ba8b5c17 |
1 | /* |
2 | ** $Id: lstrlib.c,v 1.178 2012/08/14 18:12:34 roberto Exp $ |
3 | ** Standard library for string operations and pattern-matching |
4 | ** See Copyright Notice in lua.h |
5 | */ |
6 | |
7 | |
8 | #include <ctype.h> |
9 | #include <stddef.h> |
10 | #include <stdio.h> |
11 | #include <stdlib.h> |
12 | #include <string.h> |
13 | |
14 | #define lstrlib_c |
15 | #define LUA_LIB |
16 | |
17 | #include "lua.h" |
18 | |
19 | #include "lauxlib.h" |
20 | #include "lualib.h" |
21 | |
22 | |
23 | /* |
24 | ** maximum number of captures that a pattern can do during |
25 | ** pattern-matching. This limit is arbitrary. |
26 | */ |
27 | #if !defined(LUA_MAXCAPTURES) |
28 | #define LUA_MAXCAPTURES 32 |
29 | #endif |
30 | |
31 | |
32 | /* macro to `unsign' a character */ |
33 | #define uchar(c) ((unsigned char)(c)) |
34 | |
35 | |
36 | |
37 | static int str_len (lua_State *L) { |
38 | size_t l; |
39 | luaL_checklstring(L, 1, &l); |
40 | lua_pushinteger(L, (lua_Integer)l); |
41 | return 1; |
42 | } |
43 | |
44 | |
45 | /* translate a relative string position: negative means back from end */ |
46 | static size_t posrelat (ptrdiff_t pos, size_t len) { |
47 | if (pos >= 0) return (size_t)pos; |
48 | else if (0u - (size_t)pos > len) return 0; |
49 | else return len - ((size_t)-pos) + 1; |
50 | } |
51 | |
52 | |
53 | static int str_sub (lua_State *L) { |
54 | size_t l; |
55 | const char *s = luaL_checklstring(L, 1, &l); |
56 | size_t start = posrelat(luaL_checkinteger(L, 2), l); |
57 | size_t end = posrelat(luaL_optinteger(L, 3, -1), l); |
58 | if (start < 1) start = 1; |
59 | if (end > l) end = l; |
60 | if (start <= end) |
61 | lua_pushlstring(L, s + start - 1, end - start + 1); |
62 | else lua_pushliteral(L, ""); |
63 | return 1; |
64 | } |
65 | |
66 | |
67 | static int str_reverse (lua_State *L) { |
68 | size_t l, i; |
69 | luaL_Buffer b; |
70 | const char *s = luaL_checklstring(L, 1, &l); |
71 | char *p = luaL_buffinitsize(L, &b, l); |
72 | for (i = 0; i < l; i++) |
73 | p[i] = s[l - i - 1]; |
74 | luaL_pushresultsize(&b, l); |
75 | return 1; |
76 | } |
77 | |
78 | |
79 | static int str_lower (lua_State *L) { |
80 | size_t l; |
81 | size_t i; |
82 | luaL_Buffer b; |
83 | const char *s = luaL_checklstring(L, 1, &l); |
84 | char *p = luaL_buffinitsize(L, &b, l); |
85 | for (i=0; i<l; i++) |
86 | p[i] = tolower(uchar(s[i])); |
87 | luaL_pushresultsize(&b, l); |
88 | return 1; |
89 | } |
90 | |
91 | |
92 | static int str_upper (lua_State *L) { |
93 | size_t l; |
94 | size_t i; |
95 | luaL_Buffer b; |
96 | const char *s = luaL_checklstring(L, 1, &l); |
97 | char *p = luaL_buffinitsize(L, &b, l); |
98 | for (i=0; i<l; i++) |
99 | p[i] = toupper(uchar(s[i])); |
100 | luaL_pushresultsize(&b, l); |
101 | return 1; |
102 | } |
103 | |
104 | |
105 | /* reasonable limit to avoid arithmetic overflow */ |
106 | #define MAXSIZE ((~(size_t)0) >> 1) |
107 | |
108 | static int str_rep (lua_State *L) { |
109 | size_t l, lsep; |
110 | const char *s = luaL_checklstring(L, 1, &l); |
111 | int n = luaL_checkint(L, 2); |
112 | const char *sep = luaL_optlstring(L, 3, "", &lsep); |
113 | if (n <= 0) lua_pushliteral(L, ""); |
114 | else if (l + lsep < l || l + lsep >= MAXSIZE / n) /* may overflow? */ |
115 | return luaL_error(L, "resulting string too large"); |
116 | else { |
117 | size_t totallen = n * l + (n - 1) * lsep; |
118 | luaL_Buffer b; |
119 | char *p = luaL_buffinitsize(L, &b, totallen); |
120 | while (n-- > 1) { /* first n-1 copies (followed by separator) */ |
121 | memcpy(p, s, l * sizeof(char)); p += l; |
122 | if (lsep > 0) { /* avoid empty 'memcpy' (may be expensive) */ |
123 | memcpy(p, sep, lsep * sizeof(char)); p += lsep; |
124 | } |
125 | } |
126 | memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */ |
127 | luaL_pushresultsize(&b, totallen); |
128 | } |
129 | return 1; |
130 | } |
131 | |
132 | |
133 | static int str_byte (lua_State *L) { |
134 | size_t l; |
135 | const char *s = luaL_checklstring(L, 1, &l); |
136 | size_t posi = posrelat(luaL_optinteger(L, 2, 1), l); |
137 | size_t pose = posrelat(luaL_optinteger(L, 3, posi), l); |
138 | int n, i; |
139 | if (posi < 1) posi = 1; |
140 | if (pose > l) pose = l; |
141 | if (posi > pose) return 0; /* empty interval; return no values */ |
142 | n = (int)(pose - posi + 1); |
143 | if (posi + n <= pose) /* (size_t -> int) overflow? */ |
144 | return luaL_error(L, "string slice too long"); |
145 | luaL_checkstack(L, n, "string slice too long"); |
146 | for (i=0; i<n; i++) |
147 | lua_pushinteger(L, uchar(s[posi+i-1])); |
148 | return n; |
149 | } |
150 | |
151 | |
152 | static int str_char (lua_State *L) { |
153 | int n = lua_gettop(L); /* number of arguments */ |
154 | int i; |
155 | luaL_Buffer b; |
156 | char *p = luaL_buffinitsize(L, &b, n); |
157 | for (i=1; i<=n; i++) { |
158 | int c = luaL_checkint(L, i); |
159 | luaL_argcheck(L, uchar(c) == c, i, "value out of range"); |
160 | p[i - 1] = uchar(c); |
161 | } |
162 | luaL_pushresultsize(&b, n); |
163 | return 1; |
164 | } |
165 | |
166 | |
167 | static int writer (lua_State *L, const void* b, size_t size, void* B) { |
168 | (void)L; |
169 | luaL_addlstring((luaL_Buffer*) B, (const char *)b, size); |
170 | return 0; |
171 | } |
172 | |
173 | |
174 | static int str_dump (lua_State *L) { |
175 | luaL_Buffer b; |
176 | luaL_checktype(L, 1, LUA_TFUNCTION); |
177 | lua_settop(L, 1); |
178 | luaL_buffinit(L,&b); |
179 | if (lua_dump(L, writer, &b) != 0) |
180 | return luaL_error(L, "unable to dump given function"); |
181 | luaL_pushresult(&b); |
182 | return 1; |
183 | } |
184 | |
185 | |
186 | |
187 | /* |
188 | ** {====================================================== |
189 | ** PATTERN MATCHING |
190 | ** ======================================================= |
191 | */ |
192 | |
193 | |
194 | #define CAP_UNFINISHED (-1) |
195 | #define CAP_POSITION (-2) |
196 | |
197 | |
198 | typedef struct MatchState { |
199 | int matchdepth; /* control for recursive depth (to avoid C stack overflow) */ |
200 | const char *src_init; /* init of source string */ |
201 | const char *src_end; /* end ('\0') of source string */ |
202 | const char *p_end; /* end ('\0') of pattern */ |
203 | lua_State *L; |
204 | int level; /* total number of captures (finished or unfinished) */ |
205 | struct { |
206 | const char *init; |
207 | ptrdiff_t len; |
208 | } capture[LUA_MAXCAPTURES]; |
209 | } MatchState; |
210 | |
211 | |
212 | /* recursive function */ |
213 | static const char *match (MatchState *ms, const char *s, const char *p); |
214 | |
215 | |
216 | /* maximum recursion depth for 'match' */ |
217 | #if !defined(MAXCCALLS) |
218 | #define MAXCCALLS 200 |
219 | #endif |
220 | |
221 | |
222 | #define L_ESC '%' |
223 | #define SPECIALS "^$*+?.([%-" |
224 | |
225 | |
226 | static int check_capture (MatchState *ms, int l) { |
227 | l -= '1'; |
228 | if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED) |
229 | return luaL_error(ms->L, "invalid capture index %%%d", l + 1); |
230 | return l; |
231 | } |
232 | |
233 | |
234 | static int capture_to_close (MatchState *ms) { |
235 | int level = ms->level; |
236 | for (level--; level>=0; level--) |
237 | if (ms->capture[level].len == CAP_UNFINISHED) return level; |
238 | return luaL_error(ms->L, "invalid pattern capture"); |
239 | } |
240 | |
241 | |
242 | static const char *classend (MatchState *ms, const char *p) { |
243 | switch (*p++) { |
244 | case L_ESC: { |
245 | if (p == ms->p_end) |
246 | luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")"); |
247 | return p+1; |
248 | } |
249 | case '[': { |
250 | if (*p == '^') p++; |
251 | do { /* look for a `]' */ |
252 | if (p == ms->p_end) |
253 | luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")"); |
254 | if (*(p++) == L_ESC && p < ms->p_end) |
255 | p++; /* skip escapes (e.g. `%]') */ |
256 | } while (*p != ']'); |
257 | return p+1; |
258 | } |
259 | default: { |
260 | return p; |
261 | } |
262 | } |
263 | } |
264 | |
265 | |
266 | static int match_class (int c, int cl) { |
267 | int res; |
268 | switch (tolower(cl)) { |
269 | case 'a' : res = isalpha(c); break; |
270 | case 'c' : res = iscntrl(c); break; |
271 | case 'd' : res = isdigit(c); break; |
272 | case 'g' : res = isgraph(c); break; |
273 | case 'l' : res = islower(c); break; |
274 | case 'p' : res = ispunct(c); break; |
275 | case 's' : res = isspace(c); break; |
276 | case 'u' : res = isupper(c); break; |
277 | case 'w' : res = isalnum(c); break; |
278 | case 'x' : res = isxdigit(c); break; |
279 | case 'z' : res = (c == 0); break; /* deprecated option */ |
280 | default: return (cl == c); |
281 | } |
282 | return (islower(cl) ? res : !res); |
283 | } |
284 | |
285 | |
286 | static int matchbracketclass (int c, const char *p, const char *ec) { |
287 | int sig = 1; |
288 | if (*(p+1) == '^') { |
289 | sig = 0; |
290 | p++; /* skip the `^' */ |
291 | } |
292 | while (++p < ec) { |
293 | if (*p == L_ESC) { |
294 | p++; |
295 | if (match_class(c, uchar(*p))) |
296 | return sig; |
297 | } |
298 | else if ((*(p+1) == '-') && (p+2 < ec)) { |
299 | p+=2; |
300 | if (uchar(*(p-2)) <= c && c <= uchar(*p)) |
301 | return sig; |
302 | } |
303 | else if (uchar(*p) == c) return sig; |
304 | } |
305 | return !sig; |
306 | } |
307 | |
308 | |
309 | static int singlematch (MatchState *ms, const char *s, const char *p, |
310 | const char *ep) { |
311 | if (s >= ms->src_end) |
312 | return 0; |
313 | else { |
314 | int c = uchar(*s); |
315 | switch (*p) { |
316 | case '.': return 1; /* matches any char */ |
317 | case L_ESC: return match_class(c, uchar(*(p+1))); |
318 | case '[': return matchbracketclass(c, p, ep-1); |
319 | default: return (uchar(*p) == c); |
320 | } |
321 | } |
322 | } |
323 | |
324 | |
325 | static const char *matchbalance (MatchState *ms, const char *s, |
326 | const char *p) { |
327 | if (p >= ms->p_end - 1) |
328 | luaL_error(ms->L, "malformed pattern " |
329 | "(missing arguments to " LUA_QL("%%b") ")"); |
330 | if (*s != *p) return NULL; |
331 | else { |
332 | int b = *p; |
333 | int e = *(p+1); |
334 | int cont = 1; |
335 | while (++s < ms->src_end) { |
336 | if (*s == e) { |
337 | if (--cont == 0) return s+1; |
338 | } |
339 | else if (*s == b) cont++; |
340 | } |
341 | } |
342 | return NULL; /* string ends out of balance */ |
343 | } |
344 | |
345 | |
346 | static const char *max_expand (MatchState *ms, const char *s, |
347 | const char *p, const char *ep) { |
348 | ptrdiff_t i = 0; /* counts maximum expand for item */ |
349 | while (singlematch(ms, s + i, p, ep)) |
350 | i++; |
351 | /* keeps trying to match with the maximum repetitions */ |
352 | while (i>=0) { |
353 | const char *res = match(ms, (s+i), ep+1); |
354 | if (res) return res; |
355 | i--; /* else didn't match; reduce 1 repetition to try again */ |
356 | } |
357 | return NULL; |
358 | } |
359 | |
360 | |
361 | static const char *min_expand (MatchState *ms, const char *s, |
362 | const char *p, const char *ep) { |
363 | for (;;) { |
364 | const char *res = match(ms, s, ep+1); |
365 | if (res != NULL) |
366 | return res; |
367 | else if (singlematch(ms, s, p, ep)) |
368 | s++; /* try with one more repetition */ |
369 | else return NULL; |
370 | } |
371 | } |
372 | |
373 | |
374 | static const char *start_capture (MatchState *ms, const char *s, |
375 | const char *p, int what) { |
376 | const char *res; |
377 | int level = ms->level; |
378 | if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures"); |
379 | ms->capture[level].init = s; |
380 | ms->capture[level].len = what; |
381 | ms->level = level+1; |
382 | if ((res=match(ms, s, p)) == NULL) /* match failed? */ |
383 | ms->level--; /* undo capture */ |
384 | return res; |
385 | } |
386 | |
387 | |
388 | static const char *end_capture (MatchState *ms, const char *s, |
389 | const char *p) { |
390 | int l = capture_to_close(ms); |
391 | const char *res; |
392 | ms->capture[l].len = s - ms->capture[l].init; /* close capture */ |
393 | if ((res = match(ms, s, p)) == NULL) /* match failed? */ |
394 | ms->capture[l].len = CAP_UNFINISHED; /* undo capture */ |
395 | return res; |
396 | } |
397 | |
398 | |
399 | static const char *match_capture (MatchState *ms, const char *s, int l) { |
400 | size_t len; |
401 | l = check_capture(ms, l); |
402 | len = ms->capture[l].len; |
403 | if ((size_t)(ms->src_end-s) >= len && |
404 | memcmp(ms->capture[l].init, s, len) == 0) |
405 | return s+len; |
406 | else return NULL; |
407 | } |
408 | |
409 | |
410 | static const char *match (MatchState *ms, const char *s, const char *p) { |
411 | if (ms->matchdepth-- == 0) |
412 | luaL_error(ms->L, "pattern too complex"); |
413 | init: /* using goto's to optimize tail recursion */ |
414 | if (p != ms->p_end) { /* end of pattern? */ |
415 | switch (*p) { |
416 | case '(': { /* start capture */ |
417 | if (*(p + 1) == ')') /* position capture? */ |
418 | s = start_capture(ms, s, p + 2, CAP_POSITION); |
419 | else |
420 | s = start_capture(ms, s, p + 1, CAP_UNFINISHED); |
421 | break; |
422 | } |
423 | case ')': { /* end capture */ |
424 | s = end_capture(ms, s, p + 1); |
425 | break; |
426 | } |
427 | case '$': { |
428 | if ((p + 1) != ms->p_end) /* is the `$' the last char in pattern? */ |
429 | goto dflt; /* no; go to default */ |
430 | s = (s == ms->src_end) ? s : NULL; /* check end of string */ |
431 | break; |
432 | } |
433 | case L_ESC: { /* escaped sequences not in the format class[*+?-]? */ |
434 | switch (*(p + 1)) { |
435 | case 'b': { /* balanced string? */ |
436 | s = matchbalance(ms, s, p + 2); |
437 | if (s != NULL) { |
438 | p += 4; goto init; /* return match(ms, s, p + 4); */ |
439 | } /* else fail (s == NULL) */ |
440 | break; |
441 | } |
442 | case 'f': { /* frontier? */ |
443 | const char *ep; char previous; |
444 | p += 2; |
445 | if (*p != '[') |
446 | luaL_error(ms->L, "missing " LUA_QL("[") " after " |
447 | LUA_QL("%%f") " in pattern"); |
448 | ep = classend(ms, p); /* points to what is next */ |
449 | previous = (s == ms->src_init) ? '\0' : *(s - 1); |
450 | if (!matchbracketclass(uchar(previous), p, ep - 1) && |
451 | matchbracketclass(uchar(*s), p, ep - 1)) { |
452 | p = ep; goto init; /* return match(ms, s, ep); */ |
453 | } |
454 | s = NULL; /* match failed */ |
455 | break; |
456 | } |
457 | case '0': case '1': case '2': case '3': |
458 | case '4': case '5': case '6': case '7': |
459 | case '8': case '9': { /* capture results (%0-%9)? */ |
460 | s = match_capture(ms, s, uchar(*(p + 1))); |
461 | if (s != NULL) { |
462 | p += 2; goto init; /* return match(ms, s, p + 2) */ |
463 | } |
464 | break; |
465 | } |
466 | default: goto dflt; |
467 | } |
468 | break; |
469 | } |
470 | default: dflt: { /* pattern class plus optional suffix */ |
471 | const char *ep = classend(ms, p); /* points to optional suffix */ |
472 | /* does not match at least once? */ |
473 | if (!singlematch(ms, s, p, ep)) { |
474 | if (*ep == '*' || *ep == '?' || *ep == '-') { /* accept empty? */ |
475 | p = ep + 1; goto init; /* return match(ms, s, ep + 1); */ |
476 | } |
477 | else /* '+' or no suffix */ |
478 | s = NULL; /* fail */ |
479 | } |
480 | else { /* matched once */ |
481 | switch (*ep) { /* handle optional suffix */ |
482 | case '?': { /* optional */ |
483 | const char *res; |
484 | if ((res = match(ms, s + 1, ep + 1)) != NULL) |
485 | s = res; |
486 | else { |
487 | p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */ |
488 | } |
489 | break; |
490 | } |
491 | case '+': /* 1 or more repetitions */ |
492 | s++; /* 1 match already done */ |
493 | /* go through */ |
494 | case '*': /* 0 or more repetitions */ |
495 | s = max_expand(ms, s, p, ep); |
496 | break; |
497 | case '-': /* 0 or more repetitions (minimum) */ |
498 | s = min_expand(ms, s, p, ep); |
499 | break; |
500 | default: /* no suffix */ |
501 | s++; p = ep; goto init; /* return match(ms, s + 1, ep); */ |
502 | } |
503 | } |
504 | break; |
505 | } |
506 | } |
507 | } |
508 | ms->matchdepth++; |
509 | return s; |
510 | } |
511 | |
512 | |
513 | |
514 | static const char *lmemfind (const char *s1, size_t l1, |
515 | const char *s2, size_t l2) { |
516 | if (l2 == 0) return s1; /* empty strings are everywhere */ |
517 | else if (l2 > l1) return NULL; /* avoids a negative `l1' */ |
518 | else { |
519 | const char *init; /* to search for a `*s2' inside `s1' */ |
520 | l2--; /* 1st char will be checked by `memchr' */ |
521 | l1 = l1-l2; /* `s2' cannot be found after that */ |
522 | while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) { |
523 | init++; /* 1st char is already checked */ |
524 | if (memcmp(init, s2+1, l2) == 0) |
525 | return init-1; |
526 | else { /* correct `l1' and `s1' to try again */ |
527 | l1 -= init-s1; |
528 | s1 = init; |
529 | } |
530 | } |
531 | return NULL; /* not found */ |
532 | } |
533 | } |
534 | |
535 | |
536 | static void push_onecapture (MatchState *ms, int i, const char *s, |
537 | const char *e) { |
538 | if (i >= ms->level) { |
539 | if (i == 0) /* ms->level == 0, too */ |
540 | lua_pushlstring(ms->L, s, e - s); /* add whole match */ |
541 | else |
542 | luaL_error(ms->L, "invalid capture index"); |
543 | } |
544 | else { |
545 | ptrdiff_t l = ms->capture[i].len; |
546 | if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture"); |
547 | if (l == CAP_POSITION) |
548 | lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1); |
549 | else |
550 | lua_pushlstring(ms->L, ms->capture[i].init, l); |
551 | } |
552 | } |
553 | |
554 | |
555 | static int push_captures (MatchState *ms, const char *s, const char *e) { |
556 | int i; |
557 | int nlevels = (ms->level == 0 && s) ? 1 : ms->level; |
558 | luaL_checkstack(ms->L, nlevels, "too many captures"); |
559 | for (i = 0; i < nlevels; i++) |
560 | push_onecapture(ms, i, s, e); |
561 | return nlevels; /* number of strings pushed */ |
562 | } |
563 | |
564 | |
565 | /* check whether pattern has no special characters */ |
566 | static int nospecials (const char *p, size_t l) { |
567 | size_t upto = 0; |
568 | do { |
569 | if (strpbrk(p + upto, SPECIALS)) |
570 | return 0; /* pattern has a special character */ |
571 | upto += strlen(p + upto) + 1; /* may have more after \0 */ |
572 | } while (upto <= l); |
573 | return 1; /* no special chars found */ |
574 | } |
575 | |
576 | |
577 | static int str_find_aux (lua_State *L, int find) { |
578 | size_t ls, lp; |
579 | const char *s = luaL_checklstring(L, 1, &ls); |
580 | const char *p = luaL_checklstring(L, 2, &lp); |
581 | size_t init = posrelat(luaL_optinteger(L, 3, 1), ls); |
582 | if (init < 1) init = 1; |
583 | else if (init > ls + 1) { /* start after string's end? */ |
584 | lua_pushnil(L); /* cannot find anything */ |
585 | return 1; |
586 | } |
587 | /* explicit request or no special characters? */ |
588 | if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) { |
589 | /* do a plain search */ |
590 | const char *s2 = lmemfind(s + init - 1, ls - init + 1, p, lp); |
591 | if (s2) { |
592 | lua_pushinteger(L, s2 - s + 1); |
593 | lua_pushinteger(L, s2 - s + lp); |
594 | return 2; |
595 | } |
596 | } |
597 | else { |
598 | MatchState ms; |
599 | const char *s1 = s + init - 1; |
600 | int anchor = (*p == '^'); |
601 | if (anchor) { |
602 | p++; lp--; /* skip anchor character */ |
603 | } |
604 | ms.L = L; |
605 | ms.matchdepth = MAXCCALLS; |
606 | ms.src_init = s; |
607 | ms.src_end = s + ls; |
608 | ms.p_end = p + lp; |
609 | do { |
610 | const char *res; |
611 | ms.level = 0; |
612 | lua_assert(ms.matchdepth == MAXCCALLS); |
613 | if ((res=match(&ms, s1, p)) != NULL) { |
614 | if (find) { |
615 | lua_pushinteger(L, s1 - s + 1); /* start */ |
616 | lua_pushinteger(L, res - s); /* end */ |
617 | return push_captures(&ms, NULL, 0) + 2; |
618 | } |
619 | else |
620 | return push_captures(&ms, s1, res); |
621 | } |
622 | } while (s1++ < ms.src_end && !anchor); |
623 | } |
624 | lua_pushnil(L); /* not found */ |
625 | return 1; |
626 | } |
627 | |
628 | |
629 | static int str_find (lua_State *L) { |
630 | return str_find_aux(L, 1); |
631 | } |
632 | |
633 | |
634 | static int str_match (lua_State *L) { |
635 | return str_find_aux(L, 0); |
636 | } |
637 | |
638 | |
639 | static int gmatch_aux (lua_State *L) { |
640 | MatchState ms; |
641 | size_t ls, lp; |
642 | const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls); |
643 | const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp); |
644 | const char *src; |
645 | ms.L = L; |
646 | ms.matchdepth = MAXCCALLS; |
647 | ms.src_init = s; |
648 | ms.src_end = s+ls; |
649 | ms.p_end = p + lp; |
650 | for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3)); |
651 | src <= ms.src_end; |
652 | src++) { |
653 | const char *e; |
654 | ms.level = 0; |
655 | lua_assert(ms.matchdepth == MAXCCALLS); |
656 | if ((e = match(&ms, src, p)) != NULL) { |
657 | lua_Integer newstart = e-s; |
658 | if (e == src) newstart++; /* empty match? go at least one position */ |
659 | lua_pushinteger(L, newstart); |
660 | lua_replace(L, lua_upvalueindex(3)); |
661 | return push_captures(&ms, src, e); |
662 | } |
663 | } |
664 | return 0; /* not found */ |
665 | } |
666 | |
667 | |
668 | static int gmatch (lua_State *L) { |
669 | luaL_checkstring(L, 1); |
670 | luaL_checkstring(L, 2); |
671 | lua_settop(L, 2); |
672 | lua_pushinteger(L, 0); |
673 | lua_pushcclosure(L, gmatch_aux, 3); |
674 | return 1; |
675 | } |
676 | |
677 | |
678 | static void add_s (MatchState *ms, luaL_Buffer *b, const char *s, |
679 | const char *e) { |
680 | size_t l, i; |
681 | const char *news = lua_tolstring(ms->L, 3, &l); |
682 | for (i = 0; i < l; i++) { |
683 | if (news[i] != L_ESC) |
684 | luaL_addchar(b, news[i]); |
685 | else { |
686 | i++; /* skip ESC */ |
687 | if (!isdigit(uchar(news[i]))) { |
688 | if (news[i] != L_ESC) |
689 | luaL_error(ms->L, "invalid use of " LUA_QL("%c") |
690 | " in replacement string", L_ESC); |
691 | luaL_addchar(b, news[i]); |
692 | } |
693 | else if (news[i] == '0') |
694 | luaL_addlstring(b, s, e - s); |
695 | else { |
696 | push_onecapture(ms, news[i] - '1', s, e); |
697 | luaL_addvalue(b); /* add capture to accumulated result */ |
698 | } |
699 | } |
700 | } |
701 | } |
702 | |
703 | |
704 | static void add_value (MatchState *ms, luaL_Buffer *b, const char *s, |
705 | const char *e, int tr) { |
706 | lua_State *L = ms->L; |
707 | switch (tr) { |
708 | case LUA_TFUNCTION: { |
709 | int n; |
710 | lua_pushvalue(L, 3); |
711 | n = push_captures(ms, s, e); |
712 | lua_call(L, n, 1); |
713 | break; |
714 | } |
715 | case LUA_TTABLE: { |
716 | push_onecapture(ms, 0, s, e); |
717 | lua_gettable(L, 3); |
718 | break; |
719 | } |
720 | default: { /* LUA_TNUMBER or LUA_TSTRING */ |
721 | add_s(ms, b, s, e); |
722 | return; |
723 | } |
724 | } |
725 | if (!lua_toboolean(L, -1)) { /* nil or false? */ |
726 | lua_pop(L, 1); |
727 | lua_pushlstring(L, s, e - s); /* keep original text */ |
728 | } |
729 | else if (!lua_isstring(L, -1)) |
730 | luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1)); |
731 | luaL_addvalue(b); /* add result to accumulator */ |
732 | } |
733 | |
734 | |
735 | static int str_gsub (lua_State *L) { |
736 | size_t srcl, lp; |
737 | const char *src = luaL_checklstring(L, 1, &srcl); |
738 | const char *p = luaL_checklstring(L, 2, &lp); |
739 | int tr = lua_type(L, 3); |
740 | size_t max_s = luaL_optinteger(L, 4, srcl+1); |
741 | int anchor = (*p == '^'); |
742 | size_t n = 0; |
743 | MatchState ms; |
744 | luaL_Buffer b; |
745 | luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING || |
746 | tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3, |
747 | "string/function/table expected"); |
748 | luaL_buffinit(L, &b); |
749 | if (anchor) { |
750 | p++; lp--; /* skip anchor character */ |
751 | } |
752 | ms.L = L; |
753 | ms.matchdepth = MAXCCALLS; |
754 | ms.src_init = src; |
755 | ms.src_end = src+srcl; |
756 | ms.p_end = p + lp; |
757 | while (n < max_s) { |
758 | const char *e; |
759 | ms.level = 0; |
760 | lua_assert(ms.matchdepth == MAXCCALLS); |
761 | e = match(&ms, src, p); |
762 | if (e) { |
763 | n++; |
764 | add_value(&ms, &b, src, e, tr); |
765 | } |
766 | if (e && e>src) /* non empty match? */ |
767 | src = e; /* skip it */ |
768 | else if (src < ms.src_end) |
769 | luaL_addchar(&b, *src++); |
770 | else break; |
771 | if (anchor) break; |
772 | } |
773 | luaL_addlstring(&b, src, ms.src_end-src); |
774 | luaL_pushresult(&b); |
775 | lua_pushinteger(L, n); /* number of substitutions */ |
776 | return 2; |
777 | } |
778 | |
779 | /* }====================================================== */ |
780 | |
781 | |
782 | |
783 | /* |
784 | ** {====================================================== |
785 | ** STRING FORMAT |
786 | ** ======================================================= |
787 | */ |
788 | |
789 | /* |
790 | ** LUA_INTFRMLEN is the length modifier for integer conversions in |
791 | ** 'string.format'; LUA_INTFRM_T is the integer type corresponding to |
792 | ** the previous length |
793 | */ |
794 | #if !defined(LUA_INTFRMLEN) /* { */ |
795 | #if defined(LUA_USE_LONGLONG) |
796 | |
797 | #define LUA_INTFRMLEN "ll" |
798 | #define LUA_INTFRM_T long long |
799 | |
800 | #else |
801 | |
802 | #define LUA_INTFRMLEN "l" |
803 | #define LUA_INTFRM_T long |
804 | |
805 | #endif |
806 | #endif /* } */ |
807 | |
808 | |
809 | /* |
810 | ** LUA_FLTFRMLEN is the length modifier for float conversions in |
811 | ** 'string.format'; LUA_FLTFRM_T is the float type corresponding to |
812 | ** the previous length |
813 | */ |
814 | #if !defined(LUA_FLTFRMLEN) |
815 | |
816 | #define LUA_FLTFRMLEN "" |
817 | #define LUA_FLTFRM_T double |
818 | |
819 | #endif |
820 | |
821 | |
822 | /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */ |
823 | #define MAX_ITEM 512 |
824 | /* valid flags in a format specification */ |
825 | #define FLAGS "-+ #0" |
826 | /* |
827 | ** maximum size of each format specification (such as '%-099.99d') |
828 | ** (+10 accounts for %99.99x plus margin of error) |
829 | */ |
830 | #define MAX_FORMAT (sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10) |
831 | |
832 | |
833 | static void addquoted (lua_State *L, luaL_Buffer *b, int arg) { |
834 | size_t l; |
835 | const char *s = luaL_checklstring(L, arg, &l); |
836 | luaL_addchar(b, '"'); |
837 | while (l--) { |
838 | if (*s == '"' || *s == '\\' || *s == '\n') { |
839 | luaL_addchar(b, '\\'); |
840 | luaL_addchar(b, *s); |
841 | } |
842 | else if (*s == '\0' || iscntrl(uchar(*s))) { |
843 | char buff[10]; |
844 | if (!isdigit(uchar(*(s+1)))) |
845 | sprintf(buff, "\\%d", (int)uchar(*s)); |
846 | else |
847 | sprintf(buff, "\\%03d", (int)uchar(*s)); |
848 | luaL_addstring(b, buff); |
849 | } |
850 | else |
851 | luaL_addchar(b, *s); |
852 | s++; |
853 | } |
854 | luaL_addchar(b, '"'); |
855 | } |
856 | |
857 | static const char *scanformat (lua_State *L, const char *strfrmt, char *form) { |
858 | const char *p = strfrmt; |
859 | while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */ |
860 | if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char)) |
861 | luaL_error(L, "invalid format (repeated flags)"); |
862 | if (isdigit(uchar(*p))) p++; /* skip width */ |
863 | if (isdigit(uchar(*p))) p++; /* (2 digits at most) */ |
864 | if (*p == '.') { |
865 | p++; |
866 | if (isdigit(uchar(*p))) p++; /* skip precision */ |
867 | if (isdigit(uchar(*p))) p++; /* (2 digits at most) */ |
868 | } |
869 | if (isdigit(uchar(*p))) |
870 | luaL_error(L, "invalid format (width or precision too long)"); |
871 | *(form++) = '%'; |
872 | memcpy(form, strfrmt, (p - strfrmt + 1) * sizeof(char)); |
873 | form += p - strfrmt + 1; |
874 | *form = '\0'; |
875 | return p; |
876 | } |
877 | |
878 | |
879 | /* |
880 | ** add length modifier into formats |
881 | */ |
882 | static void addlenmod (char *form, const char *lenmod) { |
883 | size_t l = strlen(form); |
884 | size_t lm = strlen(lenmod); |
885 | char spec = form[l - 1]; |
886 | strcpy(form + l - 1, lenmod); |
887 | form[l + lm - 1] = spec; |
888 | form[l + lm] = '\0'; |
889 | } |
890 | |
891 | |
892 | static int str_format (lua_State *L) { |
893 | int top = lua_gettop(L); |
894 | int arg = 1; |
895 | size_t sfl; |
896 | const char *strfrmt = luaL_checklstring(L, arg, &sfl); |
897 | const char *strfrmt_end = strfrmt+sfl; |
898 | luaL_Buffer b; |
899 | luaL_buffinit(L, &b); |
900 | while (strfrmt < strfrmt_end) { |
901 | if (*strfrmt != L_ESC) |
902 | luaL_addchar(&b, *strfrmt++); |
903 | else if (*++strfrmt == L_ESC) |
904 | luaL_addchar(&b, *strfrmt++); /* %% */ |
905 | else { /* format item */ |
906 | char form[MAX_FORMAT]; /* to store the format (`%...') */ |
907 | char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */ |
908 | int nb = 0; /* number of bytes in added item */ |
909 | if (++arg > top) |
910 | luaL_argerror(L, arg, "no value"); |
911 | strfrmt = scanformat(L, strfrmt, form); |
912 | switch (*strfrmt++) { |
913 | case 'c': { |
914 | nb = sprintf(buff, form, luaL_checkint(L, arg)); |
915 | break; |
916 | } |
917 | case 'd': case 'i': { |
918 | lua_Number n = luaL_checknumber(L, arg); |
919 | LUA_INTFRM_T ni = (LUA_INTFRM_T)n; |
920 | lua_Number diff = n - (lua_Number)ni; |
921 | luaL_argcheck(L, -1 < diff && diff < 1, arg, |
922 | "not a number in proper range"); |
923 | addlenmod(form, LUA_INTFRMLEN); |
924 | nb = sprintf(buff, form, ni); |
925 | break; |
926 | } |
927 | case 'o': case 'u': case 'x': case 'X': { |
928 | lua_Number n = luaL_checknumber(L, arg); |
929 | unsigned LUA_INTFRM_T ni = (unsigned LUA_INTFRM_T)n; |
930 | lua_Number diff = n - (lua_Number)ni; |
931 | luaL_argcheck(L, -1 < diff && diff < 1, arg, |
932 | "not a non-negative number in proper range"); |
933 | addlenmod(form, LUA_INTFRMLEN); |
934 | nb = sprintf(buff, form, ni); |
935 | break; |
936 | } |
937 | case 'e': case 'E': case 'f': |
938 | #if defined(LUA_USE_AFORMAT) |
939 | case 'a': case 'A': |
940 | #endif |
941 | case 'g': case 'G': { |
942 | addlenmod(form, LUA_FLTFRMLEN); |
943 | nb = sprintf(buff, form, (LUA_FLTFRM_T)luaL_checknumber(L, arg)); |
944 | break; |
945 | } |
946 | case 'q': { |
947 | addquoted(L, &b, arg); |
948 | break; |
949 | } |
950 | case 's': { |
951 | size_t l; |
952 | const char *s = luaL_tolstring(L, arg, &l); |
953 | if (!strchr(form, '.') && l >= 100) { |
954 | /* no precision and string is too long to be formatted; |
955 | keep original string */ |
956 | luaL_addvalue(&b); |
957 | break; |
958 | } |
959 | else { |
960 | nb = sprintf(buff, form, s); |
961 | lua_pop(L, 1); /* remove result from 'luaL_tolstring' */ |
962 | break; |
963 | } |
964 | } |
965 | default: { /* also treat cases `pnLlh' */ |
966 | return luaL_error(L, "invalid option " LUA_QL("%%%c") " to " |
967 | LUA_QL("format"), *(strfrmt - 1)); |
968 | } |
969 | } |
970 | luaL_addsize(&b, nb); |
971 | } |
972 | } |
973 | luaL_pushresult(&b); |
974 | return 1; |
975 | } |
976 | |
977 | /* }====================================================== */ |
978 | |
979 | |
980 | static const luaL_Reg strlib[] = { |
981 | {"byte", str_byte}, |
982 | {"char", str_char}, |
983 | {"dump", str_dump}, |
984 | {"find", str_find}, |
985 | {"format", str_format}, |
986 | {"gmatch", gmatch}, |
987 | {"gsub", str_gsub}, |
988 | {"len", str_len}, |
989 | {"lower", str_lower}, |
990 | {"match", str_match}, |
991 | {"rep", str_rep}, |
992 | {"reverse", str_reverse}, |
993 | {"sub", str_sub}, |
994 | {"upper", str_upper}, |
995 | {NULL, NULL} |
996 | }; |
997 | |
998 | |
999 | static void createmetatable (lua_State *L) { |
1000 | lua_createtable(L, 0, 1); /* table to be metatable for strings */ |
1001 | lua_pushliteral(L, ""); /* dummy string */ |
1002 | lua_pushvalue(L, -2); /* copy table */ |
1003 | lua_setmetatable(L, -2); /* set table as metatable for strings */ |
1004 | lua_pop(L, 1); /* pop dummy string */ |
1005 | lua_pushvalue(L, -2); /* get string library */ |
1006 | lua_setfield(L, -2, "__index"); /* metatable.__index = string */ |
1007 | lua_pop(L, 1); /* pop metatable */ |
1008 | } |
1009 | |
1010 | |
1011 | /* |
1012 | ** Open string library |
1013 | */ |
1014 | LUAMOD_API int luaopen_string (lua_State *L) { |
1015 | luaL_newlib(L, strlib); |
1016 | createmetatable(L); |
1017 | return 1; |
1018 | } |
1019 | |