]>
Commit | Line | Data |
---|---|---|
7c673cae | 1 | /* |
11fdf7f2 | 2 | ** $Id: lstrlib.c,v 1.251 2016/05/20 14:13:21 roberto Exp $ |
7c673cae FG |
3 | ** Standard library for string operations and pattern-matching |
4 | ** See Copyright Notice in lua.h | |
5 | */ | |
6 | ||
7 | #define lstrlib_c | |
8 | #define LUA_LIB | |
9 | ||
10 | #include "lprefix.h" | |
11 | ||
12 | ||
13 | #include <ctype.h> | |
14 | #include <float.h> | |
15 | #include <limits.h> | |
11fdf7f2 | 16 | #include <locale.h> |
7c673cae FG |
17 | #include <stddef.h> |
18 | #include <stdio.h> | |
19 | #include <stdlib.h> | |
20 | #include <string.h> | |
21 | ||
22 | #include "lua.h" | |
23 | ||
24 | #include "lauxlib.h" | |
25 | #include "lualib.h" | |
26 | ||
27 | ||
28 | /* | |
29 | ** maximum number of captures that a pattern can do during | |
11fdf7f2 TL |
30 | ** pattern-matching. This limit is arbitrary, but must fit in |
31 | ** an unsigned char. | |
7c673cae FG |
32 | */ |
33 | #if !defined(LUA_MAXCAPTURES) | |
34 | #define LUA_MAXCAPTURES 32 | |
35 | #endif | |
36 | ||
37 | ||
38 | /* macro to 'unsign' a character */ | |
39 | #define uchar(c) ((unsigned char)(c)) | |
40 | ||
41 | ||
42 | /* | |
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'.) | |
45 | */ | |
11fdf7f2 TL |
46 | #define MAX_SIZET ((size_t)(~(size_t)0)) |
47 | ||
7c673cae | 48 | #define MAXSIZE \ |
11fdf7f2 | 49 | (sizeof(size_t) < sizeof(int) ? MAX_SIZET : (size_t)(INT_MAX)) |
7c673cae FG |
50 | |
51 | ||
52 | ||
53 | ||
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 */ | |
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 { | |
7c673cae FG |
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 */ | |
218 | lua_State *L; | |
11fdf7f2 TL |
219 | int matchdepth; /* control for recursive depth (to avoid C stack overflow) */ |
220 | unsigned char level; /* total number of captures (finished or unfinished) */ | |
7c673cae FG |
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 | ||
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 */ | |
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 | ||
11fdf7f2 TL |
591 | static void prepstate (MatchState *ms, lua_State *L, |
592 | const char *s, size_t ls, const char *p, size_t lp) { | |
593 | ms->L = L; | |
594 | ms->matchdepth = MAXCCALLS; | |
595 | ms->src_init = s; | |
596 | ms->src_end = s + ls; | |
597 | ms->p_end = p + lp; | |
598 | } | |
599 | ||
600 | ||
601 | static void reprepstate (MatchState *ms) { | |
602 | ms->level = 0; | |
603 | lua_assert(ms->matchdepth == MAXCCALLS); | |
604 | } | |
605 | ||
606 | ||
7c673cae FG |
607 | static int str_find_aux (lua_State *L, int find) { |
608 | size_t ls, lp; | |
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 */ | |
615 | return 1; | |
616 | } | |
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); | |
621 | if (s2) { | |
622 | lua_pushinteger(L, (s2 - s) + 1); | |
623 | lua_pushinteger(L, (s2 - s) + lp); | |
624 | return 2; | |
625 | } | |
626 | } | |
627 | else { | |
628 | MatchState ms; | |
629 | const char *s1 = s + init - 1; | |
630 | int anchor = (*p == '^'); | |
631 | if (anchor) { | |
632 | p++; lp--; /* skip anchor character */ | |
633 | } | |
11fdf7f2 | 634 | prepstate(&ms, L, s, ls, p, lp); |
7c673cae FG |
635 | do { |
636 | const char *res; | |
11fdf7f2 | 637 | reprepstate(&ms); |
7c673cae FG |
638 | if ((res=match(&ms, s1, p)) != NULL) { |
639 | if (find) { | |
640 | lua_pushinteger(L, (s1 - s) + 1); /* start */ | |
641 | lua_pushinteger(L, res - s); /* end */ | |
642 | return push_captures(&ms, NULL, 0) + 2; | |
643 | } | |
644 | else | |
645 | return push_captures(&ms, s1, res); | |
646 | } | |
647 | } while (s1++ < ms.src_end && !anchor); | |
648 | } | |
649 | lua_pushnil(L); /* not found */ | |
650 | return 1; | |
651 | } | |
652 | ||
653 | ||
654 | static int str_find (lua_State *L) { | |
655 | return str_find_aux(L, 1); | |
656 | } | |
657 | ||
658 | ||
659 | static int str_match (lua_State *L) { | |
660 | return str_find_aux(L, 0); | |
661 | } | |
662 | ||
663 | ||
11fdf7f2 TL |
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 */ | |
670 | } GMatchState; | |
671 | ||
672 | ||
7c673cae | 673 | static int gmatch_aux (lua_State *L) { |
11fdf7f2 | 674 | GMatchState *gm = (GMatchState *)lua_touserdata(L, lua_upvalueindex(3)); |
7c673cae | 675 | const char *src; |
11fdf7f2 TL |
676 | gm->ms.L = L; |
677 | for (src = gm->src; src <= gm->ms.src_end; src++) { | |
7c673cae | 678 | const char *e; |
11fdf7f2 TL |
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); | |
7c673cae FG |
683 | } |
684 | } | |
685 | return 0; /* not found */ | |
686 | } | |
687 | ||
688 | ||
689 | static int gmatch (lua_State *L) { | |
11fdf7f2 TL |
690 | size_t ls, lp; |
691 | const char *s = luaL_checklstring(L, 1, &ls); | |
692 | const char *p = luaL_checklstring(L, 2, &lp); | |
693 | GMatchState *gm; | |
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; | |
7c673cae FG |
698 | lua_pushcclosure(L, gmatch_aux, 3); |
699 | return 1; | |
700 | } | |
701 | ||
702 | ||
703 | static void add_s (MatchState *ms, luaL_Buffer *b, const char *s, | |
704 | const char *e) { | |
705 | size_t l, i; | |
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]); | |
711 | else { | |
712 | i++; /* skip ESC */ | |
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]); | |
717 | } | |
718 | else if (news[i] == '0') | |
719 | luaL_addlstring(b, s, e - s); | |
720 | else { | |
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 */ | |
725 | } | |
726 | } | |
727 | } | |
728 | } | |
729 | ||
730 | ||
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; | |
734 | switch (tr) { | |
735 | case LUA_TFUNCTION: { | |
736 | int n; | |
737 | lua_pushvalue(L, 3); | |
738 | n = push_captures(ms, s, e); | |
739 | lua_call(L, n, 1); | |
740 | break; | |
741 | } | |
742 | case LUA_TTABLE: { | |
743 | push_onecapture(ms, 0, s, e); | |
744 | lua_gettable(L, 3); | |
745 | break; | |
746 | } | |
747 | default: { /* LUA_TNUMBER or LUA_TSTRING */ | |
748 | add_s(ms, b, s, e); | |
749 | return; | |
750 | } | |
751 | } | |
752 | if (!lua_toboolean(L, -1)) { /* nil or false? */ | |
753 | lua_pop(L, 1); | |
754 | lua_pushlstring(L, s, e - s); /* keep original text */ | |
755 | } | |
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 */ | |
759 | } | |
760 | ||
761 | ||
762 | static int str_gsub (lua_State *L) { | |
763 | size_t srcl, lp; | |
11fdf7f2 TL |
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 */ | |
7c673cae | 769 | int anchor = (*p == '^'); |
11fdf7f2 | 770 | lua_Integer n = 0; /* replacement count */ |
7c673cae FG |
771 | MatchState ms; |
772 | luaL_Buffer b; | |
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); | |
777 | if (anchor) { | |
778 | p++; lp--; /* skip anchor character */ | |
779 | } | |
11fdf7f2 | 780 | prepstate(&ms, L, src, srcl, p, lp); |
7c673cae FG |
781 | while (n < max_s) { |
782 | const char *e; | |
11fdf7f2 TL |
783 | reprepstate(&ms); /* (re)prepare state for new match */ |
784 | if ((e = match(&ms, src, p)) != NULL && e != lastmatch) { /* match? */ | |
7c673cae | 785 | n++; |
11fdf7f2 TL |
786 | add_value(&ms, &b, src, e, tr); /* add replacement to buffer */ |
787 | src = lastmatch = e; | |
7c673cae | 788 | } |
11fdf7f2 | 789 | else if (src < ms.src_end) /* otherwise, skip one character */ |
7c673cae | 790 | luaL_addchar(&b, *src++); |
11fdf7f2 | 791 | else break; /* end of subject */ |
7c673cae FG |
792 | if (anchor) break; |
793 | } | |
794 | luaL_addlstring(&b, src, ms.src_end-src); | |
795 | luaL_pushresult(&b); | |
796 | lua_pushinteger(L, n); /* number of substitutions */ | |
797 | return 2; | |
798 | } | |
799 | ||
800 | /* }====================================================== */ | |
801 | ||
802 | ||
803 | ||
804 | /* | |
805 | ** {====================================================== | |
806 | ** STRING FORMAT | |
807 | ** ======================================================= | |
808 | */ | |
809 | ||
810 | #if !defined(lua_number2strx) /* { */ | |
811 | ||
812 | /* | |
813 | ** Hexadecimal floating-point formatter | |
814 | */ | |
815 | ||
7c673cae FG |
816 | #include <math.h> |
817 | ||
818 | #define SIZELENMOD (sizeof(LUA_NUMBER_FRMLEN)/sizeof(char)) | |
819 | ||
820 | ||
821 | /* | |
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 | |
825 | ** multiple of 4. | |
826 | */ | |
827 | #define L_NBFD ((l_mathlim(MANT_DIG) - 1)%4 + 1) | |
828 | ||
829 | ||
830 | /* | |
831 | ** Add integer part of 'x' to buffer and return new 'x' | |
832 | */ | |
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' */ | |
835 | int d = (int)dd; | |
836 | buff[n] = (d < 10 ? d + '0' : d - 10 + 'a'); /* add to buffer */ | |
837 | return x - dd; /* return what is left */ | |
838 | } | |
839 | ||
840 | ||
11fdf7f2 | 841 | static int num2straux (char *buff, int sz, lua_Number x) { |
7c673cae | 842 | if (x != x || x == HUGE_VAL || x == -HUGE_VAL) /* inf or NaN? */ |
11fdf7f2 | 843 | return l_sprintf(buff, sz, LUA_NUMBER_FMT, x); /* equal to '%g' */ |
7c673cae | 844 | else if (x == 0) { /* can be -0... */ |
11fdf7f2 TL |
845 | /* create "0" or "-0" followed by exponent */ |
846 | return l_sprintf(buff, sz, LUA_NUMBER_FMT "x0p+0", x); | |
7c673cae FG |
847 | } |
848 | else { | |
849 | int e; | |
850 | lua_Number m = l_mathop(frexp)(x, &e); /* 'x' fraction and exponent */ | |
851 | int n = 0; /* character count */ | |
852 | if (m < 0) { /* is number negative? */ | |
853 | buff[n++] = '-'; /* add signal */ | |
854 | m = -m; /* make it positive */ | |
855 | } | |
856 | buff[n++] = '0'; buff[n++] = 'x'; /* add "0x" */ | |
857 | m = adddigit(buff, n++, m * (1 << L_NBFD)); /* add first digit */ | |
858 | e -= L_NBFD; /* this digit goes before the radix point */ | |
859 | if (m > 0) { /* more digits? */ | |
860 | buff[n++] = lua_getlocaledecpoint(); /* add radix point */ | |
861 | do { /* add as many digits as needed */ | |
862 | m = adddigit(buff, n++, m * 16); | |
863 | } while (m > 0); | |
864 | } | |
11fdf7f2 TL |
865 | n += l_sprintf(buff + n, sz - n, "p%+d", e); /* add exponent */ |
866 | lua_assert(n < sz); | |
7c673cae FG |
867 | return n; |
868 | } | |
869 | } | |
870 | ||
871 | ||
11fdf7f2 TL |
872 | static int lua_number2strx (lua_State *L, char *buff, int sz, |
873 | const char *fmt, lua_Number x) { | |
874 | int n = num2straux(buff, sz, x); | |
7c673cae FG |
875 | if (fmt[SIZELENMOD] == 'A') { |
876 | int i; | |
877 | for (i = 0; i < n; i++) | |
878 | buff[i] = toupper(uchar(buff[i])); | |
879 | } | |
880 | else if (fmt[SIZELENMOD] != 'a') | |
881 | luaL_error(L, "modifiers for format '%%a'/'%%A' not implemented"); | |
882 | return n; | |
883 | } | |
884 | ||
885 | #endif /* } */ | |
886 | ||
887 | ||
888 | /* | |
889 | ** Maximum size of each formatted item. This maximum size is produced | |
11fdf7f2 TL |
890 | ** by format('%.99f', -maxfloat), and is equal to 99 + 3 ('-', '.', |
891 | ** and '\0') + number of decimal digits to represent maxfloat (which | |
892 | ** is maximum exponent + 1). (99+3+1 then rounded to 120 for "extra | |
893 | ** expenses", such as locale-dependent stuff) | |
7c673cae | 894 | */ |
11fdf7f2 | 895 | #define MAX_ITEM (120 + l_mathlim(MAX_10_EXP)) |
7c673cae FG |
896 | |
897 | ||
898 | /* valid flags in a format specification */ | |
899 | #define FLAGS "-+ #0" | |
900 | ||
901 | /* | |
902 | ** maximum size of each format specification (such as "%-099.99d") | |
903 | */ | |
904 | #define MAX_FORMAT 32 | |
905 | ||
906 | ||
11fdf7f2 | 907 | static void addquoted (luaL_Buffer *b, const char *s, size_t len) { |
7c673cae | 908 | luaL_addchar(b, '"'); |
11fdf7f2 | 909 | while (len--) { |
7c673cae FG |
910 | if (*s == '"' || *s == '\\' || *s == '\n') { |
911 | luaL_addchar(b, '\\'); | |
912 | luaL_addchar(b, *s); | |
913 | } | |
11fdf7f2 | 914 | else if (iscntrl(uchar(*s))) { |
7c673cae FG |
915 | char buff[10]; |
916 | if (!isdigit(uchar(*(s+1)))) | |
11fdf7f2 | 917 | l_sprintf(buff, sizeof(buff), "\\%d", (int)uchar(*s)); |
7c673cae | 918 | else |
11fdf7f2 | 919 | l_sprintf(buff, sizeof(buff), "\\%03d", (int)uchar(*s)); |
7c673cae FG |
920 | luaL_addstring(b, buff); |
921 | } | |
922 | else | |
923 | luaL_addchar(b, *s); | |
924 | s++; | |
925 | } | |
926 | luaL_addchar(b, '"'); | |
927 | } | |
928 | ||
11fdf7f2 TL |
929 | |
930 | /* | |
931 | ** Ensures the 'buff' string uses a dot as the radix character. | |
932 | */ | |
933 | static void checkdp (char *buff, int nb) { | |
934 | if (memchr(buff, '.', nb) == NULL) { /* no dot? */ | |
935 | char point = lua_getlocaledecpoint(); /* try locale point */ | |
936 | char *ppoint = memchr(buff, point, nb); | |
937 | if (ppoint) *ppoint = '.'; /* change it to a dot */ | |
938 | } | |
939 | } | |
940 | ||
941 | ||
942 | static void addliteral (lua_State *L, luaL_Buffer *b, int arg) { | |
943 | switch (lua_type(L, arg)) { | |
944 | case LUA_TSTRING: { | |
945 | size_t len; | |
946 | const char *s = lua_tolstring(L, arg, &len); | |
947 | addquoted(b, s, len); | |
948 | break; | |
949 | } | |
950 | case LUA_TNUMBER: { | |
951 | char *buff = luaL_prepbuffsize(b, MAX_ITEM); | |
952 | int nb; | |
953 | if (!lua_isinteger(L, arg)) { /* float? */ | |
954 | lua_Number n = lua_tonumber(L, arg); /* write as hexa ('%a') */ | |
955 | nb = lua_number2strx(L, buff, MAX_ITEM, "%" LUA_NUMBER_FRMLEN "a", n); | |
956 | checkdp(buff, nb); /* ensure it uses a dot */ | |
957 | } | |
958 | else { /* integers */ | |
959 | lua_Integer n = lua_tointeger(L, arg); | |
960 | const char *format = (n == LUA_MININTEGER) /* corner case? */ | |
961 | ? "0x%" LUA_INTEGER_FRMLEN "x" /* use hexa */ | |
962 | : LUA_INTEGER_FMT; /* else use default format */ | |
963 | nb = l_sprintf(buff, MAX_ITEM, format, n); | |
964 | } | |
965 | luaL_addsize(b, nb); | |
966 | break; | |
967 | } | |
968 | case LUA_TNIL: case LUA_TBOOLEAN: { | |
969 | luaL_tolstring(L, arg, NULL); | |
970 | luaL_addvalue(b); | |
971 | break; | |
972 | } | |
973 | default: { | |
974 | luaL_argerror(L, arg, "value has no literal form"); | |
975 | } | |
976 | } | |
977 | } | |
978 | ||
979 | ||
7c673cae FG |
980 | static const char *scanformat (lua_State *L, const char *strfrmt, char *form) { |
981 | const char *p = strfrmt; | |
982 | while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */ | |
983 | if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char)) | |
984 | luaL_error(L, "invalid format (repeated flags)"); | |
985 | if (isdigit(uchar(*p))) p++; /* skip width */ | |
986 | if (isdigit(uchar(*p))) p++; /* (2 digits at most) */ | |
987 | if (*p == '.') { | |
988 | p++; | |
989 | if (isdigit(uchar(*p))) p++; /* skip precision */ | |
990 | if (isdigit(uchar(*p))) p++; /* (2 digits at most) */ | |
991 | } | |
992 | if (isdigit(uchar(*p))) | |
993 | luaL_error(L, "invalid format (width or precision too long)"); | |
994 | *(form++) = '%'; | |
995 | memcpy(form, strfrmt, ((p - strfrmt) + 1) * sizeof(char)); | |
996 | form += (p - strfrmt) + 1; | |
997 | *form = '\0'; | |
998 | return p; | |
999 | } | |
1000 | ||
1001 | ||
1002 | /* | |
1003 | ** add length modifier into formats | |
1004 | */ | |
1005 | static void addlenmod (char *form, const char *lenmod) { | |
1006 | size_t l = strlen(form); | |
1007 | size_t lm = strlen(lenmod); | |
1008 | char spec = form[l - 1]; | |
1009 | strcpy(form + l - 1, lenmod); | |
1010 | form[l + lm - 1] = spec; | |
1011 | form[l + lm] = '\0'; | |
1012 | } | |
1013 | ||
1014 | ||
1015 | static int str_format (lua_State *L) { | |
1016 | int top = lua_gettop(L); | |
1017 | int arg = 1; | |
1018 | size_t sfl; | |
1019 | const char *strfrmt = luaL_checklstring(L, arg, &sfl); | |
1020 | const char *strfrmt_end = strfrmt+sfl; | |
1021 | luaL_Buffer b; | |
1022 | luaL_buffinit(L, &b); | |
1023 | while (strfrmt < strfrmt_end) { | |
1024 | if (*strfrmt != L_ESC) | |
1025 | luaL_addchar(&b, *strfrmt++); | |
1026 | else if (*++strfrmt == L_ESC) | |
1027 | luaL_addchar(&b, *strfrmt++); /* %% */ | |
1028 | else { /* format item */ | |
1029 | char form[MAX_FORMAT]; /* to store the format ('%...') */ | |
1030 | char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */ | |
1031 | int nb = 0; /* number of bytes in added item */ | |
1032 | if (++arg > top) | |
1033 | luaL_argerror(L, arg, "no value"); | |
1034 | strfrmt = scanformat(L, strfrmt, form); | |
1035 | switch (*strfrmt++) { | |
1036 | case 'c': { | |
11fdf7f2 | 1037 | nb = l_sprintf(buff, MAX_ITEM, form, (int)luaL_checkinteger(L, arg)); |
7c673cae FG |
1038 | break; |
1039 | } | |
1040 | case 'd': case 'i': | |
1041 | case 'o': case 'u': case 'x': case 'X': { | |
1042 | lua_Integer n = luaL_checkinteger(L, arg); | |
1043 | addlenmod(form, LUA_INTEGER_FRMLEN); | |
11fdf7f2 | 1044 | nb = l_sprintf(buff, MAX_ITEM, form, n); |
7c673cae FG |
1045 | break; |
1046 | } | |
1047 | case 'a': case 'A': | |
1048 | addlenmod(form, LUA_NUMBER_FRMLEN); | |
11fdf7f2 TL |
1049 | nb = lua_number2strx(L, buff, MAX_ITEM, form, |
1050 | luaL_checknumber(L, arg)); | |
7c673cae FG |
1051 | break; |
1052 | case 'e': case 'E': case 'f': | |
1053 | case 'g': case 'G': { | |
1054 | addlenmod(form, LUA_NUMBER_FRMLEN); | |
11fdf7f2 | 1055 | nb = l_sprintf(buff, MAX_ITEM, form, luaL_checknumber(L, arg)); |
7c673cae FG |
1056 | break; |
1057 | } | |
1058 | case 'q': { | |
11fdf7f2 | 1059 | addliteral(L, &b, arg); |
7c673cae FG |
1060 | break; |
1061 | } | |
1062 | case 's': { | |
1063 | size_t l; | |
1064 | const char *s = luaL_tolstring(L, arg, &l); | |
11fdf7f2 TL |
1065 | if (form[2] == '\0') /* no modifiers? */ |
1066 | luaL_addvalue(&b); /* keep entire string */ | |
7c673cae | 1067 | else { |
11fdf7f2 TL |
1068 | luaL_argcheck(L, l == strlen(s), arg, "string contains zeros"); |
1069 | if (!strchr(form, '.') && l >= 100) { | |
1070 | /* no precision and string is too long to be formatted */ | |
1071 | luaL_addvalue(&b); /* keep entire string */ | |
1072 | } | |
1073 | else { /* format the string into 'buff' */ | |
1074 | nb = l_sprintf(buff, MAX_ITEM, form, s); | |
1075 | lua_pop(L, 1); /* remove result from 'luaL_tolstring' */ | |
1076 | } | |
7c673cae FG |
1077 | } |
1078 | break; | |
1079 | } | |
1080 | default: { /* also treat cases 'pnLlh' */ | |
1081 | return luaL_error(L, "invalid option '%%%c' to 'format'", | |
1082 | *(strfrmt - 1)); | |
1083 | } | |
1084 | } | |
11fdf7f2 | 1085 | lua_assert(nb < MAX_ITEM); |
7c673cae FG |
1086 | luaL_addsize(&b, nb); |
1087 | } | |
1088 | } | |
1089 | luaL_pushresult(&b); | |
1090 | return 1; | |
1091 | } | |
1092 | ||
1093 | /* }====================================================== */ | |
1094 | ||
1095 | ||
1096 | /* | |
1097 | ** {====================================================== | |
1098 | ** PACK/UNPACK | |
1099 | ** ======================================================= | |
1100 | */ | |
1101 | ||
1102 | ||
1103 | /* value used for padding */ | |
11fdf7f2 TL |
1104 | #if !defined(LUAL_PACKPADBYTE) |
1105 | #define LUAL_PACKPADBYTE 0x00 | |
7c673cae FG |
1106 | #endif |
1107 | ||
1108 | /* maximum size for the binary representation of an integer */ | |
1109 | #define MAXINTSIZE 16 | |
1110 | ||
1111 | /* number of bits in a character */ | |
1112 | #define NB CHAR_BIT | |
1113 | ||
1114 | /* mask for one character (NB 1's) */ | |
1115 | #define MC ((1 << NB) - 1) | |
1116 | ||
1117 | /* size of a lua_Integer */ | |
1118 | #define SZINT ((int)sizeof(lua_Integer)) | |
1119 | ||
1120 | ||
1121 | /* dummy union to get native endianness */ | |
1122 | static const union { | |
1123 | int dummy; | |
1124 | char little; /* true iff machine is little endian */ | |
1125 | } nativeendian = {1}; | |
1126 | ||
1127 | ||
1128 | /* dummy structure to get native alignment requirements */ | |
1129 | struct cD { | |
1130 | char c; | |
1131 | union { double d; void *p; lua_Integer i; lua_Number n; } u; | |
1132 | }; | |
1133 | ||
1134 | #define MAXALIGN (offsetof(struct cD, u)) | |
1135 | ||
1136 | ||
1137 | /* | |
1138 | ** Union for serializing floats | |
1139 | */ | |
1140 | typedef union Ftypes { | |
1141 | float f; | |
1142 | double d; | |
1143 | lua_Number n; | |
1144 | char buff[5 * sizeof(lua_Number)]; /* enough for any float type */ | |
1145 | } Ftypes; | |
1146 | ||
1147 | ||
1148 | /* | |
1149 | ** information to pack/unpack stuff | |
1150 | */ | |
1151 | typedef struct Header { | |
1152 | lua_State *L; | |
1153 | int islittle; | |
1154 | int maxalign; | |
1155 | } Header; | |
1156 | ||
1157 | ||
1158 | /* | |
1159 | ** options for pack/unpack | |
1160 | */ | |
1161 | typedef enum KOption { | |
1162 | Kint, /* signed integers */ | |
1163 | Kuint, /* unsigned integers */ | |
1164 | Kfloat, /* floating-point numbers */ | |
1165 | Kchar, /* fixed-length strings */ | |
1166 | Kstring, /* strings with prefixed length */ | |
1167 | Kzstr, /* zero-terminated strings */ | |
1168 | Kpadding, /* padding */ | |
1169 | Kpaddalign, /* padding for alignment */ | |
1170 | Knop /* no-op (configuration or spaces) */ | |
1171 | } KOption; | |
1172 | ||
1173 | ||
1174 | /* | |
1175 | ** Read an integer numeral from string 'fmt' or return 'df' if | |
1176 | ** there is no numeral | |
1177 | */ | |
1178 | static int digit (int c) { return '0' <= c && c <= '9'; } | |
1179 | ||
1180 | static int getnum (const char **fmt, int df) { | |
1181 | if (!digit(**fmt)) /* no number? */ | |
1182 | return df; /* return default value */ | |
1183 | else { | |
1184 | int a = 0; | |
1185 | do { | |
1186 | a = a*10 + (*((*fmt)++) - '0'); | |
1187 | } while (digit(**fmt) && a <= ((int)MAXSIZE - 9)/10); | |
1188 | return a; | |
1189 | } | |
1190 | } | |
1191 | ||
1192 | ||
1193 | /* | |
1194 | ** Read an integer numeral and raises an error if it is larger | |
1195 | ** than the maximum size for integers. | |
1196 | */ | |
1197 | static int getnumlimit (Header *h, const char **fmt, int df) { | |
1198 | int sz = getnum(fmt, df); | |
1199 | if (sz > MAXINTSIZE || sz <= 0) | |
1200 | luaL_error(h->L, "integral size (%d) out of limits [1,%d]", | |
1201 | sz, MAXINTSIZE); | |
1202 | return sz; | |
1203 | } | |
1204 | ||
1205 | ||
1206 | /* | |
1207 | ** Initialize Header | |
1208 | */ | |
1209 | static void initheader (lua_State *L, Header *h) { | |
1210 | h->L = L; | |
1211 | h->islittle = nativeendian.little; | |
1212 | h->maxalign = 1; | |
1213 | } | |
1214 | ||
1215 | ||
1216 | /* | |
1217 | ** Read and classify next option. 'size' is filled with option's size. | |
1218 | */ | |
1219 | static KOption getoption (Header *h, const char **fmt, int *size) { | |
1220 | int opt = *((*fmt)++); | |
1221 | *size = 0; /* default */ | |
1222 | switch (opt) { | |
1223 | case 'b': *size = sizeof(char); return Kint; | |
1224 | case 'B': *size = sizeof(char); return Kuint; | |
1225 | case 'h': *size = sizeof(short); return Kint; | |
1226 | case 'H': *size = sizeof(short); return Kuint; | |
1227 | case 'l': *size = sizeof(long); return Kint; | |
1228 | case 'L': *size = sizeof(long); return Kuint; | |
1229 | case 'j': *size = sizeof(lua_Integer); return Kint; | |
1230 | case 'J': *size = sizeof(lua_Integer); return Kuint; | |
1231 | case 'T': *size = sizeof(size_t); return Kuint; | |
1232 | case 'f': *size = sizeof(float); return Kfloat; | |
1233 | case 'd': *size = sizeof(double); return Kfloat; | |
1234 | case 'n': *size = sizeof(lua_Number); return Kfloat; | |
1235 | case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint; | |
1236 | case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint; | |
1237 | case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring; | |
1238 | case 'c': | |
1239 | *size = getnum(fmt, -1); | |
1240 | if (*size == -1) | |
1241 | luaL_error(h->L, "missing size for format option 'c'"); | |
1242 | return Kchar; | |
1243 | case 'z': return Kzstr; | |
1244 | case 'x': *size = 1; return Kpadding; | |
1245 | case 'X': return Kpaddalign; | |
1246 | case ' ': break; | |
1247 | case '<': h->islittle = 1; break; | |
1248 | case '>': h->islittle = 0; break; | |
1249 | case '=': h->islittle = nativeendian.little; break; | |
1250 | case '!': h->maxalign = getnumlimit(h, fmt, MAXALIGN); break; | |
1251 | default: luaL_error(h->L, "invalid format option '%c'", opt); | |
1252 | } | |
1253 | return Knop; | |
1254 | } | |
1255 | ||
1256 | ||
1257 | /* | |
1258 | ** Read, classify, and fill other details about the next option. | |
1259 | ** 'psize' is filled with option's size, 'notoalign' with its | |
1260 | ** alignment requirements. | |
1261 | ** Local variable 'size' gets the size to be aligned. (Kpadal option | |
1262 | ** always gets its full alignment, other options are limited by | |
1263 | ** the maximum alignment ('maxalign'). Kchar option needs no alignment | |
1264 | ** despite its size. | |
1265 | */ | |
1266 | static KOption getdetails (Header *h, size_t totalsize, | |
1267 | const char **fmt, int *psize, int *ntoalign) { | |
1268 | KOption opt = getoption(h, fmt, psize); | |
1269 | int align = *psize; /* usually, alignment follows size */ | |
1270 | if (opt == Kpaddalign) { /* 'X' gets alignment from following option */ | |
1271 | if (**fmt == '\0' || getoption(h, fmt, &align) == Kchar || align == 0) | |
1272 | luaL_argerror(h->L, 1, "invalid next option for option 'X'"); | |
1273 | } | |
1274 | if (align <= 1 || opt == Kchar) /* need no alignment? */ | |
1275 | *ntoalign = 0; | |
1276 | else { | |
1277 | if (align > h->maxalign) /* enforce maximum alignment */ | |
1278 | align = h->maxalign; | |
1279 | if ((align & (align - 1)) != 0) /* is 'align' not a power of 2? */ | |
1280 | luaL_argerror(h->L, 1, "format asks for alignment not power of 2"); | |
1281 | *ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1); | |
1282 | } | |
1283 | return opt; | |
1284 | } | |
1285 | ||
1286 | ||
1287 | /* | |
1288 | ** Pack integer 'n' with 'size' bytes and 'islittle' endianness. | |
1289 | ** The final 'if' handles the case when 'size' is larger than | |
1290 | ** the size of a Lua integer, correcting the extra sign-extension | |
1291 | ** bytes if necessary (by default they would be zeros). | |
1292 | */ | |
1293 | static void packint (luaL_Buffer *b, lua_Unsigned n, | |
1294 | int islittle, int size, int neg) { | |
1295 | char *buff = luaL_prepbuffsize(b, size); | |
1296 | int i; | |
1297 | buff[islittle ? 0 : size - 1] = (char)(n & MC); /* first byte */ | |
1298 | for (i = 1; i < size; i++) { | |
1299 | n >>= NB; | |
1300 | buff[islittle ? i : size - 1 - i] = (char)(n & MC); | |
1301 | } | |
1302 | if (neg && size > SZINT) { /* negative number need sign extension? */ | |
1303 | for (i = SZINT; i < size; i++) /* correct extra bytes */ | |
1304 | buff[islittle ? i : size - 1 - i] = (char)MC; | |
1305 | } | |
1306 | luaL_addsize(b, size); /* add result to buffer */ | |
1307 | } | |
1308 | ||
1309 | ||
1310 | /* | |
1311 | ** Copy 'size' bytes from 'src' to 'dest', correcting endianness if | |
1312 | ** given 'islittle' is different from native endianness. | |
1313 | */ | |
1314 | static void copywithendian (volatile char *dest, volatile const char *src, | |
1315 | int size, int islittle) { | |
1316 | if (islittle == nativeendian.little) { | |
1317 | while (size-- != 0) | |
1318 | *(dest++) = *(src++); | |
1319 | } | |
1320 | else { | |
1321 | dest += size - 1; | |
1322 | while (size-- != 0) | |
1323 | *(dest--) = *(src++); | |
1324 | } | |
1325 | } | |
1326 | ||
1327 | ||
1328 | static int str_pack (lua_State *L) { | |
1329 | luaL_Buffer b; | |
1330 | Header h; | |
1331 | const char *fmt = luaL_checkstring(L, 1); /* format string */ | |
1332 | int arg = 1; /* current argument to pack */ | |
1333 | size_t totalsize = 0; /* accumulate total size of result */ | |
1334 | initheader(L, &h); | |
1335 | lua_pushnil(L); /* mark to separate arguments from string buffer */ | |
1336 | luaL_buffinit(L, &b); | |
1337 | while (*fmt != '\0') { | |
1338 | int size, ntoalign; | |
1339 | KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign); | |
1340 | totalsize += ntoalign + size; | |
1341 | while (ntoalign-- > 0) | |
11fdf7f2 | 1342 | luaL_addchar(&b, LUAL_PACKPADBYTE); /* fill alignment */ |
7c673cae FG |
1343 | arg++; |
1344 | switch (opt) { | |
1345 | case Kint: { /* signed integers */ | |
1346 | lua_Integer n = luaL_checkinteger(L, arg); | |
1347 | if (size < SZINT) { /* need overflow check? */ | |
1348 | lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1); | |
1349 | luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow"); | |
1350 | } | |
1351 | packint(&b, (lua_Unsigned)n, h.islittle, size, (n < 0)); | |
1352 | break; | |
1353 | } | |
1354 | case Kuint: { /* unsigned integers */ | |
1355 | lua_Integer n = luaL_checkinteger(L, arg); | |
1356 | if (size < SZINT) /* need overflow check? */ | |
1357 | luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)), | |
1358 | arg, "unsigned overflow"); | |
1359 | packint(&b, (lua_Unsigned)n, h.islittle, size, 0); | |
1360 | break; | |
1361 | } | |
1362 | case Kfloat: { /* floating-point options */ | |
1363 | volatile Ftypes u; | |
1364 | char *buff = luaL_prepbuffsize(&b, size); | |
1365 | lua_Number n = luaL_checknumber(L, arg); /* get argument */ | |
1366 | if (size == sizeof(u.f)) u.f = (float)n; /* copy it into 'u' */ | |
1367 | else if (size == sizeof(u.d)) u.d = (double)n; | |
1368 | else u.n = n; | |
1369 | /* move 'u' to final result, correcting endianness if needed */ | |
1370 | copywithendian(buff, u.buff, size, h.islittle); | |
1371 | luaL_addsize(&b, size); | |
1372 | break; | |
1373 | } | |
1374 | case Kchar: { /* fixed-size string */ | |
1375 | size_t len; | |
1376 | const char *s = luaL_checklstring(L, arg, &len); | |
11fdf7f2 TL |
1377 | luaL_argcheck(L, len <= (size_t)size, arg, |
1378 | "string longer than given size"); | |
1379 | luaL_addlstring(&b, s, len); /* add string */ | |
1380 | while (len++ < (size_t)size) /* pad extra space */ | |
1381 | luaL_addchar(&b, LUAL_PACKPADBYTE); | |
7c673cae FG |
1382 | break; |
1383 | } | |
1384 | case Kstring: { /* strings with length count */ | |
1385 | size_t len; | |
1386 | const char *s = luaL_checklstring(L, arg, &len); | |
1387 | luaL_argcheck(L, size >= (int)sizeof(size_t) || | |
1388 | len < ((size_t)1 << (size * NB)), | |
1389 | arg, "string length does not fit in given size"); | |
1390 | packint(&b, (lua_Unsigned)len, h.islittle, size, 0); /* pack length */ | |
1391 | luaL_addlstring(&b, s, len); | |
1392 | totalsize += len; | |
1393 | break; | |
1394 | } | |
1395 | case Kzstr: { /* zero-terminated string */ | |
1396 | size_t len; | |
1397 | const char *s = luaL_checklstring(L, arg, &len); | |
1398 | luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros"); | |
1399 | luaL_addlstring(&b, s, len); | |
1400 | luaL_addchar(&b, '\0'); /* add zero at the end */ | |
1401 | totalsize += len + 1; | |
1402 | break; | |
1403 | } | |
11fdf7f2 | 1404 | case Kpadding: luaL_addchar(&b, LUAL_PACKPADBYTE); /* FALLTHROUGH */ |
7c673cae FG |
1405 | case Kpaddalign: case Knop: |
1406 | arg--; /* undo increment */ | |
1407 | break; | |
1408 | } | |
1409 | } | |
1410 | luaL_pushresult(&b); | |
1411 | return 1; | |
1412 | } | |
1413 | ||
1414 | ||
1415 | static int str_packsize (lua_State *L) { | |
1416 | Header h; | |
1417 | const char *fmt = luaL_checkstring(L, 1); /* format string */ | |
1418 | size_t totalsize = 0; /* accumulate total size of result */ | |
1419 | initheader(L, &h); | |
1420 | while (*fmt != '\0') { | |
1421 | int size, ntoalign; | |
1422 | KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign); | |
1423 | size += ntoalign; /* total space used by option */ | |
1424 | luaL_argcheck(L, totalsize <= MAXSIZE - size, 1, | |
1425 | "format result too large"); | |
1426 | totalsize += size; | |
1427 | switch (opt) { | |
1428 | case Kstring: /* strings with length count */ | |
1429 | case Kzstr: /* zero-terminated string */ | |
1430 | luaL_argerror(L, 1, "variable-length format"); | |
11fdf7f2 | 1431 | /* call never return, but to avoid warnings: *//* FALLTHROUGH */ |
7c673cae FG |
1432 | default: break; |
1433 | } | |
1434 | } | |
1435 | lua_pushinteger(L, (lua_Integer)totalsize); | |
1436 | return 1; | |
1437 | } | |
1438 | ||
1439 | ||
1440 | /* | |
1441 | ** Unpack an integer with 'size' bytes and 'islittle' endianness. | |
1442 | ** If size is smaller than the size of a Lua integer and integer | |
1443 | ** is signed, must do sign extension (propagating the sign to the | |
1444 | ** higher bits); if size is larger than the size of a Lua integer, | |
1445 | ** it must check the unread bytes to see whether they do not cause an | |
1446 | ** overflow. | |
1447 | */ | |
1448 | static lua_Integer unpackint (lua_State *L, const char *str, | |
1449 | int islittle, int size, int issigned) { | |
1450 | lua_Unsigned res = 0; | |
1451 | int i; | |
1452 | int limit = (size <= SZINT) ? size : SZINT; | |
1453 | for (i = limit - 1; i >= 0; i--) { | |
1454 | res <<= NB; | |
1455 | res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i]; | |
1456 | } | |
1457 | if (size < SZINT) { /* real size smaller than lua_Integer? */ | |
1458 | if (issigned) { /* needs sign extension? */ | |
1459 | lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1); | |
1460 | res = ((res ^ mask) - mask); /* do sign extension */ | |
1461 | } | |
1462 | } | |
1463 | else if (size > SZINT) { /* must check unread bytes */ | |
1464 | int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC; | |
1465 | for (i = limit; i < size; i++) { | |
1466 | if ((unsigned char)str[islittle ? i : size - 1 - i] != mask) | |
1467 | luaL_error(L, "%d-byte integer does not fit into Lua Integer", size); | |
1468 | } | |
1469 | } | |
1470 | return (lua_Integer)res; | |
1471 | } | |
1472 | ||
1473 | ||
1474 | static int str_unpack (lua_State *L) { | |
1475 | Header h; | |
1476 | const char *fmt = luaL_checkstring(L, 1); | |
1477 | size_t ld; | |
1478 | const char *data = luaL_checklstring(L, 2, &ld); | |
1479 | size_t pos = (size_t)posrelat(luaL_optinteger(L, 3, 1), ld) - 1; | |
1480 | int n = 0; /* number of results */ | |
1481 | luaL_argcheck(L, pos <= ld, 3, "initial position out of string"); | |
1482 | initheader(L, &h); | |
1483 | while (*fmt != '\0') { | |
1484 | int size, ntoalign; | |
1485 | KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign); | |
1486 | if ((size_t)ntoalign + size > ~pos || pos + ntoalign + size > ld) | |
1487 | luaL_argerror(L, 2, "data string too short"); | |
1488 | pos += ntoalign; /* skip alignment */ | |
1489 | /* stack space for item + next position */ | |
1490 | luaL_checkstack(L, 2, "too many results"); | |
1491 | n++; | |
1492 | switch (opt) { | |
1493 | case Kint: | |
1494 | case Kuint: { | |
1495 | lua_Integer res = unpackint(L, data + pos, h.islittle, size, | |
1496 | (opt == Kint)); | |
1497 | lua_pushinteger(L, res); | |
1498 | break; | |
1499 | } | |
1500 | case Kfloat: { | |
1501 | volatile Ftypes u; | |
1502 | lua_Number num; | |
1503 | copywithendian(u.buff, data + pos, size, h.islittle); | |
1504 | if (size == sizeof(u.f)) num = (lua_Number)u.f; | |
1505 | else if (size == sizeof(u.d)) num = (lua_Number)u.d; | |
1506 | else num = u.n; | |
1507 | lua_pushnumber(L, num); | |
1508 | break; | |
1509 | } | |
1510 | case Kchar: { | |
1511 | lua_pushlstring(L, data + pos, size); | |
1512 | break; | |
1513 | } | |
1514 | case Kstring: { | |
1515 | size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0); | |
1516 | luaL_argcheck(L, pos + len + size <= ld, 2, "data string too short"); | |
1517 | lua_pushlstring(L, data + pos + size, len); | |
1518 | pos += len; /* skip string */ | |
1519 | break; | |
1520 | } | |
1521 | case Kzstr: { | |
1522 | size_t len = (int)strlen(data + pos); | |
1523 | lua_pushlstring(L, data + pos, len); | |
1524 | pos += len + 1; /* skip string plus final '\0' */ | |
1525 | break; | |
1526 | } | |
1527 | case Kpaddalign: case Kpadding: case Knop: | |
1528 | n--; /* undo increment */ | |
1529 | break; | |
1530 | } | |
1531 | pos += size; | |
1532 | } | |
1533 | lua_pushinteger(L, pos + 1); /* next position */ | |
1534 | return n + 1; | |
1535 | } | |
1536 | ||
1537 | /* }====================================================== */ | |
1538 | ||
1539 | ||
1540 | static const luaL_Reg strlib[] = { | |
1541 | {"byte", str_byte}, | |
1542 | {"char", str_char}, | |
1543 | {"dump", str_dump}, | |
1544 | {"find", str_find}, | |
1545 | {"format", str_format}, | |
1546 | {"gmatch", gmatch}, | |
1547 | {"gsub", str_gsub}, | |
1548 | {"len", str_len}, | |
1549 | {"lower", str_lower}, | |
1550 | {"match", str_match}, | |
1551 | {"rep", str_rep}, | |
1552 | {"reverse", str_reverse}, | |
1553 | {"sub", str_sub}, | |
1554 | {"upper", str_upper}, | |
1555 | {"pack", str_pack}, | |
1556 | {"packsize", str_packsize}, | |
1557 | {"unpack", str_unpack}, | |
1558 | {NULL, NULL} | |
1559 | }; | |
1560 | ||
1561 | ||
1562 | static void createmetatable (lua_State *L) { | |
1563 | lua_createtable(L, 0, 1); /* table to be metatable for strings */ | |
1564 | lua_pushliteral(L, ""); /* dummy string */ | |
1565 | lua_pushvalue(L, -2); /* copy table */ | |
1566 | lua_setmetatable(L, -2); /* set table as metatable for strings */ | |
1567 | lua_pop(L, 1); /* pop dummy string */ | |
1568 | lua_pushvalue(L, -2); /* get string library */ | |
1569 | lua_setfield(L, -2, "__index"); /* metatable.__index = string */ | |
1570 | lua_pop(L, 1); /* pop metatable */ | |
1571 | } | |
1572 | ||
1573 | ||
1574 | /* | |
1575 | ** Open string library | |
1576 | */ | |
1577 | LUAMOD_API int luaopen_string (lua_State *L) { | |
1578 | luaL_newlib(L, strlib); | |
1579 | createmetatable(L); | |
1580 | return 1; | |
1581 | } | |
1582 |