2 ** $Id: lstrlib.c,v 1.254.1.1 2017/04/19 17:29:57 roberto Exp $
3 ** Standard library for string operations and pattern-matching
4 ** See Copyright Notice in lua.h
29 ** maximum number of captures that a pattern can do during
30 ** pattern-matching. This limit is arbitrary, but must fit in
33 #if !defined(LUA_MAXCAPTURES)
34 #define LUA_MAXCAPTURES 32
38 /* macro to 'unsign' a character */
39 #define uchar(c) ((unsigned char)(c))
43 ** Some sizes are better limited to fit in 'int', but must also fit in
44 ** 'size_t'. (We assume that 'lua_Integer' cannot be smaller than 'int'.)
46 #define MAX_SIZET ((size_t)(~(size_t)0))
49 (sizeof(size_t) < sizeof(int) ? MAX_SIZET : (size_t)(INT_MAX))
54 static int str_len (lua_State *L) {
56 luaL_checklstring(L, 1, &l);
57 lua_pushinteger(L, (lua_Integer)l);
62 /* translate a relative string position: negative means back from end */
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;
70 static int str_sub (lua_State *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;
78 lua_pushlstring(L, s + start - 1, (size_t)(end - start) + 1);
79 else lua_pushliteral(L, "");
84 static int str_reverse (lua_State *L) {
87 const char *s = luaL_checklstring(L, 1, &l);
88 char *p = luaL_buffinitsize(L, &b, l);
89 for (i = 0; i < l; i++)
91 luaL_pushresultsize(&b, l);
96 static int str_lower (lua_State *L) {
100 const char *s = luaL_checklstring(L, 1, &l);
101 char *p = luaL_buffinitsize(L, &b, l);
103 p[i] = tolower(uchar(s[i]));
104 luaL_pushresultsize(&b, l);
109 static int str_upper (lua_State *L) {
113 const char *s = luaL_checklstring(L, 1, &l);
114 char *p = luaL_buffinitsize(L, &b, l);
116 p[i] = toupper(uchar(s[i]));
117 luaL_pushresultsize(&b, l);
122 static int str_rep (lua_State *L) {
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");
131 size_t totallen = (size_t)n * l + (size_t)(n - 1) * lsep;
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));
141 memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */
142 luaL_pushresultsize(&b, totallen);
148 static int str_byte (lua_State *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);
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");
162 lua_pushinteger(L, uchar(s[posi+i-1]));
167 static int str_char (lua_State *L) {
168 int n = lua_gettop(L); /* number of arguments */
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");
177 luaL_pushresultsize(&b, n);
182 static int writer (lua_State *L, const void *b, size_t size, void *B) {
184 luaL_addlstring((luaL_Buffer *) B, (const char *)b, size);
189 static int str_dump (lua_State *L) {
191 int strip = lua_toboolean(L, 2);
192 luaL_checktype(L, 1, LUA_TFUNCTION);
195 if (lua_dump(L, writer, &b, strip) != 0)
196 return luaL_error(L, "unable to dump given function");
204 ** {======================================================
206 ** =======================================================
210 #define CAP_UNFINISHED (-1)
211 #define CAP_POSITION (-2)
214 typedef struct MatchState {
215 const char *src_init; /* init of source string */
216 const char *src_end; /* end ('\0') of source string */
217 const char *p_end; /* end ('\0') of pattern */
219 int matchdepth; /* control for recursive depth (to avoid C stack overflow) */
220 unsigned char level; /* total number of captures (finished or unfinished) */
224 } capture[LUA_MAXCAPTURES];
228 /* recursive function */
229 static const char *match (MatchState *ms, const char *s, const char *p);
232 /* maximum recursion depth for 'match' */
233 #if !defined(MAXCCALLS)
234 #define MAXCCALLS 200
239 #define SPECIALS "^$*+?.([%-"
242 static int check_capture (MatchState *ms, int l) {
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);
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");
258 static const char *classend (MatchState *ms, const char *p) {
262 luaL_error(ms->L, "malformed pattern (ends with '%%')");
267 do { /* look for a ']' */
269 luaL_error(ms->L, "malformed pattern (missing ']')");
270 if (*(p++) == L_ESC && p < ms->p_end)
271 p++; /* skip escapes (e.g. '%]') */
282 static int match_class (int c, int cl) {
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);
298 return (islower(cl) ? res : !res);
302 static int matchbracketclass (int c, const char *p, const char *ec) {
306 p++; /* skip the '^' */
311 if (match_class(c, uchar(*p)))
314 else if ((*(p+1) == '-') && (p+2 < ec)) {
316 if (uchar(*(p-2)) <= c && c <= uchar(*p))
319 else if (uchar(*p) == c) return sig;
325 static int singlematch (MatchState *ms, const char *s, const char *p,
327 if (s >= ms->src_end)
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);
341 static const char *matchbalance (MatchState *ms, const char *s,
343 if (p >= ms->p_end - 1)
344 luaL_error(ms->L, "malformed pattern (missing arguments to '%%b')");
345 if (*s != *p) return NULL;
350 while (++s < ms->src_end) {
352 if (--cont == 0) return s+1;
354 else if (*s == b) cont++;
357 return NULL; /* string ends out of balance */
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))
366 /* keeps trying to match with the maximum repetitions */
368 const char *res = match(ms, (s+i), ep+1);
370 i--; /* else didn't match; reduce 1 repetition to try again */
376 static const char *min_expand (MatchState *ms, const char *s,
377 const char *p, const char *ep) {
379 const char *res = match(ms, s, ep+1);
382 else if (singlematch(ms, s, p, ep))
383 s++; /* try with one more repetition */
389 static const char *start_capture (MatchState *ms, const char *s,
390 const char *p, int what) {
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;
397 if ((res=match(ms, s, p)) == NULL) /* match failed? */
398 ms->level--; /* undo capture */
403 static const char *end_capture (MatchState *ms, const char *s,
405 int l = capture_to_close(ms);
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 */
414 static const char *match_capture (MatchState *ms, const char *s, int l) {
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)
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? */
431 case '(': { /* start capture */
432 if (*(p + 1) == ')') /* position capture? */
433 s = start_capture(ms, s, p + 2, CAP_POSITION);
435 s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
438 case ')': { /* end capture */
439 s = end_capture(ms, s, p + 1);
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 */
448 case L_ESC: { /* escaped sequences not in the format class[*+?-]? */
450 case 'b': { /* balanced string? */
451 s = matchbalance(ms, s, p + 2);
453 p += 4; goto init; /* return match(ms, s, p + 4); */
454 } /* else fail (s == NULL) */
457 case 'f': { /* frontier? */
458 const char *ep; char previous;
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); */
468 s = NULL; /* match failed */
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)));
476 p += 2; goto init; /* return match(ms, s, p + 2) */
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); */
491 else /* '+' or no suffix */
494 else { /* matched once */
495 switch (*ep) { /* handle optional suffix */
496 case '?': { /* optional */
498 if ((res = match(ms, s + 1, ep + 1)) != NULL)
501 p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */
505 case '+': /* 1 or more repetitions */
506 s++; /* 1 match already done */
508 case '*': /* 0 or more repetitions */
509 s = max_expand(ms, s, p, ep);
511 case '-': /* 0 or more repetitions (minimum) */
512 s = min_expand(ms, s, p, ep);
514 default: /* no suffix */
515 s++; p = ep; goto init; /* return match(ms, s + 1, ep); */
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' */
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)
540 else { /* correct 'l1' and 's1' to try again */
545 return NULL; /* not found */
550 static void push_onecapture (MatchState *ms, int i, const char *s,
552 if (i >= ms->level) {
553 if (i == 0) /* ms->level == 0, too */
554 lua_pushlstring(ms->L, s, e - s); /* add whole match */
556 luaL_error(ms->L, "invalid capture index %%%d", i + 1);
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);
564 lua_pushlstring(ms->L, ms->capture[i].init, l);
569 static int push_captures (MatchState *ms, const char *s, const char *e) {
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 */
579 /* check whether pattern has no special characters */
580 static int nospecials (const char *p, size_t l) {
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 */
587 return 1; /* no special chars found */
591 static void prepstate (MatchState *ms, lua_State *L,
592 const char *s, size_t ls, const char *p, size_t lp) {
594 ms->matchdepth = MAXCCALLS;
596 ms->src_end = s + ls;
601 static void reprepstate (MatchState *ms) {
603 lua_assert(ms->matchdepth == MAXCCALLS);
607 static int str_find_aux (lua_State *L, int find) {
609 const char *s = luaL_checklstring(L, 1, &ls);
610 const char *p = luaL_checklstring(L, 2, &lp);
611 lua_Integer init = posrelat(luaL_optinteger(L, 3, 1), ls);
612 if (init < 1) init = 1;
613 else if (init > (lua_Integer)ls + 1) { /* start after string's end? */
614 lua_pushnil(L); /* cannot find anything */
617 /* explicit request or no special characters? */
618 if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {
619 /* do a plain search */
620 const char *s2 = lmemfind(s + init - 1, ls - (size_t)init + 1, p, lp);
622 lua_pushinteger(L, (s2 - s) + 1);
623 lua_pushinteger(L, (s2 - s) + lp);
629 const char *s1 = s + init - 1;
630 int anchor = (*p == '^');
632 p++; lp--; /* skip anchor character */
634 prepstate(&ms, L, s, ls, p, lp);
638 if ((res=match(&ms, s1, p)) != NULL) {
640 lua_pushinteger(L, (s1 - s) + 1); /* start */
641 lua_pushinteger(L, res - s); /* end */
642 return push_captures(&ms, NULL, 0) + 2;
645 return push_captures(&ms, s1, res);
647 } while (s1++ < ms.src_end && !anchor);
649 lua_pushnil(L); /* not found */
654 static int str_find (lua_State *L) {
655 return str_find_aux(L, 1);
659 static int str_match (lua_State *L) {
660 return str_find_aux(L, 0);
664 /* state for 'gmatch' */
665 typedef struct GMatchState {
666 const char *src; /* current position */
667 const char *p; /* pattern */
668 const char *lastmatch; /* end of last match */
669 MatchState ms; /* match state */
673 static int gmatch_aux (lua_State *L) {
674 GMatchState *gm = (GMatchState *)lua_touserdata(L, lua_upvalueindex(3));
677 for (src = gm->src; src <= gm->ms.src_end; src++) {
679 reprepstate(&gm->ms);
680 if ((e = match(&gm->ms, src, gm->p)) != NULL && e != gm->lastmatch) {
681 gm->src = gm->lastmatch = e;
682 return push_captures(&gm->ms, src, e);
685 return 0; /* not found */
689 static int gmatch (lua_State *L) {
691 const char *s = luaL_checklstring(L, 1, &ls);
692 const char *p = luaL_checklstring(L, 2, &lp);
694 lua_settop(L, 2); /* keep them on closure to avoid being collected */
695 gm = (GMatchState *)lua_newuserdata(L, sizeof(GMatchState));
696 prepstate(&gm->ms, L, s, ls, p, lp);
697 gm->src = s; gm->p = p; gm->lastmatch = NULL;
698 lua_pushcclosure(L, gmatch_aux, 3);
703 static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
706 lua_State *L = ms->L;
707 const char *news = lua_tolstring(L, 3, &l);
708 for (i = 0; i < l; i++) {
709 if (news[i] != L_ESC)
710 luaL_addchar(b, news[i]);
713 if (!isdigit(uchar(news[i]))) {
714 if (news[i] != L_ESC)
715 luaL_error(L, "invalid use of '%c' in replacement string", L_ESC);
716 luaL_addchar(b, news[i]);
718 else if (news[i] == '0')
719 luaL_addlstring(b, s, e - s);
721 push_onecapture(ms, news[i] - '1', s, e);
722 luaL_tolstring(L, -1, NULL); /* if number, convert it to string */
723 lua_remove(L, -2); /* remove original value */
724 luaL_addvalue(b); /* add capture to accumulated result */
731 static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
732 const char *e, int tr) {
733 lua_State *L = ms->L;
735 case LUA_TFUNCTION: {
738 n = push_captures(ms, s, e);
743 push_onecapture(ms, 0, s, e);
747 default: { /* LUA_TNUMBER or LUA_TSTRING */
752 if (!lua_toboolean(L, -1)) { /* nil or false? */
754 lua_pushlstring(L, s, e - s); /* keep original text */
756 else if (!lua_isstring(L, -1))
757 luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
758 luaL_addvalue(b); /* add result to accumulator */
762 static int str_gsub (lua_State *L) {
764 const char *src = luaL_checklstring(L, 1, &srcl); /* subject */
765 const char *p = luaL_checklstring(L, 2, &lp); /* pattern */
766 const char *lastmatch = NULL; /* end of last match */
767 int tr = lua_type(L, 3); /* replacement type */
768 lua_Integer max_s = luaL_optinteger(L, 4, srcl + 1); /* max replacements */
769 int anchor = (*p == '^');
770 lua_Integer n = 0; /* replacement count */
773 luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
774 tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
775 "string/function/table expected");
776 luaL_buffinit(L, &b);
778 p++; lp--; /* skip anchor character */
780 prepstate(&ms, L, src, srcl, p, lp);
783 reprepstate(&ms); /* (re)prepare state for new match */
784 if ((e = match(&ms, src, p)) != NULL && e != lastmatch) { /* match? */
786 add_value(&ms, &b, src, e, tr); /* add replacement to buffer */
789 else if (src < ms.src_end) /* otherwise, skip one character */
790 luaL_addchar(&b, *src++);
791 else break; /* end of subject */
794 luaL_addlstring(&b, src, ms.src_end-src);
796 lua_pushinteger(L, n); /* number of substitutions */
800 /* }====================================================== */
805 ** {======================================================
807 ** =======================================================
810 #if !defined(lua_number2strx) /* { */
813 ** Hexadecimal floating-point formatter
818 #define SIZELENMOD (sizeof(LUA_NUMBER_FRMLEN)/sizeof(char))
822 ** Number of bits that goes into the first digit. It can be any value
823 ** between 1 and 4; the following definition tries to align the number
824 ** to nibble boundaries by making what is left after that first digit a
827 #define L_NBFD ((l_mathlim(MANT_DIG) - 1)%4 + 1)
831 ** Add integer part of 'x' to buffer and return new 'x'
833 static lua_Number adddigit (char *buff, int n, lua_Number x) {
834 lua_Number dd = l_mathop(floor)(x); /* get integer part from 'x' */
836 buff[n] = (d < 10 ? d + '0' : d - 10 + 'a'); /* add to buffer */
837 return x - dd; /* return what is left */
841 static int num2straux (char *buff, int sz, lua_Number x) {
842 /* if 'inf' or 'NaN', format it like '%g' */
843 if (x != x || x == (lua_Number)HUGE_VAL || x == -(lua_Number)HUGE_VAL)
844 return l_sprintf(buff, sz, LUA_NUMBER_FMT, (LUAI_UACNUMBER)x);
845 else if (x == 0) { /* can be -0... */
846 /* create "0" or "-0" followed by exponent */
847 return l_sprintf(buff, sz, LUA_NUMBER_FMT "x0p+0", (LUAI_UACNUMBER)x);
851 lua_Number m = l_mathop(frexp)(x, &e); /* 'x' fraction and exponent */
852 int n = 0; /* character count */
853 if (m < 0) { /* is number negative? */
854 buff[n++] = '-'; /* add signal */
855 m = -m; /* make it positive */
857 buff[n++] = '0'; buff[n++] = 'x'; /* add "0x" */
858 m = adddigit(buff, n++, m * (1 << L_NBFD)); /* add first digit */
859 e -= L_NBFD; /* this digit goes before the radix point */
860 if (m > 0) { /* more digits? */
861 buff[n++] = lua_getlocaledecpoint(); /* add radix point */
862 do { /* add as many digits as needed */
863 m = adddigit(buff, n++, m * 16);
866 n += l_sprintf(buff + n, sz - n, "p%+d", e); /* add exponent */
873 static int lua_number2strx (lua_State *L, char *buff, int sz,
874 const char *fmt, lua_Number x) {
875 int n = num2straux(buff, sz, x);
876 if (fmt[SIZELENMOD] == 'A') {
878 for (i = 0; i < n; i++)
879 buff[i] = toupper(uchar(buff[i]));
881 else if (fmt[SIZELENMOD] != 'a')
882 return luaL_error(L, "modifiers for format '%%a'/'%%A' not implemented");
890 ** Maximum size of each formatted item. This maximum size is produced
891 ** by format('%.99f', -maxfloat), and is equal to 99 + 3 ('-', '.',
892 ** and '\0') + number of decimal digits to represent maxfloat (which
893 ** is maximum exponent + 1). (99+3+1 then rounded to 120 for "extra
894 ** expenses", such as locale-dependent stuff)
896 #define MAX_ITEM (120 + l_mathlim(MAX_10_EXP))
899 /* valid flags in a format specification */
900 #define FLAGS "-+ #0"
903 ** maximum size of each format specification (such as "%-099.99d")
905 #define MAX_FORMAT 32
908 static void addquoted (luaL_Buffer *b, const char *s, size_t len) {
909 luaL_addchar(b, '"');
911 if (*s == '"' || *s == '\\' || *s == '\n') {
912 luaL_addchar(b, '\\');
915 else if (iscntrl(uchar(*s))) {
917 if (!isdigit(uchar(*(s+1))))
918 l_sprintf(buff, sizeof(buff), "\\%d", (int)uchar(*s));
920 l_sprintf(buff, sizeof(buff), "\\%03d", (int)uchar(*s));
921 luaL_addstring(b, buff);
927 luaL_addchar(b, '"');
931 #if LUA_FLOAT_TYPE != LUA_FLOAT_INT64
933 ** Ensures the 'buff' string uses a dot as the radix character.
935 static void checkdp (char *buff, int nb) {
936 if (memchr(buff, '.', nb) == NULL) { /* no dot? */
937 char point = lua_getlocaledecpoint(); /* try locale point */
938 char *ppoint = (char *)memchr(buff, point, nb);
939 if (ppoint) *ppoint = '.'; /* change it to a dot */
944 static void addliteral (lua_State *L, luaL_Buffer *b, int arg) {
945 switch (lua_type(L, arg)) {
948 const char *s = lua_tolstring(L, arg, &len);
949 addquoted(b, s, len);
953 char *buff = luaL_prepbuffsize(b, MAX_ITEM);
955 #if LUA_FLOAT_TYPE != LUA_FLOAT_INT64
956 if (!lua_isinteger(L, arg)) { /* float? */
957 lua_Number n = lua_tonumber(L, arg); /* write as hexa ('%a') */
958 nb = lua_number2strx(L, buff, MAX_ITEM, "%" LUA_NUMBER_FRMLEN "a", n);
959 checkdp(buff, nb); /* ensure it uses a dot */
961 else { /* integers */
965 lua_Integer n = lua_tointeger(L, arg);
966 const char *format = (n == LUA_MININTEGER) /* corner case? */
967 ? "0x%" LUA_INTEGER_FRMLEN "x" /* use hexa */
968 : LUA_INTEGER_FMT; /* else use default format */
969 nb = l_sprintf(buff, MAX_ITEM, format, (LUAI_UACINT)n);
974 case LUA_TNIL: case LUA_TBOOLEAN: {
975 luaL_tolstring(L, arg, NULL);
980 luaL_argerror(L, arg, "value has no literal form");
986 static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
987 const char *p = strfrmt;
988 while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */
989 if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char))
990 luaL_error(L, "invalid format (repeated flags)");
991 if (isdigit(uchar(*p))) p++; /* skip width */
992 if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
995 if (isdigit(uchar(*p))) p++; /* skip precision */
996 if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
998 if (isdigit(uchar(*p)))
999 luaL_error(L, "invalid format (width or precision too long)");
1001 memcpy(form, strfrmt, ((p - strfrmt) + 1) * sizeof(char));
1002 form += (p - strfrmt) + 1;
1009 ** add length modifier into formats
1011 static void addlenmod (char *form, const char *lenmod) {
1012 size_t l = strlen(form);
1013 size_t lm = strlen(lenmod);
1014 char spec = form[l - 1];
1015 strcpy(form + l - 1, lenmod);
1016 form[l + lm - 1] = spec;
1017 form[l + lm] = '\0';
1021 static int str_format (lua_State *L) {
1022 int top = lua_gettop(L);
1025 const char *strfrmt = luaL_checklstring(L, arg, &sfl);
1026 const char *strfrmt_end = strfrmt+sfl;
1028 luaL_buffinit(L, &b);
1029 while (strfrmt < strfrmt_end) {
1030 if (*strfrmt != L_ESC)
1031 luaL_addchar(&b, *strfrmt++);
1032 else if (*++strfrmt == L_ESC)
1033 luaL_addchar(&b, *strfrmt++); /* %% */
1034 else { /* format item */
1035 char form[MAX_FORMAT]; /* to store the format ('%...') */
1036 char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */
1037 int nb = 0; /* number of bytes in added item */
1039 luaL_argerror(L, arg, "no value");
1040 strfrmt = scanformat(L, strfrmt, form);
1041 switch (*strfrmt++) {
1043 nb = l_sprintf(buff, MAX_ITEM, form, (int)luaL_checkinteger(L, arg));
1047 case 'o': case 'u': case 'x': case 'X': {
1048 lua_Integer n = luaL_checkinteger(L, arg);
1049 addlenmod(form, LUA_INTEGER_FRMLEN);
1050 nb = l_sprintf(buff, MAX_ITEM, form, (LUAI_UACINT)n);
1054 addlenmod(form, LUA_NUMBER_FRMLEN);
1055 nb = lua_number2strx(L, buff, MAX_ITEM, form,
1056 luaL_checknumber(L, arg));
1058 case 'e': case 'E': case 'f':
1059 case 'g': case 'G': {
1060 lua_Number n = luaL_checknumber(L, arg);
1061 addlenmod(form, LUA_NUMBER_FRMLEN);
1062 nb = l_sprintf(buff, MAX_ITEM, form, (LUAI_UACNUMBER)n);
1066 addliteral(L, &b, arg);
1071 const char *s = luaL_tolstring(L, arg, &l);
1072 if (form[2] == '\0') /* no modifiers? */
1073 luaL_addvalue(&b); /* keep entire string */
1075 luaL_argcheck(L, l == strlen(s), arg, "string contains zeros");
1076 if (!strchr(form, '.') && l >= 100) {
1077 /* no precision and string is too long to be formatted */
1078 luaL_addvalue(&b); /* keep entire string */
1080 else { /* format the string into 'buff' */
1081 nb = l_sprintf(buff, MAX_ITEM, form, s);
1082 lua_pop(L, 1); /* remove result from 'luaL_tolstring' */
1087 default: { /* also treat cases 'pnLlh' */
1088 return luaL_error(L, "invalid option '%%%c' to 'format'",
1092 lua_assert(nb < MAX_ITEM);
1093 luaL_addsize(&b, nb);
1096 luaL_pushresult(&b);
1100 /* }====================================================== */
1104 ** {======================================================
1106 ** =======================================================
1110 /* value used for padding */
1111 #if !defined(LUAL_PACKPADBYTE)
1112 #define LUAL_PACKPADBYTE 0x00
1115 /* maximum size for the binary representation of an integer */
1116 #define MAXINTSIZE 16
1118 /* number of bits in a character */
1121 /* mask for one character (NB 1's) */
1122 #define MC ((1 << NB) - 1)
1124 /* size of a lua_Integer */
1125 #define SZINT ((int)sizeof(lua_Integer))
1128 /* dummy union to get native endianness */
1129 static const union {
1131 char little; /* true iff machine is little endian */
1132 } nativeendian = {1};
1135 /* dummy structure to get native alignment requirements */
1139 #if LUA_FLOAT_TYPE != LUA_FLOAT_INT64
1142 void *p; lua_Integer i; lua_Number n; } u;
1145 #define MAXALIGN (offsetof(struct cD, u))
1149 ** Union for serializing floats
1151 typedef union Ftypes {
1152 #if LUA_FLOAT_TYPE != LUA_FLOAT_INT64
1157 char buff[5 * sizeof(lua_Number)]; /* enough for any float type */
1162 ** information to pack/unpack stuff
1164 typedef struct Header {
1172 ** options for pack/unpack
1174 typedef enum KOption {
1175 Kint, /* signed integers */
1176 Kuint, /* unsigned integers */
1177 Kfloat, /* floating-point numbers */
1178 Kchar, /* fixed-length strings */
1179 Kstring, /* strings with prefixed length */
1180 Kzstr, /* zero-terminated strings */
1181 Kpadding, /* padding */
1182 Kpaddalign, /* padding for alignment */
1183 Knop /* no-op (configuration or spaces) */
1188 ** Read an integer numeral from string 'fmt' or return 'df' if
1189 ** there is no numeral
1191 static int digit (int c) { return '0' <= c && c <= '9'; }
1193 static int getnum (const char **fmt, int df) {
1194 if (!digit(**fmt)) /* no number? */
1195 return df; /* return default value */
1199 a = a*10 + (*((*fmt)++) - '0');
1200 } while (digit(**fmt) && a <= ((int)MAXSIZE - 9)/10);
1207 ** Read an integer numeral and raises an error if it is larger
1208 ** than the maximum size for integers.
1210 static int getnumlimit (Header *h, const char **fmt, int df) {
1211 int sz = getnum(fmt, df);
1212 if (sz > MAXINTSIZE || sz <= 0)
1213 return luaL_error(h->L, "integral size (%d) out of limits [1,%d]",
1220 ** Initialize Header
1222 static void initheader (lua_State *L, Header *h) {
1224 h->islittle = nativeendian.little;
1230 ** Read and classify next option. 'size' is filled with option's size.
1232 static KOption getoption (Header *h, const char **fmt, int *size) {
1233 int opt = *((*fmt)++);
1234 *size = 0; /* default */
1236 case 'b': *size = sizeof(char); return Kint;
1237 case 'B': *size = sizeof(char); return Kuint;
1238 case 'h': *size = sizeof(short); return Kint;
1239 case 'H': *size = sizeof(short); return Kuint;
1240 case 'l': *size = sizeof(long); return Kint;
1241 case 'L': *size = sizeof(long); return Kuint;
1242 case 'j': *size = sizeof(lua_Integer); return Kint;
1243 case 'J': *size = sizeof(lua_Integer); return Kuint;
1244 case 'T': *size = sizeof(size_t); return Kuint;
1245 #if LUA_FLOAT_TYPE != LUA_FLOAT_INT64
1246 case 'f': *size = sizeof(float); return Kfloat;
1247 case 'd': *size = sizeof(double); return Kfloat;
1249 case 'n': *size = sizeof(lua_Number); return Kfloat;
1250 case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint;
1251 case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint;
1252 case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring;
1254 *size = getnum(fmt, -1);
1256 luaL_error(h->L, "missing size for format option 'c'");
1258 case 'z': return Kzstr;
1259 case 'x': *size = 1; return Kpadding;
1260 case 'X': return Kpaddalign;
1262 case '<': h->islittle = 1; break;
1263 case '>': h->islittle = 0; break;
1264 case '=': h->islittle = nativeendian.little; break;
1265 case '!': h->maxalign = getnumlimit(h, fmt, MAXALIGN); break;
1266 default: luaL_error(h->L, "invalid format option '%c'", opt);
1273 ** Read, classify, and fill other details about the next option.
1274 ** 'psize' is filled with option's size, 'notoalign' with its
1275 ** alignment requirements.
1276 ** Local variable 'size' gets the size to be aligned. (Kpadal option
1277 ** always gets its full alignment, other options are limited by
1278 ** the maximum alignment ('maxalign'). Kchar option needs no alignment
1279 ** despite its size.
1281 static KOption getdetails (Header *h, size_t totalsize,
1282 const char **fmt, int *psize, int *ntoalign) {
1283 KOption opt = getoption(h, fmt, psize);
1284 int align = *psize; /* usually, alignment follows size */
1285 if (opt == Kpaddalign) { /* 'X' gets alignment from following option */
1286 if (**fmt == '\0' || getoption(h, fmt, &align) == Kchar || align == 0)
1287 luaL_argerror(h->L, 1, "invalid next option for option 'X'");
1289 if (align <= 1 || opt == Kchar) /* need no alignment? */
1292 if (align > h->maxalign) /* enforce maximum alignment */
1293 align = h->maxalign;
1294 if ((align & (align - 1)) != 0) /* is 'align' not a power of 2? */
1295 luaL_argerror(h->L, 1, "format asks for alignment not power of 2");
1296 *ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1);
1303 ** Pack integer 'n' with 'size' bytes and 'islittle' endianness.
1304 ** The final 'if' handles the case when 'size' is larger than
1305 ** the size of a Lua integer, correcting the extra sign-extension
1306 ** bytes if necessary (by default they would be zeros).
1308 static void packint (luaL_Buffer *b, lua_Unsigned n,
1309 int islittle, int size, int neg) {
1310 char *buff = luaL_prepbuffsize(b, size);
1312 buff[islittle ? 0 : size - 1] = (char)(n & MC); /* first byte */
1313 for (i = 1; i < size; i++) {
1315 buff[islittle ? i : size - 1 - i] = (char)(n & MC);
1317 if (neg && size > SZINT) { /* negative number need sign extension? */
1318 for (i = SZINT; i < size; i++) /* correct extra bytes */
1319 buff[islittle ? i : size - 1 - i] = (char)MC;
1321 luaL_addsize(b, size); /* add result to buffer */
1326 ** Copy 'size' bytes from 'src' to 'dest', correcting endianness if
1327 ** given 'islittle' is different from native endianness.
1329 static void copywithendian (volatile char *dest, volatile const char *src,
1330 int size, int islittle) {
1331 if (islittle == nativeendian.little) {
1333 *(dest++) = *(src++);
1338 *(dest--) = *(src++);
1343 static int str_pack (lua_State *L) {
1346 const char *fmt = luaL_checkstring(L, 1); /* format string */
1347 int arg = 1; /* current argument to pack */
1348 size_t totalsize = 0; /* accumulate total size of result */
1350 lua_pushnil(L); /* mark to separate arguments from string buffer */
1351 luaL_buffinit(L, &b);
1352 while (*fmt != '\0') {
1354 KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
1355 totalsize += ntoalign + size;
1356 while (ntoalign-- > 0)
1357 luaL_addchar(&b, LUAL_PACKPADBYTE); /* fill alignment */
1360 case Kint: { /* signed integers */
1361 lua_Integer n = luaL_checkinteger(L, arg);
1362 if (size < SZINT) { /* need overflow check? */
1363 lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1);
1364 luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow");
1366 packint(&b, (lua_Unsigned)n, h.islittle, size, (n < 0));
1369 case Kuint: { /* unsigned integers */
1370 lua_Integer n = luaL_checkinteger(L, arg);
1371 if (size < SZINT) /* need overflow check? */
1372 luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)),
1373 arg, "unsigned overflow");
1374 packint(&b, (lua_Unsigned)n, h.islittle, size, 0);
1377 case Kfloat: { /* floating-point options */
1379 char *buff = luaL_prepbuffsize(&b, size);
1380 lua_Number n = luaL_checknumber(L, arg); /* get argument */
1381 #if LUA_FLOAT_TYPE != LUA_FLOAT_INT64
1382 if (size == sizeof(u.f)) u.f = (float)n; /* copy it into 'u' */
1383 else if (size == sizeof(u.d)) u.d = (double)n;
1388 /* move 'u' to final result, correcting endianness if needed */
1389 copywithendian(buff, u.buff, size, h.islittle);
1390 luaL_addsize(&b, size);
1393 case Kchar: { /* fixed-size string */
1395 const char *s = luaL_checklstring(L, arg, &len);
1396 luaL_argcheck(L, len <= (size_t)size, arg,
1397 "string longer than given size");
1398 luaL_addlstring(&b, s, len); /* add string */
1399 while (len++ < (size_t)size) /* pad extra space */
1400 luaL_addchar(&b, LUAL_PACKPADBYTE);
1403 case Kstring: { /* strings with length count */
1405 const char *s = luaL_checklstring(L, arg, &len);
1406 luaL_argcheck(L, size >= (int)sizeof(size_t) ||
1407 len < ((size_t)1 << (size * NB)),
1408 arg, "string length does not fit in given size");
1409 packint(&b, (lua_Unsigned)len, h.islittle, size, 0); /* pack length */
1410 luaL_addlstring(&b, s, len);
1414 case Kzstr: { /* zero-terminated string */
1416 const char *s = luaL_checklstring(L, arg, &len);
1417 luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros");
1418 luaL_addlstring(&b, s, len);
1419 luaL_addchar(&b, '\0'); /* add zero at the end */
1420 totalsize += len + 1;
1423 case Kpadding: luaL_addchar(&b, LUAL_PACKPADBYTE); /* FALLTHROUGH */
1424 case Kpaddalign: case Knop:
1425 arg--; /* undo increment */
1429 luaL_pushresult(&b);
1434 static int str_packsize (lua_State *L) {
1436 const char *fmt = luaL_checkstring(L, 1); /* format string */
1437 size_t totalsize = 0; /* accumulate total size of result */
1439 while (*fmt != '\0') {
1441 KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
1442 size += ntoalign; /* total space used by option */
1443 luaL_argcheck(L, totalsize <= MAXSIZE - size, 1,
1444 "format result too large");
1447 case Kstring: /* strings with length count */
1448 case Kzstr: /* zero-terminated string */
1449 luaL_argerror(L, 1, "variable-length format");
1450 /* call never return, but to avoid warnings: *//* FALLTHROUGH */
1454 lua_pushinteger(L, (lua_Integer)totalsize);
1460 ** Unpack an integer with 'size' bytes and 'islittle' endianness.
1461 ** If size is smaller than the size of a Lua integer and integer
1462 ** is signed, must do sign extension (propagating the sign to the
1463 ** higher bits); if size is larger than the size of a Lua integer,
1464 ** it must check the unread bytes to see whether they do not cause an
1467 static lua_Integer unpackint (lua_State *L, const char *str,
1468 int islittle, int size, int issigned) {
1469 lua_Unsigned res = 0;
1471 int limit = (size <= SZINT) ? size : SZINT;
1472 for (i = limit - 1; i >= 0; i--) {
1474 res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i];
1476 if (size < SZINT) { /* real size smaller than lua_Integer? */
1477 if (issigned) { /* needs sign extension? */
1478 lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1);
1479 res = ((res ^ mask) - mask); /* do sign extension */
1482 else if (size > SZINT) { /* must check unread bytes */
1483 int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC;
1484 for (i = limit; i < size; i++) {
1485 if ((unsigned char)str[islittle ? i : size - 1 - i] != mask)
1486 luaL_error(L, "%d-byte integer does not fit into Lua Integer", size);
1489 return (lua_Integer)res;
1493 static int str_unpack (lua_State *L) {
1495 const char *fmt = luaL_checkstring(L, 1);
1497 const char *data = luaL_checklstring(L, 2, &ld);
1498 size_t pos = (size_t)posrelat(luaL_optinteger(L, 3, 1), ld) - 1;
1499 int n = 0; /* number of results */
1500 luaL_argcheck(L, pos <= ld, 3, "initial position out of string");
1502 while (*fmt != '\0') {
1504 KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign);
1505 if ((size_t)ntoalign + size > ~pos || pos + ntoalign + size > ld)
1506 luaL_argerror(L, 2, "data string too short");
1507 pos += ntoalign; /* skip alignment */
1508 /* stack space for item + next position */
1509 luaL_checkstack(L, 2, "too many results");
1514 lua_Integer res = unpackint(L, data + pos, h.islittle, size,
1516 lua_pushinteger(L, res);
1522 copywithendian(u.buff, data + pos, size, h.islittle);
1523 #if LUA_FLOAT_TYPE != LUA_FLOAT_INT64
1524 if (size == sizeof(u.f)) num = (lua_Number)u.f;
1525 else if (size == sizeof(u.d)) num = (lua_Number)u.d;
1530 lua_pushnumber(L, num);
1534 lua_pushlstring(L, data + pos, size);
1538 size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0);
1539 luaL_argcheck(L, pos + len + size <= ld, 2, "data string too short");
1540 lua_pushlstring(L, data + pos + size, len);
1541 pos += len; /* skip string */
1545 size_t len = (int)strlen(data + pos);
1546 lua_pushlstring(L, data + pos, len);
1547 pos += len + 1; /* skip string plus final '\0' */
1550 case Kpaddalign: case Kpadding: case Knop:
1551 n--; /* undo increment */
1556 lua_pushinteger(L, pos + 1); /* next position */
1560 /* }====================================================== */
1563 static const luaL_Reg strlib[] = {
1568 {"format", str_format},
1572 {"lower", str_lower},
1573 {"match", str_match},
1575 {"reverse", str_reverse},
1577 {"upper", str_upper},
1579 {"packsize", str_packsize},
1580 {"unpack", str_unpack},
1585 static void createmetatable (lua_State *L) {
1586 lua_createtable(L, 0, 1); /* table to be metatable for strings */
1587 lua_pushliteral(L, ""); /* dummy string */
1588 lua_pushvalue(L, -2); /* copy table */
1589 lua_setmetatable(L, -2); /* set table as metatable for strings */
1590 lua_pop(L, 1); /* pop dummy string */
1591 lua_pushvalue(L, -2); /* get string library */
1592 lua_setfield(L, -2, "__index"); /* metatable.__index = string */
1593 lua_pop(L, 1); /* pop metatable */
1598 ** Open string library
1600 LUAMOD_API int luaopen_string (lua_State *L) {
1601 luaL_newlib(L, strlib);