2
** $Id: lstrlib.c,v 1.132.1.4 2008/07/11 17:27:21 roberto Exp $
3
** Standard library for string operations and pattern-matching
4
** See Copyright Notice in lua.h
24
/* macro to `unsign' a character */
25
#define uchar(c) ((unsigned char)(c))
29
static int str_len (lua_State *L) {
31
luaL_checklstring(L, 1, &l);
32
lua_pushinteger(L, l);
37
static ptrdiff_t posrelat (ptrdiff_t pos, size_t len) {
38
/* relative string position: negative means back from end */
39
if (pos < 0) pos += (ptrdiff_t)len + 1;
40
return (pos >= 0) ? pos : 0;
44
static int str_sub (lua_State *L) {
46
const char *s = luaL_checklstring(L, 1, &l);
47
ptrdiff_t start = posrelat(luaL_checkinteger(L, 2), l);
48
ptrdiff_t end = posrelat(luaL_optinteger(L, 3, -1), l);
49
if (start < 1) start = 1;
50
if (end > (ptrdiff_t)l) end = (ptrdiff_t)l;
52
lua_pushlstring(L, s+start-1, end-start+1);
53
else lua_pushliteral(L, "");
58
static int str_reverse (lua_State *L) {
61
const char *s = luaL_checklstring(L, 1, &l);
63
while (l--) luaL_addchar(&b, s[l]);
69
static int str_lower (lua_State *L) {
73
const char *s = luaL_checklstring(L, 1, &l);
76
luaL_addchar(&b, tolower(uchar(s[i])));
82
static int str_upper (lua_State *L) {
86
const char *s = luaL_checklstring(L, 1, &l);
89
luaL_addchar(&b, toupper(uchar(s[i])));
94
static int str_rep (lua_State *L) {
97
const char *s = luaL_checklstring(L, 1, &l);
98
int n = luaL_checkint(L, 2);
101
luaL_addlstring(&b, s, l);
107
static int str_byte (lua_State *L) {
109
const char *s = luaL_checklstring(L, 1, &l);
110
ptrdiff_t posi = posrelat(luaL_optinteger(L, 2, 1), l);
111
ptrdiff_t pose = posrelat(luaL_optinteger(L, 3, posi), l);
113
if (posi <= 0) posi = 1;
114
if ((size_t)pose > l) pose = l;
115
if (posi > pose) return 0; /* empty interval; return no values */
116
n = (int)(pose - posi + 1);
117
if (posi + n <= pose) /* overflow? */
118
luaL_error(L, "string slice too long");
119
luaL_checkstack(L, n, "string slice too long");
121
lua_pushinteger(L, uchar(s[posi+i-1]));
126
static int str_char (lua_State *L) {
127
int n = lua_gettop(L); /* number of arguments */
130
luaL_buffinit(L, &b);
131
for (i=1; i<=n; i++) {
132
int c = luaL_checkint(L, i);
133
luaL_argcheck(L, uchar(c) == c, i, "invalid value");
134
luaL_addchar(&b, uchar(c));
141
static int writer (lua_State *L, const void* b, size_t size, void* B) {
143
luaL_addlstring((luaL_Buffer*) B, (const char *)b, size);
148
static int str_dump (lua_State *L) {
150
luaL_checktype(L, 1, LUA_TFUNCTION);
153
if (lua_dump(L, writer, &b) != 0)
154
luaL_error(L, "unable to dump given function");
162
** {======================================================
164
** =======================================================
168
#define CAP_UNFINISHED (-1)
169
#define CAP_POSITION (-2)
171
typedef struct MatchState {
172
const char *src_init; /* init of source string */
173
const char *src_end; /* end (`\0') of source string */
175
int level; /* total number of captures (finished or unfinished) */
179
} capture[LUA_MAXCAPTURES];
184
#define SPECIALS "^$*+?.([%-"
187
static int check_capture (MatchState *ms, int l) {
189
if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
190
return luaL_error(ms->L, "invalid capture index");
195
static int capture_to_close (MatchState *ms) {
196
int level = ms->level;
197
for (level--; level>=0; level--)
198
if (ms->capture[level].len == CAP_UNFINISHED) return level;
199
return luaL_error(ms->L, "invalid pattern capture");
203
static const char *classend (MatchState *ms, const char *p) {
207
luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");
212
do { /* look for a `]' */
214
luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");
215
if (*(p++) == L_ESC && *p != '\0')
216
p++; /* skip escapes (e.g. `%]') */
227
static int match_class (int c, int cl) {
229
switch (tolower(cl)) {
230
case 'a' : res = isalpha(c); break;
231
case 'c' : res = iscntrl(c); break;
232
case 'd' : res = isdigit(c); break;
233
case 'l' : res = islower(c); break;
234
case 'p' : res = ispunct(c); break;
235
case 's' : res = isspace(c); break;
236
case 'u' : res = isupper(c); break;
237
case 'w' : res = isalnum(c); break;
238
case 'x' : res = isxdigit(c); break;
239
case 'z' : res = (c == 0); break;
240
default: return (cl == c);
242
return (islower(cl) ? res : !res);
246
static int matchbracketclass (int c, const char *p, const char *ec) {
250
p++; /* skip the `^' */
255
if (match_class(c, uchar(*p)))
258
else if ((*(p+1) == '-') && (p+2 < ec)) {
260
if (uchar(*(p-2)) <= c && c <= uchar(*p))
263
else if (uchar(*p) == c) return sig;
269
static int singlematch (int c, const char *p, const char *ep) {
271
case '.': return 1; /* matches any char */
272
case L_ESC: return match_class(c, uchar(*(p+1)));
273
case '[': return matchbracketclass(c, p, ep-1);
274
default: return (uchar(*p) == c);
279
static const char *match (MatchState *ms, const char *s, const char *p);
282
static const char *matchbalance (MatchState *ms, const char *s,
284
if (*p == 0 || *(p+1) == 0)
285
luaL_error(ms->L, "unbalanced pattern");
286
if (*s != *p) return NULL;
291
while (++s < ms->src_end) {
293
if (--cont == 0) return s+1;
295
else if (*s == b) cont++;
298
return NULL; /* string ends out of balance */
302
static const char *max_expand (MatchState *ms, const char *s,
303
const char *p, const char *ep) {
304
ptrdiff_t i = 0; /* counts maximum expand for item */
305
while ((s+i)<ms->src_end && singlematch(uchar(*(s+i)), p, ep))
307
/* keeps trying to match with the maximum repetitions */
309
const char *res = match(ms, (s+i), ep+1);
311
i--; /* else didn't match; reduce 1 repetition to try again */
317
static const char *min_expand (MatchState *ms, const char *s,
318
const char *p, const char *ep) {
320
const char *res = match(ms, s, ep+1);
323
else if (s<ms->src_end && singlematch(uchar(*s), p, ep))
324
s++; /* try with one more repetition */
330
static const char *start_capture (MatchState *ms, const char *s,
331
const char *p, int what) {
333
int level = ms->level;
334
if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
335
ms->capture[level].init = s;
336
ms->capture[level].len = what;
338
if ((res=match(ms, s, p)) == NULL) /* match failed? */
339
ms->level--; /* undo capture */
344
static const char *end_capture (MatchState *ms, const char *s,
346
int l = capture_to_close(ms);
348
ms->capture[l].len = s - ms->capture[l].init; /* close capture */
349
if ((res = match(ms, s, p)) == NULL) /* match failed? */
350
ms->capture[l].len = CAP_UNFINISHED; /* undo capture */
355
static const char *match_capture (MatchState *ms, const char *s, int l) {
357
l = check_capture(ms, l);
358
len = ms->capture[l].len;
359
if ((size_t)(ms->src_end-s) >= len &&
360
memcmp(ms->capture[l].init, s, len) == 0)
366
static const char *match (MatchState *ms, const char *s, const char *p) {
367
init: /* using goto's to optimize tail recursion */
369
case '(': { /* start capture */
370
if (*(p+1) == ')') /* position capture? */
371
return start_capture(ms, s, p+2, CAP_POSITION);
373
return start_capture(ms, s, p+1, CAP_UNFINISHED);
375
case ')': { /* end capture */
376
return end_capture(ms, s, p+1);
380
case 'b': { /* balanced string? */
381
s = matchbalance(ms, s, p+2);
382
if (s == NULL) return NULL;
383
p+=4; goto init; /* else return match(ms, s, p+4); */
385
case 'f': { /* frontier? */
386
const char *ep; char previous;
389
luaL_error(ms->L, "missing " LUA_QL("[") " after "
390
LUA_QL("%%f") " in pattern");
391
ep = classend(ms, p); /* points to what is next */
392
previous = (s == ms->src_init) ? '\0' : *(s-1);
393
if (matchbracketclass(uchar(previous), p, ep-1) ||
394
!matchbracketclass(uchar(*s), p, ep-1)) return NULL;
395
p=ep; goto init; /* else return match(ms, s, ep); */
398
if (isdigit(uchar(*(p+1)))) { /* capture results (%0-%9)? */
399
s = match_capture(ms, s, uchar(*(p+1)));
400
if (s == NULL) return NULL;
401
p+=2; goto init; /* else return match(ms, s, p+2) */
403
goto dflt; /* case default */
407
case '\0': { /* end of pattern */
408
return s; /* match succeeded */
411
if (*(p+1) == '\0') /* is the `$' the last char in pattern? */
412
return (s == ms->src_end) ? s : NULL; /* check end of string */
415
default: dflt: { /* it is a pattern item */
416
const char *ep = classend(ms, p); /* points to what is next */
417
int m = s<ms->src_end && singlematch(uchar(*s), p, ep);
419
case '?': { /* optional */
421
if (m && ((res=match(ms, s+1, ep+1)) != NULL))
423
p=ep+1; goto init; /* else return match(ms, s, ep+1); */
425
case '*': { /* 0 or more repetitions */
426
return max_expand(ms, s, p, ep);
428
case '+': { /* 1 or more repetitions */
429
return (m ? max_expand(ms, s+1, p, ep) : NULL);
431
case '-': { /* 0 or more repetitions (minimum) */
432
return min_expand(ms, s, p, ep);
436
s++; p=ep; goto init; /* else return match(ms, s+1, ep); */
445
static const char *lmemfind (const char *s1, size_t l1,
446
const char *s2, size_t l2) {
447
if (l2 == 0) return s1; /* empty strings are everywhere */
448
else if (l2 > l1) return NULL; /* avoids a negative `l1' */
450
const char *init; /* to search for a `*s2' inside `s1' */
451
l2--; /* 1st char will be checked by `memchr' */
452
l1 = l1-l2; /* `s2' cannot be found after that */
453
while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
454
init++; /* 1st char is already checked */
455
if (memcmp(init, s2+1, l2) == 0)
457
else { /* correct `l1' and `s1' to try again */
462
return NULL; /* not found */
467
static void push_onecapture (MatchState *ms, int i, const char *s,
469
if (i >= ms->level) {
470
if (i == 0) /* ms->level == 0, too */
471
lua_pushlstring(ms->L, s, e - s); /* add whole match */
473
luaL_error(ms->L, "invalid capture index");
476
ptrdiff_t l = ms->capture[i].len;
477
if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
478
if (l == CAP_POSITION)
479
lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
481
lua_pushlstring(ms->L, ms->capture[i].init, l);
486
static int push_captures (MatchState *ms, const char *s, const char *e) {
488
int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
489
luaL_checkstack(ms->L, nlevels, "too many captures");
490
for (i = 0; i < nlevels; i++)
491
push_onecapture(ms, i, s, e);
492
return nlevels; /* number of strings pushed */
496
static int str_find_aux (lua_State *L, int find) {
498
const char *s = luaL_checklstring(L, 1, &l1);
499
const char *p = luaL_checklstring(L, 2, &l2);
500
ptrdiff_t init = posrelat(luaL_optinteger(L, 3, 1), l1) - 1;
501
if (init < 0) init = 0;
502
else if ((size_t)(init) > l1) init = (ptrdiff_t)l1;
503
if (find && (lua_toboolean(L, 4) || /* explicit request? */
504
strpbrk(p, SPECIALS) == NULL)) { /* or no special characters? */
505
/* do a plain search */
506
const char *s2 = lmemfind(s+init, l1-init, p, l2);
508
lua_pushinteger(L, s2-s+1);
509
lua_pushinteger(L, s2-s+l2);
515
int anchor = (*p == '^') ? (p++, 1) : 0;
516
const char *s1=s+init;
523
if ((res=match(&ms, s1, p)) != NULL) {
525
lua_pushinteger(L, s1-s+1); /* start */
526
lua_pushinteger(L, res-s); /* end */
527
return push_captures(&ms, NULL, 0) + 2;
530
return push_captures(&ms, s1, res);
532
} while (s1++ < ms.src_end && !anchor);
534
lua_pushnil(L); /* not found */
539
static int str_find (lua_State *L) {
540
return str_find_aux(L, 1);
544
static int str_match (lua_State *L) {
545
return str_find_aux(L, 0);
549
static int gmatch_aux (lua_State *L) {
552
const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
553
const char *p = lua_tostring(L, lua_upvalueindex(2));
558
for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
563
if ((e = match(&ms, src, p)) != NULL) {
564
lua_Integer newstart = e-s;
565
if (e == src) newstart++; /* empty match? go at least one position */
566
lua_pushinteger(L, newstart);
567
lua_replace(L, lua_upvalueindex(3));
568
return push_captures(&ms, src, e);
571
return 0; /* not found */
575
static int gmatch (lua_State *L) {
576
luaL_checkstring(L, 1);
577
luaL_checkstring(L, 2);
579
lua_pushinteger(L, 0);
580
lua_pushcclosure(L, gmatch_aux, 3);
585
static int gfind_nodef (lua_State *L) {
586
return luaL_error(L, LUA_QL("string.gfind") " was renamed to "
587
LUA_QL("string.gmatch"));
591
static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
594
const char *news = lua_tolstring(ms->L, 3, &l);
595
for (i = 0; i < l; i++) {
596
if (news[i] != L_ESC)
597
luaL_addchar(b, news[i]);
600
if (!isdigit(uchar(news[i])))
601
luaL_addchar(b, news[i]);
602
else if (news[i] == '0')
603
luaL_addlstring(b, s, e - s);
605
push_onecapture(ms, news[i] - '1', s, e);
606
luaL_addvalue(b); /* add capture to accumulated result */
613
static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
615
lua_State *L = ms->L;
616
switch (lua_type(L, 3)) {
622
case LUA_TFUNCTION: {
625
n = push_captures(ms, s, e);
630
push_onecapture(ms, 0, s, e);
635
if (!lua_toboolean(L, -1)) { /* nil or false? */
637
lua_pushlstring(L, s, e - s); /* keep original text */
639
else if (!lua_isstring(L, -1))
640
luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
641
luaL_addvalue(b); /* add result to accumulator */
645
static int str_gsub (lua_State *L) {
647
const char *src = luaL_checklstring(L, 1, &srcl);
648
const char *p = luaL_checkstring(L, 2);
649
int tr = lua_type(L, 3);
650
int max_s = luaL_optint(L, 4, srcl+1);
651
int anchor = (*p == '^') ? (p++, 1) : 0;
655
luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
656
tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
657
"string/function/table expected");
658
luaL_buffinit(L, &b);
661
ms.src_end = src+srcl;
665
e = match(&ms, src, p);
668
add_value(&ms, &b, src, e);
670
if (e && e>src) /* non empty match? */
671
src = e; /* skip it */
672
else if (src < ms.src_end)
673
luaL_addchar(&b, *src++);
677
luaL_addlstring(&b, src, ms.src_end-src);
679
lua_pushinteger(L, n); /* number of substitutions */
683
/* }====================================================== */
686
/* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
688
/* valid flags in a format specification */
689
#define FLAGS "-+ #0"
691
** maximum size of each format specification (such as '%-099.99d')
692
** (+10 accounts for %99.99x plus margin of error)
694
#define MAX_FORMAT (sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
697
static void addquoted (lua_State *L, luaL_Buffer *b, int arg) {
699
const char *s = luaL_checklstring(L, arg, &l);
700
luaL_addchar(b, '"');
703
case '"': case '\\': case '\n': {
704
luaL_addchar(b, '\\');
709
luaL_addlstring(b, "\\r", 2);
713
luaL_addlstring(b, "\\000", 4);
723
luaL_addchar(b, '"');
726
static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
727
const char *p = strfrmt;
728
while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */
729
if ((size_t)(p - strfrmt) >= sizeof(FLAGS))
730
luaL_error(L, "invalid format (repeated flags)");
731
if (isdigit(uchar(*p))) p++; /* skip width */
732
if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
735
if (isdigit(uchar(*p))) p++; /* skip precision */
736
if (isdigit(uchar(*p))) p++; /* (2 digits at most) */
738
if (isdigit(uchar(*p)))
739
luaL_error(L, "invalid format (width or precision too long)");
741
strncpy(form, strfrmt, p - strfrmt + 1);
742
form += p - strfrmt + 1;
748
static void addintlen (char *form) {
749
size_t l = strlen(form);
750
char spec = form[l - 1];
751
strcpy(form + l - 1, LUA_INTFRMLEN);
752
form[l + sizeof(LUA_INTFRMLEN) - 2] = spec;
753
form[l + sizeof(LUA_INTFRMLEN) - 1] = '\0';
757
static int str_format (lua_State *L) {
760
const char *strfrmt = luaL_checklstring(L, arg, &sfl);
761
const char *strfrmt_end = strfrmt+sfl;
763
luaL_buffinit(L, &b);
764
while (strfrmt < strfrmt_end) {
765
if (*strfrmt != L_ESC)
766
luaL_addchar(&b, *strfrmt++);
767
else if (*++strfrmt == L_ESC)
768
luaL_addchar(&b, *strfrmt++); /* %% */
769
else { /* format item */
770
char form[MAX_FORMAT]; /* to store the format (`%...') */
771
char buff[MAX_ITEM]; /* to store the formatted item */
773
strfrmt = scanformat(L, strfrmt, form);
774
switch (*strfrmt++) {
776
snprintf(buff, sizeof (buff), form, (int)luaL_checknumber(L, arg));
779
case 'd': case 'i': {
781
snprintf(buff, sizeof (buff), form,
782
(LUA_INTFRM_T)luaL_checknumber(L, arg));
785
case 'o': case 'u': case 'x': case 'X': {
787
snprintf(buff, sizeof (buff), form,
788
(unsigned LUA_INTFRM_T)luaL_checknumber(L, arg));
791
case 'e': case 'E': case 'f':
792
case 'g': case 'G': {
793
snprintf(buff, sizeof (buff), form, (double)luaL_checknumber(L, arg));
797
addquoted(L, &b, arg);
798
continue; /* skip the 'addsize' at the end */
802
const char *s = luaL_checklstring(L, arg, &l);
803
if (!strchr(form, '.') && l >= 100) {
804
/* no precision and string is too long to be formatted;
805
keep original string */
806
lua_pushvalue(L, arg);
808
continue; /* skip the `addsize' at the end */
811
snprintf(buff, sizeof (buff), form, s);
815
default: { /* also treat cases `pnLlh' */
816
return luaL_error(L, "invalid option " LUA_QL("%%%c") " to "
817
LUA_QL("format"), *(strfrmt - 1));
820
luaL_addlstring(&b, buff, strlen(buff));
828
static const luaL_Reg strlib[] = {
833
{"format", str_format},
834
{"gfind", gfind_nodef},
838
{"lower", str_lower},
839
{"match", str_match},
841
{"reverse", str_reverse},
843
{"upper", str_upper},
848
static void createmetatable (lua_State *L) {
849
lua_createtable(L, 0, 1); /* create metatable for strings */
850
lua_pushliteral(L, ""); /* dummy string */
851
lua_pushvalue(L, -2);
852
lua_setmetatable(L, -2); /* set string metatable */
853
lua_pop(L, 1); /* pop dummy string */
854
lua_pushvalue(L, -2); /* string library... */
855
lua_setfield(L, -2, "__index"); /* ...is the __index metamethod */
856
lua_pop(L, 1); /* pop metatable */
861
** Open string library
863
LUALIB_API int luaopen_string (lua_State *L) {
864
luaL_register(L, LUA_STRLIBNAME, strlib);
865
#if defined(LUA_COMPAT_GFIND)
866
lua_getfield(L, -1, "gmatch");
867
lua_setfield(L, -2, "gfind");