1 /*
2 ** $Id: lstrlib.c,v 1.178.1.1 2013/04/12 18:48:47 roberto Exp $
3 ** Standard library for string operations and pattern-matching
4 ** See Copyright Notice in lua.h
5 */
6
7 #include <ctype.h>
8 #include <stddef.h>
9 #include <stdio.h>
10 #include <stdlib.h>
11 #include <string.h>
12
13 #define lstrlib_c
14 #define LUA_LIB
15
16 #include "lua.h"
17
18 #include "lauxlib.h"
19 #include "lualib.h"
20
21 /*
22 ** maximum number of captures that a pattern can do during
23 ** pattern-matching. This limit is arbitrary.
24 */
25 #if !defined(LUA_MAXCAPTURES)
26 #define LUA_MAXCAPTURES 32
27 #endif
28
29 /* macro to `unsign' a character */
30 #define uchar(c) ((unsigned char)(c))
31
str_len(lua_State * L)32 static int str_len(lua_State *L)
33 {
34 size_t l;
35 luaL_checklstring(L, 1, &l);
36 lua_pushinteger(L, (lua_Integer)l);
37 return 1;
38 }
39
40 /* translate a relative string position: negative means back from end */
posrelat(ptrdiff_t pos,size_t len)41 static size_t posrelat(ptrdiff_t pos, size_t len)
42 {
43 if (pos >= 0)
44 return (size_t)pos;
45 else if (0u - (size_t)pos > len)
46 return 0;
47 else
48 return len - ((size_t)-pos) + 1;
49 }
50
str_sub(lua_State * L)51 static int str_sub(lua_State *L)
52 {
53 size_t l;
54 const char *s = luaL_checklstring(L, 1, &l);
55 size_t start = posrelat(luaL_checkinteger(L, 2), l);
56 size_t end = posrelat(luaL_optinteger(L, 3, -1), l);
57 if (start < 1) start = 1;
58 if (end > l) end = l;
59 if (start <= end)
60 lua_pushlstring(L, s + start - 1, end - start + 1);
61 else
62 lua_pushliteral(L, "");
63 return 1;
64 }
65
str_reverse(lua_State * L)66 static int str_reverse(lua_State *L)
67 {
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
str_lower(lua_State * L)78 static int str_lower(lua_State *L)
79 {
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
str_upper(lua_State * L)91 static int str_upper(lua_State *L)
92 {
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 /* reasonable limit to avoid arithmetic overflow */
105 #define MAXSIZE ((~(size_t)0) >> 1)
106
str_rep(lua_State * L)107 static int str_rep(lua_State *L)
108 {
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)
114 lua_pushliteral(L, "");
115 else if (l + lsep < l || l + lsep >= MAXSIZE / n) /* may overflow? */
116 return luaL_error(L, "resulting string too large");
117 else
118 {
119 size_t totallen = n * l + (n - 1) * lsep;
120 luaL_Buffer b;
121 char *p = luaL_buffinitsize(L, &b, totallen);
122 while (n-- > 1)
123 { /* first n-1 copies (followed by separator) */
124 memcpy(p, s, l * sizeof(char));
125 p += l;
126 if (lsep > 0)
127 { /* avoid empty 'memcpy' (may be expensive) */
128 memcpy(p, sep, lsep * sizeof(char));
129 p += lsep;
130 }
131 }
132 memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */
133 luaL_pushresultsize(&b, totallen);
134 }
135 return 1;
136 }
137
str_byte(lua_State * L)138 static int str_byte(lua_State *L)
139 {
140 size_t l;
141 const char *s = luaL_checklstring(L, 1, &l);
142 size_t posi = posrelat(luaL_optinteger(L, 2, 1), l);
143 size_t pose = posrelat(luaL_optinteger(L, 3, posi), l);
144 int n, i;
145 if (posi < 1) posi = 1;
146 if (pose > l) pose = l;
147 if (posi > pose) return 0; /* empty interval; return no values */
148 n = (int)(pose - posi + 1);
149 if (posi + n <= pose) /* (size_t -> int) overflow? */
150 return luaL_error(L, "string slice too long");
151 luaL_checkstack(L, n, "string slice too long");
152 for (i = 0; i < n; i++)
153 lua_pushinteger(L, uchar(s[posi + i - 1]));
154 return n;
155 }
156
str_char(lua_State * L)157 static int str_char(lua_State *L)
158 {
159 int n = lua_gettop(L); /* number of arguments */
160 int i;
161 luaL_Buffer b;
162 char *p = luaL_buffinitsize(L, &b, n);
163 for (i = 1; i <= n; i++)
164 {
165 int c = luaL_checkint(L, i);
166 luaL_argcheck(L, uchar(c) == c, i, "value out of range");
167 p[i - 1] = uchar(c);
168 }
169 luaL_pushresultsize(&b, n);
170 return 1;
171 }
172
writer(lua_State * L,const void * b,size_t size,void * B)173 static int writer(lua_State *L, const void *b, size_t size, void *B)
174 {
175 (void)L;
176 luaL_addlstring((luaL_Buffer *)B, (const char *)b, size);
177 return 0;
178 }
179
str_dump(lua_State * L)180 static int str_dump(lua_State *L)
181 {
182 luaL_Buffer b;
183 luaL_checktype(L, 1, LUA_TFUNCTION);
184 lua_settop(L, 1);
185 luaL_buffinit(L, &b);
186 if (lua_dump(L, writer, &b) != 0)
187 return luaL_error(L, "unable to dump given function");
188 luaL_pushresult(&b);
189 return 1;
190 }
191
192 /*
193 ** {======================================================
194 ** PATTERN MATCHING
195 ** =======================================================
196 */
197
198 #define CAP_UNFINISHED (-1)
199 #define CAP_POSITION (-2)
200
201 typedef struct MatchState
202 {
203 int matchdepth; /* control for recursive depth (to avoid C stack overflow) */
204 const char *src_init; /* init of source string */
205 const char *src_end; /* end ('\0') of source string */
206 const char *p_end; /* end ('\0') of pattern */
207 lua_State *L;
208 int level; /* total number of captures (finished or unfinished) */
209 struct
210 {
211 const char *init;
212 ptrdiff_t len;
213 } capture[LUA_MAXCAPTURES];
214 } MatchState;
215
216 /* recursive function */
217 static const char *match(MatchState *ms, const char *s, const char *p);
218
219 /* maximum recursion depth for 'match' */
220 #if !defined(MAXCCALLS)
221 #define MAXCCALLS 200
222 #endif
223
224 #define L_ESC '%'
225 #define SPECIALS "^$*+?.([%-"
226
check_capture(MatchState * ms,int l)227 static int check_capture(MatchState *ms, int l)
228 {
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
capture_to_close(MatchState * ms)235 static int capture_to_close(MatchState *ms)
236 {
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
classend(MatchState * ms,const char * p)243 static const char *classend(MatchState *ms, const char *p)
244 {
245 switch (*p++)
246 {
247 case L_ESC:
248 {
249 if (p == ms->p_end)
250 luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");
251 return p + 1;
252 }
253 case '[':
254 {
255 if (*p == '^') p++;
256 do
257 { /* look for a `]' */
258 if (p == ms->p_end)
259 luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");
260 if (*(p++) == L_ESC && p < ms->p_end)
261 p++; /* skip escapes (e.g. `%]') */
262 } while (*p != ']');
263 return p + 1;
264 }
265 default:
266 {
267 return p;
268 }
269 }
270 }
271
match_class(int c,int cl)272 static int match_class(int c, int cl)
273 {
274 int res;
275 switch (tolower(cl))
276 {
277 case 'a':
278 res = isalpha(c);
279 break;
280 case 'c':
281 res = iscntrl(c);
282 break;
283 case 'd':
284 res = isdigit(c);
285 break;
286 case 'g':
287 res = isgraph(c);
288 break;
289 case 'l':
290 res = islower(c);
291 break;
292 case 'p':
293 res = ispunct(c);
294 break;
295 case 's':
296 res = isspace(c);
297 break;
298 case 'u':
299 res = isupper(c);
300 break;
301 case 'w':
302 res = isalnum(c);
303 break;
304 case 'x':
305 res = isxdigit(c);
306 break;
307 case 'z':
308 res = (c == 0);
309 break; /* deprecated option */
310 default:
311 return (cl == c);
312 }
313 return (islower(cl) ? res : !res);
314 }
315
matchbracketclass(int c,const char * p,const char * ec)316 static int matchbracketclass(int c, const char *p, const char *ec)
317 {
318 int sig = 1;
319 if (*(p + 1) == '^')
320 {
321 sig = 0;
322 p++; /* skip the `^' */
323 }
324 while (++p < ec)
325 {
326 if (*p == L_ESC)
327 {
328 p++;
329 if (match_class(c, uchar(*p)))
330 return sig;
331 }
332 else if ((*(p + 1) == '-') && (p + 2 < ec))
333 {
334 p += 2;
335 if (uchar(*(p - 2)) <= c && c <= uchar(*p))
336 return sig;
337 }
338 else if (uchar(*p) == c)
339 return sig;
340 }
341 return !sig;
342 }
343
singlematch(MatchState * ms,const char * s,const char * p,const char * ep)344 static int singlematch(MatchState *ms, const char *s, const char *p,
345 const char *ep)
346 {
347 if (s >= ms->src_end)
348 return 0;
349 else
350 {
351 int c = uchar(*s);
352 switch (*p)
353 {
354 case '.':
355 return 1; /* matches any char */
356 case L_ESC:
357 return match_class(c, uchar(*(p + 1)));
358 case '[':
359 return matchbracketclass(c, p, ep - 1);
360 default:
361 return (uchar(*p) == c);
362 }
363 }
364 }
365
matchbalance(MatchState * ms,const char * s,const char * p)366 static const char *matchbalance(MatchState *ms, const char *s,
367 const char *p)
368 {
369 if (p >= ms->p_end - 1)
370 luaL_error(ms->L,
371 "malformed pattern "
372 "(missing arguments to " LUA_QL("%%b") ")");
373 if (*s != *p)
374 return NULL;
375 else
376 {
377 int b = *p;
378 int e = *(p + 1);
379 int cont = 1;
380 while (++s < ms->src_end)
381 {
382 if (*s == e)
383 {
384 if (--cont == 0) return s + 1;
385 }
386 else if (*s == b)
387 cont++;
388 }
389 }
390 return NULL; /* string ends out of balance */
391 }
392
max_expand(MatchState * ms,const char * s,const char * p,const char * ep)393 static const char *max_expand(MatchState *ms, const char *s,
394 const char *p, const char *ep)
395 {
396 ptrdiff_t i = 0; /* counts maximum expand for item */
397 while (singlematch(ms, s + i, p, ep))
398 i++;
399 /* keeps trying to match with the maximum repetitions */
400 while (i >= 0)
401 {
402 const char *res = match(ms, (s + i), ep + 1);
403 if (res) return res;
404 i--; /* else didn't match; reduce 1 repetition to try again */
405 }
406 return NULL;
407 }
408
min_expand(MatchState * ms,const char * s,const char * p,const char * ep)409 static const char *min_expand(MatchState *ms, const char *s,
410 const char *p, const char *ep)
411 {
412 for (;;)
413 {
414 const char *res = match(ms, s, ep + 1);
415 if (res != NULL)
416 return res;
417 else if (singlematch(ms, s, p, ep))
418 s++; /* try with one more repetition */
419 else
420 return NULL;
421 }
422 }
423
start_capture(MatchState * ms,const char * s,const char * p,int what)424 static const char *start_capture(MatchState *ms, const char *s,
425 const char *p, int what)
426 {
427 const char *res;
428 int level = ms->level;
429 if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
430 ms->capture[level].init = s;
431 ms->capture[level].len = what;
432 ms->level = level + 1;
433 if ((res = match(ms, s, p)) == NULL) /* match failed? */
434 ms->level--; /* undo capture */
435 return res;
436 }
437
end_capture(MatchState * ms,const char * s,const char * p)438 static const char *end_capture(MatchState *ms, const char *s,
439 const char *p)
440 {
441 int l = capture_to_close(ms);
442 const char *res;
443 ms->capture[l].len = s - ms->capture[l].init; /* close capture */
444 if ((res = match(ms, s, p)) == NULL) /* match failed? */
445 ms->capture[l].len = CAP_UNFINISHED; /* undo capture */
446 return res;
447 }
448
match_capture(MatchState * ms,const char * s,int l)449 static const char *match_capture(MatchState *ms, const char *s, int l)
450 {
451 size_t len;
452 l = check_capture(ms, l);
453 len = ms->capture[l].len;
454 if ((size_t)(ms->src_end - s) >= len &&
455 memcmp(ms->capture[l].init, s, len) == 0)
456 return s + len;
457 else
458 return NULL;
459 }
460
match(MatchState * ms,const char * s,const char * p)461 static const char *match(MatchState *ms, const char *s, const char *p)
462 {
463 if (ms->matchdepth-- == 0)
464 luaL_error(ms->L, "pattern too complex");
465 init: /* using goto's to optimize tail recursion */
466 if (p != ms->p_end)
467 { /* end of pattern? */
468 switch (*p)
469 {
470 case '(':
471 { /* start capture */
472 if (*(p + 1) == ')') /* position capture? */
473 s = start_capture(ms, s, p + 2, CAP_POSITION);
474 else
475 s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
476 break;
477 }
478 case ')':
479 { /* end capture */
480 s = end_capture(ms, s, p + 1);
481 break;
482 }
483 case '$':
484 {
485 if ((p + 1) != ms->p_end) /* is the `$' the last char in pattern? */
486 goto dflt; /* no; go to default */
487 s = (s == ms->src_end) ? s : NULL; /* check end of string */
488 break;
489 }
490 case L_ESC:
491 { /* escaped sequences not in the format class[*+?-]? */
492 switch (*(p + 1))
493 {
494 case 'b':
495 { /* balanced string? */
496 s = matchbalance(ms, s, p + 2);
497 if (s != NULL)
498 {
499 p += 4;
500 goto init; /* return match(ms, s, p + 4); */
501 } /* else fail (s == NULL) */
502 break;
503 }
504 case 'f':
505 { /* frontier? */
506 const char *ep;
507 char previous;
508 p += 2;
509 if (*p != '[')
510 luaL_error(ms->L, "missing " LUA_QL("[") " after " LUA_QL("%%f") " in pattern");
511 ep = classend(ms, p); /* points to what is next */
512 previous = (s == ms->src_init) ? '\0' : *(s - 1);
513 if (!matchbracketclass(uchar(previous), p, ep - 1) &&
514 matchbracketclass(uchar(*s), p, ep - 1))
515 {
516 p = ep;
517 goto init; /* return match(ms, s, ep); */
518 }
519 s = NULL; /* match failed */
520 break;
521 }
522 case '0':
523 case '1':
524 case '2':
525 case '3':
526 case '4':
527 case '5':
528 case '6':
529 case '7':
530 case '8':
531 case '9':
532 { /* capture results (%0-%9)? */
533 s = match_capture(ms, s, uchar(*(p + 1)));
534 if (s != NULL)
535 {
536 p += 2;
537 goto init; /* return match(ms, s, p + 2) */
538 }
539 break;
540 }
541 default:
542 goto dflt;
543 }
544 break;
545 }
546 default:
547 dflt:
548 { /* pattern class plus optional suffix */
549 const char *ep = classend(ms, p); /* points to optional suffix */
550 /* does not match at least once? */
551 if (!singlematch(ms, s, p, ep))
552 {
553 if (*ep == '*' || *ep == '?' || *ep == '-')
554 { /* accept empty? */
555 p = ep + 1;
556 goto init; /* return match(ms, s, ep + 1); */
557 }
558 else /* '+' or no suffix */
559 s = NULL; /* fail */
560 }
561 else
562 { /* matched once */
563 switch (*ep)
564 { /* handle optional suffix */
565 case '?':
566 { /* optional */
567 const char *res;
568 if ((res = match(ms, s + 1, ep + 1)) != NULL)
569 s = res;
570 else
571 {
572 p = ep + 1;
573 goto init; /* else return match(ms, s, ep + 1); */
574 }
575 break;
576 }
577 case '+': /* 1 or more repetitions */
578 s++; /* 1 match already done */
579 /* go through */
580 case '*': /* 0 or more repetitions */
581 s = max_expand(ms, s, p, ep);
582 break;
583 case '-': /* 0 or more repetitions (minimum) */
584 s = min_expand(ms, s, p, ep);
585 break;
586 default: /* no suffix */
587 s++;
588 p = ep;
589 goto init; /* return match(ms, s + 1, ep); */
590 }
591 }
592 break;
593 }
594 }
595 }
596 ms->matchdepth++;
597 return s;
598 }
599
lmemfind(const char * s1,size_t l1,const char * s2,size_t l2)600 static const char *lmemfind(const char *s1, size_t l1,
601 const char *s2, size_t l2)
602 {
603 if (l2 == 0)
604 return s1; /* empty strings are everywhere */
605 else if (l2 > l1)
606 return NULL; /* avoids a negative `l1' */
607 else
608 {
609 const char *init; /* to search for a `*s2' inside `s1' */
610 l2--; /* 1st char will be checked by `memchr' */
611 l1 = l1 - l2; /* `s2' cannot be found after that */
612 while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL)
613 {
614 init++; /* 1st char is already checked */
615 if (memcmp(init, s2 + 1, l2) == 0)
616 return init - 1;
617 else
618 { /* correct `l1' and `s1' to try again */
619 l1 -= init - s1;
620 s1 = init;
621 }
622 }
623 return NULL; /* not found */
624 }
625 }
626
push_onecapture(MatchState * ms,int i,const char * s,const char * e)627 static void push_onecapture(MatchState *ms, int i, const char *s,
628 const char *e)
629 {
630 if (i >= ms->level)
631 {
632 if (i == 0) /* ms->level == 0, too */
633 lua_pushlstring(ms->L, s, e - s); /* add whole match */
634 else
635 luaL_error(ms->L, "invalid capture index");
636 }
637 else
638 {
639 ptrdiff_t l = ms->capture[i].len;
640 if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
641 if (l == CAP_POSITION)
642 lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
643 else
644 lua_pushlstring(ms->L, ms->capture[i].init, l);
645 }
646 }
647
push_captures(MatchState * ms,const char * s,const char * e)648 static int push_captures(MatchState *ms, const char *s, const char *e)
649 {
650 int i;
651 int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
652 luaL_checkstack(ms->L, nlevels, "too many captures");
653 for (i = 0; i < nlevels; i++)
654 push_onecapture(ms, i, s, e);
655 return nlevels; /* number of strings pushed */
656 }
657
658 /* check whether pattern has no special characters */
nospecials(const char * p,size_t l)659 static int nospecials(const char *p, size_t l)
660 {
661 size_t upto = 0;
662 do
663 {
664 if (strpbrk(p + upto, SPECIALS))
665 return 0; /* pattern has a special character */
666 upto += strlen(p + upto) + 1; /* may have more after \0 */
667 } while (upto <= l);
668 return 1; /* no special chars found */
669 }
670
str_find_aux(lua_State * L,int find)671 static int str_find_aux(lua_State *L, int find)
672 {
673 size_t ls, lp;
674 const char *s = luaL_checklstring(L, 1, &ls);
675 const char *p = luaL_checklstring(L, 2, &lp);
676 size_t init = posrelat(luaL_optinteger(L, 3, 1), ls);
677 if (init < 1)
678 init = 1;
679 else if (init > ls + 1)
680 { /* start after string's end? */
681 lua_pushnil(L); /* cannot find anything */
682 return 1;
683 }
684 /* explicit request or no special characters? */
685 if (find && (lua_toboolean(L, 4) || nospecials(p, lp)))
686 {
687 /* do a plain search */
688 const char *s2 = lmemfind(s + init - 1, ls - init + 1, p, lp);
689 if (s2)
690 {
691 lua_pushinteger(L, s2 - s + 1);
692 lua_pushinteger(L, s2 - s + lp);
693 return 2;
694 }
695 }
696 else
697 {
698 MatchState ms;
699 const char *s1 = s + init - 1;
700 int anchor = (*p == '^');
701 if (anchor)
702 {
703 p++;
704 lp--; /* skip anchor character */
705 }
706 ms.L = L;
707 ms.matchdepth = MAXCCALLS;
708 ms.src_init = s;
709 ms.src_end = s + ls;
710 ms.p_end = p + lp;
711 do
712 {
713 const char *res;
714 ms.level = 0;
715 lua_assert(ms.matchdepth == MAXCCALLS);
716 if ((res = match(&ms, s1, p)) != NULL)
717 {
718 if (find)
719 {
720 lua_pushinteger(L, s1 - s + 1); /* start */
721 lua_pushinteger(L, res - s); /* end */
722 return push_captures(&ms, NULL, 0) + 2;
723 }
724 else
725 return push_captures(&ms, s1, res);
726 }
727 } while (s1++ < ms.src_end && !anchor);
728 }
729 lua_pushnil(L); /* not found */
730 return 1;
731 }
732
str_find(lua_State * L)733 static int str_find(lua_State *L)
734 {
735 return str_find_aux(L, 1);
736 }
737
str_match(lua_State * L)738 static int str_match(lua_State *L)
739 {
740 return str_find_aux(L, 0);
741 }
742
gmatch_aux(lua_State * L)743 static int gmatch_aux(lua_State *L)
744 {
745 MatchState ms;
746 size_t ls, lp;
747 const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
748 const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp);
749 const char *src;
750 ms.L = L;
751 ms.matchdepth = MAXCCALLS;
752 ms.src_init = s;
753 ms.src_end = s + ls;
754 ms.p_end = p + lp;
755 for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
756 src <= ms.src_end;
757 src++)
758 {
759 const char *e;
760 ms.level = 0;
761 lua_assert(ms.matchdepth == MAXCCALLS);
762 if ((e = match(&ms, src, p)) != NULL)
763 {
764 lua_Integer newstart = e - s;
765 if (e == src) newstart++; /* empty match? go at least one position */
766 lua_pushinteger(L, newstart);
767 lua_replace(L, lua_upvalueindex(3));
768 return push_captures(&ms, src, e);
769 }
770 }
771 return 0; /* not found */
772 }
773
gmatch(lua_State * L)774 static int gmatch(lua_State *L)
775 {
776 luaL_checkstring(L, 1);
777 luaL_checkstring(L, 2);
778 lua_settop(L, 2);
779 lua_pushinteger(L, 0);
780 lua_pushcclosure(L, gmatch_aux, 3);
781 return 1;
782 }
783
add_s(MatchState * ms,luaL_Buffer * b,const char * s,const char * e)784 static void add_s(MatchState *ms, luaL_Buffer *b, const char *s,
785 const char *e)
786 {
787 size_t l, i;
788 const char *news = lua_tolstring(ms->L, 3, &l);
789 for (i = 0; i < l; i++)
790 {
791 if (news[i] != L_ESC)
792 luaL_addchar(b, news[i]);
793 else
794 {
795 i++; /* skip ESC */
796 if (!isdigit(uchar(news[i])))
797 {
798 if (news[i] != L_ESC)
799 luaL_error(ms->L, "invalid use of " LUA_QL("%c") " in replacement string", L_ESC);
800 luaL_addchar(b, news[i]);
801 }
802 else if (news[i] == '0')
803 luaL_addlstring(b, s, e - s);
804 else
805 {
806 push_onecapture(ms, news[i] - '1', s, e);
807 luaL_addvalue(b); /* add capture to accumulated result */
808 }
809 }
810 }
811 }
812
add_value(MatchState * ms,luaL_Buffer * b,const char * s,const char * e,int tr)813 static void add_value(MatchState *ms, luaL_Buffer *b, const char *s,
814 const char *e, int tr)
815 {
816 lua_State *L = ms->L;
817 switch (tr)
818 {
819 case LUA_TFUNCTION:
820 {
821 int n;
822 lua_pushvalue(L, 3);
823 n = push_captures(ms, s, e);
824 lua_call(L, n, 1);
825 break;
826 }
827 case LUA_TTABLE:
828 {
829 push_onecapture(ms, 0, s, e);
830 lua_gettable(L, 3);
831 break;
832 }
833 default:
834 { /* LUA_TNUMBER or LUA_TSTRING */
835 add_s(ms, b, s, e);
836 return;
837 }
838 }
839 if (!lua_toboolean(L, -1))
840 { /* nil or false? */
841 lua_pop(L, 1);
842 lua_pushlstring(L, s, e - s); /* keep original text */
843 }
844 else if (!lua_isstring(L, -1))
845 luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
846 luaL_addvalue(b); /* add result to accumulator */
847 }
848
str_gsub(lua_State * L)849 static int str_gsub(lua_State *L)
850 {
851 size_t srcl, lp;
852 const char *src = luaL_checklstring(L, 1, &srcl);
853 const char *p = luaL_checklstring(L, 2, &lp);
854 int tr = lua_type(L, 3);
855 size_t max_s = luaL_optinteger(L, 4, srcl + 1);
856 int anchor = (*p == '^');
857 size_t n = 0;
858 MatchState ms;
859 luaL_Buffer b;
860 luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING || tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
861 "string/function/table expected");
862 luaL_buffinit(L, &b);
863 if (anchor)
864 {
865 p++;
866 lp--; /* skip anchor character */
867 }
868 ms.L = L;
869 ms.matchdepth = MAXCCALLS;
870 ms.src_init = src;
871 ms.src_end = src + srcl;
872 ms.p_end = p + lp;
873 while (n < max_s)
874 {
875 const char *e;
876 ms.level = 0;
877 lua_assert(ms.matchdepth == MAXCCALLS);
878 e = match(&ms, src, p);
879 if (e)
880 {
881 n++;
882 add_value(&ms, &b, src, e, tr);
883 }
884 if (e && e > src) /* non empty match? */
885 src = e; /* skip it */
886 else if (src < ms.src_end)
887 luaL_addchar(&b, *src++);
888 else
889 break;
890 if (anchor) break;
891 }
892 luaL_addlstring(&b, src, ms.src_end - src);
893 luaL_pushresult(&b);
894 lua_pushinteger(L, n); /* number of substitutions */
895 return 2;
896 }
897
898 /* }====================================================== */
899
900 /*
901 ** {======================================================
902 ** STRING FORMAT
903 ** =======================================================
904 */
905
906 /*
907 ** LUA_INTFRMLEN is the length modifier for integer conversions in
908 ** 'string.format'; LUA_INTFRM_T is the integer type corresponding to
909 ** the previous length
910 */
911 #if !defined(LUA_INTFRMLEN) /* { */
912 #if defined(LUA_USE_LONGLONG)
913
914 #define LUA_INTFRMLEN "ll"
915 #define LUA_INTFRM_T long long
916
917 #else
918
919 #define LUA_INTFRMLEN "l"
920 #define LUA_INTFRM_T long
921
922 #endif
923 #endif /* } */
924
925 /*
926 ** LUA_FLTFRMLEN is the length modifier for float conversions in
927 ** 'string.format'; LUA_FLTFRM_T is the float type corresponding to
928 ** the previous length
929 */
930 #if !defined(LUA_FLTFRMLEN)
931
932 #define LUA_FLTFRMLEN ""
933 #define LUA_FLTFRM_T double
934
935 #endif
936
937 /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
938 #define MAX_ITEM 512
939 /* valid flags in a format specification */
940 #define FLAGS "-+ #0"
941 /*
942 ** maximum size of each format specification (such as '%-099.99d')
943 ** (+10 accounts for %99.99x plus margin of error)
944 */
945 #define MAX_FORMAT (sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
946
addquoted(lua_State * L,luaL_Buffer * b,int arg)947 static void addquoted(lua_State *L, luaL_Buffer *b, int arg)
948 {
949 size_t l;
950 const char *s = luaL_checklstring(L, arg, &l);
951 luaL_addchar(b, '"');
952 while (l--)
953 {
954 if (*s == '"' || *s == '\\' || *s == '\n')
955 {
956 luaL_addchar(b, '\\');
957 luaL_addchar(b, *s);
958 }
959 else if (*s == '\0' || iscntrl(uchar(*s)))
960 {
961 char buff[10];
962 if (!isdigit(uchar(*(s + 1))))
963 sprintf(buff, "\\%d", (int)uchar(*s));
964 else
965 sprintf(buff, "\\%03d", (int)uchar(*s));
966 luaL_addstring(b, buff);
967 }
968 else
969 luaL_addchar(b, *s);
970 s++;
971 }
972 luaL_addchar(b, '"');
973 }
974
scanformat(lua_State * L,const char * strfrmt,char * form)975 static const char *scanformat(lua_State *L, const char *strfrmt, char *form)
976 {
977 const char *p = strfrmt;
978 while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */
979 if ((size_t)(p - strfrmt) >= sizeof(FLAGS) / sizeof(char))
980 luaL_error(L, "invalid format (repeated flags)");
981 if (isdigit(uchar(*p))) p++; /* skip width */
982 if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
983 if (*p == '.')
984 {
985 p++;
986 if (isdigit(uchar(*p))) p++; /* skip precision */
987 if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
988 }
989 if (isdigit(uchar(*p)))
990 luaL_error(L, "invalid format (width or precision too long)");
991 *(form++) = '%';
992 memcpy(form, strfrmt, (p - strfrmt + 1) * sizeof(char));
993 form += p - strfrmt + 1;
994 *form = '\0';
995 return p;
996 }
997
998 /*
999 ** add length modifier into formats
1000 */
addlenmod(char * form,const char * lenmod)1001 static void addlenmod(char *form, const char *lenmod)
1002 {
1003 size_t l = strlen(form);
1004 size_t lm = strlen(lenmod);
1005 char spec = form[l - 1];
1006 strcpy(form + l - 1, lenmod);
1007 form[l + lm - 1] = spec;
1008 form[l + lm] = '\0';
1009 }
1010
str_format(lua_State * L)1011 static int str_format(lua_State *L)
1012 {
1013 int top = lua_gettop(L);
1014 int arg = 1;
1015 size_t sfl;
1016 const char *strfrmt = luaL_checklstring(L, arg, &sfl);
1017 const char *strfrmt_end = strfrmt + sfl;
1018 luaL_Buffer b;
1019 luaL_buffinit(L, &b);
1020 while (strfrmt < strfrmt_end)
1021 {
1022 if (*strfrmt != L_ESC)
1023 luaL_addchar(&b, *strfrmt++);
1024 else if (*++strfrmt == L_ESC)
1025 luaL_addchar(&b, *strfrmt++); /* %% */
1026 else
1027 { /* format item */
1028 char form[MAX_FORMAT]; /* to store the format (`%...') */
1029 char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */
1030 int nb = 0; /* number of bytes in added item */
1031 if (++arg > top)
1032 luaL_argerror(L, arg, "no value");
1033 strfrmt = scanformat(L, strfrmt, form);
1034 switch (*strfrmt++)
1035 {
1036 case 'c':
1037 {
1038 nb = sprintf(buff, form, luaL_checkint(L, arg));
1039 break;
1040 }
1041 case 'd':
1042 case 'i':
1043 {
1044 lua_Number n = luaL_checknumber(L, arg);
1045 LUA_INTFRM_T ni = (LUA_INTFRM_T)n;
1046 lua_Number diff = n - (lua_Number)ni;
1047 luaL_argcheck(L, -1 < diff && diff < 1, arg,
1048 "not a number in proper range");
1049 addlenmod(form, LUA_INTFRMLEN);
1050 nb = sprintf(buff, form, ni);
1051 break;
1052 }
1053 case 'o':
1054 case 'u':
1055 case 'x':
1056 case 'X':
1057 {
1058 lua_Number n = luaL_checknumber(L, arg);
1059 unsigned LUA_INTFRM_T ni = (unsigned LUA_INTFRM_T)n;
1060 lua_Number diff = n - (lua_Number)ni;
1061 luaL_argcheck(L, -1 < diff && diff < 1, arg,
1062 "not a non-negative number in proper range");
1063 addlenmod(form, LUA_INTFRMLEN);
1064 nb = sprintf(buff, form, ni);
1065 break;
1066 }
1067 case 'e':
1068 case 'E':
1069 case 'f':
1070 #if defined(LUA_USE_AFORMAT)
1071 case 'a':
1072 case 'A':
1073 #endif
1074 case 'g':
1075 case 'G':
1076 {
1077 addlenmod(form, LUA_FLTFRMLEN);
1078 nb = sprintf(buff, form, (LUA_FLTFRM_T)luaL_checknumber(L, arg));
1079 break;
1080 }
1081 case 'q':
1082 {
1083 addquoted(L, &b, arg);
1084 break;
1085 }
1086 case 's':
1087 {
1088 size_t l;
1089 const char *s = luaL_tolstring(L, arg, &l);
1090 if (!strchr(form, '.') && l >= 100)
1091 {
1092 /* no precision and string is too long to be formatted;
1093 keep original string */
1094 luaL_addvalue(&b);
1095 break;
1096 }
1097 else
1098 {
1099 nb = sprintf(buff, form, s);
1100 lua_pop(L, 1); /* remove result from 'luaL_tolstring' */
1101 break;
1102 }
1103 }
1104 default:
1105 { /* also treat cases `pnLlh' */
1106 return luaL_error(L, "invalid option " LUA_QL("%%%c") " to " LUA_QL("format"), *(strfrmt - 1));
1107 }
1108 }
1109 luaL_addsize(&b, nb);
1110 }
1111 }
1112 luaL_pushresult(&b);
1113 return 1;
1114 }
1115
1116 /* }====================================================== */
1117
1118 static const luaL_Reg strlib[] = {
1119 {"byte", str_byte},
1120 {"char", str_char},
1121 {"dump", str_dump},
1122 {"find", str_find},
1123 {"format", str_format},
1124 {"gmatch", gmatch},
1125 {"gsub", str_gsub},
1126 {"len", str_len},
1127 {"lower", str_lower},
1128 {"match", str_match},
1129 {"rep", str_rep},
1130 {"reverse", str_reverse},
1131 {"sub", str_sub},
1132 {"upper", str_upper},
1133 {NULL, NULL}};
1134
createmetatable(lua_State * L)1135 static void createmetatable(lua_State *L)
1136 {
1137 lua_createtable(L, 0, 1); /* table to be metatable for strings */
1138 lua_pushliteral(L, ""); /* dummy string */
1139 lua_pushvalue(L, -2); /* copy table */
1140 lua_setmetatable(L, -2); /* set table as metatable for strings */
1141 lua_pop(L, 1); /* pop dummy string */
1142 lua_pushvalue(L, -2); /* get string library */
1143 lua_setfield(L, -2, "__index"); /* metatable.__index = string */
1144 lua_pop(L, 1); /* pop metatable */
1145 }
1146
1147 /*
1148 ** Open string library
1149 */
luaopen_string(lua_State * L)1150 LUAMOD_API int luaopen_string(lua_State *L)
1151 {
1152 luaL_newlib(L, strlib);
1153 createmetatable(L);
1154 return 1;
1155 }
1156