1/*
2** $Id: lstrlib.c,v 1.178.1.1 2013/04/12 18:48:47 roberto Exp $
3** Standard library for string operations and pattern-matching
4** See Copyright Notice in lua.h
5*/
6
7
8#include <sys/ctype.h>
9#include <sys/zfs_context.h>
10
11#define lstrlib_c
12#define LUA_LIB
13
14#include "lua.h"
15
16#include "lauxlib.h"
17#include "lualib.h"
18
19
20/*
21** maximum number of captures that a pattern can do during
22** pattern-matching. This limit is arbitrary.
23*/
24#if !defined(LUA_MAXCAPTURES)
25#define LUA_MAXCAPTURES		32
26#endif
27
28
29/* macro to `unsign' a character */
30#define uchar(c)	((unsigned char)(c))
31
32/*
33 * PATCHED: add missing character macros.
34 */
35#ifdef illumos
36#define tolower(C)	(((C) >= 'A' && (C) <= 'Z') ? (C) - 'A' + 'a' : (C))
37#define toupper(C)      (((C) >= 'a' && (C) <= 'z') ? (C) - 'a' + 'A': (C))
38#define iscntrl(C)	((((C) >= 0) && ((C) <= 0x1f)) || ((C) == 0x7f))
39#else
40#define	isalnum(C)      (isalpha(C) || isdigit(C))
41#define iscntrl(C)	(uchar(C) <= 0x1f || uchar(C) == 0x7f)
42#endif
43#define isgraph(C)	((C) >= 0x21 && (C) <= 0x7E)
44#define ispunct(C)	(((C) >= 0x21 && (C) <= 0x2F) || \
45    ((C) >= 0x3A && (C) <= 0x40) || \
46    ((C) >= 0x5B && (C) <= 0x60) || \
47    ((C) >= 0x7B && (C) <= 0x7E))
48
49/*
50 * The provided version of sprintf returns a char *, but str_format expects
51 * it to return the number of characters printed. This version has the expected
52 * behavior.
53 */
54static size_t str_sprintf(char *buf, const char *fmt, ...) {
55  va_list args;
56  size_t len;
57
58  va_start(args, fmt);
59  len = vsnprintf(buf, INT_MAX, fmt, args);
60  va_end(args);
61
62  return len;
63}
64
65
66static int str_len (lua_State *L) {
67  size_t l;
68  luaL_checklstring(L, 1, &l);
69  lua_pushinteger(L, (lua_Integer)l);
70  return 1;
71}
72
73
74/* translate a relative string position: negative means back from end */
75static size_t posrelat (ptrdiff_t pos, size_t len) {
76  if (pos >= 0) return (size_t)pos;
77  else if (0u - (size_t)pos > len) return 0;
78  else return len - ((size_t)-pos) + 1;
79}
80
81
82static int str_sub (lua_State *L) {
83  size_t l;
84  const char *s = luaL_checklstring(L, 1, &l);
85  size_t start = posrelat(luaL_checkinteger(L, 2), l);
86  size_t end = posrelat(luaL_optinteger(L, 3, -1), l);
87  if (start < 1) start = 1;
88  if (end > l) end = l;
89  if (start <= end)
90    lua_pushlstring(L, s + start - 1, end - start + 1);
91  else lua_pushliteral(L, "");
92  return 1;
93}
94
95
96static int str_reverse (lua_State *L) {
97  size_t l, i;
98  luaL_Buffer b;
99  const char *s = luaL_checklstring(L, 1, &l);
100  char *p = luaL_buffinitsize(L, &b, l);
101  for (i = 0; i < l; i++)
102    p[i] = s[l - i - 1];
103  luaL_pushresultsize(&b, l);
104  return 1;
105}
106
107
108static int str_lower (lua_State *L) {
109  size_t l;
110  size_t i;
111  luaL_Buffer b;
112  const char *s = luaL_checklstring(L, 1, &l);
113  char *p = luaL_buffinitsize(L, &b, l);
114  for (i=0; i<l; i++)
115    p[i] = tolower(uchar(s[i]));
116  luaL_pushresultsize(&b, l);
117  return 1;
118}
119
120
121static int str_upper (lua_State *L) {
122  size_t l;
123  size_t i;
124  luaL_Buffer b;
125  const char *s = luaL_checklstring(L, 1, &l);
126  char *p = luaL_buffinitsize(L, &b, l);
127  for (i=0; i<l; i++)
128    p[i] = toupper(uchar(s[i]));
129  luaL_pushresultsize(&b, l);
130  return 1;
131}
132
133
134/* reasonable limit to avoid arithmetic overflow */
135#define MAXSIZE		((~(size_t)0) >> 1)
136
137static int str_rep (lua_State *L) {
138  size_t l, lsep;
139  const char *s = luaL_checklstring(L, 1, &l);
140  int n = luaL_checkint(L, 2);
141  const char *sep = luaL_optlstring(L, 3, "", &lsep);
142  if (n <= 0) lua_pushliteral(L, "");
143  else if (l + lsep < l || l + lsep >= MAXSIZE / n)  /* may overflow? */
144    return luaL_error(L, "resulting string too large");
145  else {
146    size_t totallen = n * l + (n - 1) * lsep;
147    luaL_Buffer b;
148    char *p = luaL_buffinitsize(L, &b, totallen);
149    while (n-- > 1) {  /* first n-1 copies (followed by separator) */
150      memcpy(p, s, l * sizeof(char)); p += l;
151      if (lsep > 0) {  /* avoid empty 'memcpy' (may be expensive) */
152        memcpy(p, sep, lsep * sizeof(char)); p += lsep;
153      }
154    }
155    memcpy(p, s, l * sizeof(char));  /* last copy (not followed by separator) */
156    luaL_pushresultsize(&b, totallen);
157  }
158  return 1;
159}
160
161
162static int str_byte (lua_State *L) {
163  size_t l;
164  const char *s = luaL_checklstring(L, 1, &l);
165  size_t posi = posrelat(luaL_optinteger(L, 2, 1), l);
166  size_t pose = posrelat(luaL_optinteger(L, 3, posi), l);
167  int n, i;
168  if (posi < 1) posi = 1;
169  if (pose > l) pose = l;
170  if (posi > pose) return 0;  /* empty interval; return no values */
171  n = (int)(pose -  posi + 1);
172  if (posi + n <= pose)  /* (size_t -> int) overflow? */
173    return luaL_error(L, "string slice too long");
174  luaL_checkstack(L, n, "string slice too long");
175  for (i=0; i<n; i++)
176    lua_pushinteger(L, uchar(s[posi+i-1]));
177  return n;
178}
179
180
181static int str_char (lua_State *L) {
182  int n = lua_gettop(L);  /* number of arguments */
183  int i;
184  luaL_Buffer b;
185  char *p = luaL_buffinitsize(L, &b, n);
186  for (i=1; i<=n; i++) {
187    int c = luaL_checkint(L, i);
188    luaL_argcheck(L, uchar(c) == c, i, "value out of range");
189    p[i - 1] = uchar(c);
190  }
191  luaL_pushresultsize(&b, n);
192  return 1;
193}
194
195
196static int writer (lua_State *L, const void* b, size_t size, void* B) {
197  (void)L;
198  luaL_addlstring((luaL_Buffer*) B, (const char *)b, size);
199  return 0;
200}
201
202
203static int str_dump (lua_State *L) {
204  luaL_Buffer b;
205  luaL_checktype(L, 1, LUA_TFUNCTION);
206  lua_settop(L, 1);
207  luaL_buffinit(L,&b);
208  if (lua_dump(L, writer, &b) != 0)
209    return luaL_error(L, "unable to dump given function");
210  luaL_pushresult(&b);
211  return 1;
212}
213
214
215
216/*
217** {======================================================
218** PATTERN MATCHING
219** =======================================================
220*/
221
222
223#define CAP_UNFINISHED	(-1)
224#define CAP_POSITION	(-2)
225
226
227typedef struct MatchState {
228  int matchdepth;  /* control for recursive depth (to avoid C stack overflow) */
229  const char *src_init;  /* init of source string */
230  const char *src_end;  /* end ('\0') of source string */
231  const char *p_end;  /* end ('\0') of pattern */
232  lua_State *L;
233  int level;  /* total number of captures (finished or unfinished) */
234  struct {
235    const char *init;
236    ptrdiff_t len;
237  } capture[LUA_MAXCAPTURES];
238} MatchState;
239
240
241/* recursive function */
242static const char *match (MatchState *ms, const char *s, const char *p);
243
244
245/* maximum recursion depth for 'match' */
246#if !defined(MAXCCALLS)
247#define MAXCCALLS	200
248#endif
249
250
251#define L_ESC		'%'
252#define SPECIALS	"^$*+?.([%-"
253
254
255static int check_capture (MatchState *ms, int l) {
256  l -= '1';
257  if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
258    return luaL_error(ms->L, "invalid capture index %%%d", l + 1);
259  return l;
260}
261
262
263static int capture_to_close (MatchState *ms) {
264  int level = ms->level;
265  for (level--; level>=0; level--)
266    if (ms->capture[level].len == CAP_UNFINISHED) return level;
267  return luaL_error(ms->L, "invalid pattern capture");
268}
269
270
271static const char *classend (MatchState *ms, const char *p) {
272  switch (*p++) {
273    case L_ESC: {
274      if (p == ms->p_end)
275        luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");
276      return p+1;
277    }
278    case '[': {
279      if (*p == '^') p++;
280      do {  /* look for a `]' */
281        if (p == ms->p_end)
282          luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");
283        if (*(p++) == L_ESC && p < ms->p_end)
284          p++;  /* skip escapes (e.g. `%]') */
285      } while (*p != ']');
286      return p+1;
287    }
288    default: {
289      return p;
290    }
291  }
292}
293
294
295static int match_class (int c, int cl) {
296  int res;
297  switch (tolower(cl)) {
298    case 'a' : res = isalpha(c); break;
299    case 'c' : res = iscntrl(c); break;
300    case 'd' : res = isdigit(c); break;
301    case 'g' : res = isgraph(c); break;
302    case 'l' : res = islower(c); break;
303    case 'p' : res = ispunct(c); break;
304    case 's' : res = isspace(c); break;
305    case 'u' : res = isupper(c); break;
306    case 'w' : res = isalnum(c); break;
307    case 'x' : res = isxdigit(c); break;
308    case 'z' : res = (c == 0); break;  /* deprecated option */
309    default: return (cl == c);
310  }
311  return (islower(cl) ? res : !res);
312}
313
314
315static int matchbracketclass (int c, const char *p, const char *ec) {
316  int sig = 1;
317  if (*(p+1) == '^') {
318    sig = 0;
319    p++;  /* skip the `^' */
320  }
321  while (++p < ec) {
322    if (*p == L_ESC) {
323      p++;
324      if (match_class(c, uchar(*p)))
325        return sig;
326    }
327    else if ((*(p+1) == '-') && (p+2 < ec)) {
328      p+=2;
329      if (uchar(*(p-2)) <= c && c <= uchar(*p))
330        return sig;
331    }
332    else if (uchar(*p) == c) return sig;
333  }
334  return !sig;
335}
336
337
338static int singlematch (MatchState *ms, const char *s, const char *p,
339                        const char *ep) {
340  if (s >= ms->src_end)
341    return 0;
342  else {
343    int c = uchar(*s);
344    switch (*p) {
345      case '.': return 1;  /* matches any char */
346      case L_ESC: return match_class(c, uchar(*(p+1)));
347      case '[': return matchbracketclass(c, p, ep-1);
348      default:  return (uchar(*p) == c);
349    }
350  }
351}
352
353
354static const char *matchbalance (MatchState *ms, const char *s,
355                                   const char *p) {
356  if (p >= ms->p_end - 1)
357    luaL_error(ms->L, "malformed pattern "
358                      "(missing arguments to " LUA_QL("%%b") ")");
359  if (*s != *p) return NULL;
360  else {
361    int b = *p;
362    int e = *(p+1);
363    int cont = 1;
364    while (++s < ms->src_end) {
365      if (*s == e) {
366        if (--cont == 0) return s+1;
367      }
368      else if (*s == b) cont++;
369    }
370  }
371  return NULL;  /* string ends out of balance */
372}
373
374
375static const char *max_expand (MatchState *ms, const char *s,
376                                 const char *p, const char *ep) {
377  ptrdiff_t i = 0;  /* counts maximum expand for item */
378  while (singlematch(ms, s + i, p, ep))
379    i++;
380  /* keeps trying to match with the maximum repetitions */
381  while (i>=0) {
382    const char *res = match(ms, (s+i), ep+1);
383    if (res) return res;
384    i--;  /* else didn't match; reduce 1 repetition to try again */
385  }
386  return NULL;
387}
388
389
390static const char *min_expand (MatchState *ms, const char *s,
391                                 const char *p, const char *ep) {
392  for (;;) {
393    const char *res = match(ms, s, ep+1);
394    if (res != NULL)
395      return res;
396    else if (singlematch(ms, s, p, ep))
397      s++;  /* try with one more repetition */
398    else return NULL;
399  }
400}
401
402
403static const char *start_capture (MatchState *ms, const char *s,
404                                    const char *p, int what) {
405  const char *res;
406  int level = ms->level;
407  if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
408  ms->capture[level].init = s;
409  ms->capture[level].len = what;
410  ms->level = level+1;
411  if ((res=match(ms, s, p)) == NULL)  /* match failed? */
412    ms->level--;  /* undo capture */
413  return res;
414}
415
416
417static const char *end_capture (MatchState *ms, const char *s,
418                                  const char *p) {
419  int l = capture_to_close(ms);
420  const char *res;
421  ms->capture[l].len = s - ms->capture[l].init;  /* close capture */
422  if ((res = match(ms, s, p)) == NULL)  /* match failed? */
423    ms->capture[l].len = CAP_UNFINISHED;  /* undo capture */
424  return res;
425}
426
427
428static const char *match_capture (MatchState *ms, const char *s, int l) {
429  size_t len;
430  l = check_capture(ms, l);
431  len = ms->capture[l].len;
432  if ((size_t)(ms->src_end-s) >= len &&
433      memcmp(ms->capture[l].init, s, len) == 0)
434    return s+len;
435  else return NULL;
436}
437
438
439static const char *match (MatchState *ms, const char *s, const char *p) {
440  if (ms->matchdepth-- == 0)
441    luaL_error(ms->L, "pattern too complex");
442  init: /* using goto's to optimize tail recursion */
443  if (p != ms->p_end) {  /* end of pattern? */
444    switch (*p) {
445      case '(': {  /* start capture */
446        if (*(p + 1) == ')')  /* position capture? */
447          s = start_capture(ms, s, p + 2, CAP_POSITION);
448        else
449          s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
450        break;
451      }
452      case ')': {  /* end capture */
453        s = end_capture(ms, s, p + 1);
454        break;
455      }
456      case '$': {
457        if ((p + 1) != ms->p_end)  /* is the `$' the last char in pattern? */
458          goto dflt;  /* no; go to default */
459        s = (s == ms->src_end) ? s : NULL;  /* check end of string */
460        break;
461      }
462      case L_ESC: {  /* escaped sequences not in the format class[*+?-]? */
463        switch (*(p + 1)) {
464          case 'b': {  /* balanced string? */
465            s = matchbalance(ms, s, p + 2);
466            if (s != NULL) {
467              p += 4; goto init;  /* return match(ms, s, p + 4); */
468            }  /* else fail (s == NULL) */
469            break;
470          }
471          case 'f': {  /* frontier? */
472            const char *ep; char previous;
473            p += 2;
474            if (*p != '[')
475              luaL_error(ms->L, "missing " LUA_QL("[") " after "
476                                 LUA_QL("%%f") " in pattern");
477            ep = classend(ms, p);  /* points to what is next */
478            previous = (s == ms->src_init) ? '\0' : *(s - 1);
479            if (!matchbracketclass(uchar(previous), p, ep - 1) &&
480               matchbracketclass(uchar(*s), p, ep - 1)) {
481              p = ep; goto init;  /* return match(ms, s, ep); */
482            }
483            s = NULL;  /* match failed */
484            break;
485          }
486          case '0': case '1': case '2': case '3':
487          case '4': case '5': case '6': case '7':
488          case '8': case '9': {  /* capture results (%0-%9)? */
489            s = match_capture(ms, s, uchar(*(p + 1)));
490            if (s != NULL) {
491              p += 2; goto init;  /* return match(ms, s, p + 2) */
492            }
493            break;
494          }
495          default: goto dflt;
496        }
497        break;
498      }
499      default: dflt: {  /* pattern class plus optional suffix */
500        const char *ep = classend(ms, p);  /* points to optional suffix */
501        /* does not match at least once? */
502        if (!singlematch(ms, s, p, ep)) {
503          if (*ep == '*' || *ep == '?' || *ep == '-') {  /* accept empty? */
504            p = ep + 1; goto init;  /* return match(ms, s, ep + 1); */
505          }
506          else  /* '+' or no suffix */
507            s = NULL;  /* fail */
508        }
509        else {  /* matched once */
510          switch (*ep) {  /* handle optional suffix */
511            case '?': {  /* optional */
512              const char *res;
513              if ((res = match(ms, s + 1, ep + 1)) != NULL)
514                s = res;
515              else {
516                p = ep + 1; goto init;  /* else return match(ms, s, ep + 1); */
517              }
518              break;
519            }
520            case '+':  /* 1 or more repetitions */
521              s++;  /* 1 match already done */
522              /* FALLTHROUGH */
523            case '*':  /* 0 or more repetitions */
524              s = max_expand(ms, s, p, ep);
525              break;
526            case '-':  /* 0 or more repetitions (minimum) */
527              s = min_expand(ms, s, p, ep);
528              break;
529            default:  /* no suffix */
530              s++; p = ep; goto init;  /* return match(ms, s + 1, ep); */
531          }
532        }
533        break;
534      }
535    }
536  }
537  ms->matchdepth++;
538  return s;
539}
540
541
542
543static const char *lmemfind (const char *s1, size_t l1,
544                               const char *s2, size_t l2) {
545  if (l2 == 0) return s1;  /* empty strings are everywhere */
546  else if (l2 > l1) return NULL;  /* avoids a negative `l1' */
547  else {
548    const char *init;  /* to search for a `*s2' inside `s1' */
549    l2--;  /* 1st char will be checked by `memchr' */
550    l1 = l1-l2;  /* `s2' cannot be found after that */
551    while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
552      init++;   /* 1st char is already checked */
553      if (memcmp(init, s2+1, l2) == 0)
554        return init-1;
555      else {  /* correct `l1' and `s1' to try again */
556        l1 -= init-s1;
557        s1 = init;
558      }
559    }
560    return NULL;  /* not found */
561  }
562}
563
564
565static void push_onecapture (MatchState *ms, int i, const char *s,
566                                                    const char *e) {
567  if (i >= ms->level) {
568    if (i == 0)  /* ms->level == 0, too */
569      lua_pushlstring(ms->L, s, e - s);  /* add whole match */
570    else
571      luaL_error(ms->L, "invalid capture index");
572  }
573  else {
574    ptrdiff_t l = ms->capture[i].len;
575    if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
576    if (l == CAP_POSITION)
577      lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
578    else
579      lua_pushlstring(ms->L, ms->capture[i].init, l);
580  }
581}
582
583
584static int push_captures (MatchState *ms, const char *s, const char *e) {
585  int i;
586  int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
587  luaL_checkstack(ms->L, nlevels, "too many captures");
588  for (i = 0; i < nlevels; i++)
589    push_onecapture(ms, i, s, e);
590  return nlevels;  /* number of strings pushed */
591}
592
593
594/* check whether pattern has no special characters */
595static int nospecials (const char *p, size_t l) {
596  size_t upto = 0;
597  do {
598    if (strpbrk(p + upto, SPECIALS))
599      return 0;  /* pattern has a special character */
600    upto += strlen(p + upto) + 1;  /* may have more after \0 */
601  } while (upto <= l);
602  return 1;  /* no special chars found */
603}
604
605
606static int str_find_aux (lua_State *L, int find) {
607  size_t ls, lp;
608  const char *s = luaL_checklstring(L, 1, &ls);
609  const char *p = luaL_checklstring(L, 2, &lp);
610  size_t init = posrelat(luaL_optinteger(L, 3, 1), ls);
611  if (init < 1) init = 1;
612  else if (init > ls + 1) {  /* start after string's end? */
613    lua_pushnil(L);  /* cannot find anything */
614    return 1;
615  }
616  /* explicit request or no special characters? */
617  if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {
618    /* do a plain search */
619    const char *s2 = lmemfind(s + init - 1, ls - init + 1, p, lp);
620    if (s2) {
621      lua_pushinteger(L, s2 - s + 1);
622      lua_pushinteger(L, s2 - s + lp);
623      return 2;
624    }
625  }
626  else {
627    MatchState ms;
628    const char *s1 = s + init - 1;
629    int anchor = (*p == '^');
630    if (anchor) {
631      p++; lp--;  /* skip anchor character */
632    }
633    ms.L = L;
634    ms.matchdepth = MAXCCALLS;
635    ms.src_init = s;
636    ms.src_end = s + ls;
637    ms.p_end = p + lp;
638    do {
639      const char *res;
640      ms.level = 0;
641      lua_assert(ms.matchdepth == MAXCCALLS);
642      if ((res=match(&ms, s1, p)) != NULL) {
643        if (find) {
644          lua_pushinteger(L, s1 - s + 1);  /* start */
645          lua_pushinteger(L, res - s);   /* end */
646          return push_captures(&ms, NULL, 0) + 2;
647        }
648        else
649          return push_captures(&ms, s1, res);
650      }
651    } while (s1++ < ms.src_end && !anchor);
652  }
653  lua_pushnil(L);  /* not found */
654  return 1;
655}
656
657
658static int str_find (lua_State *L) {
659  return str_find_aux(L, 1);
660}
661
662
663static int str_match (lua_State *L) {
664  return str_find_aux(L, 0);
665}
666
667
668static int gmatch_aux (lua_State *L) {
669  MatchState ms;
670  size_t ls, lp;
671  const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
672  const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp);
673  const char *src;
674  ms.L = L;
675  ms.matchdepth = MAXCCALLS;
676  ms.src_init = s;
677  ms.src_end = s+ls;
678  ms.p_end = p + lp;
679  for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
680       src <= ms.src_end;
681       src++) {
682    const char *e;
683    ms.level = 0;
684    lua_assert(ms.matchdepth == MAXCCALLS);
685    if ((e = match(&ms, src, p)) != NULL) {
686      lua_Integer newstart = e-s;
687      if (e == src) newstart++;  /* empty match? go at least one position */
688      lua_pushinteger(L, newstart);
689      lua_replace(L, lua_upvalueindex(3));
690      return push_captures(&ms, src, e);
691    }
692  }
693  return 0;  /* not found */
694}
695
696
697static int str_gmatch (lua_State *L) {
698  luaL_checkstring(L, 1);
699  luaL_checkstring(L, 2);
700  lua_settop(L, 2);
701  lua_pushinteger(L, 0);
702  lua_pushcclosure(L, gmatch_aux, 3);
703  return 1;
704}
705
706
707static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
708                                                   const char *e) {
709  size_t l, i;
710  const char *news = lua_tolstring(ms->L, 3, &l);
711  for (i = 0; i < l; i++) {
712    if (news[i] != L_ESC)
713      luaL_addchar(b, news[i]);
714    else {
715      i++;  /* skip ESC */
716      if (!isdigit(uchar(news[i]))) {
717        if (news[i] != L_ESC)
718          luaL_error(ms->L, "invalid use of " LUA_QL("%c")
719                           " in replacement string", L_ESC);
720        luaL_addchar(b, news[i]);
721      }
722      else if (news[i] == '0')
723          luaL_addlstring(b, s, e - s);
724      else {
725        push_onecapture(ms, news[i] - '1', s, e);
726        luaL_addvalue(b);  /* add capture to accumulated result */
727      }
728    }
729  }
730}
731
732
733static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
734                                       const char *e, int tr) {
735  lua_State *L = ms->L;
736  switch (tr) {
737    case LUA_TFUNCTION: {
738      int n;
739      lua_pushvalue(L, 3);
740      n = push_captures(ms, s, e);
741      lua_call(L, n, 1);
742      break;
743    }
744    case LUA_TTABLE: {
745      push_onecapture(ms, 0, s, e);
746      lua_gettable(L, 3);
747      break;
748    }
749    default: {  /* LUA_TNUMBER or LUA_TSTRING */
750      add_s(ms, b, s, e);
751      return;
752    }
753  }
754  if (!lua_toboolean(L, -1)) {  /* nil or false? */
755    lua_pop(L, 1);
756    lua_pushlstring(L, s, e - s);  /* keep original text */
757  }
758  else if (!lua_isstring(L, -1))
759    luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
760  luaL_addvalue(b);  /* add result to accumulator */
761}
762
763
764static int str_gsub (lua_State *L) {
765  size_t srcl, lp;
766  const char *src = luaL_checklstring(L, 1, &srcl);
767  const char *p = luaL_checklstring(L, 2, &lp);
768  int tr = lua_type(L, 3);
769  size_t max_s = luaL_optinteger(L, 4, srcl+1);
770  int anchor = (*p == '^');
771  size_t n = 0;
772  MatchState ms;
773  luaL_Buffer b;
774  luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
775                   tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
776                      "string/function/table expected");
777  luaL_buffinit(L, &b);
778  if (anchor) {
779    p++; lp--;  /* skip anchor character */
780  }
781  ms.L = L;
782  ms.matchdepth = MAXCCALLS;
783  ms.src_init = src;
784  ms.src_end = src+srcl;
785  ms.p_end = p + lp;
786  while (n < max_s) {
787    const char *e;
788    ms.level = 0;
789    lua_assert(ms.matchdepth == MAXCCALLS);
790    e = match(&ms, src, p);
791    if (e) {
792      n++;
793      add_value(&ms, &b, src, e, tr);
794    }
795    if (e && e>src) /* non empty match? */
796      src = e;  /* skip it */
797    else if (src < ms.src_end)
798      luaL_addchar(&b, *src++);
799    else break;
800    if (anchor) break;
801  }
802  luaL_addlstring(&b, src, ms.src_end-src);
803  luaL_pushresult(&b);
804  lua_pushinteger(L, n);  /* number of substitutions */
805  return 2;
806}
807
808/* }====================================================== */
809
810
811
812/*
813** {======================================================
814** STRING FORMAT
815** =======================================================
816*/
817
818/*
819** LUA_INTFRMLEN is the length modifier for integer conversions in
820** 'string.format'; LUA_INTFRM_T is the integer type corresponding to
821** the previous length
822*/
823#if !defined(LUA_INTFRMLEN)	/* { */
824#if defined(LUA_USE_LONGLONG)
825
826#define LUA_INTFRMLEN		"ll"
827#define LUA_INTFRM_T		long long
828
829#else
830
831#define LUA_INTFRMLEN		"l"
832#define LUA_INTFRM_T		long
833
834#endif
835#endif				/* } */
836
837
838/*
839** LUA_FLTFRMLEN is the length modifier for float conversions in
840** 'string.format'; LUA_FLTFRM_T is the float type corresponding to
841** the previous length
842*/
843#if !defined(LUA_FLTFRMLEN)
844
845#define LUA_FLTFRMLEN		""
846#define LUA_FLTFRM_T		double
847
848#endif
849
850
851/* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
852#define MAX_ITEM	512
853/* valid flags in a format specification */
854#define FLAGS	"-+ #0"
855/*
856** maximum size of each format specification (such as '%-099.99d')
857** (+10 accounts for %99.99x plus margin of error)
858*/
859#define MAX_FORMAT	(sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
860
861
862static void addquoted (lua_State *L, luaL_Buffer *b, int arg) {
863  size_t l;
864  const char *s = luaL_checklstring(L, arg, &l);
865  luaL_addchar(b, '"');
866  while (l--) {
867    if (*s == '"' || *s == '\\' || *s == '\n') {
868      luaL_addchar(b, '\\');
869      luaL_addchar(b, *s);
870    }
871    else if (*s == '\0' || iscntrl(uchar(*s))) {
872      char buff[10];
873      if (!isdigit(uchar(*(s+1))))
874        sprintf(buff, "\\%d", (int)uchar(*s));
875      else
876        sprintf(buff, "\\%03d", (int)uchar(*s));
877      luaL_addstring(b, buff);
878    }
879    else
880      luaL_addchar(b, *s);
881    s++;
882  }
883  luaL_addchar(b, '"');
884}
885
886static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
887  const char *p = strfrmt;
888  while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++;  /* skip flags */
889  if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char))
890    luaL_error(L, "invalid format (repeated flags)");
891  if (isdigit(uchar(*p))) p++;  /* skip width */
892  if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
893  if (*p == '.') {
894    p++;
895    if (isdigit(uchar(*p))) p++;  /* skip precision */
896    if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
897  }
898  if (isdigit(uchar(*p)))
899    luaL_error(L, "invalid format (width or precision too long)");
900  *(form++) = '%';
901  memcpy(form, strfrmt, (p - strfrmt + 1) * sizeof(char));
902  form += p - strfrmt + 1;
903  *form = '\0';
904  return p;
905}
906
907
908/*
909** add length modifier into formats
910*/
911static void addlenmod (char *form, const char *lenmod) {
912  size_t l = strlen(form);
913  size_t lm = strlen(lenmod);
914  char spec = form[l - 1];
915  strcpy(form + l - 1, lenmod);
916  form[l + lm - 1] = spec;
917  form[l + lm] = '\0';
918}
919
920
921static int str_format (lua_State *L) {
922  int top = lua_gettop(L);
923  int arg = 1;
924  size_t sfl;
925  const char *strfrmt = luaL_checklstring(L, arg, &sfl);
926  const char *strfrmt_end = strfrmt+sfl;
927  luaL_Buffer b;
928  luaL_buffinit(L, &b);
929  while (strfrmt < strfrmt_end) {
930    if (*strfrmt != L_ESC)
931      luaL_addchar(&b, *strfrmt++);
932    else if (*++strfrmt == L_ESC)
933      luaL_addchar(&b, *strfrmt++);  /* %% */
934    else { /* format item */
935      char form[MAX_FORMAT];  /* to store the format (`%...') */
936      char *buff = luaL_prepbuffsize(&b, MAX_ITEM);  /* to put formatted item */
937      int nb = 0;  /* number of bytes in added item */
938      if (++arg > top)
939        luaL_argerror(L, arg, "no value");
940      strfrmt = scanformat(L, strfrmt, form);
941      switch (*strfrmt++) {
942        case 'c': {
943          nb = str_sprintf(buff, form, luaL_checkint(L, arg));
944          break;
945        }
946        case 'd': case 'i': {
947          lua_Number n = luaL_checknumber(L, arg);
948          LUA_INTFRM_T ni = (LUA_INTFRM_T)n;
949          lua_Number diff = n - (lua_Number)ni;
950          luaL_argcheck(L, -1 < diff && diff < 1, arg,
951                        "not a number in proper range");
952          addlenmod(form, LUA_INTFRMLEN);
953          nb = str_sprintf(buff, form, ni);
954          break;
955        }
956        case 'o': case 'u': case 'x': case 'X': {
957          lua_Number n = luaL_checknumber(L, arg);
958          unsigned LUA_INTFRM_T ni = (unsigned LUA_INTFRM_T)n;
959          lua_Number diff = n - (lua_Number)ni;
960          luaL_argcheck(L, -1 < diff && diff < 1, arg,
961                        "not a non-negative number in proper range");
962          addlenmod(form, LUA_INTFRMLEN);
963          nb = str_sprintf(buff, form, ni);
964          break;
965        }
966#if defined(LUA_USE_FLOAT_FORMATS)
967        case 'e': case 'E': case 'f':
968#if defined(LUA_USE_AFORMAT)
969        case 'a': case 'A':
970#endif
971        case 'g': case 'G': {
972          addlenmod(form, LUA_FLTFRMLEN);
973          nb = str_sprintf(buff, form, (LUA_FLTFRM_T)luaL_checknumber(L, arg));
974          break;
975        }
976#endif
977        case 'q': {
978          addquoted(L, &b, arg);
979          break;
980        }
981        case 's': {
982          size_t l;
983          const char *s = luaL_tolstring(L, arg, &l);
984          if (!strchr(form, '.') && l >= 100) {
985            /* no precision and string is too long to be formatted;
986               keep original string */
987            luaL_addvalue(&b);
988            break;
989          }
990          else {
991            nb = str_sprintf(buff, form, s);
992            lua_pop(L, 1);  /* remove result from 'luaL_tolstring' */
993            break;
994          }
995        }
996        default: {  /* also treat cases `pnLlh' */
997          return luaL_error(L, "invalid option " LUA_QL("%%%c") " to "
998                               LUA_QL("format"), *(strfrmt - 1));
999        }
1000      }
1001      luaL_addsize(&b, nb);
1002    }
1003  }
1004  luaL_pushresult(&b);
1005  return 1;
1006}
1007
1008/* }====================================================== */
1009
1010
1011static const luaL_Reg strlib[] = {
1012  {"byte", str_byte},
1013  {"char", str_char},
1014  {"dump", str_dump},
1015  {"find", str_find},
1016  {"format", str_format},
1017  {"gmatch", str_gmatch},
1018  {"gsub", str_gsub},
1019  {"len", str_len},
1020  {"lower", str_lower},
1021  {"match", str_match},
1022  {"rep", str_rep},
1023  {"reverse", str_reverse},
1024  {"sub", str_sub},
1025  {"upper", str_upper},
1026  {NULL, NULL}
1027};
1028
1029
1030static void createmetatable (lua_State *L) {
1031  lua_createtable(L, 0, 1);  /* table to be metatable for strings */
1032  lua_pushliteral(L, "");  /* dummy string */
1033  lua_pushvalue(L, -2);  /* copy table */
1034  lua_setmetatable(L, -2);  /* set table as metatable for strings */
1035  lua_pop(L, 1);  /* pop dummy string */
1036  lua_pushvalue(L, -2);  /* get string library */
1037  lua_setfield(L, -2, "__index");  /* metatable.__index = string */
1038  lua_pop(L, 1);  /* pop metatable */
1039}
1040
1041
1042/*
1043** Open string library
1044*/
1045LUAMOD_API int luaopen_string (lua_State *L) {
1046  luaL_newlib(L, strlib);
1047  createmetatable(L);
1048  return 1;
1049}
1050
1051