• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2 ** $Id: lstrlib.c,v 1.254.1.1 2017/04/19 17:29:57 roberto Exp $
3 ** Standard library for string operations and pattern-matching
4 ** See Copyright Notice in lua.h
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>
16 #include <locale.h>
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
30 ** pattern-matching. This limit is arbitrary, but must fit in
31 ** an unsigned char.
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 */
46 #define MAX_SIZET	((size_t)(~(size_t)0))
47 
48 #define MAXSIZE  \
49 	(sizeof(size_t) < sizeof(int) ? MAX_SIZET : (size_t)(INT_MAX))
50 
51 
52 
53 
str_len(lua_State * L)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 */
posrelat(lua_Integer pos,size_t len)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 
str_sub(lua_State * L)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 
str_reverse(lua_State * L)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 
str_lower(lua_State * L)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 
str_upper(lua_State * L)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 
str_rep(lua_State * L)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 
str_byte(lua_State * L)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 
str_char(lua_State * L)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 
writer(lua_State * L,const void * b,size_t size,void * B)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 
str_dump(lua_State * L)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 {
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;
219   int matchdepth;  /* control for recursive depth (to avoid C stack overflow) */
220   unsigned char level;  /* total number of captures (finished or unfinished) */
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 
check_capture(MatchState * ms,int l)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 
capture_to_close(MatchState * ms)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 
classend(MatchState * ms,const char * p)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 
match_class(int c,int cl)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 
matchbracketclass(int c,const char * p,const char * ec)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 
singlematch(MatchState * ms,const char * s,const char * p,const char * ep)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 
matchbalance(MatchState * ms,const char * s,const char * p)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 
max_expand(MatchState * ms,const char * s,const char * p,const char * ep)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 
min_expand(MatchState * ms,const char * s,const char * p,const char * ep)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 
start_capture(MatchState * ms,const char * s,const char * p,int what)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 
end_capture(MatchState * ms,const char * s,const char * p)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 
match_capture(MatchState * ms,const char * s,int l)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 
match(MatchState * ms,const char * s,const char * p)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 
lmemfind(const char * s1,size_t l1,const char * s2,size_t l2)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 
push_onecapture(MatchState * ms,int i,const char * s,const char * e)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 
push_captures(MatchState * ms,const char * s,const char * e)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 */
nospecials(const char * p,size_t l)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 
prepstate(MatchState * ms,lua_State * L,const char * s,size_t ls,const char * p,size_t lp)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 
reprepstate(MatchState * ms)601 static void reprepstate (MatchState *ms) {
602   ms->level = 0;
603   lua_assert(ms->matchdepth == MAXCCALLS);
604 }
605 
606 
str_find_aux(lua_State * L,int find)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     }
634     prepstate(&ms, L, s, ls, p, lp);
635     do {
636       const char *res;
637       reprepstate(&ms);
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 
str_find(lua_State * L)654 static int str_find (lua_State *L) {
655   return str_find_aux(L, 1);
656 }
657 
658 
str_match(lua_State * L)659 static int str_match (lua_State *L) {
660   return str_find_aux(L, 0);
661 }
662 
663 
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 
gmatch_aux(lua_State * L)673 static int gmatch_aux (lua_State *L) {
674   GMatchState *gm = (GMatchState *)lua_touserdata(L, lua_upvalueindex(3));
675   const char *src;
676   gm->ms.L = L;
677   for (src = gm->src; src <= gm->ms.src_end; src++) {
678     const char *e;
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);
683     }
684   }
685   return 0;  /* not found */
686 }
687 
688 
gmatch(lua_State * L)689 static int gmatch (lua_State *L) {
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;
698   lua_pushcclosure(L, gmatch_aux, 3);
699   return 1;
700 }
701 
702 
add_s(MatchState * ms,luaL_Buffer * b,const char * s,const char * e)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 
add_value(MatchState * ms,luaL_Buffer * b,const char * s,const char * e,int tr)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 
str_gsub(lua_State * L)762 static int str_gsub (lua_State *L) {
763   size_t srcl, lp;
764   const char *src = luaL_checklstring(L, 1, &srcl);  /* subject */
765   const char *p = luaL_checklstring(L, 2, &lp);  /* pattern */
766   const char *lastmatch = NULL;  /* end of last match */
767   int tr = lua_type(L, 3);  /* replacement type */
768   lua_Integer max_s = luaL_optinteger(L, 4, srcl + 1);  /* max replacements */
769   int anchor = (*p == '^');
770   lua_Integer n = 0;  /* replacement count */
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   }
780   prepstate(&ms, L, src, srcl, p, lp);
781   while (n < max_s) {
782     const char *e;
783     reprepstate(&ms);  /* (re)prepare state for new match */
784     if ((e = match(&ms, src, p)) != NULL && e != lastmatch) {  /* match? */
785       n++;
786       add_value(&ms, &b, src, e, tr);  /* add replacement to buffer */
787       src = lastmatch = e;
788     }
789     else if (src < ms.src_end)  /* otherwise, skip one character */
790       luaL_addchar(&b, *src++);
791     else break;  /* end of subject */
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 
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 */
adddigit(char * buff,int n,lua_Number x)833 static lua_Number adddigit (char *buff, int n, lua_Number x) {
834   lua_Number dd = l_mathop(floor)(x);  /* get integer part from 'x' */
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 
num2straux(char * buff,int sz,lua_Number x)841 static int num2straux (char *buff, int sz, lua_Number x) {
842   /* if 'inf' or 'NaN', format it like '%g' */
843   if (x != x || x == (lua_Number)HUGE_VAL || x == -(lua_Number)HUGE_VAL)
844     return l_sprintf(buff, sz, LUA_NUMBER_FMT, (LUAI_UACNUMBER)x);
845   else if (x == 0) {  /* can be -0... */
846     /* create "0" or "-0" followed by exponent */
847     return l_sprintf(buff, sz, LUA_NUMBER_FMT "x0p+0", (LUAI_UACNUMBER)x);
848   }
849   else {
850     int e;
851     lua_Number m = l_mathop(frexp)(x, &e);  /* 'x' fraction and exponent */
852     int n = 0;  /* character count */
853     if (m < 0) {  /* is number negative? */
854       buff[n++] = '-';  /* add signal */
855       m = -m;  /* make it positive */
856     }
857     buff[n++] = '0'; buff[n++] = 'x';  /* add "0x" */
858     m = adddigit(buff, n++, m * (1 << L_NBFD));  /* add first digit */
859     e -= L_NBFD;  /* this digit goes before the radix point */
860     if (m > 0) {  /* more digits? */
861       buff[n++] = lua_getlocaledecpoint();  /* add radix point */
862       do {  /* add as many digits as needed */
863         m = adddigit(buff, n++, m * 16);
864       } while (m > 0);
865     }
866     n += l_sprintf(buff + n, sz - n, "p%+d", e);  /* add exponent */
867     lua_assert(n < sz);
868     return n;
869   }
870 }
871 
872 
lua_number2strx(lua_State * L,char * buff,int sz,const char * fmt,lua_Number x)873 static int lua_number2strx (lua_State *L, char *buff, int sz,
874                             const char *fmt, lua_Number x) {
875   int n = num2straux(buff, sz, x);
876   if (fmt[SIZELENMOD] == 'A') {
877     int i;
878     for (i = 0; i < n; i++)
879       buff[i] = toupper(uchar(buff[i]));
880   }
881   else if (fmt[SIZELENMOD] != 'a')
882     return luaL_error(L, "modifiers for format '%%a'/'%%A' not implemented");
883   return n;
884 }
885 
886 #endif				/* } */
887 
888 
889 /*
890 ** Maximum size of each formatted item. This maximum size is produced
891 ** by format('%.99f', -maxfloat), and is equal to 99 + 3 ('-', '.',
892 ** and '\0') + number of decimal digits to represent maxfloat (which
893 ** is maximum exponent + 1). (99+3+1 then rounded to 120 for "extra
894 ** expenses", such as locale-dependent stuff)
895 */
896 #define MAX_ITEM        (120 + l_mathlim(MAX_10_EXP))
897 
898 
899 /* valid flags in a format specification */
900 #define FLAGS	"-+ #0"
901 
902 /*
903 ** maximum size of each format specification (such as "%-099.99d")
904 */
905 #define MAX_FORMAT	32
906 
907 
addquoted(luaL_Buffer * b,const char * s,size_t len)908 static void addquoted (luaL_Buffer *b, const char *s, size_t len) {
909   luaL_addchar(b, '"');
910   while (len--) {
911     if (*s == '"' || *s == '\\' || *s == '\n') {
912       luaL_addchar(b, '\\');
913       luaL_addchar(b, *s);
914     }
915     else if (iscntrl(uchar(*s))) {
916       char buff[10];
917       if (!isdigit(uchar(*(s+1))))
918         l_sprintf(buff, sizeof(buff), "\\%d", (int)uchar(*s));
919       else
920         l_sprintf(buff, sizeof(buff), "\\%03d", (int)uchar(*s));
921       luaL_addstring(b, buff);
922     }
923     else
924       luaL_addchar(b, *s);
925     s++;
926   }
927   luaL_addchar(b, '"');
928 }
929 
930 
931 /*
932 ** Ensures the 'buff' string uses a dot as the radix character.
933 */
checkdp(char * buff,int nb)934 static void checkdp (char *buff, int nb) {
935   if (memchr(buff, '.', nb) == NULL) {  /* no dot? */
936     char point = lua_getlocaledecpoint();  /* try locale point */
937     char *ppoint = (char *)memchr(buff, point, nb);
938     if (ppoint) *ppoint = '.';  /* change it to a dot */
939   }
940 }
941 
942 
addliteral(lua_State * L,luaL_Buffer * b,int arg)943 static void addliteral (lua_State *L, luaL_Buffer *b, int arg) {
944   switch (lua_type(L, arg)) {
945     case LUA_TSTRING: {
946       size_t len;
947       const char *s = lua_tolstring(L, arg, &len);
948       addquoted(b, s, len);
949       break;
950     }
951     case LUA_TNUMBER: {
952       char *buff = luaL_prepbuffsize(b, MAX_ITEM);
953       int nb;
954       if (!lua_isinteger(L, arg)) {  /* float? */
955         lua_Number n = lua_tonumber(L, arg);  /* write as hexa ('%a') */
956         nb = lua_number2strx(L, buff, MAX_ITEM, "%" LUA_NUMBER_FRMLEN "a", n);
957         checkdp(buff, nb);  /* ensure it uses a dot */
958       }
959       else {  /* integers */
960         lua_Integer n = lua_tointeger(L, arg);
961         const char *format = (n == LUA_MININTEGER)  /* corner case? */
962                            ? "0x%" LUA_INTEGER_FRMLEN "x"  /* use hexa */
963                            : LUA_INTEGER_FMT;  /* else use default format */
964         nb = l_sprintf(buff, MAX_ITEM, format, (LUAI_UACINT)n);
965       }
966       luaL_addsize(b, nb);
967       break;
968     }
969     case LUA_TNIL: case LUA_TBOOLEAN: {
970       luaL_tolstring(L, arg, NULL);
971       luaL_addvalue(b);
972       break;
973     }
974     default: {
975       luaL_argerror(L, arg, "value has no literal form");
976     }
977   }
978 }
979 
980 
scanformat(lua_State * L,const char * strfrmt,char * form)981 static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
982   const char *p = strfrmt;
983   while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++;  /* skip flags */
984   if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char))
985     luaL_error(L, "invalid format (repeated flags)");
986   if (isdigit(uchar(*p))) p++;  /* skip width */
987   if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
988   if (*p == '.') {
989     p++;
990     if (isdigit(uchar(*p))) p++;  /* skip precision */
991     if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
992   }
993   if (isdigit(uchar(*p)))
994     luaL_error(L, "invalid format (width or precision too long)");
995   *(form++) = '%';
996   memcpy(form, strfrmt, ((p - strfrmt) + 1) * sizeof(char));
997   form += (p - strfrmt) + 1;
998   *form = '\0';
999   return p;
1000 }
1001 
1002 
1003 /*
1004 ** add length modifier into formats
1005 */
addlenmod(char * form,const char * lenmod)1006 static void addlenmod (char *form, const char *lenmod) {
1007   size_t l = strlen(form);
1008   size_t lm = strlen(lenmod);
1009   char spec = form[l - 1];
1010   strcpy(form + l - 1, lenmod);
1011   form[l + lm - 1] = spec;
1012   form[l + lm] = '\0';
1013 }
1014 
1015 
str_format(lua_State * L)1016 static int str_format (lua_State *L) {
1017   int top = lua_gettop(L);
1018   int arg = 1;
1019   size_t sfl;
1020   const char *strfrmt = luaL_checklstring(L, arg, &sfl);
1021   const char *strfrmt_end = strfrmt+sfl;
1022   luaL_Buffer b;
1023   luaL_buffinit(L, &b);
1024   while (strfrmt < strfrmt_end) {
1025     if (*strfrmt != L_ESC)
1026       luaL_addchar(&b, *strfrmt++);
1027     else if (*++strfrmt == L_ESC)
1028       luaL_addchar(&b, *strfrmt++);  /* %% */
1029     else { /* format item */
1030       char form[MAX_FORMAT];  /* to store the format ('%...') */
1031       char *buff = luaL_prepbuffsize(&b, MAX_ITEM);  /* to put formatted item */
1032       int nb = 0;  /* number of bytes in added item */
1033       if (++arg > top)
1034         luaL_argerror(L, arg, "no value");
1035       strfrmt = scanformat(L, strfrmt, form);
1036       switch (*strfrmt++) {
1037         case 'c': {
1038           nb = l_sprintf(buff, MAX_ITEM, form, (int)luaL_checkinteger(L, arg));
1039           break;
1040         }
1041         case 'd': case 'i':
1042         case 'o': case 'u': case 'x': case 'X': {
1043           lua_Integer n = luaL_checkinteger(L, arg);
1044           addlenmod(form, LUA_INTEGER_FRMLEN);
1045           nb = l_sprintf(buff, MAX_ITEM, form, (LUAI_UACINT)n);
1046           break;
1047         }
1048         case 'a': case 'A':
1049           addlenmod(form, LUA_NUMBER_FRMLEN);
1050           nb = lua_number2strx(L, buff, MAX_ITEM, form,
1051                                   luaL_checknumber(L, arg));
1052           break;
1053         case 'e': case 'E': case 'f':
1054         case 'g': case 'G': {
1055           lua_Number n = luaL_checknumber(L, arg);
1056           addlenmod(form, LUA_NUMBER_FRMLEN);
1057           nb = l_sprintf(buff, MAX_ITEM, form, (LUAI_UACNUMBER)n);
1058           break;
1059         }
1060         case 'q': {
1061           addliteral(L, &b, arg);
1062           break;
1063         }
1064         case 's': {
1065           size_t l;
1066           const char *s = luaL_tolstring(L, arg, &l);
1067           if (form[2] == '\0')  /* no modifiers? */
1068             luaL_addvalue(&b);  /* keep entire string */
1069           else {
1070             luaL_argcheck(L, l == strlen(s), arg, "string contains zeros");
1071             if (!strchr(form, '.') && l >= 100) {
1072               /* no precision and string is too long to be formatted */
1073               luaL_addvalue(&b);  /* keep entire string */
1074             }
1075             else {  /* format the string into 'buff' */
1076               nb = l_sprintf(buff, MAX_ITEM, form, s);
1077               lua_pop(L, 1);  /* remove result from 'luaL_tolstring' */
1078             }
1079           }
1080           break;
1081         }
1082         default: {  /* also treat cases 'pnLlh' */
1083           return luaL_error(L, "invalid option '%%%c' to 'format'",
1084                                *(strfrmt - 1));
1085         }
1086       }
1087       lua_assert(nb < MAX_ITEM);
1088       luaL_addsize(&b, nb);
1089     }
1090   }
1091   luaL_pushresult(&b);
1092   return 1;
1093 }
1094 
1095 /* }====================================================== */
1096 
1097 
1098 /*
1099 ** {======================================================
1100 ** PACK/UNPACK
1101 ** =======================================================
1102 */
1103 
1104 
1105 /* value used for padding */
1106 #if !defined(LUAL_PACKPADBYTE)
1107 #define LUAL_PACKPADBYTE		0x00
1108 #endif
1109 
1110 /* maximum size for the binary representation of an integer */
1111 #define MAXINTSIZE	16
1112 
1113 /* number of bits in a character */
1114 #define NB	CHAR_BIT
1115 
1116 /* mask for one character (NB 1's) */
1117 #define MC	((1 << NB) - 1)
1118 
1119 /* size of a lua_Integer */
1120 #define SZINT	((int)sizeof(lua_Integer))
1121 
1122 
1123 /* dummy union to get native endianness */
1124 static const union {
1125   int dummy;
1126   char little;  /* true iff machine is little endian */
1127 } nativeendian = {1};
1128 
1129 
1130 /* dummy structure to get native alignment requirements */
1131 struct cD {
1132   char c;
1133   union { double d; void *p; lua_Integer i; lua_Number n; } u;
1134 };
1135 
1136 #define MAXALIGN	(offsetof(struct cD, u))
1137 
1138 
1139 /*
1140 ** Union for serializing floats
1141 */
1142 typedef union Ftypes {
1143   float f;
1144   double d;
1145   lua_Number n;
1146   char buff[5 * sizeof(lua_Number)];  /* enough for any float type */
1147 } Ftypes;
1148 
1149 
1150 /*
1151 ** information to pack/unpack stuff
1152 */
1153 typedef struct Header {
1154   lua_State *L;
1155   int islittle;
1156   int maxalign;
1157 } Header;
1158 
1159 
1160 /*
1161 ** options for pack/unpack
1162 */
1163 typedef enum KOption {
1164   Kint,		/* signed integers */
1165   Kuint,	/* unsigned integers */
1166   Kfloat,	/* floating-point numbers */
1167   Kchar,	/* fixed-length strings */
1168   Kstring,	/* strings with prefixed length */
1169   Kzstr,	/* zero-terminated strings */
1170   Kpadding,	/* padding */
1171   Kpaddalign,	/* padding for alignment */
1172   Knop		/* no-op (configuration or spaces) */
1173 } KOption;
1174 
1175 
1176 /*
1177 ** Read an integer numeral from string 'fmt' or return 'df' if
1178 ** there is no numeral
1179 */
digit(int c)1180 static int digit (int c) { return '0' <= c && c <= '9'; }
1181 
getnum(const char ** fmt,int df)1182 static int getnum (const char **fmt, int df) {
1183   if (!digit(**fmt))  /* no number? */
1184     return df;  /* return default value */
1185   else {
1186     int a = 0;
1187     do {
1188       a = a*10 + (*((*fmt)++) - '0');
1189     } while (digit(**fmt) && a <= ((int)MAXSIZE - 9)/10);
1190     return a;
1191   }
1192 }
1193 
1194 
1195 /*
1196 ** Read an integer numeral and raises an error if it is larger
1197 ** than the maximum size for integers.
1198 */
getnumlimit(Header * h,const char ** fmt,int df)1199 static int getnumlimit (Header *h, const char **fmt, int df) {
1200   int sz = getnum(fmt, df);
1201   if (sz > MAXINTSIZE || sz <= 0)
1202     return luaL_error(h->L, "integral size (%d) out of limits [1,%d]",
1203                             sz, MAXINTSIZE);
1204   return sz;
1205 }
1206 
1207 
1208 /*
1209 ** Initialize Header
1210 */
initheader(lua_State * L,Header * h)1211 static void initheader (lua_State *L, Header *h) {
1212   h->L = L;
1213   h->islittle = nativeendian.little;
1214   h->maxalign = 1;
1215 }
1216 
1217 
1218 /*
1219 ** Read and classify next option. 'size' is filled with option's size.
1220 */
getoption(Header * h,const char ** fmt,int * size)1221 static KOption getoption (Header *h, const char **fmt, int *size) {
1222   int opt = *((*fmt)++);
1223   *size = 0;  /* default */
1224   switch (opt) {
1225     case 'b': *size = sizeof(char); return Kint;
1226     case 'B': *size = sizeof(char); return Kuint;
1227     case 'h': *size = sizeof(short); return Kint;
1228     case 'H': *size = sizeof(short); return Kuint;
1229     case 'l': *size = sizeof(long); return Kint;
1230     case 'L': *size = sizeof(long); return Kuint;
1231     case 'j': *size = sizeof(lua_Integer); return Kint;
1232     case 'J': *size = sizeof(lua_Integer); return Kuint;
1233     case 'T': *size = sizeof(size_t); return Kuint;
1234     case 'f': *size = sizeof(float); return Kfloat;
1235     case 'd': *size = sizeof(double); return Kfloat;
1236     case 'n': *size = sizeof(lua_Number); return Kfloat;
1237     case 'i': *size = getnumlimit(h, fmt, sizeof(int)); return Kint;
1238     case 'I': *size = getnumlimit(h, fmt, sizeof(int)); return Kuint;
1239     case 's': *size = getnumlimit(h, fmt, sizeof(size_t)); return Kstring;
1240     case 'c':
1241       *size = getnum(fmt, -1);
1242       if (*size == -1)
1243         luaL_error(h->L, "missing size for format option 'c'");
1244       return Kchar;
1245     case 'z': return Kzstr;
1246     case 'x': *size = 1; return Kpadding;
1247     case 'X': return Kpaddalign;
1248     case ' ': break;
1249     case '<': h->islittle = 1; break;
1250     case '>': h->islittle = 0; break;
1251     case '=': h->islittle = nativeendian.little; break;
1252     case '!': h->maxalign = getnumlimit(h, fmt, MAXALIGN); break;
1253     default: luaL_error(h->L, "invalid format option '%c'", opt);
1254   }
1255   return Knop;
1256 }
1257 
1258 
1259 /*
1260 ** Read, classify, and fill other details about the next option.
1261 ** 'psize' is filled with option's size, 'notoalign' with its
1262 ** alignment requirements.
1263 ** Local variable 'size' gets the size to be aligned. (Kpadal option
1264 ** always gets its full alignment, other options are limited by
1265 ** the maximum alignment ('maxalign'). Kchar option needs no alignment
1266 ** despite its size.
1267 */
getdetails(Header * h,size_t totalsize,const char ** fmt,int * psize,int * ntoalign)1268 static KOption getdetails (Header *h, size_t totalsize,
1269                            const char **fmt, int *psize, int *ntoalign) {
1270   KOption opt = getoption(h, fmt, psize);
1271   int align = *psize;  /* usually, alignment follows size */
1272   if (opt == Kpaddalign) {  /* 'X' gets alignment from following option */
1273     if (**fmt == '\0' || getoption(h, fmt, &align) == Kchar || align == 0)
1274       luaL_argerror(h->L, 1, "invalid next option for option 'X'");
1275   }
1276   if (align <= 1 || opt == Kchar)  /* need no alignment? */
1277     *ntoalign = 0;
1278   else {
1279     if (align > h->maxalign)  /* enforce maximum alignment */
1280       align = h->maxalign;
1281     if ((align & (align - 1)) != 0)  /* is 'align' not a power of 2? */
1282       luaL_argerror(h->L, 1, "format asks for alignment not power of 2");
1283     *ntoalign = (align - (int)(totalsize & (align - 1))) & (align - 1);
1284   }
1285   return opt;
1286 }
1287 
1288 
1289 /*
1290 ** Pack integer 'n' with 'size' bytes and 'islittle' endianness.
1291 ** The final 'if' handles the case when 'size' is larger than
1292 ** the size of a Lua integer, correcting the extra sign-extension
1293 ** bytes if necessary (by default they would be zeros).
1294 */
packint(luaL_Buffer * b,lua_Unsigned n,int islittle,int size,int neg)1295 static void packint (luaL_Buffer *b, lua_Unsigned n,
1296                      int islittle, int size, int neg) {
1297   char *buff = luaL_prepbuffsize(b, size);
1298   int i;
1299   buff[islittle ? 0 : size - 1] = (char)(n & MC);  /* first byte */
1300   for (i = 1; i < size; i++) {
1301     n >>= NB;
1302     buff[islittle ? i : size - 1 - i] = (char)(n & MC);
1303   }
1304   if (neg && size > SZINT) {  /* negative number need sign extension? */
1305     for (i = SZINT; i < size; i++)  /* correct extra bytes */
1306       buff[islittle ? i : size - 1 - i] = (char)MC;
1307   }
1308   luaL_addsize(b, size);  /* add result to buffer */
1309 }
1310 
1311 
1312 /*
1313 ** Copy 'size' bytes from 'src' to 'dest', correcting endianness if
1314 ** given 'islittle' is different from native endianness.
1315 */
copywithendian(volatile char * dest,volatile const char * src,int size,int islittle)1316 static void copywithendian (volatile char *dest, volatile const char *src,
1317                             int size, int islittle) {
1318   if (islittle == nativeendian.little) {
1319     while (size-- != 0)
1320       *(dest++) = *(src++);
1321   }
1322   else {
1323     dest += size - 1;
1324     while (size-- != 0)
1325       *(dest--) = *(src++);
1326   }
1327 }
1328 
1329 
str_pack(lua_State * L)1330 static int str_pack (lua_State *L) {
1331   luaL_Buffer b;
1332   Header h;
1333   const char *fmt = luaL_checkstring(L, 1);  /* format string */
1334   int arg = 1;  /* current argument to pack */
1335   size_t totalsize = 0;  /* accumulate total size of result */
1336   initheader(L, &h);
1337   lua_pushnil(L);  /* mark to separate arguments from string buffer */
1338   luaL_buffinit(L, &b);
1339   while (*fmt != '\0') {
1340     int size, ntoalign;
1341     KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
1342     totalsize += ntoalign + size;
1343     while (ntoalign-- > 0)
1344      luaL_addchar(&b, LUAL_PACKPADBYTE);  /* fill alignment */
1345     arg++;
1346     switch (opt) {
1347       case Kint: {  /* signed integers */
1348         lua_Integer n = luaL_checkinteger(L, arg);
1349         if (size < SZINT) {  /* need overflow check? */
1350           lua_Integer lim = (lua_Integer)1 << ((size * NB) - 1);
1351           luaL_argcheck(L, -lim <= n && n < lim, arg, "integer overflow");
1352         }
1353         packint(&b, (lua_Unsigned)n, h.islittle, size, (n < 0));
1354         break;
1355       }
1356       case Kuint: {  /* unsigned integers */
1357         lua_Integer n = luaL_checkinteger(L, arg);
1358         if (size < SZINT)  /* need overflow check? */
1359           luaL_argcheck(L, (lua_Unsigned)n < ((lua_Unsigned)1 << (size * NB)),
1360                            arg, "unsigned overflow");
1361         packint(&b, (lua_Unsigned)n, h.islittle, size, 0);
1362         break;
1363       }
1364       case Kfloat: {  /* floating-point options */
1365         volatile Ftypes u;
1366         char *buff = luaL_prepbuffsize(&b, size);
1367         lua_Number n = luaL_checknumber(L, arg);  /* get argument */
1368         if (size == sizeof(u.f)) u.f = (float)n;  /* copy it into 'u' */
1369         else if (size == sizeof(u.d)) u.d = (double)n;
1370         else u.n = n;
1371         /* move 'u' to final result, correcting endianness if needed */
1372         copywithendian(buff, u.buff, size, h.islittle);
1373         luaL_addsize(&b, size);
1374         break;
1375       }
1376       case Kchar: {  /* fixed-size string */
1377         size_t len;
1378         const char *s = luaL_checklstring(L, arg, &len);
1379         luaL_argcheck(L, len <= (size_t)size, arg,
1380                          "string longer than given size");
1381         luaL_addlstring(&b, s, len);  /* add string */
1382         while (len++ < (size_t)size)  /* pad extra space */
1383           luaL_addchar(&b, LUAL_PACKPADBYTE);
1384         break;
1385       }
1386       case Kstring: {  /* strings with length count */
1387         size_t len;
1388         const char *s = luaL_checklstring(L, arg, &len);
1389         luaL_argcheck(L, size >= (int)sizeof(size_t) ||
1390                          len < ((size_t)1 << (size * NB)),
1391                          arg, "string length does not fit in given size");
1392         packint(&b, (lua_Unsigned)len, h.islittle, size, 0);  /* pack length */
1393         luaL_addlstring(&b, s, len);
1394         totalsize += len;
1395         break;
1396       }
1397       case Kzstr: {  /* zero-terminated string */
1398         size_t len;
1399         const char *s = luaL_checklstring(L, arg, &len);
1400         luaL_argcheck(L, strlen(s) == len, arg, "string contains zeros");
1401         luaL_addlstring(&b, s, len);
1402         luaL_addchar(&b, '\0');  /* add zero at the end */
1403         totalsize += len + 1;
1404         break;
1405       }
1406       case Kpadding: luaL_addchar(&b, LUAL_PACKPADBYTE);  /* FALLTHROUGH */
1407       case Kpaddalign: case Knop:
1408         arg--;  /* undo increment */
1409         break;
1410     }
1411   }
1412   luaL_pushresult(&b);
1413   return 1;
1414 }
1415 
1416 
str_packsize(lua_State * L)1417 static int str_packsize (lua_State *L) {
1418   Header h;
1419   const char *fmt = luaL_checkstring(L, 1);  /* format string */
1420   size_t totalsize = 0;  /* accumulate total size of result */
1421   initheader(L, &h);
1422   while (*fmt != '\0') {
1423     int size, ntoalign;
1424     KOption opt = getdetails(&h, totalsize, &fmt, &size, &ntoalign);
1425     size += ntoalign;  /* total space used by option */
1426     luaL_argcheck(L, totalsize <= MAXSIZE - size, 1,
1427                      "format result too large");
1428     totalsize += size;
1429     switch (opt) {
1430       case Kstring:  /* strings with length count */
1431       case Kzstr:    /* zero-terminated string */
1432         luaL_argerror(L, 1, "variable-length format");
1433         /* call never return, but to avoid warnings: *//* FALLTHROUGH */
1434       default:  break;
1435     }
1436   }
1437   lua_pushinteger(L, (lua_Integer)totalsize);
1438   return 1;
1439 }
1440 
1441 
1442 /*
1443 ** Unpack an integer with 'size' bytes and 'islittle' endianness.
1444 ** If size is smaller than the size of a Lua integer and integer
1445 ** is signed, must do sign extension (propagating the sign to the
1446 ** higher bits); if size is larger than the size of a Lua integer,
1447 ** it must check the unread bytes to see whether they do not cause an
1448 ** overflow.
1449 */
unpackint(lua_State * L,const char * str,int islittle,int size,int issigned)1450 static lua_Integer unpackint (lua_State *L, const char *str,
1451                               int islittle, int size, int issigned) {
1452   lua_Unsigned res = 0;
1453   int i;
1454   int limit = (size  <= SZINT) ? size : SZINT;
1455   for (i = limit - 1; i >= 0; i--) {
1456     res <<= NB;
1457     res |= (lua_Unsigned)(unsigned char)str[islittle ? i : size - 1 - i];
1458   }
1459   if (size < SZINT) {  /* real size smaller than lua_Integer? */
1460     if (issigned) {  /* needs sign extension? */
1461       lua_Unsigned mask = (lua_Unsigned)1 << (size*NB - 1);
1462       res = ((res ^ mask) - mask);  /* do sign extension */
1463     }
1464   }
1465   else if (size > SZINT) {  /* must check unread bytes */
1466     int mask = (!issigned || (lua_Integer)res >= 0) ? 0 : MC;
1467     for (i = limit; i < size; i++) {
1468       if ((unsigned char)str[islittle ? i : size - 1 - i] != mask)
1469         luaL_error(L, "%d-byte integer does not fit into Lua Integer", size);
1470     }
1471   }
1472   return (lua_Integer)res;
1473 }
1474 
1475 
str_unpack(lua_State * L)1476 static int str_unpack (lua_State *L) {
1477   Header h;
1478   const char *fmt = luaL_checkstring(L, 1);
1479   size_t ld;
1480   const char *data = luaL_checklstring(L, 2, &ld);
1481   size_t pos = (size_t)posrelat(luaL_optinteger(L, 3, 1), ld) - 1;
1482   int n = 0;  /* number of results */
1483   luaL_argcheck(L, pos <= ld, 3, "initial position out of string");
1484   initheader(L, &h);
1485   while (*fmt != '\0') {
1486     int size, ntoalign;
1487     KOption opt = getdetails(&h, pos, &fmt, &size, &ntoalign);
1488     if ((size_t)ntoalign + size > ~pos || pos + ntoalign + size > ld)
1489       luaL_argerror(L, 2, "data string too short");
1490     pos += ntoalign;  /* skip alignment */
1491     /* stack space for item + next position */
1492     luaL_checkstack(L, 2, "too many results");
1493     n++;
1494     switch (opt) {
1495       case Kint:
1496       case Kuint: {
1497         lua_Integer res = unpackint(L, data + pos, h.islittle, size,
1498                                        (opt == Kint));
1499         lua_pushinteger(L, res);
1500         break;
1501       }
1502       case Kfloat: {
1503         volatile Ftypes u;
1504         lua_Number num;
1505         copywithendian(u.buff, data + pos, size, h.islittle);
1506         if (size == sizeof(u.f)) num = (lua_Number)u.f;
1507         else if (size == sizeof(u.d)) num = (lua_Number)u.d;
1508         else num = u.n;
1509         lua_pushnumber(L, num);
1510         break;
1511       }
1512       case Kchar: {
1513         lua_pushlstring(L, data + pos, size);
1514         break;
1515       }
1516       case Kstring: {
1517         size_t len = (size_t)unpackint(L, data + pos, h.islittle, size, 0);
1518         luaL_argcheck(L, pos + len + size <= ld, 2, "data string too short");
1519         lua_pushlstring(L, data + pos + size, len);
1520         pos += len;  /* skip string */
1521         break;
1522       }
1523       case Kzstr: {
1524         size_t len = (int)strlen(data + pos);
1525         lua_pushlstring(L, data + pos, len);
1526         pos += len + 1;  /* skip string plus final '\0' */
1527         break;
1528       }
1529       case Kpaddalign: case Kpadding: case Knop:
1530         n--;  /* undo increment */
1531         break;
1532     }
1533     pos += size;
1534   }
1535   lua_pushinteger(L, pos + 1);  /* next position */
1536   return n + 1;
1537 }
1538 
1539 /* }====================================================== */
1540 
1541 
1542 static const luaL_Reg strlib[] = {
1543   {"byte", str_byte},
1544   {"char", str_char},
1545   {"dump", str_dump},
1546   {"find", str_find},
1547   {"format", str_format},
1548   {"gmatch", gmatch},
1549   {"gsub", str_gsub},
1550   {"len", str_len},
1551   {"lower", str_lower},
1552   {"match", str_match},
1553   {"rep", str_rep},
1554   {"reverse", str_reverse},
1555   {"sub", str_sub},
1556   {"upper", str_upper},
1557   {"pack", str_pack},
1558   {"packsize", str_packsize},
1559   {"unpack", str_unpack},
1560   {NULL, NULL}
1561 };
1562 
1563 
createmetatable(lua_State * L)1564 static void createmetatable (lua_State *L) {
1565   lua_createtable(L, 0, 1);  /* table to be metatable for strings */
1566   lua_pushliteral(L, "");  /* dummy string */
1567   lua_pushvalue(L, -2);  /* copy table */
1568   lua_setmetatable(L, -2);  /* set table as metatable for strings */
1569   lua_pop(L, 1);  /* pop dummy string */
1570   lua_pushvalue(L, -2);  /* get string library */
1571   lua_setfield(L, -2, "__index");  /* metatable.__index = string */
1572   lua_pop(L, 1);  /* pop metatable */
1573 }
1574 
1575 
1576 /*
1577 ** Open string library
1578 */
luaopen_string(lua_State * L)1579 LUAMOD_API int luaopen_string (lua_State *L) {
1580   luaL_newlib(L, strlib);
1581   createmetatable(L);
1582   return 1;
1583 }
1584 
1585