]>
Commit | Line | Data |
---|---|---|
16a5fed6 | 1 | /*\r |
2 | ** $Id: lstrlib.c,v 1.178.1.1 2013/04/12 18:48:47 roberto Exp $\r | |
3 | ** Standard library for string operations and pattern-matching\r | |
4 | ** See Copyright Notice in lua.h\r | |
5 | */\r | |
6 | \r | |
7 | \r | |
8 | #include <ctype.h>\r | |
9 | #include <stddef.h>\r | |
10 | #include <stdio.h>\r | |
11 | #include <stdlib.h>\r | |
12 | #include <string.h>\r | |
13 | \r | |
14 | #define lstrlib_c\r | |
15 | #define LUA_LIB\r | |
16 | \r | |
17 | #include "lua.h"\r | |
18 | \r | |
19 | #include "lauxlib.h"\r | |
20 | #include "lualib.h"\r | |
21 | \r | |
22 | \r | |
23 | /*\r | |
24 | ** maximum number of captures that a pattern can do during\r | |
25 | ** pattern-matching. This limit is arbitrary.\r | |
26 | */\r | |
27 | #if !defined(LUA_MAXCAPTURES)\r | |
28 | #define LUA_MAXCAPTURES 32\r | |
29 | #endif\r | |
30 | \r | |
31 | \r | |
32 | /* macro to `unsign' a character */\r | |
33 | #define uchar(c) ((unsigned char)(c))\r | |
34 | \r | |
35 | \r | |
36 | \r | |
37 | static int str_len (lua_State *L) {\r | |
38 | size_t l;\r | |
39 | luaL_checklstring(L, 1, &l);\r | |
40 | lua_pushinteger(L, (lua_Integer)l);\r | |
41 | return 1;\r | |
42 | }\r | |
43 | \r | |
44 | \r | |
45 | /* translate a relative string position: negative means back from end */\r | |
46 | static size_t posrelat (ptrdiff_t pos, size_t len) {\r | |
47 | if (pos >= 0) return (size_t)pos;\r | |
48 | else if (0u - (size_t)pos > len) return 0;\r | |
49 | else return len - ((size_t)-pos) + 1;\r | |
50 | }\r | |
51 | \r | |
52 | \r | |
53 | static int str_sub (lua_State *L) {\r | |
54 | size_t l;\r | |
55 | const char *s = luaL_checklstring(L, 1, &l);\r | |
56 | size_t start = posrelat(luaL_checkinteger(L, 2), l);\r | |
57 | size_t end = posrelat(luaL_optinteger(L, 3, -1), l);\r | |
58 | if (start < 1) start = 1;\r | |
59 | if (end > l) end = l;\r | |
60 | if (start <= end)\r | |
61 | lua_pushlstring(L, s + start - 1, end - start + 1);\r | |
62 | else lua_pushliteral(L, "");\r | |
63 | return 1;\r | |
64 | }\r | |
65 | \r | |
66 | \r | |
67 | static int str_reverse (lua_State *L) {\r | |
68 | size_t l, i;\r | |
69 | luaL_Buffer b;\r | |
70 | const char *s = luaL_checklstring(L, 1, &l);\r | |
71 | char *p = luaL_buffinitsize(L, &b, l);\r | |
72 | for (i = 0; i < l; i++)\r | |
73 | p[i] = s[l - i - 1];\r | |
74 | luaL_pushresultsize(&b, l);\r | |
75 | return 1;\r | |
76 | }\r | |
77 | \r | |
78 | \r | |
79 | static int str_lower (lua_State *L) {\r | |
80 | size_t l;\r | |
81 | size_t i;\r | |
82 | luaL_Buffer b;\r | |
83 | const char *s = luaL_checklstring(L, 1, &l);\r | |
84 | char *p = luaL_buffinitsize(L, &b, l);\r | |
85 | for (i=0; i<l; i++)\r | |
86 | p[i] = tolower(uchar(s[i]));\r | |
87 | luaL_pushresultsize(&b, l);\r | |
88 | return 1;\r | |
89 | }\r | |
90 | \r | |
91 | \r | |
92 | static int str_upper (lua_State *L) {\r | |
93 | size_t l;\r | |
94 | size_t i;\r | |
95 | luaL_Buffer b;\r | |
96 | const char *s = luaL_checklstring(L, 1, &l);\r | |
97 | char *p = luaL_buffinitsize(L, &b, l);\r | |
98 | for (i=0; i<l; i++)\r | |
99 | p[i] = toupper(uchar(s[i]));\r | |
100 | luaL_pushresultsize(&b, l);\r | |
101 | return 1;\r | |
102 | }\r | |
103 | \r | |
104 | \r | |
105 | /* reasonable limit to avoid arithmetic overflow */\r | |
106 | #define MAXSIZE ((~(size_t)0) >> 1)\r | |
107 | \r | |
108 | static int str_rep (lua_State *L) {\r | |
109 | size_t l, lsep;\r | |
110 | const char *s = luaL_checklstring(L, 1, &l);\r | |
111 | int n = luaL_checkint(L, 2);\r | |
112 | const char *sep = luaL_optlstring(L, 3, "", &lsep);\r | |
113 | if (n <= 0) lua_pushliteral(L, "");\r | |
114 | else if (l + lsep < l || l + lsep >= MAXSIZE / n) /* may overflow? */\r | |
115 | return luaL_error(L, "resulting string too large");\r | |
116 | else {\r | |
117 | size_t totallen = n * l + (n - 1) * lsep;\r | |
118 | luaL_Buffer b;\r | |
119 | char *p = luaL_buffinitsize(L, &b, totallen);\r | |
120 | while (n-- > 1) { /* first n-1 copies (followed by separator) */\r | |
121 | memcpy(p, s, l * sizeof(char)); p += l;\r | |
122 | if (lsep > 0) { /* avoid empty 'memcpy' (may be expensive) */\r | |
123 | memcpy(p, sep, lsep * sizeof(char)); p += lsep;\r | |
124 | }\r | |
125 | }\r | |
126 | memcpy(p, s, l * sizeof(char)); /* last copy (not followed by separator) */\r | |
127 | luaL_pushresultsize(&b, totallen);\r | |
128 | }\r | |
129 | return 1;\r | |
130 | }\r | |
131 | \r | |
132 | \r | |
133 | static int str_byte (lua_State *L) {\r | |
134 | size_t l;\r | |
135 | const char *s = luaL_checklstring(L, 1, &l);\r | |
136 | size_t posi = posrelat(luaL_optinteger(L, 2, 1), l);\r | |
137 | size_t pose = posrelat(luaL_optinteger(L, 3, posi), l);\r | |
138 | int n, i;\r | |
139 | if (posi < 1) posi = 1;\r | |
140 | if (pose > l) pose = l;\r | |
141 | if (posi > pose) return 0; /* empty interval; return no values */\r | |
142 | n = (int)(pose - posi + 1);\r | |
143 | if (posi + n <= pose) /* (size_t -> int) overflow? */\r | |
144 | return luaL_error(L, "string slice too long");\r | |
145 | luaL_checkstack(L, n, "string slice too long");\r | |
146 | for (i=0; i<n; i++)\r | |
147 | lua_pushinteger(L, uchar(s[posi+i-1]));\r | |
148 | return n;\r | |
149 | }\r | |
150 | \r | |
151 | \r | |
152 | static int str_char (lua_State *L) {\r | |
153 | int n = lua_gettop(L); /* number of arguments */\r | |
154 | int i;\r | |
155 | luaL_Buffer b;\r | |
156 | char *p = luaL_buffinitsize(L, &b, n);\r | |
157 | for (i=1; i<=n; i++) {\r | |
158 | int c = luaL_checkint(L, i);\r | |
159 | luaL_argcheck(L, uchar(c) == c, i, "value out of range");\r | |
160 | p[i - 1] = uchar(c);\r | |
161 | }\r | |
162 | luaL_pushresultsize(&b, n);\r | |
163 | return 1;\r | |
164 | }\r | |
165 | \r | |
166 | \r | |
167 | static int writer (lua_State *L, const void* b, size_t size, void* B) {\r | |
168 | (void)L;\r | |
169 | luaL_addlstring((luaL_Buffer*) B, (const char *)b, size);\r | |
170 | return 0;\r | |
171 | }\r | |
172 | \r | |
173 | \r | |
174 | static int str_dump (lua_State *L) {\r | |
175 | luaL_Buffer b;\r | |
176 | luaL_checktype(L, 1, LUA_TFUNCTION);\r | |
177 | lua_settop(L, 1);\r | |
178 | luaL_buffinit(L,&b);\r | |
179 | if (lua_dump(L, writer, &b) != 0)\r | |
180 | return luaL_error(L, "unable to dump given function");\r | |
181 | luaL_pushresult(&b);\r | |
182 | return 1;\r | |
183 | }\r | |
184 | \r | |
185 | \r | |
186 | \r | |
187 | /*\r | |
188 | ** {======================================================\r | |
189 | ** PATTERN MATCHING\r | |
190 | ** =======================================================\r | |
191 | */\r | |
192 | \r | |
193 | \r | |
194 | #define CAP_UNFINISHED (-1)\r | |
195 | #define CAP_POSITION (-2)\r | |
196 | \r | |
197 | \r | |
198 | typedef struct MatchState {\r | |
199 | int matchdepth; /* control for recursive depth (to avoid C stack overflow) */\r | |
200 | const char *src_init; /* init of source string */\r | |
201 | const char *src_end; /* end ('\0') of source string */\r | |
202 | const char *p_end; /* end ('\0') of pattern */\r | |
203 | lua_State *L;\r | |
204 | int level; /* total number of captures (finished or unfinished) */\r | |
205 | struct {\r | |
206 | const char *init;\r | |
207 | ptrdiff_t len;\r | |
208 | } capture[LUA_MAXCAPTURES];\r | |
209 | } MatchState;\r | |
210 | \r | |
211 | \r | |
212 | /* recursive function */\r | |
213 | static const char *match (MatchState *ms, const char *s, const char *p);\r | |
214 | \r | |
215 | \r | |
216 | /* maximum recursion depth for 'match' */\r | |
217 | #if !defined(MAXCCALLS)\r | |
218 | #define MAXCCALLS 200\r | |
219 | #endif\r | |
220 | \r | |
221 | \r | |
222 | #define L_ESC '%'\r | |
223 | #define SPECIALS "^$*+?.([%-"\r | |
224 | \r | |
225 | \r | |
226 | static int check_capture (MatchState *ms, int l) {\r | |
227 | l -= '1';\r | |
228 | if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)\r | |
229 | return luaL_error(ms->L, "invalid capture index %%%d", l + 1);\r | |
230 | return l;\r | |
231 | }\r | |
232 | \r | |
233 | \r | |
234 | static int capture_to_close (MatchState *ms) {\r | |
235 | int level = ms->level;\r | |
236 | for (level--; level>=0; level--)\r | |
237 | if (ms->capture[level].len == CAP_UNFINISHED) return level;\r | |
238 | return luaL_error(ms->L, "invalid pattern capture");\r | |
239 | }\r | |
240 | \r | |
241 | \r | |
242 | static const char *classend (MatchState *ms, const char *p) {\r | |
243 | switch (*p++) {\r | |
244 | case L_ESC: {\r | |
245 | if (p == ms->p_end)\r | |
246 | luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");\r | |
247 | return p+1;\r | |
248 | }\r | |
249 | case '[': {\r | |
250 | if (*p == '^') p++;\r | |
251 | do { /* look for a `]' */\r | |
252 | if (p == ms->p_end)\r | |
253 | luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");\r | |
254 | if (*(p++) == L_ESC && p < ms->p_end)\r | |
255 | p++; /* skip escapes (e.g. `%]') */\r | |
256 | } while (*p != ']');\r | |
257 | return p+1;\r | |
258 | }\r | |
259 | default: {\r | |
260 | return p;\r | |
261 | }\r | |
262 | }\r | |
263 | }\r | |
264 | \r | |
265 | \r | |
266 | static int match_class (int c, int cl) {\r | |
267 | int res;\r | |
268 | switch (tolower(cl)) {\r | |
269 | case 'a' : res = isalpha(c); break;\r | |
270 | case 'c' : res = iscntrl(c); break;\r | |
271 | case 'd' : res = isdigit(c); break;\r | |
272 | case 'g' : res = isgraph(c); break;\r | |
273 | case 'l' : res = islower(c); break;\r | |
274 | case 'p' : res = ispunct(c); break;\r | |
275 | case 's' : res = isspace(c); break;\r | |
276 | case 'u' : res = isupper(c); break;\r | |
277 | case 'w' : res = isalnum(c); break;\r | |
278 | case 'x' : res = isxdigit(c); break;\r | |
279 | case 'z' : res = (c == 0); break; /* deprecated option */\r | |
280 | default: return (cl == c);\r | |
281 | }\r | |
282 | return (islower(cl) ? res : !res);\r | |
283 | }\r | |
284 | \r | |
285 | \r | |
286 | static int matchbracketclass (int c, const char *p, const char *ec) {\r | |
287 | int sig = 1;\r | |
288 | if (*(p+1) == '^') {\r | |
289 | sig = 0;\r | |
290 | p++; /* skip the `^' */\r | |
291 | }\r | |
292 | while (++p < ec) {\r | |
293 | if (*p == L_ESC) {\r | |
294 | p++;\r | |
295 | if (match_class(c, uchar(*p)))\r | |
296 | return sig;\r | |
297 | }\r | |
298 | else if ((*(p+1) == '-') && (p+2 < ec)) {\r | |
299 | p+=2;\r | |
300 | if (uchar(*(p-2)) <= c && c <= uchar(*p))\r | |
301 | return sig;\r | |
302 | }\r | |
303 | else if (uchar(*p) == c) return sig;\r | |
304 | }\r | |
305 | return !sig;\r | |
306 | }\r | |
307 | \r | |
308 | \r | |
309 | static int singlematch (MatchState *ms, const char *s, const char *p,\r | |
310 | const char *ep) {\r | |
311 | if (s >= ms->src_end)\r | |
312 | return 0;\r | |
313 | else {\r | |
314 | int c = uchar(*s);\r | |
315 | switch (*p) {\r | |
316 | case '.': return 1; /* matches any char */\r | |
317 | case L_ESC: return match_class(c, uchar(*(p+1)));\r | |
318 | case '[': return matchbracketclass(c, p, ep-1);\r | |
319 | default: return (uchar(*p) == c);\r | |
320 | }\r | |
321 | }\r | |
322 | }\r | |
323 | \r | |
324 | \r | |
325 | static const char *matchbalance (MatchState *ms, const char *s,\r | |
326 | const char *p) {\r | |
327 | if (p >= ms->p_end - 1)\r | |
328 | luaL_error(ms->L, "malformed pattern "\r | |
329 | "(missing arguments to " LUA_QL("%%b") ")");\r | |
330 | if (*s != *p) return NULL;\r | |
331 | else {\r | |
332 | int b = *p;\r | |
333 | int e = *(p+1);\r | |
334 | int cont = 1;\r | |
335 | while (++s < ms->src_end) {\r | |
336 | if (*s == e) {\r | |
337 | if (--cont == 0) return s+1;\r | |
338 | }\r | |
339 | else if (*s == b) cont++;\r | |
340 | }\r | |
341 | }\r | |
342 | return NULL; /* string ends out of balance */\r | |
343 | }\r | |
344 | \r | |
345 | \r | |
346 | static const char *max_expand (MatchState *ms, const char *s,\r | |
347 | const char *p, const char *ep) {\r | |
348 | ptrdiff_t i = 0; /* counts maximum expand for item */\r | |
349 | while (singlematch(ms, s + i, p, ep))\r | |
350 | i++;\r | |
351 | /* keeps trying to match with the maximum repetitions */\r | |
352 | while (i>=0) {\r | |
353 | const char *res = match(ms, (s+i), ep+1);\r | |
354 | if (res) return res;\r | |
355 | i--; /* else didn't match; reduce 1 repetition to try again */\r | |
356 | }\r | |
357 | return NULL;\r | |
358 | }\r | |
359 | \r | |
360 | \r | |
361 | static const char *min_expand (MatchState *ms, const char *s,\r | |
362 | const char *p, const char *ep) {\r | |
363 | for (;;) {\r | |
364 | const char *res = match(ms, s, ep+1);\r | |
365 | if (res != NULL)\r | |
366 | return res;\r | |
367 | else if (singlematch(ms, s, p, ep))\r | |
368 | s++; /* try with one more repetition */\r | |
369 | else return NULL;\r | |
370 | }\r | |
371 | }\r | |
372 | \r | |
373 | \r | |
374 | static const char *start_capture (MatchState *ms, const char *s,\r | |
375 | const char *p, int what) {\r | |
376 | const char *res;\r | |
377 | int level = ms->level;\r | |
378 | if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");\r | |
379 | ms->capture[level].init = s;\r | |
380 | ms->capture[level].len = what;\r | |
381 | ms->level = level+1;\r | |
382 | if ((res=match(ms, s, p)) == NULL) /* match failed? */\r | |
383 | ms->level--; /* undo capture */\r | |
384 | return res;\r | |
385 | }\r | |
386 | \r | |
387 | \r | |
388 | static const char *end_capture (MatchState *ms, const char *s,\r | |
389 | const char *p) {\r | |
390 | int l = capture_to_close(ms);\r | |
391 | const char *res;\r | |
392 | ms->capture[l].len = s - ms->capture[l].init; /* close capture */\r | |
393 | if ((res = match(ms, s, p)) == NULL) /* match failed? */\r | |
394 | ms->capture[l].len = CAP_UNFINISHED; /* undo capture */\r | |
395 | return res;\r | |
396 | }\r | |
397 | \r | |
398 | \r | |
399 | static const char *match_capture (MatchState *ms, const char *s, int l) {\r | |
400 | size_t len;\r | |
401 | l = check_capture(ms, l);\r | |
402 | len = ms->capture[l].len;\r | |
403 | if ((size_t)(ms->src_end-s) >= len &&\r | |
404 | memcmp(ms->capture[l].init, s, len) == 0)\r | |
405 | return s+len;\r | |
406 | else return NULL;\r | |
407 | }\r | |
408 | \r | |
409 | \r | |
410 | static const char *match (MatchState *ms, const char *s, const char *p) {\r | |
411 | if (ms->matchdepth-- == 0)\r | |
412 | luaL_error(ms->L, "pattern too complex");\r | |
413 | init: /* using goto's to optimize tail recursion */\r | |
414 | if (p != ms->p_end) { /* end of pattern? */\r | |
415 | switch (*p) {\r | |
416 | case '(': { /* start capture */\r | |
417 | if (*(p + 1) == ')') /* position capture? */\r | |
418 | s = start_capture(ms, s, p + 2, CAP_POSITION);\r | |
419 | else\r | |
420 | s = start_capture(ms, s, p + 1, CAP_UNFINISHED);\r | |
421 | break;\r | |
422 | }\r | |
423 | case ')': { /* end capture */\r | |
424 | s = end_capture(ms, s, p + 1);\r | |
425 | break;\r | |
426 | }\r | |
427 | case '$': {\r | |
428 | if ((p + 1) != ms->p_end) /* is the `$' the last char in pattern? */\r | |
429 | goto dflt; /* no; go to default */\r | |
430 | s = (s == ms->src_end) ? s : NULL; /* check end of string */\r | |
431 | break;\r | |
432 | }\r | |
433 | case L_ESC: { /* escaped sequences not in the format class[*+?-]? */\r | |
434 | switch (*(p + 1)) {\r | |
435 | case 'b': { /* balanced string? */\r | |
436 | s = matchbalance(ms, s, p + 2);\r | |
437 | if (s != NULL) {\r | |
438 | p += 4; goto init; /* return match(ms, s, p + 4); */\r | |
439 | } /* else fail (s == NULL) */\r | |
440 | break;\r | |
441 | }\r | |
442 | case 'f': { /* frontier? */\r | |
443 | const char *ep; char previous;\r | |
444 | p += 2;\r | |
445 | if (*p != '[')\r | |
446 | luaL_error(ms->L, "missing " LUA_QL("[") " after "\r | |
447 | LUA_QL("%%f") " in pattern");\r | |
448 | ep = classend(ms, p); /* points to what is next */\r | |
449 | previous = (s == ms->src_init) ? '\0' : *(s - 1);\r | |
450 | if (!matchbracketclass(uchar(previous), p, ep - 1) &&\r | |
451 | matchbracketclass(uchar(*s), p, ep - 1)) {\r | |
452 | p = ep; goto init; /* return match(ms, s, ep); */\r | |
453 | }\r | |
454 | s = NULL; /* match failed */\r | |
455 | break;\r | |
456 | }\r | |
457 | case '0': case '1': case '2': case '3':\r | |
458 | case '4': case '5': case '6': case '7':\r | |
459 | case '8': case '9': { /* capture results (%0-%9)? */\r | |
460 | s = match_capture(ms, s, uchar(*(p + 1)));\r | |
461 | if (s != NULL) {\r | |
462 | p += 2; goto init; /* return match(ms, s, p + 2) */\r | |
463 | }\r | |
464 | break;\r | |
465 | }\r | |
466 | default: goto dflt;\r | |
467 | }\r | |
468 | break;\r | |
469 | }\r | |
470 | default: dflt: { /* pattern class plus optional suffix */\r | |
471 | const char *ep = classend(ms, p); /* points to optional suffix */\r | |
472 | /* does not match at least once? */\r | |
473 | if (!singlematch(ms, s, p, ep)) {\r | |
474 | if (*ep == '*' || *ep == '?' || *ep == '-') { /* accept empty? */\r | |
475 | p = ep + 1; goto init; /* return match(ms, s, ep + 1); */\r | |
476 | }\r | |
477 | else /* '+' or no suffix */\r | |
478 | s = NULL; /* fail */\r | |
479 | }\r | |
480 | else { /* matched once */\r | |
481 | switch (*ep) { /* handle optional suffix */\r | |
482 | case '?': { /* optional */\r | |
483 | const char *res;\r | |
484 | if ((res = match(ms, s + 1, ep + 1)) != NULL)\r | |
485 | s = res;\r | |
486 | else {\r | |
487 | p = ep + 1; goto init; /* else return match(ms, s, ep + 1); */\r | |
488 | }\r | |
489 | break;\r | |
490 | }\r | |
491 | case '+': /* 1 or more repetitions */\r | |
492 | s++; /* 1 match already done */\r | |
493 | /* go through */\r | |
494 | case '*': /* 0 or more repetitions */\r | |
495 | s = max_expand(ms, s, p, ep);\r | |
496 | break;\r | |
497 | case '-': /* 0 or more repetitions (minimum) */\r | |
498 | s = min_expand(ms, s, p, ep);\r | |
499 | break;\r | |
500 | default: /* no suffix */\r | |
501 | s++; p = ep; goto init; /* return match(ms, s + 1, ep); */\r | |
502 | }\r | |
503 | }\r | |
504 | break;\r | |
505 | }\r | |
506 | }\r | |
507 | }\r | |
508 | ms->matchdepth++;\r | |
509 | return s;\r | |
510 | }\r | |
511 | \r | |
512 | \r | |
513 | \r | |
514 | static const char *lmemfind (const char *s1, size_t l1,\r | |
515 | const char *s2, size_t l2) {\r | |
516 | if (l2 == 0) return s1; /* empty strings are everywhere */\r | |
517 | else if (l2 > l1) return NULL; /* avoids a negative `l1' */\r | |
518 | else {\r | |
519 | const char *init; /* to search for a `*s2' inside `s1' */\r | |
520 | l2--; /* 1st char will be checked by `memchr' */\r | |
521 | l1 = l1-l2; /* `s2' cannot be found after that */\r | |
522 | while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {\r | |
523 | init++; /* 1st char is already checked */\r | |
524 | if (memcmp(init, s2+1, l2) == 0)\r | |
525 | return init-1;\r | |
526 | else { /* correct `l1' and `s1' to try again */\r | |
527 | l1 -= init-s1;\r | |
528 | s1 = init;\r | |
529 | }\r | |
530 | }\r | |
531 | return NULL; /* not found */\r | |
532 | }\r | |
533 | }\r | |
534 | \r | |
535 | \r | |
536 | static void push_onecapture (MatchState *ms, int i, const char *s,\r | |
537 | const char *e) {\r | |
538 | if (i >= ms->level) {\r | |
539 | if (i == 0) /* ms->level == 0, too */\r | |
540 | lua_pushlstring(ms->L, s, e - s); /* add whole match */\r | |
541 | else\r | |
542 | luaL_error(ms->L, "invalid capture index");\r | |
543 | }\r | |
544 | else {\r | |
545 | ptrdiff_t l = ms->capture[i].len;\r | |
546 | if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");\r | |
547 | if (l == CAP_POSITION)\r | |
548 | lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);\r | |
549 | else\r | |
550 | lua_pushlstring(ms->L, ms->capture[i].init, l);\r | |
551 | }\r | |
552 | }\r | |
553 | \r | |
554 | \r | |
555 | static int push_captures (MatchState *ms, const char *s, const char *e) {\r | |
556 | int i;\r | |
557 | int nlevels = (ms->level == 0 && s) ? 1 : ms->level;\r | |
558 | luaL_checkstack(ms->L, nlevels, "too many captures");\r | |
559 | for (i = 0; i < nlevels; i++)\r | |
560 | push_onecapture(ms, i, s, e);\r | |
561 | return nlevels; /* number of strings pushed */\r | |
562 | }\r | |
563 | \r | |
564 | \r | |
565 | /* check whether pattern has no special characters */\r | |
566 | static int nospecials (const char *p, size_t l) {\r | |
567 | size_t upto = 0;\r | |
568 | do {\r | |
569 | if (strpbrk(p + upto, SPECIALS))\r | |
570 | return 0; /* pattern has a special character */\r | |
571 | upto += strlen(p + upto) + 1; /* may have more after \0 */\r | |
572 | } while (upto <= l);\r | |
573 | return 1; /* no special chars found */\r | |
574 | }\r | |
575 | \r | |
576 | \r | |
577 | static int str_find_aux (lua_State *L, int find) {\r | |
578 | size_t ls, lp;\r | |
579 | const char *s = luaL_checklstring(L, 1, &ls);\r | |
580 | const char *p = luaL_checklstring(L, 2, &lp);\r | |
581 | size_t init = posrelat(luaL_optinteger(L, 3, 1), ls);\r | |
582 | if (init < 1) init = 1;\r | |
583 | else if (init > ls + 1) { /* start after string's end? */\r | |
584 | lua_pushnil(L); /* cannot find anything */\r | |
585 | return 1;\r | |
586 | }\r | |
587 | /* explicit request or no special characters? */\r | |
588 | if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {\r | |
589 | /* do a plain search */\r | |
590 | const char *s2 = lmemfind(s + init - 1, ls - init + 1, p, lp);\r | |
591 | if (s2) {\r | |
592 | lua_pushinteger(L, s2 - s + 1);\r | |
593 | lua_pushinteger(L, s2 - s + lp);\r | |
594 | return 2;\r | |
595 | }\r | |
596 | }\r | |
597 | else {\r | |
598 | MatchState ms;\r | |
599 | const char *s1 = s + init - 1;\r | |
600 | int anchor = (*p == '^');\r | |
601 | if (anchor) {\r | |
602 | p++; lp--; /* skip anchor character */\r | |
603 | }\r | |
604 | ms.L = L;\r | |
605 | ms.matchdepth = MAXCCALLS;\r | |
606 | ms.src_init = s;\r | |
607 | ms.src_end = s + ls;\r | |
608 | ms.p_end = p + lp;\r | |
609 | do {\r | |
610 | const char *res;\r | |
611 | ms.level = 0;\r | |
612 | lua_assert(ms.matchdepth == MAXCCALLS);\r | |
613 | if ((res=match(&ms, s1, p)) != NULL) {\r | |
614 | if (find) {\r | |
615 | lua_pushinteger(L, s1 - s + 1); /* start */\r | |
616 | lua_pushinteger(L, res - s); /* end */\r | |
617 | return push_captures(&ms, NULL, 0) + 2;\r | |
618 | }\r | |
619 | else\r | |
620 | return push_captures(&ms, s1, res);\r | |
621 | }\r | |
622 | } while (s1++ < ms.src_end && !anchor);\r | |
623 | }\r | |
624 | lua_pushnil(L); /* not found */\r | |
625 | return 1;\r | |
626 | }\r | |
627 | \r | |
628 | \r | |
629 | static int str_find (lua_State *L) {\r | |
630 | return str_find_aux(L, 1);\r | |
631 | }\r | |
632 | \r | |
633 | \r | |
634 | static int str_match (lua_State *L) {\r | |
635 | return str_find_aux(L, 0);\r | |
636 | }\r | |
637 | \r | |
638 | \r | |
639 | static int gmatch_aux (lua_State *L) {\r | |
640 | MatchState ms;\r | |
641 | size_t ls, lp;\r | |
642 | const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);\r | |
643 | const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp);\r | |
644 | const char *src;\r | |
645 | ms.L = L;\r | |
646 | ms.matchdepth = MAXCCALLS;\r | |
647 | ms.src_init = s;\r | |
648 | ms.src_end = s+ls;\r | |
649 | ms.p_end = p + lp;\r | |
650 | for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));\r | |
651 | src <= ms.src_end;\r | |
652 | src++) {\r | |
653 | const char *e;\r | |
654 | ms.level = 0;\r | |
655 | lua_assert(ms.matchdepth == MAXCCALLS);\r | |
656 | if ((e = match(&ms, src, p)) != NULL) {\r | |
657 | lua_Integer newstart = e-s;\r | |
658 | if (e == src) newstart++; /* empty match? go at least one position */\r | |
659 | lua_pushinteger(L, newstart);\r | |
660 | lua_replace(L, lua_upvalueindex(3));\r | |
661 | return push_captures(&ms, src, e);\r | |
662 | }\r | |
663 | }\r | |
664 | return 0; /* not found */\r | |
665 | }\r | |
666 | \r | |
667 | \r | |
668 | static int gmatch (lua_State *L) {\r | |
669 | luaL_checkstring(L, 1);\r | |
670 | luaL_checkstring(L, 2);\r | |
671 | lua_settop(L, 2);\r | |
672 | lua_pushinteger(L, 0);\r | |
673 | lua_pushcclosure(L, gmatch_aux, 3);\r | |
674 | return 1;\r | |
675 | }\r | |
676 | \r | |
677 | \r | |
678 | static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,\r | |
679 | const char *e) {\r | |
680 | size_t l, i;\r | |
681 | const char *news = lua_tolstring(ms->L, 3, &l);\r | |
682 | for (i = 0; i < l; i++) {\r | |
683 | if (news[i] != L_ESC)\r | |
684 | luaL_addchar(b, news[i]);\r | |
685 | else {\r | |
686 | i++; /* skip ESC */\r | |
687 | if (!isdigit(uchar(news[i]))) {\r | |
688 | if (news[i] != L_ESC)\r | |
689 | luaL_error(ms->L, "invalid use of " LUA_QL("%c")\r | |
690 | " in replacement string", L_ESC);\r | |
691 | luaL_addchar(b, news[i]);\r | |
692 | }\r | |
693 | else if (news[i] == '0')\r | |
694 | luaL_addlstring(b, s, e - s);\r | |
695 | else {\r | |
696 | push_onecapture(ms, news[i] - '1', s, e);\r | |
697 | luaL_addvalue(b); /* add capture to accumulated result */\r | |
698 | }\r | |
699 | }\r | |
700 | }\r | |
701 | }\r | |
702 | \r | |
703 | \r | |
704 | static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,\r | |
705 | const char *e, int tr) {\r | |
706 | lua_State *L = ms->L;\r | |
707 | switch (tr) {\r | |
708 | case LUA_TFUNCTION: {\r | |
709 | int n;\r | |
710 | lua_pushvalue(L, 3);\r | |
711 | n = push_captures(ms, s, e);\r | |
712 | lua_call(L, n, 1);\r | |
713 | break;\r | |
714 | }\r | |
715 | case LUA_TTABLE: {\r | |
716 | push_onecapture(ms, 0, s, e);\r | |
717 | lua_gettable(L, 3);\r | |
718 | break;\r | |
719 | }\r | |
720 | default: { /* LUA_TNUMBER or LUA_TSTRING */\r | |
721 | add_s(ms, b, s, e);\r | |
722 | return;\r | |
723 | }\r | |
724 | }\r | |
725 | if (!lua_toboolean(L, -1)) { /* nil or false? */\r | |
726 | lua_pop(L, 1);\r | |
727 | lua_pushlstring(L, s, e - s); /* keep original text */\r | |
728 | }\r | |
729 | else if (!lua_isstring(L, -1))\r | |
730 | luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));\r | |
731 | luaL_addvalue(b); /* add result to accumulator */\r | |
732 | }\r | |
733 | \r | |
734 | \r | |
735 | static int str_gsub (lua_State *L) {\r | |
736 | size_t srcl, lp;\r | |
737 | const char *src = luaL_checklstring(L, 1, &srcl);\r | |
738 | const char *p = luaL_checklstring(L, 2, &lp);\r | |
739 | int tr = lua_type(L, 3);\r | |
740 | size_t max_s = luaL_optinteger(L, 4, srcl+1);\r | |
741 | int anchor = (*p == '^');\r | |
742 | size_t n = 0;\r | |
743 | MatchState ms;\r | |
744 | luaL_Buffer b;\r | |
745 | luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||\r | |
746 | tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,\r | |
747 | "string/function/table expected");\r | |
748 | luaL_buffinit(L, &b);\r | |
749 | if (anchor) {\r | |
750 | p++; lp--; /* skip anchor character */\r | |
751 | }\r | |
752 | ms.L = L;\r | |
753 | ms.matchdepth = MAXCCALLS;\r | |
754 | ms.src_init = src;\r | |
755 | ms.src_end = src+srcl;\r | |
756 | ms.p_end = p + lp;\r | |
757 | while (n < max_s) {\r | |
758 | const char *e;\r | |
759 | ms.level = 0;\r | |
760 | lua_assert(ms.matchdepth == MAXCCALLS);\r | |
761 | e = match(&ms, src, p);\r | |
762 | if (e) {\r | |
763 | n++;\r | |
764 | add_value(&ms, &b, src, e, tr);\r | |
765 | }\r | |
766 | if (e && e>src) /* non empty match? */\r | |
767 | src = e; /* skip it */\r | |
768 | else if (src < ms.src_end)\r | |
769 | luaL_addchar(&b, *src++);\r | |
770 | else break;\r | |
771 | if (anchor) break;\r | |
772 | }\r | |
773 | luaL_addlstring(&b, src, ms.src_end-src);\r | |
774 | luaL_pushresult(&b);\r | |
775 | lua_pushinteger(L, n); /* number of substitutions */\r | |
776 | return 2;\r | |
777 | }\r | |
778 | \r | |
779 | /* }====================================================== */\r | |
780 | \r | |
781 | \r | |
782 | \r | |
783 | /*\r | |
784 | ** {======================================================\r | |
785 | ** STRING FORMAT\r | |
786 | ** =======================================================\r | |
787 | */\r | |
788 | \r | |
789 | /*\r | |
790 | ** LUA_INTFRMLEN is the length modifier for integer conversions in\r | |
791 | ** 'string.format'; LUA_INTFRM_T is the integer type corresponding to\r | |
792 | ** the previous length\r | |
793 | */\r | |
794 | #if !defined(LUA_INTFRMLEN) /* { */\r | |
795 | #if defined(LUA_USE_LONGLONG)\r | |
796 | \r | |
797 | #define LUA_INTFRMLEN "ll"\r | |
798 | #define LUA_INTFRM_T long long\r | |
799 | \r | |
800 | #else\r | |
801 | \r | |
802 | #define LUA_INTFRMLEN "l"\r | |
803 | #define LUA_INTFRM_T long\r | |
804 | \r | |
805 | #endif\r | |
806 | #endif /* } */\r | |
807 | \r | |
808 | \r | |
809 | /*\r | |
810 | ** LUA_FLTFRMLEN is the length modifier for float conversions in\r | |
811 | ** 'string.format'; LUA_FLTFRM_T is the float type corresponding to\r | |
812 | ** the previous length\r | |
813 | */\r | |
814 | #if !defined(LUA_FLTFRMLEN)\r | |
815 | \r | |
816 | #define LUA_FLTFRMLEN ""\r | |
817 | #define LUA_FLTFRM_T double\r | |
818 | \r | |
819 | #endif\r | |
820 | \r | |
821 | \r | |
822 | /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */\r | |
823 | #define MAX_ITEM 512\r | |
824 | /* valid flags in a format specification */\r | |
825 | #define FLAGS "-+ #0"\r | |
826 | /*\r | |
827 | ** maximum size of each format specification (such as '%-099.99d')\r | |
828 | ** (+10 accounts for %99.99x plus margin of error)\r | |
829 | */\r | |
830 | #define MAX_FORMAT (sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)\r | |
831 | \r | |
832 | \r | |
833 | static void addquoted (lua_State *L, luaL_Buffer *b, int arg) {\r | |
834 | size_t l;\r | |
835 | const char *s = luaL_checklstring(L, arg, &l);\r | |
836 | luaL_addchar(b, '"');\r | |
837 | while (l--) {\r | |
838 | if (*s == '"' || *s == '\\' || *s == '\n') {\r | |
839 | luaL_addchar(b, '\\');\r | |
840 | luaL_addchar(b, *s);\r | |
841 | }\r | |
842 | else if (*s == '\0' || iscntrl(uchar(*s))) {\r | |
843 | char buff[10];\r | |
844 | if (!isdigit(uchar(*(s+1))))\r | |
845 | sprintf(buff, "\\%d", (int)uchar(*s));\r | |
846 | else\r | |
847 | sprintf(buff, "\\%03d", (int)uchar(*s));\r | |
848 | luaL_addstring(b, buff);\r | |
849 | }\r | |
850 | else\r | |
851 | luaL_addchar(b, *s);\r | |
852 | s++;\r | |
853 | }\r | |
854 | luaL_addchar(b, '"');\r | |
855 | }\r | |
856 | \r | |
857 | static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {\r | |
858 | const char *p = strfrmt;\r | |
859 | while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++; /* skip flags */\r | |
860 | if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char))\r | |
861 | luaL_error(L, "invalid format (repeated flags)");\r | |
862 | if (isdigit(uchar(*p))) p++; /* skip width */\r | |
863 | if (isdigit(uchar(*p))) p++; /* (2 digits at most) */\r | |
864 | if (*p == '.') {\r | |
865 | p++;\r | |
866 | if (isdigit(uchar(*p))) p++; /* skip precision */\r | |
867 | if (isdigit(uchar(*p))) p++; /* (2 digits at most) */\r | |
868 | }\r | |
869 | if (isdigit(uchar(*p)))\r | |
870 | luaL_error(L, "invalid format (width or precision too long)");\r | |
871 | *(form++) = '%';\r | |
872 | memcpy(form, strfrmt, (p - strfrmt + 1) * sizeof(char));\r | |
873 | form += p - strfrmt + 1;\r | |
874 | *form = '\0';\r | |
875 | return p;\r | |
876 | }\r | |
877 | \r | |
878 | \r | |
879 | /*\r | |
880 | ** add length modifier into formats\r | |
881 | */\r | |
882 | static void addlenmod (char *form, const char *lenmod) {\r | |
883 | size_t l = strlen(form);\r | |
884 | size_t lm = strlen(lenmod);\r | |
885 | char spec = form[l - 1];\r | |
886 | strcpy(form + l - 1, lenmod);\r | |
887 | form[l + lm - 1] = spec;\r | |
888 | form[l + lm] = '\0';\r | |
889 | }\r | |
890 | \r | |
891 | \r | |
892 | static int str_format (lua_State *L) {\r | |
893 | int top = lua_gettop(L);\r | |
894 | int arg = 1;\r | |
895 | size_t sfl;\r | |
896 | const char *strfrmt = luaL_checklstring(L, arg, &sfl);\r | |
897 | const char *strfrmt_end = strfrmt+sfl;\r | |
898 | luaL_Buffer b;\r | |
899 | luaL_buffinit(L, &b);\r | |
900 | while (strfrmt < strfrmt_end) {\r | |
901 | if (*strfrmt != L_ESC)\r | |
902 | luaL_addchar(&b, *strfrmt++);\r | |
903 | else if (*++strfrmt == L_ESC)\r | |
904 | luaL_addchar(&b, *strfrmt++); /* %% */\r | |
905 | else { /* format item */\r | |
906 | char form[MAX_FORMAT]; /* to store the format (`%...') */\r | |
907 | char *buff = luaL_prepbuffsize(&b, MAX_ITEM); /* to put formatted item */\r | |
908 | int nb = 0; /* number of bytes in added item */\r | |
909 | if (++arg > top)\r | |
910 | luaL_argerror(L, arg, "no value");\r | |
911 | strfrmt = scanformat(L, strfrmt, form);\r | |
912 | switch (*strfrmt++) {\r | |
913 | case 'c': {\r | |
914 | nb = sprintf(buff, form, luaL_checkint(L, arg));\r | |
915 | break;\r | |
916 | }\r | |
917 | case 'd': case 'i': {\r | |
918 | lua_Number n = luaL_checknumber(L, arg);\r | |
919 | LUA_INTFRM_T ni = (LUA_INTFRM_T)n;\r | |
920 | lua_Number diff = n - (lua_Number)ni;\r | |
921 | luaL_argcheck(L, -1 < diff && diff < 1, arg,\r | |
922 | "not a number in proper range");\r | |
923 | addlenmod(form, LUA_INTFRMLEN);\r | |
924 | nb = sprintf(buff, form, ni);\r | |
925 | break;\r | |
926 | }\r | |
927 | case 'o': case 'u': case 'x': case 'X': {\r | |
928 | lua_Number n = luaL_checknumber(L, arg);\r | |
929 | unsigned LUA_INTFRM_T ni = (unsigned LUA_INTFRM_T)n;\r | |
930 | lua_Number diff = n - (lua_Number)ni;\r | |
931 | luaL_argcheck(L, -1 < diff && diff < 1, arg,\r | |
932 | "not a non-negative number in proper range");\r | |
933 | addlenmod(form, LUA_INTFRMLEN);\r | |
934 | nb = sprintf(buff, form, ni);\r | |
935 | break;\r | |
936 | }\r | |
937 | case 'e': case 'E': case 'f':\r | |
938 | #if defined(LUA_USE_AFORMAT)\r | |
939 | case 'a': case 'A':\r | |
940 | #endif\r | |
941 | case 'g': case 'G': {\r | |
942 | addlenmod(form, LUA_FLTFRMLEN);\r | |
943 | nb = sprintf(buff, form, (LUA_FLTFRM_T)luaL_checknumber(L, arg));\r | |
944 | break;\r | |
945 | }\r | |
946 | case 'q': {\r | |
947 | addquoted(L, &b, arg);\r | |
948 | break;\r | |
949 | }\r | |
950 | case 's': {\r | |
951 | size_t l;\r | |
952 | const char *s = luaL_tolstring(L, arg, &l);\r | |
953 | if (!strchr(form, '.') && l >= 100) {\r | |
954 | /* no precision and string is too long to be formatted;\r | |
955 | keep original string */\r | |
956 | luaL_addvalue(&b);\r | |
957 | break;\r | |
958 | }\r | |
959 | else {\r | |
960 | nb = sprintf(buff, form, s);\r | |
961 | lua_pop(L, 1); /* remove result from 'luaL_tolstring' */\r | |
962 | break;\r | |
963 | }\r | |
964 | }\r | |
965 | default: { /* also treat cases `pnLlh' */\r | |
966 | return luaL_error(L, "invalid option " LUA_QL("%%%c") " to "\r | |
967 | LUA_QL("format"), *(strfrmt - 1));\r | |
968 | }\r | |
969 | }\r | |
970 | luaL_addsize(&b, nb);\r | |
971 | }\r | |
972 | }\r | |
973 | luaL_pushresult(&b);\r | |
974 | return 1;\r | |
975 | }\r | |
976 | \r | |
977 | /* }====================================================== */\r | |
978 | \r | |
979 | \r | |
980 | static const luaL_Reg strlib[] = {\r | |
981 | {"byte", str_byte},\r | |
982 | {"char", str_char},\r | |
983 | {"dump", str_dump},\r | |
984 | {"find", str_find},\r | |
985 | {"format", str_format},\r | |
986 | {"gmatch", gmatch},\r | |
987 | {"gsub", str_gsub},\r | |
988 | {"len", str_len},\r | |
989 | {"lower", str_lower},\r | |
990 | {"match", str_match},\r | |
991 | {"rep", str_rep},\r | |
992 | {"reverse", str_reverse},\r | |
993 | {"sub", str_sub},\r | |
994 | {"upper", str_upper},\r | |
995 | {NULL, NULL}\r | |
996 | };\r | |
997 | \r | |
998 | \r | |
999 | static void createmetatable (lua_State *L) {\r | |
1000 | lua_createtable(L, 0, 1); /* table to be metatable for strings */\r | |
1001 | lua_pushliteral(L, ""); /* dummy string */\r | |
1002 | lua_pushvalue(L, -2); /* copy table */\r | |
1003 | lua_setmetatable(L, -2); /* set table as metatable for strings */\r | |
1004 | lua_pop(L, 1); /* pop dummy string */\r | |
1005 | lua_pushvalue(L, -2); /* get string library */\r | |
1006 | lua_setfield(L, -2, "__index"); /* metatable.__index = string */\r | |
1007 | lua_pop(L, 1); /* pop metatable */\r | |
1008 | }\r | |
1009 | \r | |
1010 | \r | |
1011 | /*\r | |
1012 | ** Open string library\r | |
1013 | */\r | |
1014 | LUAMOD_API int luaopen_string (lua_State *L) {\r | |
1015 | luaL_newlib(L, strlib);\r | |
1016 | createmetatable(L);\r | |
1017 | return 1;\r | |
1018 | }\r | |
1019 | \r |