ms.src_end && !anchor);
 534   }
 535   lua_pushnil(L);  /* not found */
 536   return 1;
 537 }
 538
 539
 540 static int str_find (lua_State *L) {
 541   return str_find_aux(L, 1);
 542 }
 543
 544
 545 static int str_match (lua_State *L) {
 546   return str_find_aux(L, 0);
 547 }
 548
 549
 550 static int gmatch_aux (lua_State *L) {
 551   MatchState ms;
 552   size_t ls;
 553   const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
 554   const char *p = lua_tostring(L, lua_upvalueindex(2));
 555   const char *src;
 556   ms.L = L;
 557   ms.src_init = s;
 558   ms.src_end = s+ls;
 559   for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
 560        src <= ms.src_end;
 561        src++) {
 562     const char *e;
 563     ms.level = 0;
 564     if ((e = match(&ms, src, p)) != NULL) {
 565       lua_Integer newstart = e-s;
 566       if (e == src) newstart++;  /* empty match? go at least one position */
 567       lua_pushinteger(L, newstart);
 568       lua_replace(L, lua_upvalueindex(3));
 569       return push_captures(&ms, src, e);
 570     }
 571   }
 572   return 0;  /* not found */
 573 }
 574
 575
 576 static int gmatch (lua_State *L) {
 577   luaL_checkstring(L, 1);
 578   luaL_checkstring(L, 2);
 579   lua_settop(L, 2);
 580   lua_pushinteger(L, 0);
 581   lua_pushcclosure(L, gmatch_aux, 3);
 582   return 1;
 583 }
 584
 585
 586 static int gfind_nodef (lua_State *L) {
 587   return luaL_error(L, LUA_QL("string.gfind") " was renamed to "
 588                        LUA_QL("string.gmatch"));
 589 }
 590
 591
 592 static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
 593                                                    const char *e) {
 594   size_t l, i;
 595   const char *news = lua_tolstring(ms->L, 3, &l);
 596   for (i = 0; i < l; i++) {
 597     if (news[i] != L_ESC)
 598       luaL_addchar(b, news[i]);
 599     else {
 600       i++;  /* skip ESC */
 601       if (!isdigit(uchar(news[i])))
 602         luaL_addchar(b, news[i]);
 603       else if (news[i] == '0')
 604           luaL_addlstring(b, s, e - s);
 605       else {
 606         push_onecapture(ms, news[i] - '1', s, e);
 607         luaL_addvalue(b);  /* add capture to accumulated result */
 608       }
 609     }
 610   }
 611 }
 612
 613
 614 static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
 615                                                        const char *e) {
 616   lua_State *L = ms->L;
 617   switch (lua_type(L, 3)) {
 618     case LUA_TNUMBER:
 619     case LUA_TSTRING: {
 620       add_s(ms, b, s, e);
 621       return;
 622     }
 623     case LUA_TFUNCTION: {
 624       int n;
 625       lua_pushvalue(L, 3);
 626       n = push_captures(ms, s, e);
 627       lua_call(L, n, 1);
 628       break;
 629     }
 630     case LUA_TTABLE: {
 631       push_onecapture(ms, 0, s, e);
 632       lua_gettable(L, 3);
 633       break;
 634     }
 635   }
 636   if (!lua_toboolean(L, -1)) {  /* nil or false? */
 637     lua_pop(L, 1);
 638     lua_pushlstring(L, s, e - s);  /* keep original text */
 639   }
 640   else if (!lua_isstring(L, -1))
 641     luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
 642   luaL_addvalue(b);  /* add result to accumulator */
 643 }
 644
 645
 646 static int str_gsub (lua_State *L) {
 647   size_t srcl;
 648   const char *src = luaL_checklstring(L, 1, &srcl);
 649   const char *p = luaL_checkstring(L, 2);
 650   int  tr = lua_type(L, 3);
 651   int max_s = luaL_optint(L, 4, srcl+1);
 652   int anchor = (*p == '^') ? (p++, 1) : 0;
 653   int n = 0;
 654   MatchState ms;
 655   luaL_Buffer b;
 656   luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
 657                    tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
 658                       "string/function/table expected");
 659   luaL_buffinit(L, &b);
 660   ms.L = L;
 661   ms.src_init = src;
 662   ms.src_end = src+srcl;
 663   while (n < max_s) {
 664     const char *e;
 665     ms.level = 0;
 666     e = match(&ms, src, p);
 667     if (e) {
 668       n++;
 669       add_value(&ms, &b, src, e);
 670     }
 671     if (e && e>src) /* non empty match? */
 672       src = e;  /* skip it */
 673     else if (src < ms.src_end)
 674       luaL_addchar(&b, *src++);
 675     else break;
 676     if (anchor) break;
 677   }
 678   luaL_addlstring(&b, src, ms.src_end-src);
 679   luaL_pushresult(&b);
 680   lua_pushinteger(L, n);  /* number of substitutions */
 681   return 2;
 682 }
 683
 684 /* }====================================================== */
 685
 686
 687 /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
 688 #define MAX_ITEM        512
 689 /* valid flags in a format specification */
 690 #define FLAGS   "-+ #0"
 691 /*
 692 ** maximum size of each format specification (such as '%-099.99d')
 693 ** (+10 accounts for %99.99x plus margin of error)
 694 */
 695 #define MAX_FORMAT      (sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
 696
 697
 698 static void addquoted (lua_State *L, luaL_Buffer *b, int arg) {
 699   size_t l;
 700   const char *s = luaL_checklstring(L, arg, &l);
 701   luaL_addchar(b, '"');
 702   while (l--) {
 703     switch (*s) {
 704       case '"': case '\\': case '\n': {
 705         luaL_addchar(b, '\\');
 706         luaL_addchar(b, *s);
 707         break;
 708       }
 709       case '\r': {
 710         luaL_addlstring(b, "\\r", 2);
 711         break;
 712       }
 713       case '\0': {
 714         luaL_addlstring(b, "\\000", 4);
 715         break;
 716       }
 717       default: {
 718         luaL_addchar(b, *s);
 719         break;
 720       }
 721     }
 722     s++;
 723   }
 724   luaL_addchar(b, '"');
 725 }
 726
 727 static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
 728   const char *p = strfrmt;
 729   while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++;  /* skip flags */
 730   if ((size_t)(p - strfrmt) >= sizeof(FLAGS))
 731     luaL_error(L, "invalid format (repeated flags)");
 732   if (isdigit(uchar(*p))) p++;  /* skip width */
 733   if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 734   if (*p == '.') {
 735     p++;
 736     if (isdigit(uchar(*p))) p++;  /* skip precision */
 737     if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
 738   }
 739   if (isdigit(uchar(*p)))
 740     luaL_error(L, "invalid format (width or precision too long)");
 741   *(form++) = '%';
 742   strncpy(form, strfrmt, p - strfrmt + 1);
 743   form += p - strfrmt + 1;
 744   *form = '\0';
 745   return p;
 746 }
 747
 748
 749 static void addintlen (char *form) {
 750   size_t l = strlen(form);
 751   char spec = form[l - 1];
 752   strcpy(form + l - 1, LUA_INTFRMLEN);
 753   form[l + sizeof(LUA_INTFRMLEN) - 2] = spec;
 754   form[l + sizeof(LUA_INTFRMLEN) - 1] = '\0';
 755 }
 756
 757
 758 static int str_format (lua_State *L) {
 759   int top = lua_gettop(L);
 760   int arg = 1;
 761   size_t sfl;
 762   const char *strfrmt = luaL_checklstring(L, arg, &sfl);
 763   const char *strfrmt_end = strfrmt+sfl;
 764   luaL_Buffer b;
 765   luaL_buffinit(L, &b);
 766   while (strfrmt < strfrmt_end) {
 767     if (*strfrmt != L_ESC)
 768       luaL_addchar(&b, *strfrmt++);
 769     else if (*++strfrmt == L_ESC)
 770       luaL_addchar(&b, *strfrmt++);  /* %% */
 771     else { /* format item */
 772       char form[MAX_FORMAT];  /* to store the format (`%...') */
 773       char buff[MAX_ITEM];  /* to store the formatted item */
 774       if (++arg > top)
 775         luaL_argerror(L, arg, "no value");
 776       strfrmt = scanformat(L, strfrmt, form);
 777       switch (*strfrmt++) {
 778         case 'c': {
 779           sprintf(buff, form, (int)luaL_checknumber(L, arg));
 780           break;
 781         }
 782         case 'd':  case 'i': {
 783           addintlen(form);
 784           sprintf(buff, form, (LUA_INTFRM_T)luaL_checknumber(L, arg));
 785           break;
 786         }
 787         case 'o':  case 'u':  case 'x':  case 'X': {
 788           addintlen(form);
 789           sprintf(buff, form, (unsigned LUA_INTFRM_T)luaL_checknumber(L, arg));
 790           break;
 791         }
 792         case 'e':  case 'E': case 'f':
 793         case 'g': case 'G': {
 794           sprintf(buff, form, (double)luaL_checknumber(L, arg));
 795           break;
 796         }
 797         case 'q': {
 798           addquoted(L, &b, arg);
 799           continue;  /* skip the 'addsize' at the end */
 800         }
 801         case 's': {
 802           size_t l;
 803           const char *s = luaL_checklstring(L, arg, &l);
 804           if (!strchr(form, '.') && l >= 100) {
 805             /* no precision and string is too long to be formatted;
 806                keep original string */
 807             lua_pushvalue(L, arg);
 808             luaL_addvalue(&b);
 809             continue;  /* skip the `addsize' at the end */
 810           }
 811           else {
 812             sprintf(buff, form, s);
 813             break;
 814           }
 815         }
 816         default: {  /* also treat cases `pnLlh' */
 817           return luaL_error(L, "invalid option " LUA_QL("%%%c") " to "
 818                                LUA_QL("format"), *(strfrmt - 1));
 819         }
 820       }
 821       luaL_addlstring(&b, buff, strlen(buff));
 822     }
 823   }
 824   luaL_pushresult(&b);
 825   return 1;
 826 }
 827
 828
 829 static const luaL_Reg strlib[] = {
 830   {"byte", str_byte},
 831   {"char", str_char},
 832   {"dump", str_dump},
 833   {"find", str_find},
 834   {"format", str_format},
 835   {"gfind", gfind_nodef},
 836   {"gmatch", gmatch},
 837   {"gsub", str_gsub},
 838   {"len", str_len},
 839   {"lower", str_lower},
 840   {"match", str_match},
 841   {"rep", str_rep},
 842   {"reverse", str_reverse},
 843   {"sub", str_sub},
 844   {"upper", str_upper},
 845   {NULL, NULL}
 846 };
 847
 848
 849 static void createmetatable (lua_State *L) {
 850   lua_createtable(L, 0, 1);  /* create metatable for strings */
 851   lua_pushliteral(L, "");  /* dummy string */
 852   lua_pushvalue(L, -2);
 853   lua_setmetatable(L, -2);  /* set string metatable */
 854   lua_pop(L, 1);  /* pop dummy string */
 855   lua_pushvalue(L, -2);  /* string library... */
 856   lua_setfield(L, -2, "__index");  /* ...is the __index metamethod */
 857   lua_pop(L, 1);  /* pop metatable */
 858 }
 859
 860
 861 /*
 862 ** Open string library
 863 */
 864 LUALIB_API int luaopen_string (lua_State *L) {
 865   luaL_register(L, LUA_STRLIBNAME, strlib);
 866 #if defined(LUA_COMPAT_GFIND)
 867   lua_getfield(L, -1, "gmatch");
 868   lua_setfield(L, -2, "gfind");
 869 #endif
 870   createmetatable(L);
 871   return 1;
 872 }
 873