xref: /illumos-gate/usr/src/uts/common/fs/zfs/lua/lstrlib.c (revision 46ac8fdf)
1dfc11533SChris Williamson /*
2dfc11533SChris Williamson ** $Id: lstrlib.c,v 1.178.1.1 2013/04/12 18:48:47 roberto Exp $
3dfc11533SChris Williamson ** Standard library for string operations and pattern-matching
4dfc11533SChris Williamson ** See Copyright Notice in lua.h
5dfc11533SChris Williamson */
6dfc11533SChris Williamson 
7dfc11533SChris Williamson 
8dfc11533SChris Williamson #include <sys/ctype.h>
9dfc11533SChris Williamson #include <sys/zfs_context.h>
10dfc11533SChris Williamson 
11dfc11533SChris Williamson #define lstrlib_c
12dfc11533SChris Williamson #define LUA_LIB
13dfc11533SChris Williamson 
14dfc11533SChris Williamson #include "lua.h"
15dfc11533SChris Williamson 
16dfc11533SChris Williamson #include "lauxlib.h"
17dfc11533SChris Williamson #include "lualib.h"
18dfc11533SChris Williamson 
19dfc11533SChris Williamson 
20dfc11533SChris Williamson /*
21dfc11533SChris Williamson ** maximum number of captures that a pattern can do during
22dfc11533SChris Williamson ** pattern-matching. This limit is arbitrary.
23dfc11533SChris Williamson */
24dfc11533SChris Williamson #if !defined(LUA_MAXCAPTURES)
25dfc11533SChris Williamson #define LUA_MAXCAPTURES		32
26dfc11533SChris Williamson #endif
27dfc11533SChris Williamson 
28dfc11533SChris Williamson 
29dfc11533SChris Williamson /* macro to `unsign' a character */
30dfc11533SChris Williamson #define uchar(c)	((unsigned char)(c))
31dfc11533SChris Williamson 
32dfc11533SChris Williamson /*
33dfc11533SChris Williamson  * PATCHED: add missing character macros.
34dfc11533SChris Williamson  */
35dfc11533SChris Williamson #define tolower(C)	(((C) >= 'A' && (C) <= 'Z') ? (C) - 'A' + 'a' : (C))
36dfc11533SChris Williamson #define toupper(C)      (((C) >= 'a' && (C) <= 'z') ? (C) - 'a' + 'A': (C))
37dfc11533SChris Williamson #define iscntrl(C)	((((C) >= 0) && ((C) <= 0x1f)) || ((C) == 0x7f))
38dfc11533SChris Williamson #define isgraph(C)	((C) >= 0x21 && (C) <= 0x7E)
39dfc11533SChris Williamson #define ispunct(C)	(((C) >= 0x21 && (C) <= 0x2F) || \
40dfc11533SChris Williamson     ((C) >= 0x3A && (C) <= 0x40) || \
41dfc11533SChris Williamson     ((C) >= 0x5B && (C) <= 0x60) || \
42dfc11533SChris Williamson     ((C) >= 0x7B && (C) <= 0x7E))
43dfc11533SChris Williamson 
44dfc11533SChris Williamson /*
45dfc11533SChris Williamson  * The provided version of sprintf returns a char *, but str_format expects
46dfc11533SChris Williamson  * it to return the number of characters printed. This version has the expected
47dfc11533SChris Williamson  * behavior.
48dfc11533SChris Williamson  */
49dfc11533SChris Williamson static size_t str_sprintf(char *buf, const char *fmt, ...) {
50dfc11533SChris Williamson   va_list args;
51dfc11533SChris Williamson   size_t len;
52dfc11533SChris Williamson 
53dfc11533SChris Williamson   va_start(args, fmt);
54dfc11533SChris Williamson   len = vsnprintf(buf, INT_MAX, fmt, args);
55dfc11533SChris Williamson   va_end(args);
56dfc11533SChris Williamson 
57dfc11533SChris Williamson   return len;
58dfc11533SChris Williamson }
59dfc11533SChris Williamson 
60dfc11533SChris Williamson 
61dfc11533SChris Williamson static int str_len (lua_State *L) {
62dfc11533SChris Williamson   size_t l;
63dfc11533SChris Williamson   luaL_checklstring(L, 1, &l);
64dfc11533SChris Williamson   lua_pushinteger(L, (lua_Integer)l);
65dfc11533SChris Williamson   return 1;
66dfc11533SChris Williamson }
67dfc11533SChris Williamson 
68dfc11533SChris Williamson 
69dfc11533SChris Williamson /* translate a relative string position: negative means back from end */
70dfc11533SChris Williamson static size_t posrelat (ptrdiff_t pos, size_t len) {
71dfc11533SChris Williamson   if (pos >= 0) return (size_t)pos;
72dfc11533SChris Williamson   else if (0u - (size_t)pos > len) return 0;
73dfc11533SChris Williamson   else return len - ((size_t)-pos) + 1;
74dfc11533SChris Williamson }
75dfc11533SChris Williamson 
76dfc11533SChris Williamson 
77dfc11533SChris Williamson static int str_sub (lua_State *L) {
78dfc11533SChris Williamson   size_t l;
79dfc11533SChris Williamson   const char *s = luaL_checklstring(L, 1, &l);
80dfc11533SChris Williamson   size_t start = posrelat(luaL_checkinteger(L, 2), l);
81dfc11533SChris Williamson   size_t end = posrelat(luaL_optinteger(L, 3, -1), l);
82dfc11533SChris Williamson   if (start < 1) start = 1;
83dfc11533SChris Williamson   if (end > l) end = l;
84dfc11533SChris Williamson   if (start <= end)
85dfc11533SChris Williamson     lua_pushlstring(L, s + start - 1, end - start + 1);
86dfc11533SChris Williamson   else lua_pushliteral(L, "");
87dfc11533SChris Williamson   return 1;
88dfc11533SChris Williamson }
89dfc11533SChris Williamson 
90dfc11533SChris Williamson 
91dfc11533SChris Williamson static int str_reverse (lua_State *L) {
92dfc11533SChris Williamson   size_t l, i;
93dfc11533SChris Williamson   luaL_Buffer b;
94dfc11533SChris Williamson   const char *s = luaL_checklstring(L, 1, &l);
95dfc11533SChris Williamson   char *p = luaL_buffinitsize(L, &b, l);
96dfc11533SChris Williamson   for (i = 0; i < l; i++)
97dfc11533SChris Williamson     p[i] = s[l - i - 1];
98dfc11533SChris Williamson   luaL_pushresultsize(&b, l);
99dfc11533SChris Williamson   return 1;
100dfc11533SChris Williamson }
101dfc11533SChris Williamson 
102dfc11533SChris Williamson 
103dfc11533SChris Williamson static int str_lower (lua_State *L) {
104dfc11533SChris Williamson   size_t l;
105dfc11533SChris Williamson   size_t i;
106dfc11533SChris Williamson   luaL_Buffer b;
107dfc11533SChris Williamson   const char *s = luaL_checklstring(L, 1, &l);
108dfc11533SChris Williamson   char *p = luaL_buffinitsize(L, &b, l);
109dfc11533SChris Williamson   for (i=0; i<l; i++)
110dfc11533SChris Williamson     p[i] = tolower(uchar(s[i]));
111dfc11533SChris Williamson   luaL_pushresultsize(&b, l);
112dfc11533SChris Williamson   return 1;
113dfc11533SChris Williamson }
114dfc11533SChris Williamson 
115dfc11533SChris Williamson 
116dfc11533SChris Williamson static int str_upper (lua_State *L) {
117dfc11533SChris Williamson   size_t l;
118dfc11533SChris Williamson   size_t i;
119dfc11533SChris Williamson   luaL_Buffer b;
120dfc11533SChris Williamson   const char *s = luaL_checklstring(L, 1, &l);
121dfc11533SChris Williamson   char *p = luaL_buffinitsize(L, &b, l);
122dfc11533SChris Williamson   for (i=0; i<l; i++)
123dfc11533SChris Williamson     p[i] = toupper(uchar(s[i]));
124dfc11533SChris Williamson   luaL_pushresultsize(&b, l);
125dfc11533SChris Williamson   return 1;
126dfc11533SChris Williamson }
127dfc11533SChris Williamson 
128dfc11533SChris Williamson 
129dfc11533SChris Williamson /* reasonable limit to avoid arithmetic overflow */
130dfc11533SChris Williamson #define MAXSIZE		((~(size_t)0) >> 1)
131dfc11533SChris Williamson 
132dfc11533SChris Williamson static int str_rep (lua_State *L) {
133dfc11533SChris Williamson   size_t l, lsep;
134dfc11533SChris Williamson   const char *s = luaL_checklstring(L, 1, &l);
135dfc11533SChris Williamson   int n = luaL_checkint(L, 2);
136dfc11533SChris Williamson   const char *sep = luaL_optlstring(L, 3, "", &lsep);
137dfc11533SChris Williamson   if (n <= 0) lua_pushliteral(L, "");
138dfc11533SChris Williamson   else if (l + lsep < l || l + lsep >= MAXSIZE / n)  /* may overflow? */
139dfc11533SChris Williamson     return luaL_error(L, "resulting string too large");
140dfc11533SChris Williamson   else {
141dfc11533SChris Williamson     size_t totallen = n * l + (n - 1) * lsep;
142dfc11533SChris Williamson     luaL_Buffer b;
143dfc11533SChris Williamson     char *p = luaL_buffinitsize(L, &b, totallen);
144dfc11533SChris Williamson     while (n-- > 1) {  /* first n-1 copies (followed by separator) */
145dfc11533SChris Williamson       memcpy(p, s, l * sizeof(char)); p += l;
146dfc11533SChris Williamson       if (lsep > 0) {  /* avoid empty 'memcpy' (may be expensive) */
147dfc11533SChris Williamson         memcpy(p, sep, lsep * sizeof(char)); p += lsep;
148dfc11533SChris Williamson       }
149dfc11533SChris Williamson     }
150dfc11533SChris Williamson     memcpy(p, s, l * sizeof(char));  /* last copy (not followed by separator) */
151dfc11533SChris Williamson     luaL_pushresultsize(&b, totallen);
152dfc11533SChris Williamson   }
153dfc11533SChris Williamson   return 1;
154dfc11533SChris Williamson }
155dfc11533SChris Williamson 
156dfc11533SChris Williamson 
157dfc11533SChris Williamson static int str_byte (lua_State *L) {
158dfc11533SChris Williamson   size_t l;
159dfc11533SChris Williamson   const char *s = luaL_checklstring(L, 1, &l);
160dfc11533SChris Williamson   size_t posi = posrelat(luaL_optinteger(L, 2, 1), l);
161dfc11533SChris Williamson   size_t pose = posrelat(luaL_optinteger(L, 3, posi), l);
162dfc11533SChris Williamson   int n, i;
163dfc11533SChris Williamson   if (posi < 1) posi = 1;
164dfc11533SChris Williamson   if (pose > l) pose = l;
165dfc11533SChris Williamson   if (posi > pose) return 0;  /* empty interval; return no values */
166dfc11533SChris Williamson   n = (int)(pose -  posi + 1);
167dfc11533SChris Williamson   if (posi + n <= pose)  /* (size_t -> int) overflow? */
168dfc11533SChris Williamson     return luaL_error(L, "string slice too long");
169dfc11533SChris Williamson   luaL_checkstack(L, n, "string slice too long");
170dfc11533SChris Williamson   for (i=0; i<n; i++)
171dfc11533SChris Williamson     lua_pushinteger(L, uchar(s[posi+i-1]));
172dfc11533SChris Williamson   return n;
173dfc11533SChris Williamson }
174dfc11533SChris Williamson 
175dfc11533SChris Williamson 
176dfc11533SChris Williamson static int str_char (lua_State *L) {
177dfc11533SChris Williamson   int n = lua_gettop(L);  /* number of arguments */
178dfc11533SChris Williamson   int i;
179dfc11533SChris Williamson   luaL_Buffer b;
180dfc11533SChris Williamson   char *p = luaL_buffinitsize(L, &b, n);
181dfc11533SChris Williamson   for (i=1; i<=n; i++) {
182dfc11533SChris Williamson     int c = luaL_checkint(L, i);
183dfc11533SChris Williamson     luaL_argcheck(L, uchar(c) == c, i, "value out of range");
184dfc11533SChris Williamson     p[i - 1] = uchar(c);
185dfc11533SChris Williamson   }
186dfc11533SChris Williamson   luaL_pushresultsize(&b, n);
187dfc11533SChris Williamson   return 1;
188dfc11533SChris Williamson }
189dfc11533SChris Williamson 
190dfc11533SChris Williamson 
191dfc11533SChris Williamson static int writer (lua_State *L, const void* b, size_t size, void* B) {
192dfc11533SChris Williamson   (void)L;
193dfc11533SChris Williamson   luaL_addlstring((luaL_Buffer*) B, (const char *)b, size);
194dfc11533SChris Williamson   return 0;
195dfc11533SChris Williamson }
196dfc11533SChris Williamson 
197dfc11533SChris Williamson 
198dfc11533SChris Williamson static int str_dump (lua_State *L) {
199dfc11533SChris Williamson   luaL_Buffer b;
200dfc11533SChris Williamson   luaL_checktype(L, 1, LUA_TFUNCTION);
201dfc11533SChris Williamson   lua_settop(L, 1);
202dfc11533SChris Williamson   luaL_buffinit(L,&b);
203dfc11533SChris Williamson   if (lua_dump(L, writer, &b) != 0)
204dfc11533SChris Williamson     return luaL_error(L, "unable to dump given function");
205dfc11533SChris Williamson   luaL_pushresult(&b);
206dfc11533SChris Williamson   return 1;
207dfc11533SChris Williamson }
208dfc11533SChris Williamson 
209dfc11533SChris Williamson 
210dfc11533SChris Williamson 
211dfc11533SChris Williamson /*
212dfc11533SChris Williamson ** {======================================================
213dfc11533SChris Williamson ** PATTERN MATCHING
214dfc11533SChris Williamson ** =======================================================
215dfc11533SChris Williamson */
216dfc11533SChris Williamson 
217dfc11533SChris Williamson 
218dfc11533SChris Williamson #define CAP_UNFINISHED	(-1)
219dfc11533SChris Williamson #define CAP_POSITION	(-2)
220dfc11533SChris Williamson 
221dfc11533SChris Williamson 
222dfc11533SChris Williamson typedef struct MatchState {
223dfc11533SChris Williamson   int matchdepth;  /* control for recursive depth (to avoid C stack overflow) */
224dfc11533SChris Williamson   const char *src_init;  /* init of source string */
225dfc11533SChris Williamson   const char *src_end;  /* end ('\0') of source string */
226dfc11533SChris Williamson   const char *p_end;  /* end ('\0') of pattern */
227dfc11533SChris Williamson   lua_State *L;
228dfc11533SChris Williamson   int level;  /* total number of captures (finished or unfinished) */
229dfc11533SChris Williamson   struct {
230dfc11533SChris Williamson     const char *init;
231dfc11533SChris Williamson     ptrdiff_t len;
232dfc11533SChris Williamson   } capture[LUA_MAXCAPTURES];
233dfc11533SChris Williamson } MatchState;
234dfc11533SChris Williamson 
235dfc11533SChris Williamson 
236dfc11533SChris Williamson /* recursive function */
237dfc11533SChris Williamson static const char *match (MatchState *ms, const char *s, const char *p);
238dfc11533SChris Williamson 
239dfc11533SChris Williamson 
240dfc11533SChris Williamson /* maximum recursion depth for 'match' */
241dfc11533SChris Williamson #if !defined(MAXCCALLS)
242dfc11533SChris Williamson #define MAXCCALLS	200
243dfc11533SChris Williamson #endif
244dfc11533SChris Williamson 
245dfc11533SChris Williamson 
246dfc11533SChris Williamson #define L_ESC		'%'
247dfc11533SChris Williamson #define SPECIALS	"^$*+?.([%-"
248dfc11533SChris Williamson 
249dfc11533SChris Williamson 
250dfc11533SChris Williamson static int check_capture (MatchState *ms, int l) {
251dfc11533SChris Williamson   l -= '1';
252dfc11533SChris Williamson   if (l < 0 || l >= ms->level || ms->capture[l].len == CAP_UNFINISHED)
253dfc11533SChris Williamson     return luaL_error(ms->L, "invalid capture index %%%d", l + 1);
254dfc11533SChris Williamson   return l;
255dfc11533SChris Williamson }
256dfc11533SChris Williamson 
257dfc11533SChris Williamson 
258dfc11533SChris Williamson static int capture_to_close (MatchState *ms) {
259dfc11533SChris Williamson   int level = ms->level;
260dfc11533SChris Williamson   for (level--; level>=0; level--)
261dfc11533SChris Williamson     if (ms->capture[level].len == CAP_UNFINISHED) return level;
262dfc11533SChris Williamson   return luaL_error(ms->L, "invalid pattern capture");
263dfc11533SChris Williamson }
264dfc11533SChris Williamson 
265dfc11533SChris Williamson 
266dfc11533SChris Williamson static const char *classend (MatchState *ms, const char *p) {
267dfc11533SChris Williamson   switch (*p++) {
268dfc11533SChris Williamson     case L_ESC: {
269dfc11533SChris Williamson       if (p == ms->p_end)
270dfc11533SChris Williamson         luaL_error(ms->L, "malformed pattern (ends with " LUA_QL("%%") ")");
271dfc11533SChris Williamson       return p+1;
272dfc11533SChris Williamson     }
273dfc11533SChris Williamson     case '[': {
274dfc11533SChris Williamson       if (*p == '^') p++;
275dfc11533SChris Williamson       do {  /* look for a `]' */
276dfc11533SChris Williamson         if (p == ms->p_end)
277dfc11533SChris Williamson           luaL_error(ms->L, "malformed pattern (missing " LUA_QL("]") ")");
278dfc11533SChris Williamson         if (*(p++) == L_ESC && p < ms->p_end)
279dfc11533SChris Williamson           p++;  /* skip escapes (e.g. `%]') */
280dfc11533SChris Williamson       } while (*p != ']');
281dfc11533SChris Williamson       return p+1;
282dfc11533SChris Williamson     }
283dfc11533SChris Williamson     default: {
284dfc11533SChris Williamson       return p;
285dfc11533SChris Williamson     }
286dfc11533SChris Williamson   }
287dfc11533SChris Williamson }
288dfc11533SChris Williamson 
289dfc11533SChris Williamson 
290dfc11533SChris Williamson static int match_class (int c, int cl) {
291dfc11533SChris Williamson   int res;
292dfc11533SChris Williamson   switch (tolower(cl)) {
293dfc11533SChris Williamson     case 'a' : res = isalpha(c); break;
294dfc11533SChris Williamson     case 'c' : res = iscntrl(c); break;
295dfc11533SChris Williamson     case 'd' : res = isdigit(c); break;
296dfc11533SChris Williamson     case 'g' : res = isgraph(c); break;
297dfc11533SChris Williamson     case 'l' : res = islower(c); break;
298dfc11533SChris Williamson     case 'p' : res = ispunct(c); break;
299dfc11533SChris Williamson     case 's' : res = isspace(c); break;
300dfc11533SChris Williamson     case 'u' : res = isupper(c); break;
301dfc11533SChris Williamson     case 'w' : res = isalnum(c); break;
302dfc11533SChris Williamson     case 'x' : res = isxdigit(c); break;
303dfc11533SChris Williamson     case 'z' : res = (c == 0); break;  /* deprecated option */
304dfc11533SChris Williamson     default: return (cl == c);
305dfc11533SChris Williamson   }
306dfc11533SChris Williamson   return (islower(cl) ? res : !res);
307dfc11533SChris Williamson }
308dfc11533SChris Williamson 
309dfc11533SChris Williamson 
310dfc11533SChris Williamson static int matchbracketclass (int c, const char *p, const char *ec) {
311dfc11533SChris Williamson   int sig = 1;
312dfc11533SChris Williamson   if (*(p+1) == '^') {
313dfc11533SChris Williamson     sig = 0;
314dfc11533SChris Williamson     p++;  /* skip the `^' */
315dfc11533SChris Williamson   }
316dfc11533SChris Williamson   while (++p < ec) {
317dfc11533SChris Williamson     if (*p == L_ESC) {
318dfc11533SChris Williamson       p++;
319dfc11533SChris Williamson       if (match_class(c, uchar(*p)))
320dfc11533SChris Williamson         return sig;
321dfc11533SChris Williamson     }
322dfc11533SChris Williamson     else if ((*(p+1) == '-') && (p+2 < ec)) {
323dfc11533SChris Williamson       p+=2;
324dfc11533SChris Williamson       if (uchar(*(p-2)) <= c && c <= uchar(*p))
325dfc11533SChris Williamson         return sig;
326dfc11533SChris Williamson     }
327dfc11533SChris Williamson     else if (uchar(*p) == c) return sig;
328dfc11533SChris Williamson   }
329dfc11533SChris Williamson   return !sig;
330dfc11533SChris Williamson }
331dfc11533SChris Williamson 
332dfc11533SChris Williamson 
333dfc11533SChris Williamson static int singlematch (MatchState *ms, const char *s, const char *p,
334dfc11533SChris Williamson                         const char *ep) {
335dfc11533SChris Williamson   if (s >= ms->src_end)
336dfc11533SChris Williamson     return 0;
337dfc11533SChris Williamson   else {
338dfc11533SChris Williamson     int c = uchar(*s);
339dfc11533SChris Williamson     switch (*p) {
340dfc11533SChris Williamson       case '.': return 1;  /* matches any char */
341dfc11533SChris Williamson       case L_ESC: return match_class(c, uchar(*(p+1)));
342dfc11533SChris Williamson       case '[': return matchbracketclass(c, p, ep-1);
343dfc11533SChris Williamson       default:  return (uchar(*p) == c);
344dfc11533SChris Williamson     }
345dfc11533SChris Williamson   }
346dfc11533SChris Williamson }
347dfc11533SChris Williamson 
348dfc11533SChris Williamson 
349dfc11533SChris Williamson static const char *matchbalance (MatchState *ms, const char *s,
350dfc11533SChris Williamson                                    const char *p) {
351dfc11533SChris Williamson   if (p >= ms->p_end - 1)
352dfc11533SChris Williamson     luaL_error(ms->L, "malformed pattern "
353dfc11533SChris Williamson                       "(missing arguments to " LUA_QL("%%b") ")");
354dfc11533SChris Williamson   if (*s != *p) return NULL;
355dfc11533SChris Williamson   else {
356dfc11533SChris Williamson     int b = *p;
357dfc11533SChris Williamson     int e = *(p+1);
358dfc11533SChris Williamson     int cont = 1;
359dfc11533SChris Williamson     while (++s < ms->src_end) {
360dfc11533SChris Williamson       if (*s == e) {
361dfc11533SChris Williamson         if (--cont == 0) return s+1;
362dfc11533SChris Williamson       }
363dfc11533SChris Williamson       else if (*s == b) cont++;
364dfc11533SChris Williamson     }
365dfc11533SChris Williamson   }
366dfc11533SChris Williamson   return NULL;  /* string ends out of balance */
367dfc11533SChris Williamson }
368dfc11533SChris Williamson 
369dfc11533SChris Williamson 
370dfc11533SChris Williamson static const char *max_expand (MatchState *ms, const char *s,
371dfc11533SChris Williamson                                  const char *p, const char *ep) {
372dfc11533SChris Williamson   ptrdiff_t i = 0;  /* counts maximum expand for item */
373dfc11533SChris Williamson   while (singlematch(ms, s + i, p, ep))
374dfc11533SChris Williamson     i++;
375dfc11533SChris Williamson   /* keeps trying to match with the maximum repetitions */
376dfc11533SChris Williamson   while (i>=0) {
377dfc11533SChris Williamson     const char *res = match(ms, (s+i), ep+1);
378dfc11533SChris Williamson     if (res) return res;
379dfc11533SChris Williamson     i--;  /* else didn't match; reduce 1 repetition to try again */
380dfc11533SChris Williamson   }
381dfc11533SChris Williamson   return NULL;
382dfc11533SChris Williamson }
383dfc11533SChris Williamson 
384dfc11533SChris Williamson 
385dfc11533SChris Williamson static const char *min_expand (MatchState *ms, const char *s,
386dfc11533SChris Williamson                                  const char *p, const char *ep) {
387dfc11533SChris Williamson   for (;;) {
388dfc11533SChris Williamson     const char *res = match(ms, s, ep+1);
389dfc11533SChris Williamson     if (res != NULL)
390dfc11533SChris Williamson       return res;
391dfc11533SChris Williamson     else if (singlematch(ms, s, p, ep))
392dfc11533SChris Williamson       s++;  /* try with one more repetition */
393dfc11533SChris Williamson     else return NULL;
394dfc11533SChris Williamson   }
395dfc11533SChris Williamson }
396dfc11533SChris Williamson 
397dfc11533SChris Williamson 
398dfc11533SChris Williamson static const char *start_capture (MatchState *ms, const char *s,
399dfc11533SChris Williamson                                     const char *p, int what) {
400dfc11533SChris Williamson   const char *res;
401dfc11533SChris Williamson   int level = ms->level;
402dfc11533SChris Williamson   if (level >= LUA_MAXCAPTURES) luaL_error(ms->L, "too many captures");
403dfc11533SChris Williamson   ms->capture[level].init = s;
404dfc11533SChris Williamson   ms->capture[level].len = what;
405dfc11533SChris Williamson   ms->level = level+1;
406dfc11533SChris Williamson   if ((res=match(ms, s, p)) == NULL)  /* match failed? */
407dfc11533SChris Williamson     ms->level--;  /* undo capture */
408dfc11533SChris Williamson   return res;
409dfc11533SChris Williamson }
410dfc11533SChris Williamson 
411dfc11533SChris Williamson 
412dfc11533SChris Williamson static const char *end_capture (MatchState *ms, const char *s,
413dfc11533SChris Williamson                                   const char *p) {
414dfc11533SChris Williamson   int l = capture_to_close(ms);
415dfc11533SChris Williamson   const char *res;
416dfc11533SChris Williamson   ms->capture[l].len = s - ms->capture[l].init;  /* close capture */
417dfc11533SChris Williamson   if ((res = match(ms, s, p)) == NULL)  /* match failed? */
418dfc11533SChris Williamson     ms->capture[l].len = CAP_UNFINISHED;  /* undo capture */
419dfc11533SChris Williamson   return res;
420dfc11533SChris Williamson }
421dfc11533SChris Williamson 
422dfc11533SChris Williamson 
423dfc11533SChris Williamson static const char *match_capture (MatchState *ms, const char *s, int l) {
424dfc11533SChris Williamson   size_t len;
425dfc11533SChris Williamson   l = check_capture(ms, l);
426dfc11533SChris Williamson   len = ms->capture[l].len;
427dfc11533SChris Williamson   if ((size_t)(ms->src_end-s) >= len &&
428dfc11533SChris Williamson       memcmp(ms->capture[l].init, s, len) == 0)
429dfc11533SChris Williamson     return s+len;
430dfc11533SChris Williamson   else return NULL;
431dfc11533SChris Williamson }
432dfc11533SChris Williamson 
433dfc11533SChris Williamson 
434dfc11533SChris Williamson static const char *match (MatchState *ms, const char *s, const char *p) {
435dfc11533SChris Williamson   if (ms->matchdepth-- == 0)
436dfc11533SChris Williamson     luaL_error(ms->L, "pattern too complex");
437dfc11533SChris Williamson   init: /* using goto's to optimize tail recursion */
438dfc11533SChris Williamson   if (p != ms->p_end) {  /* end of pattern? */
439dfc11533SChris Williamson     switch (*p) {
440dfc11533SChris Williamson       case '(': {  /* start capture */
441dfc11533SChris Williamson         if (*(p + 1) == ')')  /* position capture? */
442dfc11533SChris Williamson           s = start_capture(ms, s, p + 2, CAP_POSITION);
443dfc11533SChris Williamson         else
444dfc11533SChris Williamson           s = start_capture(ms, s, p + 1, CAP_UNFINISHED);
445dfc11533SChris Williamson         break;
446dfc11533SChris Williamson       }
447dfc11533SChris Williamson       case ')': {  /* end capture */
448dfc11533SChris Williamson         s = end_capture(ms, s, p + 1);
449dfc11533SChris Williamson         break;
450dfc11533SChris Williamson       }
451dfc11533SChris Williamson       case '$': {
452dfc11533SChris Williamson         if ((p + 1) != ms->p_end)  /* is the `$' the last char in pattern? */
453dfc11533SChris Williamson           goto dflt;  /* no; go to default */
454dfc11533SChris Williamson         s = (s == ms->src_end) ? s : NULL;  /* check end of string */
455dfc11533SChris Williamson         break;
456dfc11533SChris Williamson       }
457dfc11533SChris Williamson       case L_ESC: {  /* escaped sequences not in the format class[*+?-]? */
458dfc11533SChris Williamson         switch (*(p + 1)) {
459dfc11533SChris Williamson           case 'b': {  /* balanced string? */
460dfc11533SChris Williamson             s = matchbalance(ms, s, p + 2);
461dfc11533SChris Williamson             if (s != NULL) {
462dfc11533SChris Williamson               p += 4; goto init;  /* return match(ms, s, p + 4); */
463dfc11533SChris Williamson             }  /* else fail (s == NULL) */
464dfc11533SChris Williamson             break;
465dfc11533SChris Williamson           }
466dfc11533SChris Williamson           case 'f': {  /* frontier? */
467dfc11533SChris Williamson             const char *ep; char previous;
468dfc11533SChris Williamson             p += 2;
469dfc11533SChris Williamson             if (*p != '[')
470dfc11533SChris Williamson               luaL_error(ms->L, "missing " LUA_QL("[") " after "
471dfc11533SChris Williamson                                  LUA_QL("%%f") " in pattern");
472dfc11533SChris Williamson             ep = classend(ms, p);  /* points to what is next */
473dfc11533SChris Williamson             previous = (s == ms->src_init) ? '\0' : *(s - 1);
474dfc11533SChris Williamson             if (!matchbracketclass(uchar(previous), p, ep - 1) &&
475dfc11533SChris Williamson                matchbracketclass(uchar(*s), p, ep - 1)) {
476dfc11533SChris Williamson               p = ep; goto init;  /* return match(ms, s, ep); */
477dfc11533SChris Williamson             }
478dfc11533SChris Williamson             s = NULL;  /* match failed */
479dfc11533SChris Williamson             break;
480dfc11533SChris Williamson           }
481dfc11533SChris Williamson           case '0': case '1': case '2': case '3':
482dfc11533SChris Williamson           case '4': case '5': case '6': case '7':
483dfc11533SChris Williamson           case '8': case '9': {  /* capture results (%0-%9)? */
484dfc11533SChris Williamson             s = match_capture(ms, s, uchar(*(p + 1)));
485dfc11533SChris Williamson             if (s != NULL) {
486dfc11533SChris Williamson               p += 2; goto init;  /* return match(ms, s, p + 2) */
487dfc11533SChris Williamson             }
488dfc11533SChris Williamson             break;
489dfc11533SChris Williamson           }
490dfc11533SChris Williamson           default: goto dflt;
491dfc11533SChris Williamson         }
492dfc11533SChris Williamson         break;
493dfc11533SChris Williamson       }
494dfc11533SChris Williamson       default: dflt: {  /* pattern class plus optional suffix */
495dfc11533SChris Williamson         const char *ep = classend(ms, p);  /* points to optional suffix */
496dfc11533SChris Williamson         /* does not match at least once? */
497dfc11533SChris Williamson         if (!singlematch(ms, s, p, ep)) {
498dfc11533SChris Williamson           if (*ep == '*' || *ep == '?' || *ep == '-') {  /* accept empty? */
499dfc11533SChris Williamson             p = ep + 1; goto init;  /* return match(ms, s, ep + 1); */
500dfc11533SChris Williamson           }
501dfc11533SChris Williamson           else  /* '+' or no suffix */
502dfc11533SChris Williamson             s = NULL;  /* fail */
503dfc11533SChris Williamson         }
504dfc11533SChris Williamson         else {  /* matched once */
505dfc11533SChris Williamson           switch (*ep) {  /* handle optional suffix */
506dfc11533SChris Williamson             case '?': {  /* optional */
507dfc11533SChris Williamson               const char *res;
508dfc11533SChris Williamson               if ((res = match(ms, s + 1, ep + 1)) != NULL)
509dfc11533SChris Williamson                 s = res;
510dfc11533SChris Williamson               else {
511dfc11533SChris Williamson                 p = ep + 1; goto init;  /* else return match(ms, s, ep + 1); */
512dfc11533SChris Williamson               }
513dfc11533SChris Williamson               break;
514dfc11533SChris Williamson             }
515dfc11533SChris Williamson             case '+':  /* 1 or more repetitions */
516dfc11533SChris Williamson               s++;  /* 1 match already done */
517*46ac8fdfSToomas Soome               /* FALLTHROUGH */
518dfc11533SChris Williamson             case '*':  /* 0 or more repetitions */
519dfc11533SChris Williamson               s = max_expand(ms, s, p, ep);
520dfc11533SChris Williamson               break;
521dfc11533SChris Williamson             case '-':  /* 0 or more repetitions (minimum) */
522dfc11533SChris Williamson               s = min_expand(ms, s, p, ep);
523dfc11533SChris Williamson               break;
524dfc11533SChris Williamson             default:  /* no suffix */
525dfc11533SChris Williamson               s++; p = ep; goto init;  /* return match(ms, s + 1, ep); */
526dfc11533SChris Williamson           }
527dfc11533SChris Williamson         }
528dfc11533SChris Williamson         break;
529dfc11533SChris Williamson       }
530dfc11533SChris Williamson     }
531dfc11533SChris Williamson   }
532dfc11533SChris Williamson   ms->matchdepth++;
533dfc11533SChris Williamson   return s;
534dfc11533SChris Williamson }
535dfc11533SChris Williamson 
536dfc11533SChris Williamson 
537dfc11533SChris Williamson 
538dfc11533SChris Williamson static const char *lmemfind (const char *s1, size_t l1,
539dfc11533SChris Williamson                                const char *s2, size_t l2) {
540dfc11533SChris Williamson   if (l2 == 0) return s1;  /* empty strings are everywhere */
541dfc11533SChris Williamson   else if (l2 > l1) return NULL;  /* avoids a negative `l1' */
542dfc11533SChris Williamson   else {
543dfc11533SChris Williamson     const char *init;  /* to search for a `*s2' inside `s1' */
544dfc11533SChris Williamson     l2--;  /* 1st char will be checked by `memchr' */
545dfc11533SChris Williamson     l1 = l1-l2;  /* `s2' cannot be found after that */
546dfc11533SChris Williamson     while (l1 > 0 && (init = (const char *)memchr(s1, *s2, l1)) != NULL) {
547dfc11533SChris Williamson       init++;   /* 1st char is already checked */
548dfc11533SChris Williamson       if (memcmp(init, s2+1, l2) == 0)
549dfc11533SChris Williamson         return init-1;
550dfc11533SChris Williamson       else {  /* correct `l1' and `s1' to try again */
551dfc11533SChris Williamson         l1 -= init-s1;
552dfc11533SChris Williamson         s1 = init;
553dfc11533SChris Williamson       }
554dfc11533SChris Williamson     }
555dfc11533SChris Williamson     return NULL;  /* not found */
556dfc11533SChris Williamson   }
557dfc11533SChris Williamson }
558dfc11533SChris Williamson 
559dfc11533SChris Williamson 
560dfc11533SChris Williamson static void push_onecapture (MatchState *ms, int i, const char *s,
561dfc11533SChris Williamson                                                     const char *e) {
562dfc11533SChris Williamson   if (i >= ms->level) {
563dfc11533SChris Williamson     if (i == 0)  /* ms->level == 0, too */
564dfc11533SChris Williamson       lua_pushlstring(ms->L, s, e - s);  /* add whole match */
565dfc11533SChris Williamson     else
566dfc11533SChris Williamson       luaL_error(ms->L, "invalid capture index");
567dfc11533SChris Williamson   }
568dfc11533SChris Williamson   else {
569dfc11533SChris Williamson     ptrdiff_t l = ms->capture[i].len;
570dfc11533SChris Williamson     if (l == CAP_UNFINISHED) luaL_error(ms->L, "unfinished capture");
571dfc11533SChris Williamson     if (l == CAP_POSITION)
572dfc11533SChris Williamson       lua_pushinteger(ms->L, ms->capture[i].init - ms->src_init + 1);
573dfc11533SChris Williamson     else
574dfc11533SChris Williamson       lua_pushlstring(ms->L, ms->capture[i].init, l);
575dfc11533SChris Williamson   }
576dfc11533SChris Williamson }
577dfc11533SChris Williamson 
578dfc11533SChris Williamson 
579dfc11533SChris Williamson static int push_captures (MatchState *ms, const char *s, const char *e) {
580dfc11533SChris Williamson   int i;
581dfc11533SChris Williamson   int nlevels = (ms->level == 0 && s) ? 1 : ms->level;
582dfc11533SChris Williamson   luaL_checkstack(ms->L, nlevels, "too many captures");
583dfc11533SChris Williamson   for (i = 0; i < nlevels; i++)
584dfc11533SChris Williamson     push_onecapture(ms, i, s, e);
585dfc11533SChris Williamson   return nlevels;  /* number of strings pushed */
586dfc11533SChris Williamson }
587dfc11533SChris Williamson 
588dfc11533SChris Williamson 
589dfc11533SChris Williamson /* check whether pattern has no special characters */
590dfc11533SChris Williamson static int nospecials (const char *p, size_t l) {
591dfc11533SChris Williamson   size_t upto = 0;
592dfc11533SChris Williamson   do {
593dfc11533SChris Williamson     if (strpbrk(p + upto, SPECIALS))
594dfc11533SChris Williamson       return 0;  /* pattern has a special character */
595dfc11533SChris Williamson     upto += strlen(p + upto) + 1;  /* may have more after \0 */
596dfc11533SChris Williamson   } while (upto <= l);
597dfc11533SChris Williamson   return 1;  /* no special chars found */
598dfc11533SChris Williamson }
599dfc11533SChris Williamson 
600dfc11533SChris Williamson 
601dfc11533SChris Williamson static int str_find_aux (lua_State *L, int find) {
602dfc11533SChris Williamson   size_t ls, lp;
603dfc11533SChris Williamson   const char *s = luaL_checklstring(L, 1, &ls);
604dfc11533SChris Williamson   const char *p = luaL_checklstring(L, 2, &lp);
605dfc11533SChris Williamson   size_t init = posrelat(luaL_optinteger(L, 3, 1), ls);
606dfc11533SChris Williamson   if (init < 1) init = 1;
607dfc11533SChris Williamson   else if (init > ls + 1) {  /* start after string's end? */
608dfc11533SChris Williamson     lua_pushnil(L);  /* cannot find anything */
609dfc11533SChris Williamson     return 1;
610dfc11533SChris Williamson   }
611dfc11533SChris Williamson   /* explicit request or no special characters? */
612dfc11533SChris Williamson   if (find && (lua_toboolean(L, 4) || nospecials(p, lp))) {
613dfc11533SChris Williamson     /* do a plain search */
614dfc11533SChris Williamson     const char *s2 = lmemfind(s + init - 1, ls - init + 1, p, lp);
615dfc11533SChris Williamson     if (s2) {
616dfc11533SChris Williamson       lua_pushinteger(L, s2 - s + 1);
617dfc11533SChris Williamson       lua_pushinteger(L, s2 - s + lp);
618dfc11533SChris Williamson       return 2;
619dfc11533SChris Williamson     }
620dfc11533SChris Williamson   }
621dfc11533SChris Williamson   else {
622dfc11533SChris Williamson     MatchState ms;
623dfc11533SChris Williamson     const char *s1 = s + init - 1;
624dfc11533SChris Williamson     int anchor = (*p == '^');
625dfc11533SChris Williamson     if (anchor) {
626dfc11533SChris Williamson       p++; lp--;  /* skip anchor character */
627dfc11533SChris Williamson     }
628dfc11533SChris Williamson     ms.L = L;
629dfc11533SChris Williamson     ms.matchdepth = MAXCCALLS;
630dfc11533SChris Williamson     ms.src_init = s;
631dfc11533SChris Williamson     ms.src_end = s + ls;
632dfc11533SChris Williamson     ms.p_end = p + lp;
633dfc11533SChris Williamson     do {
634dfc11533SChris Williamson       const char *res;
635dfc11533SChris Williamson       ms.level = 0;
636dfc11533SChris Williamson       lua_assert(ms.matchdepth == MAXCCALLS);
637dfc11533SChris Williamson       if ((res=match(&ms, s1, p)) != NULL) {
638dfc11533SChris Williamson         if (find) {
639dfc11533SChris Williamson           lua_pushinteger(L, s1 - s + 1);  /* start */
640dfc11533SChris Williamson           lua_pushinteger(L, res - s);   /* end */
641dfc11533SChris Williamson           return push_captures(&ms, NULL, 0) + 2;
642dfc11533SChris Williamson         }
643dfc11533SChris Williamson         else
644dfc11533SChris Williamson           return push_captures(&ms, s1, res);
645dfc11533SChris Williamson       }
646dfc11533SChris Williamson     } while (s1++ < ms.src_end && !anchor);
647dfc11533SChris Williamson   }
648dfc11533SChris Williamson   lua_pushnil(L);  /* not found */
649dfc11533SChris Williamson   return 1;
650dfc11533SChris Williamson }
651dfc11533SChris Williamson 
652dfc11533SChris Williamson 
653dfc11533SChris Williamson static int str_find (lua_State *L) {
654dfc11533SChris Williamson   return str_find_aux(L, 1);
655dfc11533SChris Williamson }
656dfc11533SChris Williamson 
657dfc11533SChris Williamson 
658dfc11533SChris Williamson static int str_match (lua_State *L) {
659dfc11533SChris Williamson   return str_find_aux(L, 0);
660dfc11533SChris Williamson }
661dfc11533SChris Williamson 
662dfc11533SChris Williamson 
663dfc11533SChris Williamson static int gmatch_aux (lua_State *L) {
664dfc11533SChris Williamson   MatchState ms;
665dfc11533SChris Williamson   size_t ls, lp;
666dfc11533SChris Williamson   const char *s = lua_tolstring(L, lua_upvalueindex(1), &ls);
667dfc11533SChris Williamson   const char *p = lua_tolstring(L, lua_upvalueindex(2), &lp);
668dfc11533SChris Williamson   const char *src;
669dfc11533SChris Williamson   ms.L = L;
670dfc11533SChris Williamson   ms.matchdepth = MAXCCALLS;
671dfc11533SChris Williamson   ms.src_init = s;
672dfc11533SChris Williamson   ms.src_end = s+ls;
673dfc11533SChris Williamson   ms.p_end = p + lp;
674dfc11533SChris Williamson   for (src = s + (size_t)lua_tointeger(L, lua_upvalueindex(3));
675dfc11533SChris Williamson        src <= ms.src_end;
676dfc11533SChris Williamson        src++) {
677dfc11533SChris Williamson     const char *e;
678dfc11533SChris Williamson     ms.level = 0;
679dfc11533SChris Williamson     lua_assert(ms.matchdepth == MAXCCALLS);
680dfc11533SChris Williamson     if ((e = match(&ms, src, p)) != NULL) {
681dfc11533SChris Williamson       lua_Integer newstart = e-s;
682dfc11533SChris Williamson       if (e == src) newstart++;  /* empty match? go at least one position */
683dfc11533SChris Williamson       lua_pushinteger(L, newstart);
684dfc11533SChris Williamson       lua_replace(L, lua_upvalueindex(3));
685dfc11533SChris Williamson       return push_captures(&ms, src, e);
686dfc11533SChris Williamson     }
687dfc11533SChris Williamson   }
688dfc11533SChris Williamson   return 0;  /* not found */
689dfc11533SChris Williamson }
690dfc11533SChris Williamson 
691dfc11533SChris Williamson 
692dfc11533SChris Williamson static int str_gmatch (lua_State *L) {
693dfc11533SChris Williamson   luaL_checkstring(L, 1);
694dfc11533SChris Williamson   luaL_checkstring(L, 2);
695dfc11533SChris Williamson   lua_settop(L, 2);
696dfc11533SChris Williamson   lua_pushinteger(L, 0);
697dfc11533SChris Williamson   lua_pushcclosure(L, gmatch_aux, 3);
698dfc11533SChris Williamson   return 1;
699dfc11533SChris Williamson }
700dfc11533SChris Williamson 
701dfc11533SChris Williamson 
702dfc11533SChris Williamson static void add_s (MatchState *ms, luaL_Buffer *b, const char *s,
703dfc11533SChris Williamson                                                    const char *e) {
704dfc11533SChris Williamson   size_t l, i;
705dfc11533SChris Williamson   const char *news = lua_tolstring(ms->L, 3, &l);
706dfc11533SChris Williamson   for (i = 0; i < l; i++) {
707dfc11533SChris Williamson     if (news[i] != L_ESC)
708dfc11533SChris Williamson       luaL_addchar(b, news[i]);
709dfc11533SChris Williamson     else {
710dfc11533SChris Williamson       i++;  /* skip ESC */
711dfc11533SChris Williamson       if (!isdigit(uchar(news[i]))) {
712dfc11533SChris Williamson         if (news[i] != L_ESC)
713dfc11533SChris Williamson           luaL_error(ms->L, "invalid use of " LUA_QL("%c")
714dfc11533SChris Williamson                            " in replacement string", L_ESC);
715dfc11533SChris Williamson         luaL_addchar(b, news[i]);
716dfc11533SChris Williamson       }
717dfc11533SChris Williamson       else if (news[i] == '0')
718dfc11533SChris Williamson           luaL_addlstring(b, s, e - s);
719dfc11533SChris Williamson       else {
720dfc11533SChris Williamson         push_onecapture(ms, news[i] - '1', s, e);
721dfc11533SChris Williamson         luaL_addvalue(b);  /* add capture to accumulated result */
722dfc11533SChris Williamson       }
723dfc11533SChris Williamson     }
724dfc11533SChris Williamson   }
725dfc11533SChris Williamson }
726dfc11533SChris Williamson 
727dfc11533SChris Williamson 
728dfc11533SChris Williamson static void add_value (MatchState *ms, luaL_Buffer *b, const char *s,
729dfc11533SChris Williamson                                        const char *e, int tr) {
730dfc11533SChris Williamson   lua_State *L = ms->L;
731dfc11533SChris Williamson   switch (tr) {
732dfc11533SChris Williamson     case LUA_TFUNCTION: {
733dfc11533SChris Williamson       int n;
734dfc11533SChris Williamson       lua_pushvalue(L, 3);
735dfc11533SChris Williamson       n = push_captures(ms, s, e);
736dfc11533SChris Williamson       lua_call(L, n, 1);
737dfc11533SChris Williamson       break;
738dfc11533SChris Williamson     }
739dfc11533SChris Williamson     case LUA_TTABLE: {
740dfc11533SChris Williamson       push_onecapture(ms, 0, s, e);
741dfc11533SChris Williamson       lua_gettable(L, 3);
742dfc11533SChris Williamson       break;
743dfc11533SChris Williamson     }
744dfc11533SChris Williamson     default: {  /* LUA_TNUMBER or LUA_TSTRING */
745dfc11533SChris Williamson       add_s(ms, b, s, e);
746dfc11533SChris Williamson       return;
747dfc11533SChris Williamson     }
748dfc11533SChris Williamson   }
749dfc11533SChris Williamson   if (!lua_toboolean(L, -1)) {  /* nil or false? */
750dfc11533SChris Williamson     lua_pop(L, 1);
751dfc11533SChris Williamson     lua_pushlstring(L, s, e - s);  /* keep original text */
752dfc11533SChris Williamson   }
753dfc11533SChris Williamson   else if (!lua_isstring(L, -1))
754dfc11533SChris Williamson     luaL_error(L, "invalid replacement value (a %s)", luaL_typename(L, -1));
755dfc11533SChris Williamson   luaL_addvalue(b);  /* add result to accumulator */
756dfc11533SChris Williamson }
757dfc11533SChris Williamson 
758dfc11533SChris Williamson 
759dfc11533SChris Williamson static int str_gsub (lua_State *L) {
760dfc11533SChris Williamson   size_t srcl, lp;
761dfc11533SChris Williamson   const char *src = luaL_checklstring(L, 1, &srcl);
762dfc11533SChris Williamson   const char *p = luaL_checklstring(L, 2, &lp);
763dfc11533SChris Williamson   int tr = lua_type(L, 3);
764dfc11533SChris Williamson   size_t max_s = luaL_optinteger(L, 4, srcl+1);
765dfc11533SChris Williamson   int anchor = (*p == '^');
766dfc11533SChris Williamson   size_t n = 0;
767dfc11533SChris Williamson   MatchState ms;
768dfc11533SChris Williamson   luaL_Buffer b;
769dfc11533SChris Williamson   luaL_argcheck(L, tr == LUA_TNUMBER || tr == LUA_TSTRING ||
770dfc11533SChris Williamson                    tr == LUA_TFUNCTION || tr == LUA_TTABLE, 3,
771dfc11533SChris Williamson                       "string/function/table expected");
772dfc11533SChris Williamson   luaL_buffinit(L, &b);
773dfc11533SChris Williamson   if (anchor) {
774dfc11533SChris Williamson     p++; lp--;  /* skip anchor character */
775dfc11533SChris Williamson   }
776dfc11533SChris Williamson   ms.L = L;
777dfc11533SChris Williamson   ms.matchdepth = MAXCCALLS;
778dfc11533SChris Williamson   ms.src_init = src;
779dfc11533SChris Williamson   ms.src_end = src+srcl;
780dfc11533SChris Williamson   ms.p_end = p + lp;
781dfc11533SChris Williamson   while (n < max_s) {
782dfc11533SChris Williamson     const char *e;
783dfc11533SChris Williamson     ms.level = 0;
784dfc11533SChris Williamson     lua_assert(ms.matchdepth == MAXCCALLS);
785dfc11533SChris Williamson     e = match(&ms, src, p);
786dfc11533SChris Williamson     if (e) {
787dfc11533SChris Williamson       n++;
788dfc11533SChris Williamson       add_value(&ms, &b, src, e, tr);
789dfc11533SChris Williamson     }
790dfc11533SChris Williamson     if (e && e>src) /* non empty match? */
791dfc11533SChris Williamson       src = e;  /* skip it */
792dfc11533SChris Williamson     else if (src < ms.src_end)
793dfc11533SChris Williamson       luaL_addchar(&b, *src++);
794dfc11533SChris Williamson     else break;
795dfc11533SChris Williamson     if (anchor) break;
796dfc11533SChris Williamson   }
797dfc11533SChris Williamson   luaL_addlstring(&b, src, ms.src_end-src);
798dfc11533SChris Williamson   luaL_pushresult(&b);
799dfc11533SChris Williamson   lua_pushinteger(L, n);  /* number of substitutions */
800dfc11533SChris Williamson   return 2;
801dfc11533SChris Williamson }
802dfc11533SChris Williamson 
803dfc11533SChris Williamson /* }====================================================== */
804dfc11533SChris Williamson 
805dfc11533SChris Williamson 
806dfc11533SChris Williamson 
807dfc11533SChris Williamson /*
808dfc11533SChris Williamson ** {======================================================
809dfc11533SChris Williamson ** STRING FORMAT
810dfc11533SChris Williamson ** =======================================================
811dfc11533SChris Williamson */
812dfc11533SChris Williamson 
813dfc11533SChris Williamson /*
814dfc11533SChris Williamson ** LUA_INTFRMLEN is the length modifier for integer conversions in
815dfc11533SChris Williamson ** 'string.format'; LUA_INTFRM_T is the integer type corresponding to
816dfc11533SChris Williamson ** the previous length
817dfc11533SChris Williamson */
818dfc11533SChris Williamson #if !defined(LUA_INTFRMLEN)	/* { */
819dfc11533SChris Williamson #if defined(LUA_USE_LONGLONG)
820dfc11533SChris Williamson 
821dfc11533SChris Williamson #define LUA_INTFRMLEN		"ll"
822dfc11533SChris Williamson #define LUA_INTFRM_T		long long
823dfc11533SChris Williamson 
824dfc11533SChris Williamson #else
825dfc11533SChris Williamson 
826dfc11533SChris Williamson #define LUA_INTFRMLEN		"l"
827dfc11533SChris Williamson #define LUA_INTFRM_T		long
828dfc11533SChris Williamson 
829dfc11533SChris Williamson #endif
830dfc11533SChris Williamson #endif				/* } */
831dfc11533SChris Williamson 
832dfc11533SChris Williamson 
833dfc11533SChris Williamson /*
834dfc11533SChris Williamson ** LUA_FLTFRMLEN is the length modifier for float conversions in
835dfc11533SChris Williamson ** 'string.format'; LUA_FLTFRM_T is the float type corresponding to
836dfc11533SChris Williamson ** the previous length
837dfc11533SChris Williamson */
838dfc11533SChris Williamson #if !defined(LUA_FLTFRMLEN)
839dfc11533SChris Williamson 
840dfc11533SChris Williamson #define LUA_FLTFRMLEN		""
841dfc11533SChris Williamson #define LUA_FLTFRM_T		double
842dfc11533SChris Williamson 
843dfc11533SChris Williamson #endif
844dfc11533SChris Williamson 
845dfc11533SChris Williamson 
846dfc11533SChris Williamson /* maximum size of each formatted item (> len(format('%99.99f', -1e308))) */
847dfc11533SChris Williamson #define MAX_ITEM	512
848dfc11533SChris Williamson /* valid flags in a format specification */
849dfc11533SChris Williamson #define FLAGS	"-+ #0"
850dfc11533SChris Williamson /*
851dfc11533SChris Williamson ** maximum size of each format specification (such as '%-099.99d')
852dfc11533SChris Williamson ** (+10 accounts for %99.99x plus margin of error)
853dfc11533SChris Williamson */
854dfc11533SChris Williamson #define MAX_FORMAT	(sizeof(FLAGS) + sizeof(LUA_INTFRMLEN) + 10)
855dfc11533SChris Williamson 
856dfc11533SChris Williamson 
857dfc11533SChris Williamson static void addquoted (lua_State *L, luaL_Buffer *b, int arg) {
858dfc11533SChris Williamson   size_t l;
859dfc11533SChris Williamson   const char *s = luaL_checklstring(L, arg, &l);
860dfc11533SChris Williamson   luaL_addchar(b, '"');
861dfc11533SChris Williamson   while (l--) {
862dfc11533SChris Williamson     if (*s == '"' || *s == '\\' || *s == '\n') {
863dfc11533SChris Williamson       luaL_addchar(b, '\\');
864dfc11533SChris Williamson       luaL_addchar(b, *s);
865dfc11533SChris Williamson     }
866dfc11533SChris Williamson     else if (*s == '\0' || iscntrl(uchar(*s))) {
867dfc11533SChris Williamson       char buff[10];
868dfc11533SChris Williamson       if (!isdigit(uchar(*(s+1))))
869dfc11533SChris Williamson         sprintf(buff, "\\%d", (int)uchar(*s));
870dfc11533SChris Williamson       else
871dfc11533SChris Williamson         sprintf(buff, "\\%03d", (int)uchar(*s));
872dfc11533SChris Williamson       luaL_addstring(b, buff);
873dfc11533SChris Williamson     }
874dfc11533SChris Williamson     else
875dfc11533SChris Williamson       luaL_addchar(b, *s);
876dfc11533SChris Williamson     s++;
877dfc11533SChris Williamson   }
878dfc11533SChris Williamson   luaL_addchar(b, '"');
879dfc11533SChris Williamson }
880dfc11533SChris Williamson 
881dfc11533SChris Williamson static const char *scanformat (lua_State *L, const char *strfrmt, char *form) {
882dfc11533SChris Williamson   const char *p = strfrmt;
883dfc11533SChris Williamson   while (*p != '\0' && strchr(FLAGS, *p) != NULL) p++;  /* skip flags */
884dfc11533SChris Williamson   if ((size_t)(p - strfrmt) >= sizeof(FLAGS)/sizeof(char))
885dfc11533SChris Williamson     luaL_error(L, "invalid format (repeated flags)");
886dfc11533SChris Williamson   if (isdigit(uchar(*p))) p++;  /* skip width */
887dfc11533SChris Williamson   if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
888dfc11533SChris Williamson   if (*p == '.') {
889dfc11533SChris Williamson     p++;
890dfc11533SChris Williamson     if (isdigit(uchar(*p))) p++;  /* skip precision */
891dfc11533SChris Williamson     if (isdigit(uchar(*p))) p++;  /* (2 digits at most) */
892dfc11533SChris Williamson   }
893dfc11533SChris Williamson   if (isdigit(uchar(*p)))
894dfc11533SChris Williamson     luaL_error(L, "invalid format (width or precision too long)");
895dfc11533SChris Williamson   *(form++) = '%';
896dfc11533SChris Williamson   memcpy(form, strfrmt, (p - strfrmt + 1) * sizeof(char));
897dfc11533SChris Williamson   form += p - strfrmt + 1;
898dfc11533SChris Williamson   *form = '\0';
899dfc11533SChris Williamson   return p;
900dfc11533SChris Williamson }
901dfc11533SChris Williamson 
902dfc11533SChris Williamson 
903dfc11533SChris Williamson /*
904dfc11533SChris Williamson ** add length modifier into formats
905dfc11533SChris Williamson */
906dfc11533SChris Williamson static void addlenmod (char *form, const char *lenmod) {
907dfc11533SChris Williamson   size_t l = strlen(form);
908dfc11533SChris Williamson   size_t lm = strlen(lenmod);
909dfc11533SChris Williamson   char spec = form[l - 1];
910dfc11533SChris Williamson   strcpy(form + l - 1, lenmod);
911dfc11533SChris Williamson   form[l + lm - 1] = spec;
912dfc11533SChris Williamson   form[l + lm] = '\0';
913dfc11533SChris Williamson }
914dfc11533SChris Williamson 
915dfc11533SChris Williamson 
916dfc11533SChris Williamson static int str_format (lua_State *L) {
917dfc11533SChris Williamson   int top = lua_gettop(L);
918dfc11533SChris Williamson   int arg = 1;
919dfc11533SChris Williamson   size_t sfl;
920dfc11533SChris Williamson   const char *strfrmt = luaL_checklstring(L, arg, &sfl);
921dfc11533SChris Williamson   const char *strfrmt_end = strfrmt+sfl;
922dfc11533SChris Williamson   luaL_Buffer b;
923dfc11533SChris Williamson   luaL_buffinit(L, &b);
924dfc11533SChris Williamson   while (strfrmt < strfrmt_end) {
925dfc11533SChris Williamson     if (*strfrmt != L_ESC)
926dfc11533SChris Williamson       luaL_addchar(&b, *strfrmt++);
927dfc11533SChris Williamson     else if (*++strfrmt == L_ESC)
928dfc11533SChris Williamson       luaL_addchar(&b, *strfrmt++);  /* %% */
929dfc11533SChris Williamson     else { /* format item */
930dfc11533SChris Williamson       char form[MAX_FORMAT];  /* to store the format (`%...') */
931dfc11533SChris Williamson       char *buff = luaL_prepbuffsize(&b, MAX_ITEM);  /* to put formatted item */
932dfc11533SChris Williamson       int nb = 0;  /* number of bytes in added item */
933dfc11533SChris Williamson       if (++arg > top)
934dfc11533SChris Williamson         luaL_argerror(L, arg, "no value");
935dfc11533SChris Williamson       strfrmt = scanformat(L, strfrmt, form);
936dfc11533SChris Williamson       switch (*strfrmt++) {
937dfc11533SChris Williamson         case 'c': {
938dfc11533SChris Williamson           nb = str_sprintf(buff, form, luaL_checkint(L, arg));
939dfc11533SChris Williamson           break;
940dfc11533SChris Williamson         }
941dfc11533SChris Williamson         case 'd': case 'i': {
942dfc11533SChris Williamson           lua_Number n = luaL_checknumber(L, arg);
943dfc11533SChris Williamson           LUA_INTFRM_T ni = (LUA_INTFRM_T)n;
944dfc11533SChris Williamson           lua_Number diff = n - (lua_Number)ni;
945dfc11533SChris Williamson           luaL_argcheck(L, -1 < diff && diff < 1, arg,
946dfc11533SChris Williamson                         "not a number in proper range");
947dfc11533SChris Williamson           addlenmod(form, LUA_INTFRMLEN);
948dfc11533SChris Williamson           nb = str_sprintf(buff, form, ni);
949dfc11533SChris Williamson           break;
950dfc11533SChris Williamson         }
951dfc11533SChris Williamson         case 'o': case 'u': case 'x': case 'X': {
952dfc11533SChris Williamson           lua_Number n = luaL_checknumber(L, arg);
953dfc11533SChris Williamson           unsigned LUA_INTFRM_T ni = (unsigned LUA_INTFRM_T)n;
954dfc11533SChris Williamson           lua_Number diff = n - (lua_Number)ni;
955dfc11533SChris Williamson           luaL_argcheck(L, -1 < diff && diff < 1, arg,
956dfc11533SChris Williamson                         "not a non-negative number in proper range");
957dfc11533SChris Williamson           addlenmod(form, LUA_INTFRMLEN);
958dfc11533SChris Williamson           nb = str_sprintf(buff, form, ni);
959dfc11533SChris Williamson           break;
960dfc11533SChris Williamson         }
961916c8d88SMatthew Ahrens #if defined(LUA_USE_FLOAT_FORMATS)
962dfc11533SChris Williamson         case 'e': case 'E': case 'f':
963dfc11533SChris Williamson #if defined(LUA_USE_AFORMAT)
964dfc11533SChris Williamson         case 'a': case 'A':
965dfc11533SChris Williamson #endif
966dfc11533SChris Williamson         case 'g': case 'G': {
967dfc11533SChris Williamson           addlenmod(form, LUA_FLTFRMLEN);
968dfc11533SChris Williamson           nb = str_sprintf(buff, form, (LUA_FLTFRM_T)luaL_checknumber(L, arg));
969dfc11533SChris Williamson           break;
970dfc11533SChris Williamson         }
971916c8d88SMatthew Ahrens #endif
972dfc11533SChris Williamson         case 'q': {
973dfc11533SChris Williamson           addquoted(L, &b, arg);
974dfc11533SChris Williamson           break;
975dfc11533SChris Williamson         }
976dfc11533SChris Williamson         case 's': {
977dfc11533SChris Williamson           size_t l;
978dfc11533SChris Williamson           const char *s = luaL_tolstring(L, arg, &l);
979dfc11533SChris Williamson           if (!strchr(form, '.') && l >= 100) {
980dfc11533SChris Williamson             /* no precision and string is too long to be formatted;
981dfc11533SChris Williamson                keep original string */
982dfc11533SChris Williamson             luaL_addvalue(&b);
983dfc11533SChris Williamson             break;
984dfc11533SChris Williamson           }
985dfc11533SChris Williamson           else {
986dfc11533SChris Williamson             nb = str_sprintf(buff, form, s);
987dfc11533SChris Williamson             lua_pop(L, 1);  /* remove result from 'luaL_tolstring' */
988dfc11533SChris Williamson             break;
989dfc11533SChris Williamson           }
990dfc11533SChris Williamson         }
991dfc11533SChris Williamson         default: {  /* also treat cases `pnLlh' */
992dfc11533SChris Williamson           return luaL_error(L, "invalid option " LUA_QL("%%%c") " to "
993dfc11533SChris Williamson                                LUA_QL("format"), *(strfrmt - 1));
994dfc11533SChris Williamson         }
995dfc11533SChris Williamson       }
996dfc11533SChris Williamson       luaL_addsize(&b, nb);
997dfc11533SChris Williamson     }
998dfc11533SChris Williamson   }
999dfc11533SChris Williamson   luaL_pushresult(&b);
1000dfc11533SChris Williamson   return 1;
1001dfc11533SChris Williamson }
1002dfc11533SChris Williamson 
1003dfc11533SChris Williamson /* }====================================================== */
1004dfc11533SChris Williamson 
1005dfc11533SChris Williamson 
1006dfc11533SChris Williamson static const luaL_Reg strlib[] = {
1007dfc11533SChris Williamson   {"byte", str_byte},
1008dfc11533SChris Williamson   {"char", str_char},
1009dfc11533SChris Williamson   {"dump", str_dump},
1010dfc11533SChris Williamson   {"find", str_find},
1011dfc11533SChris Williamson   {"format", str_format},
1012dfc11533SChris Williamson   {"gmatch", str_gmatch},
1013dfc11533SChris Williamson   {"gsub", str_gsub},
1014dfc11533SChris Williamson   {"len", str_len},
1015dfc11533SChris Williamson   {"lower", str_lower},
1016dfc11533SChris Williamson   {"match", str_match},
1017dfc11533SChris Williamson   {"rep", str_rep},
1018dfc11533SChris Williamson   {"reverse", str_reverse},
1019dfc11533SChris Williamson   {"sub", str_sub},
1020dfc11533SChris Williamson   {"upper", str_upper},
1021dfc11533SChris Williamson   {NULL, NULL}
1022dfc11533SChris Williamson };
1023dfc11533SChris Williamson 
1024dfc11533SChris Williamson 
1025dfc11533SChris Williamson static void createmetatable (lua_State *L) {
1026dfc11533SChris Williamson   lua_createtable(L, 0, 1);  /* table to be metatable for strings */
1027dfc11533SChris Williamson   lua_pushliteral(L, "");  /* dummy string */
1028dfc11533SChris Williamson   lua_pushvalue(L, -2);  /* copy table */
1029dfc11533SChris Williamson   lua_setmetatable(L, -2);  /* set table as metatable for strings */
1030dfc11533SChris Williamson   lua_pop(L, 1);  /* pop dummy string */
1031dfc11533SChris Williamson   lua_pushvalue(L, -2);  /* get string library */
1032dfc11533SChris Williamson   lua_setfield(L, -2, "__index");  /* metatable.__index = string */
1033dfc11533SChris Williamson   lua_pop(L, 1);  /* pop metatable */
1034dfc11533SChris Williamson }
1035dfc11533SChris Williamson 
1036dfc11533SChris Williamson 
1037dfc11533SChris Williamson /*
1038dfc11533SChris Williamson ** Open string library
1039dfc11533SChris Williamson */
1040dfc11533SChris Williamson LUAMOD_API int luaopen_string (lua_State *L) {
1041dfc11533SChris Williamson   luaL_newlib(L, strlib);
1042dfc11533SChris Williamson   createmetatable(L);
1043dfc11533SChris Williamson   return 1;
1044dfc11533SChris Williamson }
1045dfc11533SChris Williamson 
1046