1 /* $NetBSD: lstrlib.c,v 1.9 2015/10/08 13:40:16 mbalmer Exp $ */
2
3 /*
4 ** Id: lstrlib.c,v 1.229 2015/05/20 17:39:23 roberto Exp
5 ** Standard library for string operations and pattern-matching
6 ** See Copyright Notice in lua.h
7 */
8
9 #define lstrlib_c
10 #define LUA_LIB
11
12 #include "lprefix.h"
13
14
15 #ifndef _KERNEL
16 #include <ctype.h>
17 #include <float.h>
18 #include <limits.h>
19 #include <stddef.h>
20 #include <stdio.h>
21 #include <stdlib.h>
22 #include <string.h>
23 #endif
24
25 #include "lua.h"
26
27 #include "lauxlib.h"
28 #include "lualib.h"
29
30
31 /*
32 ** maximum number of captures that a pattern can do during
33 ** pattern-matching. This limit is arbitrary.
34 */
35 #if !defined(LUA_MAXCAPTURES)
36 #define LUA_MAXCAPTURES 32
37 #endif
38
39
40 /* macro to 'unsign' a character */
41 #define uchar(c) ((unsigned char)(c))
42
43
44 /*
45 ** Some sizes are better limited to fit in 'int', but must also fit in
46 ** 'size_t'. (We assume that 'lua_Integer' cannot be smaller than 'int'.)
47 */
48 #define MAXSIZE \
49 (sizeof(size_t) < sizeof(int) ? (~(size_t)0) : (size_t)(INT_MAX))
50
51
52
53
str_len(lua_State * L)54 static int str_len (lua_State *L) {
55 size_t l;
56 luaL_checklstring(L, 1, &l);
57 lua_pushinteger(L, (lua_Integer)l);
58 return 1;
59 }
60
61
62 /* translate a relative string position: negative means back from end */
posrelat(lua_Integer pos,size_t len)63 static lua_Integer posrelat (lua_Integer pos, size_t len) {
64 if (pos >= 0) return pos;
65 else if (0u - (size_t)pos > len) return 0;
66 else return (lua_Integer)len + pos + 1;
67 }
68
69
str_sub(lua_State * L)70 static int str_sub (lua_State *L) {
71 size_t l;
72 const char *s = luaL_checklstring(L, 1, &l);
73 lua_Integer start = posrelat(luaL_checkinteger(L, 2), l);
74 lua_Integer end = posrelat(luaL_optinteger(L, 3, -1), l);
75 if (start < 1) start = 1;
76 if (end > (lua_Integer)l) end = l;
77 if (start <= end)
78 lua_pushlstring(L, s + start - 1, (size_t)(end - start) + 1);
79 else lua_pushliteral(L, "");
80 return 1;
81 }
82
83
str_reverse(lua_State * L)84 static int str_reverse (lua_State *L) {
85 size_t l, i;
86 luaL_Buffer b;
87 const char *s = luaL_checklstring(L, 1, &l);
88 char *p = luaL_buffinitsize(L, &b, l);
89 for (i = 0; i < l; i++)
90 p[i] = s[l - i - 1];
91 luaL_pushresultsize(&b, l);
92 return 1;
93 }
94
95
str_lower(lua_State * L)96 static int str_lower (lua_State *L) {
97 size_t l;
98 size_t i;
99 luaL_Buffer b;
100 const char *s = luaL_checklstring(L, 1, &l);
101 char *p = luaL_buffinitsize(L, &b, l);
102 for (i=0; i<l; i++)
103 p[i] = tolower(uchar(s[i]));
104 luaL_pushresultsize(&b, l);
105 return 1;
106 }
107
108
str_upper(lua_State * L)109 static int str_upper (lua_State *L) {
110 size_t l;
111 size_t i;
112 luaL_Buffer b;
113 const char *s = luaL_checklstring(L, 1, &l);
114 char *p = luaL_buffinitsize(L, &b, l);
115 for (i=0; i<l; i++)
116 p[i] = toupper(uchar(s[i]));
117 luaL_pushresultsize(&b, l);
118 return 1;
119 }
120
121
str_rep(lua_State * L)122 static int str_rep (lua_State *L) {
123 size_t l, lsep;
124 const char *s = luaL_checklstring(L, 1, &l);
125 lua_Integer n = luaL_checkinteger(L, 2);
126 const char *sep = luaL_optlstring(L, 3, "", &lsep);
127 if (n <= 0) lua_pushliteral(L, "");
128 else if (l + lsep < l || l + lsep > MAXSIZE / n) /* may overflow? */
129 return luaL_error(L, "resulting string too large");
130 else {
131 size_t totallen = (size_t)n * l + (size_t)(n - 1) * lsep;
132 luaL_Buffer b;
133 char *p = luaL_buffinitsize(L, &b, totallen);
134 while (n-- > 1) { /* first n-1 copies (followed by separator) */
135 memcpy(p, s, l * sizeof(char)); p += l;
136 if (lsep > 0) { /* empty 'memcpy' is not that cheap */
137 memcpy(p, sep, lsep * sizeof(char));
138 p += lsep;
139 }
140 }
141 memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */
142 luaL_pushresultsize(&b, totallen);
143 }
144 return 1;
145 }
146
147
str_byte(lua_State * L)148 static int str_byte (lua_State *L) {
149 size_t l;
150 const char *s = luaL_checklstring(L, 1, &l);
151 lua_Integer posi = posrelat(luaL_optinteger(L, 2, 1), l);
152 lua_Integer pose = posrelat(luaL_optinteger(L, 3, posi), l);
153 int n, i;
154 if (posi < 1) posi = 1;
155 if (pose > (lua_Integer)l) pose = l;
156 if (posi > pose) return 0; /* empty interval; return no values */
157 if (pose - posi >= INT_MAX) /* arithmetic overflow? */
158 return luaL_error(L, "string slice too long");
159 n = (int)(pose - posi) + 1;
160 luaL_checkstack(L, n, "string slice too long");
161 for (i=0; i<n; i++)
162 lua_pushinteger(L, uchar(s[posi+i-1]));
163 return n;
164 }
165
166
str_char(lua_State * L)167 static int str_char (lua_State *L) {
168 int n = lua_gettop(L); /* number of arguments */
169 int i;
170 luaL_Buffer b;
171 char *p = luaL_buffinitsize(L, &b, n);
172 for (i=1; i<=n; i++) {
173 lua_Integer c = luaL_checkinteger(L, i);
174 luaL_argcheck(L, uchar(c) == c, i, "value out of range");
175 p[i - 1] = uchar(c);
176 }
177 luaL_pushresultsize(&b, n);
178 return 1;
179 }
180
181
writer(lua_State * L,const void * b,size_t size,void * B)182 static int writer (lua_State *L, const void *b, size_t size, void *B) {
183 (void)L;
184 luaL_addlstring((luaL_Buffer *) B, (const char *)b, size);
185 return 0;
186 }
187
188
str_dump(lua_State * L)189 static int str_dump (lua_State *L) {
190 luaL_Buffer b;
191 int strip = lua_toboolean(L, 2);
192 luaL_checktype(L, 1, LUA_TFUNCTION);
193 lua_settop(L, 1);
194 luaL_buffinit(L,&b);
195 if (lua_dump(L, writer, &b, strip) != 0)
196 return luaL_error(L, "unable to dump given function");
197 luaL_pushresult(&b);
198 return 1;
199 }
200
201
202
203 /*
204 ** {======================================================
205 ** PATTERN MATCHING
206 ** =======================================================
207 */
208
209
210 #define CAP_UNFINISHED (-1)
211 #define CAP_POSITION (-2)
212
213
214 typedef struct MatchState {
215 int matchdepth; /* control for recursive depth (to avoid C stack overflow) */
216 const char *src_init; /* init of source string */
217 const char *src_end; /* end ('\0') of source string */
218 const char *p_end; /* end ('\0') of pattern */
219 lua_State *L;
220 int level; /* total number of captures (finished or unfinished) */
221 struct {
222 const char *init;
223 ptrdiff_t len;
224 } capture[LUA_MAXCAPTURES];
225 } MatchState;
226
227
228 /* recursive function */
229 static const char *match (MatchState *ms, const char *s, const char *p);
230
231
232 /* maximum recursion depth for 'match' */
233 #if !defined(MAXCCALLS)
234 #define MAXCCALLS 200
235 #endif
236
237
238 #define L_ESC '%'
239 #define SPECIALS "^$*+?.([%-"
240
241
check_capture(MatchState * ms,int l)242 static int check_capture (MatchState *ms, int l) {
243 l -= '1';
244 if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
245 return luaL_error(ms->L, "invalid capture index %%%d", l + 1);
246 return l;
247 }
248
249
capture_to_close(MatchState * ms)250 static int capture_to_close (MatchState *ms) {
251 int level = ms->level;
252 for (level--; level>=0; level--)
253 if (ms->capture[level].len == CAP_UNFINISHED) return level;
254 return luaL_error(ms->L, "invalid pattern capture");
255 }
256
257
classend(MatchState * ms,const char * p)258 static const char *classend (MatchState *ms, const char *p) {
259 switch (*p++) {
260 case L_ESC: {
261 if (p == ms->p_end)
262 luaL_error(ms->L, "malformed pattern (ends with '%%')");
263 return p+1;
264 }
265 case '[': {
266 if (*p == '^') p++;
267 do { /* look for a ']' */
268 if (p == ms->p_end)
269 luaL_error(ms->L, "malformed pattern (missing ']')");
270 if (*(p++) == L_ESC && p < ms->p_end)
271 p++; /* skip escapes (e.g. '%]') */
272 } while (*p != ']');
273 return p+1;
274 }
275 default: {
276 return p;
277 }
278 }
279 }
280
281
match_class(int c,int cl)282 static int match_class (int c, int cl) {
283 int res;
284 switch (tolower(cl)) {
285 case 'a' : res = isalpha(c); break;
286 case 'c' : res = iscntrl(c); break;
287 case 'd' : res = isdigit(c); break;
288 case 'g' : res = isgraph(c); break;
289 case 'l' : res = islower(c); break;
290 case 'p' : res = ispunct(c); break;
291 case 's' : res = isspace(c); break;
292 case 'u' : res = isupper(c); break;
293 case 'w' : res = isalnum(c); break;
294 case 'x' : res = isxdigit(c); break;
295 case 'z' : res = (c == 0); break; /* deprecated option */
296 default: return (cl == c);
297 }
298 return (islower(cl) ? res : !res);
299 }
300
301
matchbracketclass(int c,const char * p,const char * ec)302 static int matchbracketclass (int c, const char *p, const char *ec) {
303 int sig = 1;
304 if (*(p+1) == '^') {
305 sig = 0;
306 p++; /* skip the '^' */
307 }
308 while (++p < ec) {
309 if (*p == L_ESC) {
310 p++;
311 if (match_class(c, uchar(*p)))
312 return sig;
313 }
314 else if ((*(p+1) == '-') && (p+2 < ec)) {
315 p+=2;
316 if (uchar(*(p-2)) <= c && c <= uchar(*p))
317 return sig;
318 }
319 else if (uchar(*p) == c) return sig;
320 }
321 return !sig;
322 }
323
324
singlematch(MatchState * ms,const char * s,const char * p,const char * ep)325 static int singlematch (MatchState *ms, const char *s, const char *p,
326 const char *ep) {
327 if (s >= ms->src_end)
328 return 0;
329 else {
330 int c = uchar(*s);
331 switch (*p) {
332 case '.': return 1; /* matches any char */
333 case L_ESC: return match_class(c, uchar(*(p+1)));
334 case '[': return matchbracketclass(c, p, ep-1);
335 default: return (uchar(*p) == c);
336 }
337 }
338 }
339
340
matchbalance(MatchState * ms,const char * s,const char * p)341 static const char *matchbalance (MatchState *ms, const char *s,
342 const char *p) {
343 if (p >= ms->p_end - 1)
344 luaL_error(ms->L, "malformed pattern (missing arguments to '%%b')");
345 if (*s != *p) return NULL;
346 else {
347 int b = *p;
348 int e = *(p+1);
349 int cont = 1;
350 while (++s < ms->src_end) {
351 if (*s == e) {
352 if (--cont == 0) return s+1;
353 }
354 else if (*s == b) cont++;
355 }
356 }
357 return NULL; /* string ends out of balance */
358 }
359
360
max_expand(MatchState * ms,const char * s,const char * p,const char * ep)361 static const char *max_expand (MatchState *ms, const char *s,
362 const char *p, const char *ep) {
363 ptrdiff_t i = 0; /* counts maximum expand for item */
364 while (singlematch(ms, s + i, p, ep))
365 i++;
366 /* keeps trying to match with the maximum repetitions */
367 while (i>=0) {
368 const char *res = match(ms, (s+i), ep+1);
369 if (res) return res;
370 i--; /* else didn't match; reduce 1 repetition to try again */
371 }
372 return NULL;
373 }
374
375
min_expand(MatchState * ms,const char * s,const char * p,const char * ep)376 static const char *min_expand (MatchState *ms, const char *s,
377 const char *p, const char *ep) {
378 for (;;) {
379 const char *res = match(ms, s, ep+1);
380 if (res != NULL)
381 return res;
382 else if (singlematch(ms, s, p, ep))
383 s++; /* try with one more repetition */
384 else return NULL;
385 }
386 }
387
388
start_capture(MatchState * ms,const char * s,const char * p,int what)389 static const char *start_capture (MatchState *ms, const char *s,
390 const char *p, int what) {
391 const char *res;
392 int level = ms->level;
393 if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
394 ms->capture[level].init = s;
395 ms->capture[level].len = what;
396 ms->level = level+1;
397 if ((res=match(ms, s, p)) == NULL) /* match failed? */
398 ms->level--; /* undo capture */
399 return res;
400 }
401
402
end_capture(MatchState * ms,const char * s,const char * p)403 static const char *end_capture (MatchState *ms, const char *s,
404 const char *p) {
405 int l = capture_to_close(ms);
406 const char *res;
407 ms->capture[l].len = s - ms->capture[l].init; /* close capture */
408 if ((res = match(ms, s, p)) == NULL) /* match failed? */
409 ms->capture[l].len = CAP_UNFINISHED; /* undo capture */
410 return res;
411 }
412
413
match_capture(MatchState * ms,const char * s,int l)414 static const char *match_capture (MatchState *ms, const char *s, int l) {
415 size_t len;
416 l = check_capture(ms, l);
417 len = ms->capture[l].len;
418 if ((size_t)(ms->src_end-s) >= len &&
419 memcmp(ms->capture[l].init, s, len) == 0)
420 return s+len;
421 else return NULL;
422 }
423
424
match(MatchState * ms,const char * s,const char * p)425 static const char *match (MatchState *ms, const char *s, const char *p) {
426 if (ms->matchdepth-- == 0)
427 luaL_error(ms->L, "pattern too complex");
428 init: /* using goto's to optimize tail recursion */
429 if (p != ms->p_end) { /* end of pattern? */
430 switch (*p) {
431 case '(': { /* start capture */
432 if (*(p + 1) == ')') /* position capture? */
433 s = start_capture(ms, s, p + 2, CAP_POSITION);
434 else
435 s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
436 break;
437 }
438 case ')': { /* end capture */
439 s = end_capture(ms, s, p + 1);
440 break;
441 }
442 case '$': {
443 if ((p + 1) != ms->p_end) /* is the '$' the last char in pattern? */
444 goto dflt; /* no; go to default */
445 s = (s == ms->src_end) ? s : NULL; /* check end of string */
446 break;
447 }
448 case L_ESC: { /* escaped sequences not in the format class[*+?-]? */
449 switch (*(p + 1)) {
450 case 'b': { /* balanced string? */
451 s = matchbalance(ms, s, p + 2);
452 if (s != NULL) {
453 p += 4; goto init; /* return match(ms, s, p + 4); */
454 } /* else fail (s == NULL) */
455 break;
456 }
457 case 'f': { /* frontier? */
458 const char *ep; char previous;
459 p += 2;
460 if (*p != '[')
461 luaL_error(ms->L, "missing '[' after '%%f' in pattern");
462 ep = classend(ms, p); /* points to what is next */
463 previous = (s == ms->src_init) ? '\0' : *(s - 1);
464 if (!matchbracketclass(uchar(previous), p, ep - 1) &&
465 matchbracketclass(uchar(*s), p, ep - 1)) {
466 p = ep; goto init; /* return match(ms, s, ep); */
467 }
468 s = NULL; /* match failed */
469 break;
470 }
471 case '0': case '1': case '2': case '3':
472 case '4': case '5': case '6': case '7':
473 case '8': case '9': { /* capture results (%0-%9)? */
474 s = match_capture(ms, s, uchar(*(p + 1)));
475 if (s != NULL) {
476 p += 2; goto init; /* return match(ms, s, p + 2) */
477 }
478 break;
479 }
480 default: goto dflt;
481 }
482 break;
483 }
484 default: dflt: { /* pattern class plus optional suffix */
485 const char *ep = classend(ms, p); /* points to optional suffix */
486 /* does not match at least once? */
487 if (!singlematch(ms, s, p, ep)) {
488 if (*ep == '*' || *ep == '?' || *ep == '-') { /* accept empty? */
489 p = ep + 1; goto init; /* return match(ms, s, ep + 1); */
490 }
491 else /* '+' or no suffix */
492 s = NULL; /* fail */
493 }
494 else { /* matched once */
495 switch (*ep) { /* handle optional suffix */
496 case '?': { /* optional */
497 const char *res;
498 if ((res = match(ms, s + 1, ep + 1)) != NULL)
499 s = res;
500 else {
501 p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */
502 }
503 break;
504 }
505 case '+': /* 1 or more repetitions */
506 s++; /* 1 match already done */
507 /* FALLTHROUGH */
508 case '*': /* 0 or more repetitions */
509 s = max_expand(ms, s, p, ep);
510 break;
511 case '-': /* 0 or more repetitions (minimum) */
512 s = min_expand(ms, s, p, ep);
513 break;
514 default: /* no suffix */
515 s++; p = ep; goto init; /* return match(ms, s + 1, ep); */
516 }
517 }
518 break;
519 }
520 }
521 }
522 ms->matchdepth++;
523 return s;
524 }
525
526
527
lmemfind(const char * s1,size_t l1,const char * s2,size_t l2)528 static const char *lmemfind (const char *s1, size_t l1,
529 const char *s2, size_t l2) {
530 if (l2 == 0) return s1; /* empty strings are everywhere */
531 else if (l2 > l1) return NULL; /* avoids a negative 'l1' */
532 else {
533 const char *init; /* to search for a '*s2' inside 's1' */
534 l2--; /* 1st char will be checked by 'memchr' */
535 l1 = l1-l2; /* 's2' cannot be found after that */
536 while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
537 init++; /* 1st char is already checked */
538 if (memcmp(init, s2+1, l2) == 0)
539 return init-1;
540 else { /* correct 'l1' and 's1' to try again */
541 l1 -= init-s1;
542 s1 = init;
543 }
544 }
545 return NULL; /* not found */
546 }
547 }
548
549
push_onecapture(MatchState * ms,int i,const char * s,const char * e)550 static void push_onecapture (MatchState *ms, int i, const char *s,
551 const char *e) {
552 if (i >= ms->level) {
553 if (i == 0) /* ms->level == 0, too */
554 lua_pushlstring(ms->L, s, e - s); /* add whole match */
555 else
556 luaL_error(ms->L, "invalid capture index %%%d", i + 1);
557 }
558 else {
559 ptrdiff_t l = ms->capture[i].len;
560 if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
561 if (l == CAP_POSITION)
562 lua_pushinteger(ms->L, (ms->capture[i].init - ms->src_init) + 1);
563 else
564 lua_pushlstring(ms->L, ms->capture[i].init, l);
565 }
566 }
567
568
push_captures(MatchState * ms,const char * s,const char * e)569 static int push_captures (MatchState *ms, const char *s, const char *e) {
570 int i;
571 int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
572 luaL_checkstack(ms->L, nlevels, "too many captures");
573 for (i = 0; i < nlevels; i++)
574 push_onecapture(ms, i, s, e);
575 return nlevels; /* number of strings pushed */
576 }
577
578
579 /* check whether pattern has no special characters */
nospecials(const char * p,size_t l)580 static int nospecials (const char *p, size_t l) {
581 size_t upto = 0;
582 do {
583 if (strpbrk(p + upto, SPECIALS))
584 return 0; /* pattern has a special character */
585 upto += strlen(p + upto) + 1; /* may have more after \0 */
586 } while (upto <= l);
587 return 1; /* no special chars found */
588 }
589
590
str_find_aux(lua_State * L,int find)591 static int str_find_aux (lua_State *L, int find) {
592 size_t ls, lp;
593 const char *s = luaL_checklstring(L, 1, &ls);
594 const char *p = luaL_checklstring(L, 2, &lp);
595 lua_Integer init = posrelat(luaL_optinteger(L, 3, 1), ls);
596 if (init < 1) init = 1;
597 else if (init > (lua_Integer)ls + 1) { /* start after string's end? */
598 lua_pushnil(L); /* cannot find anything */
599 return 1;
600 }
601 /* explicit request or no special characters? */
602 if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {
603 /* do a plain search */
604 const char *s2 = lmemfind(s + init - 1, ls - (size_t)init + 1, p, lp);
605 if (s2) {
606 lua_pushinteger(L, (s2 - s) + 1);
607 lua_pushinteger(L, (s2 - s) + lp);
608 return 2;
609 }
610 }
611 else {
612 MatchState ms;
613 const char *s1 = s + init - 1;
614 int anchor = (*p == '^');
615 if (anchor) {
616 p++; lp--; /* skip anchor character */
617 }
618 ms.L = L;
619 ms.matchdepth = MAXCCALLS;
620 ms.src_init = s;
621 ms.src_end = s + ls;
622 ms.p_end = p + lp;
623 do {
624 const char *res;
625 ms.level = 0;
626 lua_assert(ms.matchdepth == MAXCCALLS);
627 if ((res=match(&ms, s1, p)) != NULL) {
628 if (find) {
629 lua_pushinteger(L, (s1 - s) + 1); /* start */
630 lua_pushinteger(L, res - s); /* end */
631 return push_captures(&ms, NULL, 0) + 2;
632 }
633 else
634 return push_captures(&ms, s1, res);
635 }
636 } while (s1++ < ms.src_end && !anchor);
637 }
638 lua_pushnil(L); /* not found */
639 return 1;
640 }
641
642
str_find(lua_State * L)643 static int str_find (lua_State *L) {
644 return str_find_aux(L, 1);
645 }
646
647
str_match(lua_State * L)648 static int str_match (lua_State *L) {
649 return str_find_aux(L, 0);
650 }
651
652
gmatch_aux(lua_State * L)653 static int gmatch_aux (lua_State *L) {
654 MatchState ms;
655 size_t ls, lp;
656 const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
657 const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp);
658 const char *src;
659 ms.L = L;
660 ms.matchdepth = MAXCCALLS;
661 ms.src_init = s;
662 ms.src_end = s+ls;
663 ms.p_end = p + lp;
664 for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
665 src <= ms.src_end;
666 src++) {
667 const char *e;
668 ms.level = 0;
669 lua_assert(ms.matchdepth == MAXCCALLS);
670 if ((e = match(&ms, src, p)) != NULL) {
671 lua_Integer newstart = e-s;
672 if (e == src) newstart++; /* empty match? go at least one position */
673 lua_pushinteger(L, newstart);
674 lua_replace(L, lua_upvalueindex(3));
675 return push_captures(&ms, src, e);
676 }
677 }
678 return 0; /* not found */
679 }
680
681
gmatch(lua_State * L)682 static int gmatch (lua_State *L) {
683 luaL_checkstring(L, 1);
684 luaL_checkstring(L, 2);
685 lua_settop(L, 2);
686 lua_pushinteger(L, 0);
687 lua_pushcclosure(L, gmatch_aux, 3);
688 return 1;
689 }
690
691
add_s(MatchState * ms,luaL_Buffer * b,const char * s,const char * e)692 static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
693 const char *e) {
694 size_t l, i;
695 lua_State *L = ms->L;
696 const char *news = lua_tolstring(L, 3, &l);
697 for (i = 0; i < l; i++) {
698 if (news[i] != L_ESC)
699 luaL_addchar(b, news[i]);
700 else {
701 i++; /* skip ESC */
702 if (!isdigit(uchar(news[i]))) {
703 if (news[i] != L_ESC)
704 luaL_error(L, "invalid use of '%c' in replacement string", L_ESC);
705 luaL_addchar(b, news[i]);
706 }
707 else if (news[i] == '0')
708 luaL_addlstring(b, s, e - s);
709 else {
710 push_onecapture(ms, news[i] - '1', s, e);
711 luaL_tolstring(L, -1, NULL); /* if number, convert it to string */
712 lua_remove(L, -2); /* remove original value */
713 luaL_addvalue(b); /* add capture to accumulated result */
714 }
715 }
716 }
717 }
718
719
add_value(MatchState * ms,luaL_Buffer * b,const char * s,const char * e,int tr)720 static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
721 const char *e, int tr) {
722 lua_State *L = ms->L;
723 switch (tr) {
724 case LUA_TFUNCTION: {
725 int n;
726 lua_pushvalue(L, 3);
727 n = push_captures(ms, s, e);
728 lua_call(L, n, 1);
729 break;
730 }
731 case LUA_TTABLE: {
732 push_onecapture(ms, 0, s, e);
733 lua_gettable(L, 3);
734 break;
735 }
736 default: { /* LUA_TNUMBER or LUA_TSTRING */
737 add_s(ms, b, s, e);
738 return;
739 }
740 }
741 if (!lua_toboolean(L, -1)) { /* nil or false? */
742 lua_pop(L, 1);
743 lua_pushlstring(L, s, e - s); /* keep original text */
744 }
745 else if (!lua_isstring(L, -1))
746 luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
747 luaL_addvalue(b); /* add result to accumulator */
748 }
749
750
str_gsub(lua_State * L)751 static int str_gsub (lua_State *L) {
752 size_t srcl, lp;
753 const char *src = luaL_checklstring(L, 1, &srcl);
754 const char *p = luaL_checklstring(L, 2, &lp);
755 int tr = lua_type(L, 3);
756 lua_Integer max_s = luaL_optinteger(L, 4, srcl + 1);
757 int anchor = (*p == '^');
758 lua_Integer n = 0;
759 MatchState ms;
760 luaL_Buffer b;
761 luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
762 tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
763 "string/function/table expected");
764 luaL_buffinit(L, &b);
765 if (anchor) {
766 p++; lp--; /* skip anchor character */
767 }
768 ms.L = L;
769 ms.matchdepth = MAXCCALLS;
770 ms.src_init = src;
771 ms.src_end = src+srcl;
772 ms.p_end = p + lp;
773 while (n < max_s) {
774 const char *e;
775 ms.level = 0;
776 lua_assert(ms.matchdepth == MAXCCALLS);
777 e = match(&ms, src, p);
778 if (e) {
779 n++;
780 add_value(&ms, &b, src, e, tr);
781 }
782 if (e && e>src) /* non empty match? */
783 src = e; /* skip it */
784 else if (src < ms.src_end)
785 luaL_addchar(&b, *src++);
786 else break;
787 if (anchor) break;
788 }
789 luaL_addlstring(&b, src, ms.src_end-src);
790 luaL_pushresult(&b);
791 lua_pushinteger(L, n); /* number of substitutions */
792 return 2;
793 }
794
795 /* }====================================================== */
796
797
798
799 /*
800 ** {======================================================
801 ** STRING FORMAT
802 ** =======================================================
803 */
804
805 #if !defined(lua_number2strx) /* { */
806
807 /*
808 ** Hexadecimal floating-point formatter
809 */
810
811 #include <locale.h>
812 #include <math.h>
813
814 #define SIZELENMOD (sizeof(LUA_NUMBER_FRMLEN)/sizeof(char))
815
816
817 /*
818 ** Number of bits that goes into the first digit. It can be any value
819 ** between 1 and 4; the following definition tries to align the number
820 ** to nibble boundaries by making what is left after that first digit a
821 ** multiple of 4.
822 */
823 #define L_NBFD ((l_mathlim(MANT_DIG) - 1)%4 + 1)
824
825
826 /*
827 ** Add integer part of 'x' to buffer and return new 'x'
828 */
adddigit(char * buff,int n,lua_Number x)829 static lua_Number adddigit (char *buff, int n, lua_Number x) {
830 lua_Number dd = l_mathop(floor)(x); /* get integer part from 'x' */
831 int d = (int)dd;
832 buff[n] = (d < 10 ? d + '0' : d - 10 + 'a'); /* add to buffer */
833 return x - dd; /* return what is left */
834 }
835
836
num2straux(char * buff,lua_Number x)837 static int num2straux (char *buff, lua_Number x) {
838 if (x != x || x == HUGE_VAL || x == -HUGE_VAL) /* inf or NaN? */
839 return sprintf(buff, LUA_NUMBER_FMT, x); /* equal to '%g' */
840 else if (x == 0) { /* can be -0... */
841 sprintf(buff, LUA_NUMBER_FMT, x);
842 strcat(buff, "x0p+0"); /* reuses '0/-0' from 'sprintf'... */
843 return strlen(buff);
844 }
845 else {
846 int e;
847 lua_Number m = l_mathop(frexp)(x, &e); /* 'x' fraction and exponent */
848 int n = 0; /* character count */
849 if (m < 0) { /* is number negative? */
850 buff[n++] = '-'; /* add signal */
851 m = -m; /* make it positive */
852 }
853 buff[n++] = '0'; buff[n++] = 'x'; /* add "0x" */
854 m = adddigit(buff, n++, m * (1 << L_NBFD)); /* add first digit */
855 e -= L_NBFD; /* this digit goes before the radix point */
856 if (m > 0) { /* more digits? */
857 buff[n++] = lua_getlocaledecpoint(); /* add radix point */
858 do { /* add as many digits as needed */
859 m = adddigit(buff, n++, m * 16);
860 } while (m > 0);
861 }
862 n += sprintf(buff + n, "p%+d", e); /* add exponent */
863 return n;
864 }
865 }
866
867
lua_number2strx(lua_State * L,char * buff,const char * fmt,lua_Number x)868 static int lua_number2strx (lua_State *L, char *buff, const char *fmt,
869 lua_Number x) {
870 int n = num2straux(buff, x);
871 if (fmt[SIZELENMOD] == 'A') {
872 int i;
873 for (i = 0; i < n; i++)
874 buff[i] = toupper(uchar(buff[i]));
875 }
876 else if (fmt[SIZELENMOD] != 'a')
877 luaL_error(L, "modifiers for format '%%a'/'%%A' not implemented");
878 return n;
879 }
880
881 #endif /* } */
882
883
884 /*
885 ** Maximum size of each formatted item. This maximum size is produced
886 ** by format('%.99f', minfloat), and is equal to 99 + 2 ('-' and '.') +
887 ** number of decimal digits to represent minfloat.
888 */
889 #ifndef _KERNEL
890 #define MAX_ITEM (120 + l_mathlim(MAX_10_EXP))
891 #else
892 #define MAX_ITEM (120)
893 #endif
894
895 /* valid flags in a format specification */
896 #define FLAGS "-+ #0"
897
898 /*
899 ** maximum size of each format specification (such as "%-099.99d")
900 */
901 #define MAX_FORMAT 32
902
903
addquoted(lua_State * L,luaL_Buffer * b,int arg)904 static void addquoted (lua_State *L, luaL_Buffer *b, int arg) {
905 size_t l;
906 const char *s = luaL_checklstring(L, arg, &l);
907 luaL_addchar(b, '"');
908 while (l--) {
909 if (*s == '"' || *s == '\\' || *s == '\n') {
910 luaL_addchar(b, '\\');
911 luaL_addchar(b, *s);
912 }
913 else if (*s == '\0' || iscntrl(uchar(*s))) {
914 char buff[10];
915 if (!isdigit(uchar(*(s+1))))
916 sprintf(buff, "\\%d", (int)uchar(*s));
917 else
918 sprintf(buff, "\\%03d", (int)uchar(*s));
919 luaL_addstring(b, buff);
920 }
921 else
922 luaL_addchar(b, *s);
923 s++;
924 }
925 luaL_addchar(b, '"');
926 }
927
scanformat(lua_State * L,const char * strfrmt,char * form)928 static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
929 const char *p = strfrmt;
930 while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */
931 if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char))
932 luaL_error(L, "invalid format (repeated flags)");
933 if (isdigit(uchar(*p))) p++; /* skip width */
934 if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
935 if (*p == '.') {
936 p++;
937 if (isdigit(uchar(*p))) p++; /* skip precision */
938 if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
939 }
940 if (isdigit(uchar(*p)))
941 luaL_error(L, "invalid format (width or precision too long)");
942 *(form++) = '%';
943 memcpy(form, strfrmt, ((p - strfrmt) + 1) * sizeof(char));
944 form += (p - strfrmt) + 1;
945 *form = '\0';
946 return p;
947 }
948
949
950 /*
951 ** add length modifier into formats
952 */
addlenmod(char * form,const char * lenmod)953 static void addlenmod (char *form, const char *lenmod) {
954 size_t l = strlen(form);
955 size_t lm = strlen(lenmod);
956 char spec = form[l - 1];
957 strcpy(form + l - 1, lenmod);
958 form[l + lm - 1] = spec;
959 form[l + lm] = '\0';
960 }
961
962
str_format(lua_State * L)963 static int str_format (lua_State *L) {
964 int top = lua_gettop(L);
965 int arg = 1;
966 size_t sfl;
967 const char *strfrmt = luaL_checklstring(L, arg, &sfl);
968 const char *strfrmt_end = strfrmt+sfl;
969 luaL_Buffer b;
970 luaL_buffinit(L, &b);
971 while (strfrmt < strfrmt_end) {
972 if (*strfrmt != L_ESC)
973 luaL_addchar(&b, *strfrmt++);
974 else if (*++strfrmt == L_ESC)
975 luaL_addchar(&b, *strfrmt++); /* %% */
976 else { /* format item */
977 char form[MAX_FORMAT]; /* to store the format ('%...') */
978 char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */
979 int nb = 0; /* number of bytes in added item */
980 if (++arg > top)
981 luaL_argerror(L, arg, "no value");
982 strfrmt = scanformat(L, strfrmt, form);
983 switch (*strfrmt++) {
984 case 'c': {
985 nb = sprintf(buff, form, (int)luaL_checkinteger(L, arg));
986 break;
987 }
988 case 'd': case 'i':
989 case 'o': case 'u': case 'x': case 'X': {
990 lua_Integer n = luaL_checkinteger(L, arg);
991 addlenmod(form, LUA_INTEGER_FRMLEN);
992 nb = sprintf(buff, form, n);
993 break;
994 }
995 #ifndef _KERNEL
996 case 'a': case 'A':
997 addlenmod(form, LUA_NUMBER_FRMLEN);
998 nb = lua_number2strx(L, buff, form, luaL_checknumber(L, arg));
999 break;
1000 case 'e': case 'E': case 'f':
1001 case 'g': case 'G': {
1002 addlenmod(form, LUA_NUMBER_FRMLEN);
1003 nb = sprintf(buff, form, luaL_checknumber(L, arg));
1004 break;
1005 }
1006 #endif
1007 case 'q': {
1008 addquoted(L, &b, arg);
1009 break;
1010 }
1011 case 's': {
1012 size_t l;
1013 const char *s = luaL_tolstring(L, arg, &l);
1014 if (!strchr(form, '.') && l >= 100) {
1015 /* no precision and string is too long to be formatted;
1016 keep original string */
1017 luaL_addvalue(&b);
1018 }
1019 else {
1020 nb = sprintf(buff, form, s);
1021 lua_pop(L, 1); /* remove result from 'luaL_tolstring' */
1022 }
1023 break;
1024 }
1025 default: { /* also treat cases 'pnLlh' */
1026 return luaL_error(L, "invalid option '%%%c' to 'format'",
1027 *(strfrmt - 1));
1028 }
1029 }
1030 luaL_addsize(&b, nb);
1031 }
1032 }
1033 luaL_pushresult(&b);
1034 return 1;
1035 }
1036
1037 /* }====================================================== */
1038
1039
1040 /*
1041 ** {======================================================
1042 ** PACK/UNPACK
1043 ** =======================================================
1044 */
1045
1046
1047 /* value used for padding */
1048 #if !defined(LUA_PACKPADBYTE)
1049 #define LUA_PACKPADBYTE 0x00
1050 #endif
1051
1052 /* maximum size for the binary representation of an integer */
1053 #define MAXINTSIZE 16
1054
1055 /* number of bits in a character */
1056 #define NB CHAR_BIT
1057
1058 /* mask for one character (NB 1's) */
1059 #define MC ((1 << NB) - 1)
1060
1061 /* size of a lua_Integer */
1062 #define SZINT ((int)sizeof(lua_Integer))
1063
1064
1065 /* dummy union to get native endianness */
1066 static const union {
1067 int dummy;
1068 char little; /* true iff machine is little endian */
1069 } nativeendian = {1};
1070
1071
1072 /* dummy structure to get native alignment requirements */
1073 struct cD {
1074 char c;
1075 #ifndef _KERNEL
1076 union { double d; void *p; lua_Integer i; lua_Number n; } u;
1077 #else /* _KERNEL */
1078 union { void *p; lua_Integer i; lua_Number n; } u;
1079 #endif
1080 };
1081
1082 #define MAXALIGN (offsetof(struct cD, u))
1083
1084
1085 #ifndef _KERNEL
1086 /*
1087 ** Union for serializing floats
1088 */
1089 typedef union Ftypes {
1090 float f;
1091 double d;
1092 lua_Number n;
1093 char buff[5 * sizeof(lua_Number)]; /* enough for any float type */
1094 } Ftypes;
1095 #endif
1096
1097
1098 /*
1099 ** information to pack/unpack stuff
1100 */
1101 typedef struct Header {
1102 lua_State *L;
1103 int islittle;
1104 int maxalign;
1105 } Header;
1106
1107
1108 /*
1109 ** options for pack/unpack
1110 */
1111 typedef enum KOption {
1112 Kint, /* signed integers */
1113 Kuint, /* unsigned integers */
1114 #ifndef _KERNEL
1115 Kfloat, /* floating-point numbers */
1116 #endif
1117 Kchar, /* fixed-length strings */
1118 Kstring, /* strings with prefixed length */
1119 Kzstr, /* zero-terminated strings */
1120 Kpadding, /* padding */
1121 Kpaddalign, /* padding for alignment */
1122 Knop /* no-op (configuration or spaces) */
1123 } KOption;
1124
1125
1126 /*
1127 ** Read an integer numeral from string 'fmt' or return 'df' if
1128 ** there is no numeral
1129 */
digit(int c)1130 static int digit (int c) { return '0' <= c && c <= '9'; }
1131
getnum(const char ** fmt,int df)1132 static int getnum (const char **fmt, int df) {
1133 if (!digit(**fmt)) /* no number? */
1134 return df; /* return default value */
1135 else {
1136 int a = 0;
1137 do {
1138 a = a*10 + (*((*fmt)++) - '0');
1139 } while (digit(**fmt) && a <= ((int)MAXSIZE - 9)/10);
1140 return a;
1141 }
1142 }
1143
1144
1145 /*
1146 ** Read an integer numeral and raises an error if it is larger
1147 ** than the maximum size for integers.
1148 */
getnumlimit(Header * h,const char ** fmt,int df)1149 static int getnumlimit (Header *h, const char **fmt, int df) {
1150 int sz = getnum(fmt, df);
1151 if (sz > MAXINTSIZE || sz <= 0)
1152 luaL_error(h->L, "integral size (%d) out of limits [1,%d]",
1153 sz, MAXINTSIZE);
1154 return sz;
1155 }
1156
1157
1158 /*
1159 ** Initialize Header
1160 */
initheader(lua_State * L,Header * h)1161 static void initheader (lua_State *L, Header *h) {
1162 h->L = L;
1163 h->islittle = nativeendian.little;
1164 h->maxalign = 1;
1165 }
1166
1167
1168 /*
1169 ** Read and classify next option. 'size' is filled with option's size.
1170 */
getoption(Header * h,const char ** fmt,int * size)1171 static KOption getoption (Header *h, const char **fmt, int *size) {
1172 int opt = *((*fmt)++);
1173 *size = 0; /* default */
1174 switch (opt) {
1175 case 'b': *size = sizeof(char); return Kint;
1176 case 'B': *size = sizeof(char); return Kuint;
1177 case 'h': *size = sizeof(short); return Kint;
1178 case 'H': *size = sizeof(short); return Kuint;
1179 case 'l': *size = sizeof(long); return Kint;
1180 case 'L': *size = sizeof(long); return Kuint;
1181 case 'j': *size = sizeof(lua_Integer); return Kint;
1182 case 'J': *size = sizeof(lua_Integer); return Kuint;
1183 case 'T': *size = sizeof(size_t); return Kuint;
1184 #ifndef _KERNEL
1185 case 'f': *size = sizeof(float); return Kfloat;
1186 case 'd': *size = sizeof(double); return Kfloat;
1187 case 'n': *size = sizeof(lua_Number); return Kfloat;
1188 #else /* _KERNEL */
1189 case 'n': *size = sizeof(lua_Number); return Kint;
1190 #endif
1191 case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint;
1192 case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint;
1193 case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring;
1194 case 'c':
1195 *size = getnum(fmt, -1);
1196 if (*size == -1)
1197 luaL_error(h->L, "missing size for format option 'c'");
1198 return Kchar;
1199 case 'z': return Kzstr;
1200 case 'x': *size = 1; return Kpadding;
1201 case 'X': return Kpaddalign;
1202 case ' ': break;
1203 case '<': h->islittle = 1; break;
1204 case '>': h->islittle = 0; break;
1205 case '=': h->islittle = nativeendian.little; break;
1206 case '!': h->maxalign = getnumlimit(h, fmt, MAXALIGN); break;
1207 default: luaL_error(h->L, "invalid format option '%c'", opt);
1208 }
1209 return Knop;
1210 }
1211
1212
1213 /*
1214 ** Read, classify, and fill other details about the next option.
1215 ** 'psize' is filled with option's size, 'notoalign' with its
1216 ** alignment requirements.
1217 ** Local variable 'size' gets the size to be aligned. (Kpadal option
1218 ** always gets its full alignment, other options are limited by
1219 ** the maximum alignment ('maxalign'). Kchar option needs no alignment
1220 ** despite its size.
1221 */
getdetails(Header * h,size_t totalsize,const char ** fmt,int * psize,int * ntoalign)1222 static KOption getdetails (Header *h, size_t totalsize,
1223 const char **fmt, int *psize, int *ntoalign) {
1224 KOption opt = getoption(h, fmt, psize);
1225 int align = *psize; /* usually, alignment follows size */
1226 if (opt == Kpaddalign) { /* 'X' gets alignment from following option */
1227 if (**fmt == '\0' || getoption(h, fmt, &align) == Kchar || align == 0)
1228 luaL_argerror(h->L, 1, "invalid next option for option 'X'");
1229 }
1230 if (align <= 1 || opt == Kchar) /* need no alignment? */
1231 *ntoalign = 0;
1232 else {
1233 if (align > h->maxalign) /* enforce maximum alignment */
1234 align = h->maxalign;
1235 if ((align & (align - 1)) != 0) /* is 'align' not a power of 2? */
1236 luaL_argerror(h->L, 1, "format asks for alignment not power of 2");
1237 *ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1);
1238 }
1239 return opt;
1240 }
1241
1242
1243 /*
1244 ** Pack integer 'n' with 'size' bytes and 'islittle' endianness.
1245 ** The final 'if' handles the case when 'size' is larger than
1246 ** the size of a Lua integer, correcting the extra sign-extension
1247 ** bytes if necessary (by default they would be zeros).
1248 */
packint(luaL_Buffer * b,lua_Unsigned n,int islittle,int size,int neg)1249 static void packint (luaL_Buffer *b, lua_Unsigned n,
1250 int islittle, int size, int neg) {
1251 char *buff = luaL_prepbuffsize(b, size);
1252 int i;
1253 buff[islittle ? 0 : size - 1] = (char)(n & MC); /* first byte */
1254 for (i = 1; i < size; i++) {
1255 n >>= NB;
1256 buff[islittle ? i : size - 1 - i] = (char)(n & MC);
1257 }
1258 if (neg && size > SZINT) { /* negative number need sign extension? */
1259 for (i = SZINT; i < size; i++) /* correct extra bytes */
1260 buff[islittle ? i : size - 1 - i] = (char)MC;
1261 }
1262 luaL_addsize(b, size); /* add result to buffer */
1263 }
1264
1265
1266 #ifndef _KERNEL
1267 /*
1268 ** Copy 'size' bytes from 'src' to 'dest', correcting endianness if
1269 ** given 'islittle' is different from native endianness.
1270 */
copywithendian(volatile char * dest,volatile const char * src,int size,int islittle)1271 static void copywithendian (volatile char *dest, volatile const char *src,
1272 int size, int islittle) {
1273 if (islittle == nativeendian.little) {
1274 while (size-- != 0)
1275 *(dest++) = *(src++);
1276 }
1277 else {
1278 dest += size - 1;
1279 while (size-- != 0)
1280 *(dest--) = *(src++);
1281 }
1282 }
1283 #endif
1284
1285
str_pack(lua_State * L)1286 static int str_pack (lua_State *L) {
1287 luaL_Buffer b;
1288 Header h;
1289 const char *fmt = luaL_checkstring(L, 1); /* format string */
1290 int arg = 1; /* current argument to pack */
1291 size_t totalsize = 0; /* accumulate total size of result */
1292 initheader(L, &h);
1293 lua_pushnil(L); /* mark to separate arguments from string buffer */
1294 luaL_buffinit(L, &b);
1295 while (*fmt != '\0') {
1296 int size, ntoalign;
1297 KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
1298 totalsize += ntoalign + size;
1299 while (ntoalign-- > 0)
1300 luaL_addchar(&b, LUA_PACKPADBYTE); /* fill alignment */
1301 arg++;
1302 switch (opt) {
1303 case Kint: { /* signed integers */
1304 lua_Integer n = luaL_checkinteger(L, arg);
1305 if (size < SZINT) { /* need overflow check? */
1306 lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1);
1307 luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow");
1308 }
1309 packint(&b, (lua_Unsigned)n, h.islittle, size, (n < 0));
1310 break;
1311 }
1312 case Kuint: { /* unsigned integers */
1313 lua_Integer n = luaL_checkinteger(L, arg);
1314 if (size < SZINT) /* need overflow check? */
1315 luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)),
1316 arg, "unsigned overflow");
1317 packint(&b, (lua_Unsigned)n, h.islittle, size, 0);
1318 break;
1319 }
1320 #ifndef _KERNEL
1321 case Kfloat: { /* floating-point options */
1322 volatile Ftypes u;
1323 char *buff = luaL_prepbuffsize(&b, size);
1324 lua_Number n = luaL_checknumber(L, arg); /* get argument */
1325 if (size == sizeof(u.f)) u.f = (float)n; /* copy it into 'u' */
1326 else if (size == sizeof(u.d)) u.d = (double)n;
1327 else u.n = n;
1328 /* move 'u' to final result, correcting endianness if needed */
1329 copywithendian(buff, u.buff, size, h.islittle);
1330 luaL_addsize(&b, size);
1331 break;
1332 }
1333 #endif
1334 case Kchar: { /* fixed-size string */
1335 size_t len;
1336 const char *s = luaL_checklstring(L, arg, &len);
1337 luaL_argcheck(L, len == (size_t)size, arg, "wrong length");
1338 luaL_addlstring(&b, s, size);
1339 break;
1340 }
1341 case Kstring: { /* strings with length count */
1342 size_t len;
1343 const char *s = luaL_checklstring(L, arg, &len);
1344 luaL_argcheck(L, size >= (int)sizeof(size_t) ||
1345 len < ((size_t)1 << (size * NB)),
1346 arg, "string length does not fit in given size");
1347 packint(&b, (lua_Unsigned)len, h.islittle, size, 0); /* pack length */
1348 luaL_addlstring(&b, s, len);
1349 totalsize += len;
1350 break;
1351 }
1352 case Kzstr: { /* zero-terminated string */
1353 size_t len;
1354 const char *s = luaL_checklstring(L, arg, &len);
1355 luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros");
1356 luaL_addlstring(&b, s, len);
1357 luaL_addchar(&b, '\0'); /* add zero at the end */
1358 totalsize += len + 1;
1359 break;
1360 }
1361 case Kpadding: luaL_addchar(&b, LUA_PACKPADBYTE); /* FALLTHROUGH */
1362 case Kpaddalign: case Knop:
1363 arg--; /* undo increment */
1364 break;
1365 }
1366 }
1367 luaL_pushresult(&b);
1368 return 1;
1369 }
1370
1371
str_packsize(lua_State * L)1372 static int str_packsize (lua_State *L) {
1373 Header h;
1374 const char *fmt = luaL_checkstring(L, 1); /* format string */
1375 size_t totalsize = 0; /* accumulate total size of result */
1376 initheader(L, &h);
1377 while (*fmt != '\0') {
1378 int size, ntoalign;
1379 KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
1380 size += ntoalign; /* total space used by option */
1381 luaL_argcheck(L, totalsize <= MAXSIZE - size, 1,
1382 "format result too large");
1383 totalsize += size;
1384 switch (opt) {
1385 case Kstring: /* strings with length count */
1386 case Kzstr: /* zero-terminated string */
1387 luaL_argerror(L, 1, "variable-length format");
1388 break;
1389 default: break;
1390 }
1391 }
1392 lua_pushinteger(L, (lua_Integer)totalsize);
1393 return 1;
1394 }
1395
1396
1397 /*
1398 ** Unpack an integer with 'size' bytes and 'islittle' endianness.
1399 ** If size is smaller than the size of a Lua integer and integer
1400 ** is signed, must do sign extension (propagating the sign to the
1401 ** higher bits); if size is larger than the size of a Lua integer,
1402 ** it must check the unread bytes to see whether they do not cause an
1403 ** overflow.
1404 */
unpackint(lua_State * L,const char * str,int islittle,int size,int issigned)1405 static lua_Integer unpackint (lua_State *L, const char *str,
1406 int islittle, int size, int issigned) {
1407 lua_Unsigned res = 0;
1408 int i;
1409 int limit = (size <= SZINT) ? size : SZINT;
1410 for (i = limit - 1; i >= 0; i--) {
1411 res <<= NB;
1412 res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i];
1413 }
1414 if (size < SZINT) { /* real size smaller than lua_Integer? */
1415 if (issigned) { /* needs sign extension? */
1416 lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1);
1417 res = ((res ^ mask) - mask); /* do sign extension */
1418 }
1419 }
1420 else if (size > SZINT) { /* must check unread bytes */
1421 int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC;
1422 for (i = limit; i < size; i++) {
1423 if ((unsigned char)str[islittle ? i : size - 1 - i] != mask)
1424 luaL_error(L, "%d-byte integer does not fit into Lua Integer", size);
1425 }
1426 }
1427 return (lua_Integer)res;
1428 }
1429
1430
str_unpack(lua_State * L)1431 static int str_unpack (lua_State *L) {
1432 Header h;
1433 const char *fmt = luaL_checkstring(L, 1);
1434 size_t ld;
1435 const char *data = luaL_checklstring(L, 2, &ld);
1436 size_t pos = (size_t)posrelat(luaL_optinteger(L, 3, 1), ld) - 1;
1437 int n = 0; /* number of results */
1438 luaL_argcheck(L, pos <= ld, 3, "initial position out of string");
1439 initheader(L, &h);
1440 while (*fmt != '\0') {
1441 int size, ntoalign;
1442 KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign);
1443 if ((size_t)ntoalign + size > ~pos || pos + ntoalign + size > ld)
1444 luaL_argerror(L, 2, "data string too short");
1445 pos += ntoalign; /* skip alignment */
1446 /* stack space for item + next position */
1447 luaL_checkstack(L, 2, "too many results");
1448 n++;
1449 switch (opt) {
1450 case Kint:
1451 case Kuint: {
1452 lua_Integer res = unpackint(L, data + pos, h.islittle, size,
1453 (opt == Kint));
1454 lua_pushinteger(L, res);
1455 break;
1456 }
1457 #ifndef _KERNEL
1458 case Kfloat: {
1459 volatile Ftypes u;
1460 lua_Number num;
1461 copywithendian(u.buff, data + pos, size, h.islittle);
1462 if (size == sizeof(u.f)) num = (lua_Number)u.f;
1463 else if (size == sizeof(u.d)) num = (lua_Number)u.d;
1464 else num = u.n;
1465 lua_pushnumber(L, num);
1466 break;
1467 }
1468 #endif
1469 case Kchar: {
1470 lua_pushlstring(L, data + pos, size);
1471 break;
1472 }
1473 case Kstring: {
1474 size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0);
1475 luaL_argcheck(L, pos + len + size <= ld, 2, "data string too short");
1476 lua_pushlstring(L, data + pos + size, len);
1477 pos += len; /* skip string */
1478 break;
1479 }
1480 case Kzstr: {
1481 size_t len = (int)strlen(data + pos);
1482 lua_pushlstring(L, data + pos, len);
1483 pos += len + 1; /* skip string plus final '\0' */
1484 break;
1485 }
1486 case Kpaddalign: case Kpadding: case Knop:
1487 n--; /* undo increment */
1488 break;
1489 }
1490 pos += size;
1491 }
1492 lua_pushinteger(L, pos + 1); /* next position */
1493 return n + 1;
1494 }
1495
1496 /* }====================================================== */
1497
1498
1499 static const luaL_Reg strlib[] = {
1500 {"byte", str_byte},
1501 {"char", str_char},
1502 {"dump", str_dump},
1503 {"find", str_find},
1504 {"format", str_format},
1505 {"gmatch", gmatch},
1506 {"gsub", str_gsub},
1507 {"len", str_len},
1508 {"lower", str_lower},
1509 {"match", str_match},
1510 {"rep", str_rep},
1511 {"reverse", str_reverse},
1512 {"sub", str_sub},
1513 {"upper", str_upper},
1514 {"pack", str_pack},
1515 {"packsize", str_packsize},
1516 {"unpack", str_unpack},
1517 {NULL, NULL}
1518 };
1519
1520
createmetatable(lua_State * L)1521 static void createmetatable (lua_State *L) {
1522 lua_createtable(L, 0, 1); /* table to be metatable for strings */
1523 lua_pushliteral(L, ""); /* dummy string */
1524 lua_pushvalue(L, -2); /* copy table */
1525 lua_setmetatable(L, -2); /* set table as metatable for strings */
1526 lua_pop(L, 1); /* pop dummy string */
1527 lua_pushvalue(L, -2); /* get string library */
1528 lua_setfield(L, -2, "__index"); /* metatable.__index = string */
1529 lua_pop(L, 1); /* pop metatable */
1530 }
1531
1532
1533 /*
1534 ** Open string library
1535 */
luaopen_string(lua_State * L)1536 LUAMOD_API int luaopen_string (lua_State *L) {
1537 luaL_newlib(L, strlib);
1538 createmetatable(L);
1539 return 1;
1540 }
1541
1542